33
Wildlife Monographs 203:335; 2019; DOI: 10.1002/wmon.1041 Dynamics, Persistence, and Genetic Management of the Endangered Florida Panther Population MADELON VAN DE KERK, 1,2 Department of Wildlife Ecology and Conservation, University of Florida, 110 NewinsZiegler Hall, Gainesville, FL 326110430, USA DAVID P. ONORATO, 2 Fish and Wildlife Research Institute, Florida Fish and Wildlife Conservation Commission, 298 Sabal Palm Road, Naples, FL 34114, USA JEFFREY A. HOSTETLER, Fish and Wildlife Research Institute, Florida Fish and Wildlife Conservation Commission, 100 8th Avenue SE, St. Petersburg, FL 33701, USA BENJAMIN M. BOLKER, Departments of Mathematics and Statistics and Biology, McMaster University, 314 Hamilton Hall, Hamilton, ON L8S 4K1, Canada MADAN K. OLI, 3 Department of Wildlife Ecology and Conservation, University of Florida, 110 NewinsZiegler Hall, Gainesville, FL 326110430, USA ABSTRACT Abundant evidence supports the benets accrued to the Florida panther (Puma concolor coryi) population via the genetic introgression project implemented in South Florida, USA, in 1995. Since then, genetic diversity has improved, the frequency of morphological and biomedical correlates of inbreeding de- pression have declined, and the population size has increased. Nevertheless, the panther population remains small and isolated and faces substantial challenges due to deterministic and stochastic forces. Our goals were 1) to comprehensively assess the demographics of the Florida panther population using longterm (19812015) eld data and modeling to gauge the persistence of benets accrued via genetic introgression and 2) to evaluate the eectiveness of various potential genetic management strategies. Translocation and introduction of female pumas (Puma concolor stanleyana) from Texas, USA, substantially improved genetic diversity. The average individual heterozygosity of canonical (nonintrogressed) panthers was 0.386 ± 0.012 (SE); for admixed pan- thers, it was 0.615 ± 0.007. Survival rates were strongly agedependent (kittens had the lowest survival rates), were positively aected by individual heterozygosity, and decreased with increasing population abundance. Overall annual kitten survival was 0.32 ± 0.09; sex did not have a clear eect on kitten survival. Annual survival of subadult and adult panthers diered by sex; regardless of age, females exhibited higher survival than males. Annual survival rates of subadult, prime adult, and old adult females were 0.97 ± 0.02, 0.86 ± 0.03, and 0.78 ± 0.09, respectively. Survival rates of subadult, prime adult, and old adult males were 0.66 ± 0.06, 0.77 ± 0.05, and 0.65 ± 0.10, respectively. For panthers of all ages, genetic ancestry strongly aected survival rate, where rst lial generation (F1) admixed panthers of all ages exhibited the highest rates and canonical (mostly preintrogression panthers and their postintrogression descendants) individuals exhibited the lowest rates. The most frequently observed causes of death of radiocollared panthers were intraspecic aggression and vehicle collision. Causespecic mortality analyses revealed that mortality rates from vehicle collision, in- traspecic aggression, other causes, and unknown causes were generally similar for males and females, although males were more likely to die from intraspecic aggression than females. The probability of reproduction and the annual number of kittens produced varied by age; evidence that ancestry or abundance inuenced these para- meters was weak. Predicted annual probabilities of reproduction were 0.35 ± 0.08, 0.50 ± 0.05, and 0.25 ± 0.06 for subadult, prime adult, and old adult females, respectively. The number of kittens predicted to be produced annually by subadult, prime adult, and old adult females were 2.80 ± 0.75, 2.67 ± 0.43, and 2.28 ± 0.83, re- spectively. The stochastic annual population growth rate estimated using a matrix population model was 1.04 (95% CI = 0.721.41). An individualbased population model predicted that the probability that the population would fall below 10 panthers within 100 years (quasiextinction) was 1.4% (00.8%) if the adverse eects of genetic erosion were ignored. However, when the eect of genetic erosion was considered, the probability of quasiextinction within 100 years increased to 17% (0100%). Mean times to quasiextinction, conditioned on going quasiextinct within 100 years, was 22 (075) years when the eect of genetic erosion was considered. Sensitivity analyses revealed that the probability of quasiextinction and expected time until quasiextinction were most sensitive to changes in kitten survival parameters. Without genetic management intervention, the Florida panther population would face a substantially increased risk of quasiextinction. The question, therefore, -------------------------------------------------------------------------------------------- This is an open access article under the terms of the Creative Commons AttributionNonCommercialNoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is noncommercial and no modications or adaptations are made. Received: 18 April 2017; Accepted: 4 January 2019 1 Present address: Department of Plant and Wildlife Sciences, Brigham Young University, 5047 Life Sciences Building, Provo, UT 84602, USA. 2 These authors contributed equally to this work. 3 Email: olim@u.edu van de Kerk et al. Florida Panther Population and Genetic Viability 3

Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

  • Upload
    others

  • View
    3

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

Wildlife Monographs 2033ndash35 2019 DOI 101002wmon1041

Dynamics Persistence and Genetic Managementof the Endangered Florida Panther Population

MADELON VAN DE KERK12 Department of Wildlife Ecology and Conservation University of Florida 110 Newins‐Ziegler Hall Gainesville FL 32611‐0430 USA

DAVID P ONORATO2 Fish and Wildlife Research Institute Florida Fish and Wildlife Conservation Commission 298 Sabal Palm Road Naples FL34114 USA

JEFFREY A HOSTETLER Fish and Wildlife Research Institute Florida Fish and Wildlife Conservation Commission 100 8th Avenue SE St Petersburg FL33701 USA

BENJAMIN M BOLKER Departments of Mathematics and Statistics and Biology McMaster University 314 Hamilton Hall Hamilton ON L8S 4K1Canada

MADAN K OLI3 Department of Wildlife Ecology and Conservation University of Florida 110 Newins‐Ziegler Hall Gainesville FL 32611‐0430 USA

ABSTRACT Abundant evidence supports the benefits accrued to the Florida panther (Puma concolor coryi)population via the genetic introgression project implemented in South Florida USA in 1995 Since thengenetic diversity has improved the frequency of morphological and biomedical correlates of inbreeding de-pression have declined and the population size has increased Nevertheless the panther population remainssmall and isolated and faces substantial challenges due to deterministic and stochastic forces Our goals were 1)to comprehensively assess the demographics of the Florida panther population using long‐term (1981ndash2015)field data and modeling to gauge the persistence of benefits accrued via genetic introgression and 2) to evaluatethe effectiveness of various potential genetic management strategies Translocation and introduction of femalepumas (Puma concolor stanleyana) from Texas USA substantially improved genetic diversity The averageindividual heterozygosity of canonical (non‐introgressed) panthers was 0386plusmn 0012 (SE) for admixed pan-thers it was 0615 plusmn 0007 Survival rates were strongly age‐dependent (kittens had the lowest survival rates)were positively affected by individual heterozygosity and decreased with increasing population abundanceOverall annual kitten survival was 032plusmn 009 sex did not have a clear effect on kitten survival Annual survivalof subadult and adult panthers differed by sex regardless of age females exhibited higher survival than malesAnnual survival rates of subadult prime adult and old adult females were 097plusmn 002 086plusmn 003 and078plusmn 009 respectively Survival rates of subadult prime adult and old adult males were 066plusmn 006077plusmn 005 and 065plusmn 010 respectively For panthers of all ages genetic ancestry strongly affected survivalrate where first filial generation (F1) admixed panthers of all ages exhibited the highest rates and canonical(mostly pre‐introgression panthers and their post‐introgression descendants) individuals exhibited the lowestrates The most frequently observed causes of death of radio‐collared panthers were intraspecific aggression andvehicle collision Cause‐specific mortality analyses revealed that mortality rates from vehicle collision in-traspecific aggression other causes and unknown causes were generally similar for males and females althoughmales were more likely to die from intraspecific aggression than females The probability of reproduction and theannual number of kittens produced varied by age evidence that ancestry or abundance influenced these para-meters was weak Predicted annual probabilities of reproduction were 035plusmn 008 050 plusmn 005 and 025 plusmn 006for subadult prime adult and old adult females respectively The number of kittens predicted to be producedannually by subadult prime adult and old adult females were 280plusmn 075 267 plusmn 043 and 228plusmn 083 re-spectively The stochastic annual population growth rate estimated using a matrix population model was 104(95 CI = 072ndash141) An individual‐based population model predicted that the probability that the populationwould fall below 10 panthers within 100 years (quasi‐extinction) was 14 (0ndash08) if the adverse effects ofgenetic erosion were ignored However when the effect of genetic erosion was considered the probability ofquasi‐extinction within 100 years increased to 17 (0ndash100) Mean times to quasi‐extinction conditioned ongoing quasi‐extinct within 100 years was 22 (0ndash75) years when the effect of genetic erosion was consideredSensitivity analyses revealed that the probability of quasi‐extinction and expected time until quasi‐extinctionwere most sensitive to changes in kitten survival parameters Without genetic management intervention theFlorida panther population would face a substantially increased risk of quasi‐extinction The question therefore

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -This is an open access article under the terms of the Creative Commons Attribution‐NonCommercial‐NoDerivs License which permits use and distribution in any medium provided the

original work is properly cited the use is non‐commercial and no modifications or adaptations are made

Received 18 April 2017 Accepted 4 January 2019

1Present address Department of Plant and Wildlife Sciences Brigham Young University 5047 Life Sciences Building Provo UT 84602 USA

2These authors contributed equally to this work

3E‐mail olimufledu

van de Kerk et al bull Florida Panther Population and Genetic Viability 3

is not whether genetic management of the Florida panther population is needed but when and how it should beimplemented Thus we evaluated genetic and population consequences of alternative genetic introgressionstrategies to identify optimal management actions using individual‐based simulation models Releasing 5 pumasevery 20 years would cost much less ($200000 over 100 years) than releasing 15 pumas every 10 years($1200000 over 100 years) yet would reduce the risk of quasi‐extinction by comparable amount (44ndash59 vs40ndash58) Generally releasing more females per introgression attempt provided little added benefit The positiveeffects of the genetic introgression project persist in the panther population after 20 years We suggest thatmanagers contemplate repeating genetic introgression by releasing 5ndash10 individuals from other puma popula-tions every 20ndash40 years We also recommend that managers continue to collect data that will permit estimationand monitoring of kitten adult and subadult survival We identified these parameters via sensitivity analyses asmost critical in terms of their impact on the probability of and expected times to quasi‐extinction The con-tinuation of long‐term monitoring should permit the adaptation of genetic management strategies as necessarywhile collecting data that have proved essential in assessing the genetic and demographic health of the popu-lation The prospects for recovery of the panther will certainly be improved by following these guidelines copy2019 The Authors Wildlife Monographs published by Wiley Periodicals Inc on behalf of The Wildlife Society

KEY WORDS Florida panther demography genetic introgression genetic rescue heterosis inbreeding depressionindividual‐based population models population viability analysis Puma concolor coryi

Dinaacutemicas Persistencia y Manejo Geneacutetico de la Poblacioacutenen Peligro de Extincioacuten de Pantera de Florida

RESUMEN La poblacioacuten de pantera de Florida (Puma concolor coryi) mejoroacute tras la implementacioacuten en 1995 delproyecto de introgresioacuten geneacutetica en el sur de Florida USA como lo demuestran varias liacuteneas de evidencia Desdeentonces su diversidad geneacutetica ha mejorado la frecuencia de iacutendices morfoloacutegicos y biomeacutedicos correlacionadoscon depresioacuten endogaacutemica ha disminuido y el tamantildeo de la poblacioacuten ha aumentado Sin embargo la poblacioacuten depanteras permanece pequentildea aislada y se enfrenta a retos sustanciales producidos por fuerzas determiniacutesticas yestocaacutesticas Los objetivos de este estudio fueron 1) evaluar exhaustivamente la demografiacutea de la poblacioacuten depanteras de Florida usando datos de campo (del periodo 1981ndash2015) y modelos con el fin de calibrar en que medidapersisten los beneficios adquiridos a traveacutes de la introgresioacuten geneacutetica y 2) evaluar la efectividad de varias estrategiasde manejo geneacutetico La diversidad geneacutetica de la poblacioacuten mejoroacute sustancialmente con la introduccioacuten de pumashembra (Puma concolor stanleyana) procedentes de Texas USA En panteras canoacutenicas (no procedentes de in-trogresioacuten) el valor medio de heterocigosidad individual fue 0386plusmn 0012 (SE) y en panteras mezcladas0615plusmn 0007 En gran medida las tasas de supervivencia dependieron de la edad (los cachorros teniacutean las tasas desupervivencia maacutes bajas) estuvieron afectadas positivamente por la heterocigosidad individual y disminuyeroncuando la poblacioacuten aumentoacute La tasa de supervivencia total independientemente del sexo del cachorro fue de032plusmn 009 La tasa de supervivencia anual de panteras adultas y subadultas varioacute seguacuten el sexo independientementede la edad las hembras vivieron maacutes que los machos Las tasas anuales de supervivencia de hembras subadultasadultas y adultas mayores fueron 097plusmn 002 086plusmn 003 y 078plusmn 009 respectivamente Las tasas de supervivenciade machos subadultos adultos y adultos mayores fueron 066plusmn 006 077plusmn 005 y 065plusmn 010 respectivamenteLa ascendencia geneacutetica determinoacute en gran medida la tasa de supervivencia de panteras de cualquier edad siendomayor en la primera generacioacuten filiar (F1) de panteras mezcladas en todas las edades y menor en los individuoscanoacutenicos (sobre todo panteras pre‐introgresioacuten y sus descendientes post‐introgresioacuten) En panteras con collares deradio telemetriacutea las causas de mortalidad maacutes frecuentes fueron la agresioacuten intraespeciacutefica y la colisioacuten convehiacuteculos El anaacutelisis de las causas de mortalidad reveloacute que en las categoriacuteas colisioacuten con vehiacuteculos agresioacutenintraespeciacutefica otras causas y motivos desconocidos la tasa de mortalidad de machos y hembras era similar aunquelos machos teniacutean maacutes posibilidades de morir por agresioacuten intraespeciacutefica que las hembras Las probabilidades dereproduccioacuten y el nuacutemero anual de cachorros dependieron de la edad pero no de los ancestros o el tamantildeo de lapoblacioacuten Las probabilidades de reproduccioacuten de hembras subadultas adultas y adultas mayores se estimaron en035plusmn 008 050plusmn 005 y 025plusmn 006 respectivamente El nuacutemero de cachorros por antildeo y por pantera subadultaadulta y adulta mayor se estimoacute en 280plusmn 075 267plusmn 043 y 228plusmn 083 respectivamente Usando un modelodemograacutefico matricial se estimoacute la tasa anual de crecimiento estocaacutestico poblacional en 104 (95 CI= 072ndash141)Usando un modelo de poblacioacuten basado en el individuo e ignorando el impacto adverso de la erosioacuten geneacutetica seestimoacute la probabilidad de que la poblacioacuten disminuyese a menos de 10 panteras en 100 antildeos (cuasi‐extincioacuten) en14 (0ndash08) Sin embargo incluyendo el impacto de la erosioacuten geneacutetica la probabilidad de cuasi‐extincioacuten en 100antildeos aumentoacute al 17 (0ndash100) El plazo medio para la cuasi‐extincioacuten asumiendo que la cuasi‐extincioacuten ocurre en

4 Wildlife Monographs bull 203

100 antildeos e incluyendo el impacto de la erosioacuten geneacutetica fue de 22 (0ndash75) antildeos Anaacutelisis de sensibilidad demostraronque la probabilidad de cuasi‐extincioacuten y el plazo hasta alcanzarla dependiacutean de los valores utilizados para losparaacutemetros de supervivencia de cachorros Sin manejo geneacutetico la poblacioacuten de panteras de Florida se enfrentariacutea aun aumento sustancial del riesgo de cuasi‐extincioacuten Por lo tanto la pregunta no es si es necesario el manejo geneacuteticode la poblacioacuten de las panteras de Florida sino cuaacutendo y coacutemo implementarlo Usando modelos de simulacioacutenbasados en individuos evaluamos diferentes estrategias de introgresioacuten geneacutetica y sus posibles impactos en lapoblacioacuten y en su geneacutetica La reduccioacuten del riesgo de cuasi‐extincioacuten fue similar introduciendo 5 pumas cada 20antildeos o 15 pumas cada 10 antildeos (44ndash59 vs 40ndash58) pero la primera opcioacuten resulta maacutes econoacutemica ($200000 en100 antildeos) que la segunda ($12000000 en 100 antildeos) Introducir maacutes hembras en cada intento de introgresioacuten noprodujo beneficios adicionales El impacto positivo del proyecto de introgresioacuten geneacutetica persiste en la poblacioacuten depanteras veinte antildeos despueacutes de su implementacioacuten Recomendamos a los gestores que consideren repetir la in-trogresioacuten geneacutetica introduciendo 5ndash10 individuos de otras poblaciones de puma cada 20ndash40 antildeos Adicionalmenterecomendamos que se continuacutee con la recopilacioacuten de datos lo que es crucial para poder estimar y monitorizar lasupervivencia de cachorros adultos y subadultos Estos paraacutemetros son los maacutes criacuteticos para estimar la probabilidad yel periodo de cuasi‐extincioacuten seguacuten se demostroacute con el anaacutelisis de sensibilidad Se debe continuar con la mon-itorizacioacuten de la poblacioacuten a largo plazo lo que permitiraacute adaptar las estrategias de manejo seguacuten sea necesario a lavez que recopilar informacioacuten esencial para evaluar la salud demograacutefica de la poblacioacuten Siguiendo estas re-comendaciones las perspectivas de recuperacioacuten de las panteras mejoraraacuten

Contents

INTRODUCTION 5

STUDY AREA 9

METHODS 9

Field Methods 9

Genetic Analysis 10

Demographic Parameters 11

Survival of kittens 11

Subadult and adult survival 11

Probability of reproduction 11

Reproductive output 12

Covariates 12

Cause‐specific mortality 12Statistical inference 12

Matrix Population Models 13

IBMs and PVA 14

Sensitivity Analysis 15

Genetic Management 15

Introgression strategies 15

Cost‐benefit analysis 16RESULTS 17

Genetic Variation 17

Survival and Cause‐Specific Mortality Rates 17

Reproductive Parameters 18

Population Dynamics and Persistence 18

Deterministic and stochastic demography 18

Minimum population count scenario 18

Motor vehicle mortality scenario 18

Sensitivity of extinction probability to demographic parameters 19

Benefits and Costs of Genetic Management 19

Minimum population count scenario 19

Motor vehicle mortality scenario 19

Financial cost and persistence benefits 21

DISCUSSION 21

Genetic Variation Demographic Parameters

and Persistence of Heterotic Benefits from Genetic Introgression 24

Population Dynamics and Persistence 28

Genetic Management When and How to Intervene 28

MANAGEMENT IMPLICATIONS 30

ACKNOWLEDGMENTS 31

LITERATURE CITED 31

INTRODUCTION

Promoting genetic admixture between individuals from differentpopulations commonly referred to as genetic introgression hasthe potential to serve as a powerful management and con-servation tool (Keller and Waller 2002 Tallmon et al 2004Whiteley et al 2015) Genetic introgression can occur naturallythrough immigration of individuals into small or isolated po-pulations (Marr et al 2002 Vila et al 2003 Adams et al 2011)but it can also be implemented as a management strategy via theintentional translocation of individuals from large viable popu-lations to small endangered populations Intentional geneticintrogression of populations of wild animals has prevented thedemise of species such as adders (Vipera berus Madsen et al2004) prairie chickens (Tympanuchus cupido Bateson et al

2014) and bighorn sheep (Ovis canadensis Madsen et al 1999Hogg et al 2006) In most of these cases genetic introgressionhas been effective at improving demographic performance ofpopulations affected by inbreeding depressionInbreeding depression a reduction in fitness resulting from

the production of offspring by individuals related by ancestryis a conservation challenge common to small isolated popu-lations Inbreeding depression has historically been docu-mented in captive populations (Lacy et al 1993) The dele-terious impact of inbreeding depression on wild populationswas somewhat controversial (Frankham 2010a) until theseminal work of Crnokrak and Roff (1999) which verified thenegative effects of inbreeding on the fitness traits of severalwildlife populations including cheetahs (Acinonyx jubatus)

van de Kerk et al bull Florida Panther Population and Genetic Viability 5

black‐tailed prairie dogs (Cynomys ludovicianus) and white‐footed mice (Peromyscus leucopus) Subsequently the effect ofinbreeding depression has been extensively documented inseveral wild animal populations including mammals such aswolves (Canis lupus Liberg et al 2005) African lions (Pan-thera leo Trinkel et al 2010 2011) and Iberian lynx (Lynxpardinus Ruiz‐Loacutepez et al 2012) In these examples in-breeding depression had negative effects on a variety of fitnesstraits including survival of neonates susceptibility to diseaseand male fertility The demographic and genetic impacts ofinbreeding on population growth in carnivores is highlightedbest in one of the quintessential long‐term conservation stu-dies on the gray wolves of Isle Royale in northern MichiganUSA (Peterson 1977 Peterson and Page 1988 Peterson et al1998 Adams et al 2011) Research on this population hasproved informative for gaining a better understanding of po-pulation consequences of loss of genetic variation and impactsof intrinsic and extrinsic factors that have led to its decline toonly 2 individuals as of 2017 (Hedrick et al 2017)Conversely outbreeding depression is a scenario when mating

between members of genetically divergent populations might resultin a reduction in fitness correlated with the loss of local adaptationsor genetic incompatibilities (Moritz 1999) Outbreeding depressionhas been hypothesized to be a possible side effect of genetic in-trogression (Reinert 1991 Maehr and Caddick 1995) albeit in-frequently and with less and weaker empirical evidence for verte-brate species Negative fitness responses resulting from matingbetween individuals from highly divergent populations of copepodswere noted by Hwang et al (2012) Other examples of outbreedingdepression in varied species of plants are provided by Fenster andGallaway (2000) Waser et al (2000) Goto et al (2011) and Brysand Jacquemyn (2016) In general however outbreeding depres-sion is thought to be less of a concern when implementing con-servation measures such as genetic introgression (Miller et al 1999Whiteley et al 2015)In the last 3 decades conservation biologists have exponentially

increased their use of molecular techniques to assist with mon-itoring managing and conserving wildlife populations (DeYoungand Honeycutt 2005) Although early research relied on methodssuch as restriction fragment length polymorphisms allozymeelectrophoresis and minisatellites (Smith and Wayne 1996) thelast 20 years have been dominated by the implementation of im-proved automated sequencing technology to provide more in-formative and fine‐scale data from both nuclear and mitochondrialDNA sequences (eg mtDNA sequencing microsatellitesDeYoung and Honeycutt 2005) Microsatellites in particular havebecome a common tool for conservation projects to assess geneticparameters including inbreeding depression and the impacts ofgenetic introgression in imperiled populations They continue tobe widely used molecular markers in conservation researchThe field of molecular ecology is changing rapidly and new

technology is constantly being developed Currently there is in-terest in applying genomics or so‐called whole‐genome sequencingtechniques to assist with conserving imperiled species Two recentexamples of the application of whole‐genome sequencing to en-dangered species of carnivores include the work of Robinson et al(2016) on the Channel Island fox (Urocyon littoralis) and Murch-ison et al (2012) on the Tasmanian devil (Sarcophilus harrisii) For

instance Robinson et al (2016) reported a near absence ofgenomic variation in Channel Island foxes from the island of SanNicolas demonstrating an extreme reduction in heterozygositycompared to mainland populations Similarly Murchison et al(2012) applied whole genome sequencing to ascertain whether theTasmanian devil facial tumor disease originated from a male or afemale devil and further used whole genome sequencing techniquesto assess transmission of the disease across Tasmania The appli-cation of whole genome sequencing in conservation biology andwildlife management will continue to increase (Whiteley et al2015) Although costs associated with implementing wholegenome sequencing are steadily declining (Whiteley et al 2015)other factorsmdashsuch as sample preparation and the need forbioinformatic expertise to deal with voluminous amounts of datamdashare still an impediment to their widespread use in conservation andmanagement of wildlife populations Therefore long‐term datasets frommore traditional molecular markers such as microsatelliteswill continue to have utility for providing insight into the genetichealth of endangered species and for assessing impacts of man-agement and recovery initiativesThe identification and quantification of inbreeding depression

using molecular techniques (Kardos et al 2016) along with pro-tocols for implementing translocations to promote genetic in-trogression have been documented for varied endangered popu-lations (Whiteley et al 2015) Conversely studies that provide anassessment of the longevity of benefits accrued to wild populationsvia genetic introgression are uncommon Hedrick et al (2014)provided evidence for the declining effect of genetic introgressionon the Isle Royale population of wolves after only 2 or 3 genera-tions Whereas heterosis is expected to increase individual fitness itcan also result in outbreeding depression (Fenster and Gallaway2000 Waser et al 2000 Goto et al 2011 Waller 2015 Brys andJacquemyn 2016) Depending on the mechanisms outbreedingdepression may not become evident until generations subsequent tothe first filial (F1) generation (Pickup et al 2013 Waller 2015)For example ancestral chromosomes and coevolved genes remainintact in the F1 but recombination may break up favorable genecomplexes in F2 and later generations (Waser et al 2000 Waller2015 Frankham 2016) Investigating the timeframe over whichthe positive impacts of genetic introgression are sustained wouldlikely necessitate long‐term monitoring but for most studies of theprocess populations are typically monitored only for 1 or 2 gen-erations following genetic introgression (Willi et al 2007 Hedrickand Fredrickson 2009 Hedrick et al 2014 Whiteley et al 2015)Consequently data are rarely gathered that would allow us to es-timate the time over which populations continue to benefit fromgenetic introgression (Hedrick et al 2014 Whiteley et al 2015)The Florida panther (Puma concolor coryi) provides a textbook

example of how inbreeding depression in a small isolated po-pulation can result in low levels of genetic variation that areassociated with morphological and biomedical abnormalities andprecipitous population decline The timeline associated with thedecline of panthers parallels that of most large carnivores inNorth America (Onorato et al 2010) A robust population ofpanthers was thought to extend across Florida USA inpre‐Columbian times (Fig 1 Alvarez 1993) Subsequentlydocumented declines in the numbers of panthers were noted inthe nineteenth and twentieth centuries mainly as a result of

6 Wildlife Monographs bull 203

anthropogenic alteration of the landscape and unregulatedhunting As early as the 1870s panthers were rarely encounteredand considered mythical in some portions of their range (Beverly1874) The decline in the population of panthers eventuallycaptured the attention of lawmakers subsequently they receivedpartial legal protection as a game animal in 1950 and completelegal protection in Florida by 1958 (Onorato et al 2010) By the1960s the core of remaining wilderness within the Big Cypressregion of South Florida where the last group of breedingpanthers was thought to persist was further affected by theconstruction of Alligator Alley This roadway along with theTamiami Trail allowed access to these once inaccessible areasAll these factors ultimately led to the listing of the Floridapanther as endangered on the inaugural list of endangeredspecies on 11 March 1967 (US Federal Register 1967) andsubsequent protection under the Endangered Species Act of1973 (Public Law 93‐205) This designation invariably raisedawareness of the plight of the panther and served as a catalyst toinitiate research to avoid what appeared to be their imminentextinction (Onorato et al 2010)In 1973 the World Wildlife Fund initiated a survey that re-

sulted in the capture of a single female panther and doc-umentation of their sign in just a few locations in South Florida(Nowak and McBride 1974) The information accrued duringthis survey would ultimately lead to the commencement of along‐term research project on the Florida panther in 1981 by

what is now known as the Florida Fish and Wildlife Con-servation Commission (FWC) Over the next decade a multi-tude of studies were published that helped improve knowledgeof panther reproduction (Maehr et al 1989 Barone et al 1994)diet (Maehr 1990) and genetics (OrsquoBrien et al 1990 Roelkeet al 1993) among other topics Early research was essential foridentifying challenges faced by the remaining panthers mostimportantly the impacts of the continued loss of habitat isola-tion of the population and inbreeding depression (Onoratoet al 2010) In combination these factors would force wildlifemanagers to contemplate the implementation of genetic in-trogression to avert the extinction of the pantherBy the early 1990s the Florida panther population had declined

to a minimum of 20ndash30 individuals (McBride et al 2008) Severalstudies documented low levels of genetic diversity and expressedtraits thought to be correlated with inbreeding depression (Roelkeet al 1993 Johnson et al 2010 Onorato et al 2010) includingcowlicks (mid‐dorsal pelage whorls) and kinked tails (Wilkinset al 1997) These morphological traits probably had little directimpact on fitness of Florida panthers More concerning were theproportion of individuals in the population in the early 1990s thatwere unilaterally or bilaterally cryptorchid (Roelke et al 1993)possessed poor sperm quality (Barone et al 1994) sufferedcompromised immune systems (OrsquoBrien 1994) exhibited atrialseptal defects (Roelke et al 1993) and possessed some of thelowest levels of genetic variation observed in wild felids (Driscoll

Figure 1 Historical and current (2015) breeding range of the Florida panther in the southeastern United States Delineation of current breeding range is describedin Frakes et al (2015)

van de Kerk et al bull Florida Panther Population and Genetic Viability 7

et al 2002) Taken together these factors suggested that thelong‐term prospects for the persistence of the Florida pantherwere poorSubsequently the United States Fish and Wildlife Service

(USFWS) conducted an assessment that led to the delineation of 4possible management options 1) initiate a captive breeding pro-gram 2) introduce genetic material from captive stockpurported to contain both pure Florida panther and SouthAmerican puma lineages 3) translocate wild female pumas (Pumaconcolor stanleyana) from Texas USA into South Florida or 4)take no action at all (USFWS 1994) Collaborative discussionsinvolving academics non‐governmental organizations and stateand federal wildlife agencies eventually led to the selection of op-tion 3 and a plan for genetic introgression that involved the releaseof 8 female pumas from Texas into South Florida was im-plemented in 1995 (Seal 1994 Onorato et al 2010)Continued research conducted after the introgression interven-

tion provided clear evidence of the benefits accrued to the pantherpopulation including increased levels of genetic diversity declinesin the frequency of morphological and biomedical correlates ofinbreeding depression and the substantial increase of the popu-lation size (McBride et al 2008 Johnson et al 2010 McClintocket al 2015) Furthermore Hostetler et al (2013) showed that thepost‐introgression population growth was driven primarily byhigher survival rates associated with admixed panthers (offspringof pairings between the female Texas pumas and male Floridapanthers and subsequent generations of those progeny) followinggenetic introgression The apparent success of genetic introgres-sion in preventing the imminent extinction of the Florida pantherpopulation is encouraging but the population remains small andisolated and continues to face substantial challenges especially thelarge‐scale loss of habitatAlmost 5 panther generations (generation time= 45 years

Hostetler et al 2013) have passed since the implementation of thegenetic introgression program The panther population has beenmonitored continuously during this time period leading to asubstantial amount of additional data and biological insights sincethe demographic assessment made immediately following the in-trogression (Hostetler et al 2010 2012 2013 Benson et al 2011)Given that the effects of genetic introgression are not permanent itis important to discern whether the positive impacts of this man-agement initiative are continuing or waning Decline in the benefitsof the introgression could reduce the population growth rate andprobability of persistence At the same time panther habitat israpidly lost to a variety of human activities (eg land development)and environmental changes associated with invasive species andclimate change (Land et al 2008 Dorcas et al 2012 Frakes et al2015) When long‐term monitoring data are available it is prudentto periodically evaluate demographic rates population growth andpersistence parameters so that changes are detected and timelymanagement action can be implementedTwo population modeling approaches have become popular

among ecologists and wildlife managers for studying the dynamicsand persistence of age‐ or stage‐structured populations matrixpopulation models and individual‐based models (IBMs Brooket al 2000 Morris and Doak 2002 Morrison et al 2016) Matrixpopulation models are the generalized discrete time‐version of theLotka‐Euler equation which describes the dynamics of age‐ or

stage‐structured populations of plants animals and humans (Lotka1924 Caswell 2001) They have become standard tools for mod-eling human and wildlife populations (Tuljapurkar 1990 Caswell2001 Keyfitz and Caswell 2005) because 1) they are powerful andflexible permitting adequate representation of life history of a widevariety of organisms including those with complex life histories 2)parameters for these models can be empirically estimated usingstandard analytical methods such as multi‐state capture‐recapturemethods (eg Williams et al 2002) 3) it is relatively straight‐forward to include the effects of factors such as environmental anddemographic stochasticity and density dependence 4) importantdemographic quantities such as asymptotic population growth ratestable age‐ or stage distribution and generation time can be easilycalculated from these models 5) they permit prospective and ret-rospective perturbation analyses which have been proven useful inwildlife conservation and management 6) they are computer‐friendly do not require extensive programming experience andtheir implementation using software packages such as R (R CoreTeam 2016) and MATLAB (MathWorks 2014) is straight‐for-ward and 7) they have sound theoretical foundation offeringanalytical solutions to many aspects of population modeling Weused matrix population models for 1) calculating the aforemen-tioned demographic quantities 2) performing sensitivity analysesinvolving asymptotic population growth rate and 3) comparativepurposes to check the results obtained from equivalent simulation‐based modelsDespite many advantages offered by matrix population models it

is difficult to incorporate attributes of individuals such as geneticsor behavior within that framework Agent‐based or IBMs (Grimm1999 Grimm and Railsback 2005 McLane et al 2011 Railsbackand Grimm 2011 Albeke et al 2015) offer an alternative modelingframework with tremendous flexibility Individual‐based modelsare computer simulation models that rely on a bottom‐up approachthat begins by explicitly considering the components of a system(ie individuals) population‐level properties emerge from the be-havior of and interactions among discrete individuals UsingIBMs it is possible to explicitly represent attributes of individualssuch as movement mating or genetics (McLane et al 2011Railsback and Grimm 2011 Albeke et al 2015) We used in-dividual‐based population models as the primary populationmodeling approach because they offered a framework for an ex-plicit consideration of genetics both at individual and populationlevelsOur overall goal was to provide a comprehensive demographic

assessment of the sole extant population of the Florida pantherusing long‐term (1981ndash2015) field data and novel modelingtechniques Our objectives were 1) to estimate demographicparameters for the Florida panther and investigate whetherpositive impacts of the 1995 genetic introgression remained inthe population and if so to what extent 2) to estimate popu-lation growth and persistence parameters using an IBM tailoredto the Florida pantherrsquos life history and 3) to evaluate thesensitivity of population growth and persistence parameters tovital demographic ratesThe Florida panther population remains small (le230 adults and

subadults FWC 2017) with no realistic possibility of natural geneflow so the loss of genetic variation and concomitant increase ininbreeding‐related problems are inevitable Ensuring the long‐term

8 Wildlife Monographs bull 203

persistence of the population will likely necessitate periodic geneticmanagement interventions Therefore our final objective was to 4)evaluate genetic and population‐dynamic consequences of variedgenetic introgression strategies and from these identify manage-ment strategies that are affordable and effective at improvingprospects of Florida panther persistence To achieve this objectivewe extended the individual‐based population model to track in-dividual genetics based on empirical allele frequencies taking ad-vantage of the long‐term demographic and genetic data that havebeen collected on Florida panthers since 1981 We used individualheterozygosity to inform vital rates based on empirically estimatedrelationships between individual heterozygosity and demographicparameters We then simulated population trajectories trackedindividual heterozygosity over time and estimated quasi‐extinctiontimes and probabilities without introgression and with introgres-sion for a range of intervals and number of immigrants perintrogression event We compared the genetic and demographicbenefits of implementing alternative genetic introgression scenariosand ranked them by the level of improvement in population per-sistence parameters and estimated costs because funding is a per-sistent challenge to conservation planningManagement of imperiled species is a complex process invol-

ving many stakeholders and plagued by layers of uncertainties(Runge 2011) It would therefore seem that management ofthreatened and endangered species would be a perfect case foradaptive resource management approach (ARM) because ARMfocuses on structured decision making for recurrent decisionsmade under uncertainty (Williams et al 2002 Runge 2011)Because it combines structured decision making with learningprudent application of ARM reduces uncertainty over time andcan lead to optimal management actions (Williams and Brown2012 2016) Key ingredients of a successful adaptive manage-ment program include stakeholder engagement clearly definedand mutually agreed‐upon objectives alternative managementactions and objective decision rules (Nichols et al 2007Williams and Brown 2012 2016) Unfortunately Florida pan-ther stakeholders have divergent views about the state of thesystem (eg panther abundance) and it is a challenge to havemutually agreed‐upon objectives management actions or deci-sion rules acceptable to all or most stakeholders AlthoughFlorida panther conservation efforts would be best served withinthe ARM framework in the long run impediments to its im-plementation (Runge 2011 Williams and Brown 2012) areunlikely to be overcome in the near future

STUDY AREA

The breeding population of Florida panthers mainly persists as asingle population South of the Caloosahatchee River (approxi-mately 267133degN latitude 815566degW longitude see Fig 1)One exception is a female that was documented just north of theCaloosahatchee River in 2016 (and subsequently documentedwith kittens in 2017) by FWC the first validation of a femalepanther north of the River since 1973 (FWC unpublisheddata) The breeding range in South Florida is topographicallyflat has a subtropical climate and is characterized by permanentand ephemeral wetlands influenced by rains from May throughOctober (Duever et al 1986) The study area contains a variety

of wildland land cover types including hardwood hammockscypress forests pine flatwoods freshwater marshes prairies andgrasslands (Davis 1943) and land characterized by human ac-tivity including citrus groves croplands pastureland rockmining and residential developments (Onorato et al 2011)The area is intersected by a multitude of roads that fragmentpanther habitat Most notable among these is Interstate 75(Alligator Alley) which was expanded from 2 to 4 lanes in 1990when fencing and wildlife crossings were constructed with theintention of minimizing harmful effects of road expansion onlocal wildlife (Foster and Humphrey 1995) A large portion ofthe area that comprises South Florida is under public ownershipmainly as federal or state lands Everglades National Park BigCypress National Preserve Florida Panther National WildlifeRefuge Picayune Strand State Forest and Fakahatchee StrandPreserve State Park alone encompass 9745 km2 though not allof the area is considered suitable for panthers (eg large ex-panses of sawgrass marshes in the western Everglades) Frakeset al (2015) delineated 5579 km2 of adult panther habitat inSouth Florida 1399 km2 of which are in private ownership Amajority of these private lands are located in the northern extentof the breeding range Although some private lands may beprotected (eg conservation easements) other areas are sus-ceptible to incompatible land uses such as rock mining or re-sidential developments The breeding range in South Florida isbounded to the east and west by large metropolitan areas in-clusive of Miami‐Fort Lauderdale‐Pompano Beach and CapeCoral‐Fort Myers‐Naples‐Marco Island‐Immokalee respec-tively that are populated by gt6000000 people (httpswwwcensusgov accessed 24 Jan 2019) These characteristics of thestudy area highlight the recovery challenges faced by panthers

METHODS

Field MethodsSince 1981 Florida panthers have been captured radiocollared andtracked by the FWC and National Park Service staff (Table 1)Capture and handling protocols followed guidelines of theAmerican Society of Mammalogists (Sikes 2016) Livestock Pro-tection Company (Alpine TX USA) provided trained hounds andhoundsmen for captures Teams either treed or bayed panthers onthe ground and then darted them with a 3‐ml compressed‐air dartfired from a CO2‐powered rifle Immobilization drugs have variedduring the tenure of the project but most recently teamsimmobilized panthers with a combination of ketamine HCl (10mgkg Doc Lanersquos Veterinary Pharmacy Lexington KY USA) andxylazine HCl (1mgkg Doc Lanersquos Veterinary Pharmacy) Fol-lowing immobilization teams lowered treed panthers to the groundby a rope or caught panthers with a net in some cases they used aportable cushion (McCown et al 1990) to further mitigate theimpact of a fall Capture teams administered propofol (PropoFlotradeAbbott Laboratories Abbott Park IL USA) intravenously either asa bolus or continuous drip to maintain anesthesia They adminis-tered midazolam HCl (003mgkg) intramuscularly or intravenouslyto supplement anesthesia in some panthers Panthers recovered in ashaded area away from water In some cases capture teams reversedxylazine HCl with yohimbine HCl (Yobinereg Lloyd IncShenandoah IA USA) at 25 its recommended dose

van de Kerk et al bull Florida Panther Population and Genetic Viability 9

Capture teams determined sex and age of panthers marked themwith an ear tattoo and a subcutaneous passive integrated trans-ponder (PIT‐tag) and fitted them with a very high frequency(VHF) or global positioning system (GPS) radio‐collar equippedwith a mortality sensor The GPS collars were programmed toattempt to collect a position at a variety of time intervals(range= 1ndash12 hr) depending on objectives of concurrent studiesobservers routinely tracked panthers with VHF radiocollars from afixed‐wing aircraft 3 times per week When a radio‐collar trans-mitted a mortality signal or when a location did not change forseveral days observers investigated the site to establish the pan-therrsquos fate When observers found a dead panther they assessed thecause of death based on an examination of the carcass and sur-rounding area if possible They then transported carcasses to anexperienced veterinarian for necropsy and a histopathological ex-amination to attempt to assign the cause of death Starting in 1995observers continually assessed successive locations of radio‐collaredfemales to determine if they initiated denning behavior Three to 4fixes at the same location (VHF collars) or successive returns to thesame location over a weeklong period (GPS collars) served as anindicator of a possible den Observers then located dens moreprecisely via triangulation with ground telemetry They visited denswhen the dam left the site typically to make a kill Teams capturedkittens by hand captured kittens ranged in age from 4 to 35 dayspost‐partum Teams counted sexed weighed dewormed andpermanently marked kittens with a subcutaneous PIT‐tag

Genetic AnalysisTeams collected blood tissue or hair from the majority ofcaptured kittens subadult and adult Florida panthers for DNAsamples The United States Department of Agriculture ForestService National Genomics Center for Wildlife and FishConservation (Missoula MT USA) processed samplesTechnicians used Qiagen DNeasy Blood and Tissue Kits(Qiagen Valencia CA USA) to extract whole genomic DNAfrom blood and tissue samples and used a slight modification(overnight incubation of sample in lysis buffer and proteinase Kon a rocker at 60deg elution of DNA in 50 μl of buffer) of theQiagen DNA extraction protocol (Mills et al 2000) to extractDNA from hair samples Technicians used a panel of 16 mi-crosatellite loci (Fca090 Fca133 Fca243 F124 F37 Fca075Fca559 Fca057 Fca081 Fca566 F42 Fca043 Fca161

Fca293 Fca369 and Fca668) identified by Menotti‐Raymondet al (1999) which had been previously applied to Floridapanther samples (Johnson et al 2010) Technicians amplifiedDNA samples in 10‐microl reaction volumes that included 10 microl ofDNA 1times reaction buffer (Applied Biosystems Life Technolo-gies Corporation Grand Island NY USA) 20 mM of MgCl2200 microM of each dNTP 1 microM of reverse primer 1 microM of dye‐labeled forward primer 15 mgml of bovine serum albumin(BSA) and 1U Taq polymerase (Applied Biosystems) Thepolymerase chain reactions (PCRs) included a thermal profile of94degC for 5 minutes followed by 36 cycles of 94degC for 60 seconds55degC for 60 seconds and 72degC for 30 seconds Techniciansvisualized the resulting PCR products on a LI‐COR DNAanalyzer (LI‐COR Biotechnology Lincoln NE USA) Tech-nicians collected genotypes from hair samples (n= 16) using amultitube approach (Taberlet et al 1996) error‐checked geno-types using Program DROPOUT (McKelvey and Schwartz2005) and combined them with genotypes from tissue andblood samples (n= 487) We evaluated all 16 loci for variousdescriptive statistics for the radio‐collared sample of adult andsubadult panthers We excluded genotypes of kittens from thatanalysis because related individuals are known to result in theoverrepresentation of certain alleles (Marshall et al 1998) Wedetermined the number of alleles observed heterozygosity andexpected heterozygosity using GenAlex (Peakall and Smouse2006 2012) We assessed allelic richness at each locus in Fstat293 (Goudet 1995) We assessed conformance of genotypedata from these 16 loci to HardyndashWeinberg assumptionslinkage disequilibrium and null alleles (Appendix A availableonline in Supporting Information) The Genomics Center alsoprovided us with genotype data from the same 16 loci for 41pumas from western Texas the source population for thegenetic rescue programEvidence of inbreeding depression in wild populations is

commonly determined on the basis of heterozygosityndashfitnesscorrelations (Silva et al 2005 Ortego et al 2007 Mainguyet al 2009 Johnson et al 2011) so we also estimated theheterozygosity of each individual panther We calculatedhomozygosity by loci (HL) which varies between 0 (all lociheterozygous) and 1 (all loci homozygous Aparicio et al 2006)using the Rhh package (Alho et al 2010) in Program R (R CoreTeam 2016) A microsatellite locus has more weight in HL

Table 1 Number of Florida panthers by age sex and genetic ancestry categories used in subsequent demographic analyses during the study period (1981ndash2013) inSouth Florida USA

Kittensa Subadults and adultsb

Males Females Total Males Females Total

PIT‐taggedc 215 180 395 Radiocollaredc 108 101 209Recovered 11 4 15 Subadult 72 50 122Litter failed 47 38 85 Prime adult 65 95 160Recaptured 25 30 55 Old adult 13 20 33Canonical 27 20 47 Canonical 43 29 72F1 admixed 6 12 18 F1 admixed 2 6 8Other admixed 130 105 235 Other admixed 45 53 98

a Kitten samples included those that were PIT‐tagged recovered (ie panthers that were found dead) part of a failed litter or recaptured as subadults or adultsb Radiocollared subadult and adult panthers are presented as sample size within each age class and ancestry classc Combined number of panthers of different ancestries is less than the number radiocollared or tagged with a subcutaneous passive integrated transponder (PIT‐tagged) because we did not conduct genetic sampling on all panthers Combined number of panthers in different age classes is greater than the number ofradiocollared panthers because some individuals were counted in ge2 age classes as animals aged

10 Wildlife Monographs bull 203

when it is more informative (ie has more alleles) and has aneven distribution of allele frequencies For our demographicanalyses we calculated individual heterozygosity for each in-dividual panther as 1 minusHL As do most indices of individualheterozygosity HL takes into account the average allele fre-quency in the population (Aparicio et al 2006) Thus duringsimulations the allele frequency in the population changesthroughout an individualrsquos lifetime which causes its hetero-zygosity also to change This is not biologically plausible becausegenetic properties are retained throughout an individualrsquos life-time Therefore for use in model simulations we simply cal-culated heterozygosity for each individual as the proportion ofheterozygous lociWe also used genotype data to implement a Bayesian clus-

tering analysis in Program STRUCTURE (Pritchard et al2000) to infer ancestral clusters of panthers This method uses aMarkov chain Monte Carlo (MCMC) method to determine thenumber of genetic clusters (K) in the sample while also pro-viding an individualrsquos percentage of ancestry (q values) allocatedto each cluster Runs for this analysis included a 100000MCMC burn‐in period followed by 500000 MCMC iterationsfor 1ndash10 ancestral genetic clusters (K= 1 to 10) with 10 re-petitions for each K To determine the most likely number of Kgenetic clusters we used the logarithm of the probability of thedata (log Pr(D)∣K Pritchard et al 2000) and estimates of ΔK(Evanno et al 2005) in Program STRUCTURE HAR-VESTER (Earl and VonHoldt 2011) We then used q valuesprovided in runs for the best K to assign individual panthers aseither canonical (in most cases ge90 pre‐introgression ancestrysee Johnson et al 2010) or admixed (lt90 pre‐introgressionancestry) We recognized admixed panthers that were the off-spring of matings between a Texas female puma and a canonicalmale panther as F1 admixed panthers for analyses

Demographic ParametersSurvival of kittensmdashMost kittens PIT‐tagged at den sites were

never encountered again A small proportion of the PIT‐taggedkittens were recovered dead or were recaptured as subadults oradults and radiocollared so that their fate could be monitoredWe also identified instances of complete litter failure when afemale died within 9 months after giving birth (the minimumage of independence for kittens) or when a female wasdocumented to have another litter less than 12 months aftergiving birth (the minimum age of independence plus a 3‐monthgestation period Hostetler et al 2010) Thus our data setconsisted of 1) live recaptures and subsequent radio‐tracking ofpanthers of various ages that had been PIT‐tagged as kittens 2)recovery of dead panthers of various ages that had been PIT‐tagged as kitten 3) live captures and subsequent radio‐trackingof other panthers and 4) instances of complete litter failure(Table 1) We analyzed these data using Burnhamrsquos live‐recapture dead‐recovery modeling framework (Burnham 1993Williams et al 2002) to estimate survival of kittens (0ndash1‐yearold) and to test covariate effects on this parameter We includeddata on individuals encountered dead or alive at an older age inour analyses because their encounter histories also providedinformation on kitten survival We used a live‐dead data inputformat coded with a 1‐year time step (Cooch and White 2018)

All known litter failures occurred within the first year of thekittensrsquo lives so we treated all litter failures as recoveries withinthe first year We treated panthers that would have died if wehad not removed them to captivity as having died on the date ofremoval Data preparation and analytical approach are describedin more detail by Hostetler et al (2010)In addition to survival probability (S) the Burnham model

incorporates parameters for recapture probability (p) recoveryprobability (r) and site fidelity (f) We set r and p to 10 forradio‐tracked individuals because their status was known eachyear We also fixed f at 10 for all individuals because the re-capture and recovery areas were the same and encompassed theentire range of the Florida panther Based on findings of Hos-tetler et al (2010) we used a base model that 1) allowed survivalto differ between kittens and older panthers 2) allowed survivalto differ between sexes and among age classes (females ages 1and 2 ge3 males ages 1 2 and 3 ge4) 3) constrained recaptureprobability to be the same for all uncollared panthers and 4)allowed recovery rates to differ between kittens and uncollaredolder panthers Even with our additional data this model re-mained the best fit for a range of models for recapture recoveryand survival of kittens subadults and adults of various age‐classcategories (Hostetler et al 2010)Subadult and adult survivalmdashTo estimate survival for panthers

ge1 year of age we analyzed radio‐tracking data using a Coxproportional hazard‐modeling framework (Cox 1972 Therneauand Grambsch 2000) We followed procedures outlined byBenson et al (2011) for data preparation and analysis Brieflywe right‐censored panthers on the last day of observation beforetheir collar failed When an individual aged into a new age classwhile being tracked we created 2 data entries 1 ending the dayit entered the new age class the other starting on that day Weused the FlemingndashHarrington method to generate survivalestimates from the Cox analysis and the Huber sandwichmethod to estimate robust standard errors (Therneau andGrambsch 2000 Benson et al 2011)We considered 3 age classes subadults (1ndash25 yr for females and

1ndash35 yr for males) prime adults (25ndash10 yr for females and35ndash10 yr for males) and old adults (gt10 yr old for both sexes)after Benson et al (2011) We estimated overall survival using abase model that allowed survival to differ between subadultfemales subadult males prime adult females prime adult malesand old adults of either sex (old age was additive with sex other-wise age and sex were interactive) This model provided the best fitto the data relative to a range of models considering differences insurvival between prime‐adult and older‐adult age classes both ad-ditively and interactively with sex (Benson et al 2011)To examine whether survival rates observed in recent years

differed from those observed immediately following introgres-sion we also estimated age‐specific survival rates for 2 nearlyequal periods of time immediately post‐introgression (1995ndash2004) and the more recent period (2005ndash2013) For theseanalyses we separated the radio‐tracking data into the 2 periodsand estimated survival rates for each period separately using thebase model Similar analyses for kitten survival were not possiblebecause of data limitationsProbability of reproductionmdashWe used telemetry data obtained

from radio‐tracked females that reproduced at least once to

van de Kerk et al bull Florida Panther Population and Genetic Viability 11

estimate annual probability of reproduction of subadult prime‐adult and old‐adult females using complementary logndashlogregression (Agresti 2013) following methods described indetail by Hostetler et al (2012) Briefly we modeled theprobability of a female panther giving birth (b) as a function ofcovariates as

γ= minus [minus ( )]b z1 exp exp i i

where zi is a row vector of covariates (see below for covariateinformation) for female panther i and γ is a column vector ofmodel coefficients (Appendix B Table B2) To account fordifferences in observation time we included log(m) as an offsetin the model where m is the number of months a femalepanther was radio‐tracked (Hostetler et al 2012)

Reproductive outputmdashTo estimate the number of kittensproduced by reproductive females we used cumulative logitregression (Min and Agresti 2005 Agresti 2013) Weconsidered a model that includes J categories for the numberof offspring with j denoting the number of offspring ( j= 1 2hellip J and J= 6) provided that a female reproduces Weconsidered J= 6 which is greater than the largest litter sizeobserved in our study (4 kittens) because females can produce 2or more litters per year if litters fail We modeled the probabilitythat a female i produces a litter of size at most j (Yi) as

le⎧

⎨⎩

δ( ) = == hellip minus

=

( )+ θ βminus +Y jj J

j JPr

1 2 1

1i ij e

1

1 xj i

where θj is the intercept for the annual number of kittens pro-duced by a female i being at most j xi is a row vector of cov-ariates for female panther i (see below) and β is a column vectorof model coefficients The probability that Yi will be exactly j isgiven by

⎧⎨⎩

πδ

δ δ( = ) = =

=

minus = hellip( minus )Y j

jj J

Pr 1

2 i ij

i

ij i j

1

1

Finally the expected annual number of kittens produced byfemale i (νi) is given by

sumν π==

j i

j

J

ij

1

Additional details regarding the estimation and modelingof the annual number of kittens produced can be found inHostetler et al (2012)

CovariatesmdashIn addition to sex and age we tested for the effect ofabundance genetic ancestry and individual heterozygosity on theaforementioned demographic parameters (model coefficients forkitten and adult survival are given by η and ι respectively AppendixB Table B2) We used a minimum population count (MPC) basedon radio‐tracking data and field evidence of subadult and adult (ieindependent age) panthers as an index of abundance (McBride et al2008) For these analyses we did not use the population sizeestimates reported by McClintock et al (2015) because unlike theminimum counts they did not cover the entire study period

(McBride et al 2008 R T McBride and C McBride RancherrsquosSupply Incorporated unpublished report Fig 2)To test if demographic parameters differed depending on an-

cestry we considered 3 ancestry models dividing the panthers into1) F1 admixed and all other panthers 2) canonical and admixed(including F1 admixed) panthers and 3) F1 admixed canonicaland other admixed panthers We quantified genetic diversity usingindividual heterozygosity as described previously We tested for theeffect of genetic covariates using a subset of kittens that were ge-netically sampled We calculated model‐averaged estimates ofmodel parameters on the scale in which they were estimated(Appendix B available online in Supporting Information)

Cause‐specific mortalitymdashWe performed cause‐specificmortality analysis only on dead radio‐collared panthers becausedetection rates of uncollared panther mortalities are highlycorrelated with the cause of death (eg vehicle collision)Following Benson et al (2011) we attributed each mortality to1 of 4 causes 1) vehicle collision 2) intraspecific aggression 3)other causes (including known causes such as diseases injuriesand infections unrelated to the first 2 causes) and 4) unknown(ie mortalities for which we could not assign a cause) Wethen used the non‐parametric cumulative incidence functionestimator a generalization of the staggered‐entry KaplanndashMeiermethod of survival estimation to estimate cause‐specificmortality rates for male and female panthers in different ageclasses (Pollock et al 1989 Heisey and Patterson 2006 Bensonet al 2011) We compared cause‐specific mortality rates betweensexes age classes and ancestries

Statistical inferencemdashWe used an information‐theoreticapproach (Akaikersquos information criterion [AIC]) for modelselection and statistical inference (Burnham and Anderson 2002)For the analysis of kitten survival we adjusted the AIC foroverdispersion and small sample size (QAICc details in Hostetleret al 2010) We performed all statistical analyses in Program R (RCore Team 2016) We used Burnhamrsquos live recapturendashdead

0

100

200

300

1985 1990 1995 2000 2005 2010Year

Indi

vidu

als Method

MPC

MVM

Figure 2 The Florida panther minimum population count (MPC) and populationsize estimated using motor vehicle collision mortalities (MVM) during our studyperiod Minimum population counts are from McBride et al (2008) and R TMcBride and C McBride (Rancherrsquos Supply Incorporated unpublished report)Estimates based on MVM are from McClintock et al (2015) The solid vertical lineindicates the year of genetic introgression

12 Wildlife Monographs bull 203

recovery model (Burnham 1993) to estimate and model kittensurvival using the RMark package version 2113 (Laake 2013) asan interface for Program MARK (White and Burnham 1999)We implemented the Cox proportional hazard model for esti-

mation and modeling of survival of subadults and adults using the Rpackage SURVIVAL (Therneau and Grambsch 2000) We per-formed cause‐specific mortality analyses using an R version of S‐PLUS code provided by Heisey and Patterson (2006)To account for model selection uncertainty we calculated

model‐averaged parameter estimates across all models using theR package AICcmodavg (Mazerolle 2017) using AIC weights asmodel weights (Burnham and Anderson 2002) We used modelswithout the categorical ancestry variables for model averagingand estimated parameters based on the average value for thecontinuous covariates (individual heterozygosity and abundanceindex) Because only a subset of the kittens was geneticallysampled we estimated 2 kitten survival rates First we estimatedkitten survival based on the data set that included all kittens(overall kitten survival) Second we estimated kitten survivalusing a subset of the data that only included kittens that weregenetically sampled (survival of genetically sampled kittens)

Matrix Population ModelsWe investigated Florida panther population dynamics andpersistence using 2 complementary modeling frameworks ma-trix population models (Caswell 2001) and IBMs (Grimm andRailsback 2005 McLane et al 2011 Railsback and Grimm2011 Albeke et al 2015) Matrix population models offer aflexible and powerful framework for investigating the dynamicsand persistence of age‐ or stage‐structured populations (egCaswell 2001 Robinson et al 2008 Kohira et al 2009 Hunteret al 2010 Hostetler et al 2012) With a fully developed theorybehind them these models also serve as a check for resultsobtained from simulation‐based models such as IBMs We usedan age‐structured‐matrix population model for deterministic andstochastic demographic analyses and for preliminary investiga-tions of Florida panther population viability (Caswell 2001Morris and Doak 2002)For deterministic and stochastic demographic analyses we

parameterized a female‐only 19 times 19 age‐specific populationprojection matrix of the form

⎢⎢⎢⎢⎢

⋯ ⋯

⋯ ⋯

⋱ ⋱ ⋮

⋮ ⋱ ⋱ ⋱ ⋮

⎥⎥⎥⎥⎥

( )=t

F F F

P

P

P

A0 0

0

0 0 0

t t t

t

t

t

1 2 19

1

2

18

where Fi and Pi are the age‐specific fertility and survival ratesrespectively and ( )tA is a time‐varying (or stochastic) popula-tion projection matrix We assumed that the longevity and ageof last reproduction of female Florida panthers was 185 yearswe based this assumption on the observation that the oldestfemale in our study was 186 years old Florida panthers re-produce year‐round thus we used birth‐flow methods to esti-mate Fi and Pi values following Caswell (2001) and Hostetleret al (2013) We estimated Pi as

= +P S S i i i 1

where Si is the age‐specific survival probability We estimated Fi as

⎛⎝

⎞⎠

=+ +F S

m Pm

2i k

i i i 1

where Sk is the kitten survival probability and mi is the age‐specific fecundity rate which was estimated as

ν=m b f i i i k

where bi is the breeding probability for a female in age class iduring time step t νi is the annual number of kittens producedby a breeding female in age class i and fk is the proportionfemale kittens at birth (assumed to be 05)We estimated the finite deterministic (λ) and stochastic annual

population growth rate (λs) and stochastic sensitivities and elasti-cities following methods described in detail by Caswell (2001) Toestimate λs we assumed identically and independently distributedenvironments (Caswell 2001 Morris and Doak 2002 Haridas andTuljapurkar 2005) and used a parametric bootstrapping approachto obtain a sequence of demographic parameters for 50000 ma-trices We then estimated log(λs) from this sequence of matrices(Tuljapurkar 1990 Caswell 2001) and exponentiated it to obtain λsWe calculated the sensitivity and elasticity of λs to changes in lower‐level vital demographic rates as per Caswell (2001) and Haridas andTuljapurkar (2005)For population projection and population viability analysis (PVA)

simulations we expanded the population projection matrix into a34 times 34 2‐sex age‐structured matrix to include female and maleFlorida panthers (Caswell 2001 Hostetler et al 2013) Malescontribute to the population only through survival in this modelingframework We assumed maximum longevity for males to be 145years of age because the oldest male in our study was 144 years oldThe population projection matrix was of the following form

⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢

⋯ ⋯ ⋯ ⋯

⋯ ⋯ ⋯ ⋯

⋱ ⋱ ⋮ ⋮ ⋱ ⋱ ⋮

⋮ ⋱ ⋱ ⋱ ⋮ ⋮ ⋱ ⋱ ⋮

⋯ ⋯ ⋯

⋯ ⋯ ⋯ ⋯

⋯ ⋯ ⋱ ⋱ ⋮

⋮ ⋮ ⋱ ⋱ ⋮ ⋱ ⋱ ⋮

⋯ ⋯ ⋯

⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥

( )=

( ) ( ) ( )

( )

( )

( )

( ) ( ) ( )

( )

( )

t

F N F N F N

P N

P N

P N

M N M N M N

Q N

Q N

A n

0 0

0 0 0 0

0

0 0 0 0 0

0 0

0 0 0

00 0 0 0 0

t t t

t

t

t

t t t

t

t

1 2 19

1

2

18

1 2 19

1

14

where Pi and Qi are age‐specific survival rates of female and maleFlorida panthers respectively and Fi and Mi are the rates atwhich female and male kittens are produced by females of ageclass i respectively Here the population projection matrix isboth time‐varying and density‐dependent (Caswell 2001)We incorporated the influence of parameter uncertainty en-

vironmental stochasticity and density dependence on Floridapanther population dynamics and persistence following the ap-proach outlined by Hostetler et al (2013) we used the sameapproach in the IBM analysis which is described below UnlikeIBMs matrix models do not intrinsically include demographicstochasticity so we explicitly included the influence of

van de Kerk et al bull Florida Panther Population and Genetic Viability 13

demographic stochasticity by simulating the fates of individualsas described in Caswell (2001)Population projections using matrix models necessitate a

vector of initial sex‐ and age‐specific abundance Because suchdata are currently unavailable for the Florida panther we esti-mated it based on the stable sex and age distribution and MPCin 2013 Using point estimates of the demographic parameterswe estimated the stable sex and age distribution for the 2‐sexdeterministic and density‐independent population projectionmatrix We then multiplied the stable sex and age distributionby the MPC in 2013 to get the starting population vector n(0)We projected the population over time as n(t+ 1)=A(tn)n(t)where A(tn) indicates the time‐specific density‐dependentpopulation projection matrix (Caswell 2001)We had 2 indices of abundance available the MPC (McBride

et al 2008) and population size estimated using motor vehiclecollision mortalities (MVM McClintock et al 2015) These 2indices are substantially different because MPC does not ac-count for imperfect detection whereas MVM accounts forimperfect detection and provides an estimate of population size(Fig 2) Therefore we examined 2 scenarios one using densitydependence based on MPC and the other using density de-pendence based on MVM We ran 1000 bootstraps of para-meter values and 1000 simulations per bootstrap for each sce-nario We projected population trajectories over 200 years andthen estimated population growth rates probabilities ofquasi‐extinction and time until quasi‐extinction and comparedthese with the results obtained from the IBM (see below) Weran scenarios with a quasi‐extinction threshold of 10 and 30panthers We used the mean of probabilities of quasi‐extinction(PQE) across simulations as the point estimate for PQE and the5th and 95th percentiles across simulations as the 90 con-fidence interval Because the confidence intervals werequantile‐based but the point estimates were not it is possiblewith skewed distributions for the point estimates to be outsidethe confidence interval We implemented the matrix

population models in the R computing environment (R CoreTeam 2016)

IBMs and PVABecause of well‐developed theory ease of implementation andthe modeling flexibility that they offer matrix populationmodels have become the model of choice for investigating thedynamics and persistence of age‐ and stage‐structured popula-tions (Caswell 2001) However it is difficult to incorporateattributes of individuals such as genetics and behavior usingmatrix models Quantifying the effects of genetic erosion on theFlorida panther population dynamics and persistence andevaluating benefits and costs of alternative genetic managementscenarios were important goals of our study Because IBMs relyon a bottom‐up approach that begins by explicitly consideringindividuals (McLane et al 2011 Railsback and Grimm 2011Albeke et al 2015) it is possible to represent genetic attributesof each individual and to track genetic variation over time Weused IBMs as the primary modeling framework in this studybecause they allowed for explicit consideration of genetics bothat individual and population levels and population viabilityassessment under alternative genetic management scenariosWe developed an IBM (Grimm 1999 Grimm and Railsback

2005 McLane et al 2011 Railsback and Grimm 2011 Albekeet al 2015) tailored to the life history of the Florida panther(Fig 3) We simulated every individual from birth until deathbased on a set of rules describing fates (eg birth and death) andattributes (eg genetics) of individuals at each annual time stepAs in other IBMs population‐level processes (eg populationsize and genetic variation) emerged from fates of and interac-tions among individuals (Grimm 1999 Railsback and Grimm2011) We developed an overview design concepts and details(ODD) protocol (Grimm et al 2006 2010) to describe ourIBM (Appendix C available online in Supporting Information)and provide pseudocodes for individual‐based simulations(Appendix D available online in Supporting Information)

For each panther at time = t

Kitten (lt1 yr)

Reproduce

Survive

Kitten(s) born

Kitten

Sex and age class

Subadult male

Prime adult male

Old adult male

Subadult female

Prime adult female

Old adult female

Age one year

No Male

Female

Yes

No

Yes

t=t+1

Yes

lt35 yrs

35-10 yrs

gt10 yrs

lt25 yrs

25-10 yrs

gt10 yrs

Figure 3 Schematic representation of Florida panther life history We tailored the individual‐based population model to represent the life‐history patterndepicted here

14 Wildlife Monographs bull 203

Dynamics and persistence of populations are influenced bymany factors such as demographic and environmental sto-chasticity and density dependence these effects and parametricuncertainty must be considered in population models whenpossible (Caswell 2001 Boyce et al 2006 Bakker et al 2009Hostetler et al 2013) Because by definition IBMs simulate thelife history of individuals in a population they inherently in-corporate the effect of demographic stochasticity Our analysesrevealed that survival of kittens subadults and adults and theprobability of reproduction varied over time so we incorporatedenvironmental stochasticity in the model for these variablesfollowing Hostetler et al (2013) Likewise we found evidenceof density‐dependent effects on kitten survival Because theMPC might be an underestimate of population size (McBrideet al 2008) we performed simulations using both the MPC andMVM abundance indicesTo account for model selection and parametric uncertainty we

used a Monte Carlo simulation approach For each bootstrapsample we selected a model based on the AIC (or QAICc)weights We then sampled the intercept and slope for kittensurvival (on the logit scale) subadult and adult survival (on thelogit scale with log‐hazard effect sizes) and probability of re-production (on the complementary logndashlog scale Appendix B)from multivariate normal distributions with mean vectors equalto the estimated parameters and variance‐covariance matricesequal to the sampling variance‐covariance matrices We thentransformed the results to nominal scale and converted the es-timates to age‐specific survival and reproduction parameters asdescribed in Hostetler et al (2013)We ran 1000 simulations for 1000 parametric bootstraps for

200 years for the MPC and the MVM scenario for both thestochastic 2‐sex matrix‐population model and the IBM Wetracked the population size at each time step and estimatedquasi‐extinction probabilities and times based on the populationtrajectory We considered the population to be quasi‐extinct ifindividuals of only 1 sex remained or if the population size fellbelow critical thresholds set at 10 and 30 individuals We alsoestimated expected time to quasi‐extinction for both thresholdsdefined as the mean and median time until a population be-comes quasi‐extinct conditioned on quasi‐extinction We im-plemented the IBM using the RNetLogo package version 10ndash2(Thiele et al 2012 Thiele 2014) as an interface for the IBMplatform NetLogo (Wilensky 1999)

Sensitivity AnalysisAn important component of demographic analysis is sensitivityanalysis which is the quantification of the sensitivity of amodelrsquos outcome to changes in the input parameters (Caswell2001 Bakker et al 2009 Thiele et al 2014) Using an IBM weperformed global sensitivity analyses for 2 (ie MPC andMVM) density‐dependent scenarios to assess how sensitivequasi‐extinction times and probabilities were to changes (anduncertainties) in the estimated demographic rates We usedLatin hypercube sampling to sample parameters from the entireparameter space (Marino et al 2008 Thiele et al 2014) Wesampled intercept and slope for kitten survival (on logit scale)subadult and adult survival (on logit scale with log‐hazard effectsizes) and probability of reproduction (on complementary

logndashlog scale) from normal distributions We sampled theprobability distribution for the number of kittens producedannually (on the real scale) from a Dirichlet distribution themultivariate generalization of the beta distribution (Kotz et al2000) We sampled 1000 different parameter sets and ran 1000simulations for each scenario We calculated the probability ofquasi‐extinction within 200 years for each density‐dependencescenario (MPC or MVM) using a quasi‐extinction threshold of10 individuals and then calculated the partial rank correlationcoefficient between each of those and each parameter trans-formed to the real scale (Bakker et al 2009 Hostetler et al2013) Partial rank correlation coefficients quantify the sensi-tivity of the probability of quasi‐extinction to input variables(ie survival and reproductive parameters) with higher absolutevalues indicating stronger influence of that variable on quasi‐extinction probabilities

Genetic ManagementOne of our goals was to estimate the benefits (improvements ingenetic variation demographic parameters and populationgrowth and persistence parameters) and costs (financial costs ofcapture transportation quarantine release and monitoring ofintroduced panthers) of genetic management To achieve thisobjective we extended the IBM to include a genetic componentTo simulate individual‐ and population‐level heterozygosity overtime we assigned each panther alleles for 16 microsatellite locibased on the empirical allele frequency in the population (esti-mated from genetic samples collected during 2008ndash2015) whenwe initialized the model (Pierson et al 2015) At each time stepwe simulated Mendelian inheritance for kittens produced suchthat each kitten randomly inherited alleles from its mother andfrom a male panther randomly chosen to have putatively siredthe litter We did not explicitly force inbreeding to occur butthe probability of inbreeding naturally increased as the popula-tion size decreasedIntrogression strategiesmdashWe modeled genetic introgression as a

result of the release of 2‐year‐old female pumas from Texas intothe simulated Florida panther population For the geneticintrogression implemented in 1995 Texas females between 15and 25 years of age were released because females at that agehave lower mortality rates (Benson et al 2011) higherreproductive potential and also because it is much easier todetect with certainty the successful reproduction by females thanby males The ability to more closely monitor reproduction andthe birth of kittens is an important consideration in reducing theprobability of a genomic sweep of Texas alleles into the Floridapanther population Wildlife managers removed the last 2surviving Texas females from the wild population in Florida in2003 to reduce the possibility of genomic sweep To simulatethis strategy in the model we removed any Texas females thatwere still alive 5 years after each introgression event Weassigned Texas females alleles based on the allele frequency ofthe Texas population (based on genotypes from 48 samplescollected in west Texas that was inclusive of the pumas releasedin South Florida in 1995) when they entered the Florida pantherpopulation We could not estimate demographic rates separatelyfor the released Texas females because of small sample size (only8 females were released in 1995) therefore we assumed that

van de Kerk et al bull Florida Panther Population and Genetic Viability 15

survival and reproductive rates and the effects of covariates onthese rates for the released Texas females were identical to thoseof female Florida panthersThe impact of genetic introgression depends on the frequency

of introgression events and the number of individuals introducedat each attempt Thus we defined introgression strategies basedon the number of Texas females to be released (0 5 10 or 15)and the interval between genetic introgressions (10 20 40 and80 yr) for a total of 13 strategies We simulated each manage-ment strategy with density dependence estimated using theMPC and MVM abundance indices

We sampled 1000 parameter sets and for each of these setswe ran all introgression strategies 1000 times for 200 years Weapplied the same parametric uncertainty and environmentalstochasticity across all management scenarios to allow for bettercomparison of introgression strategies At each time step werecorded the projected population size and average individualheterozygosity We calculated the heterozygosity for each in-dividual at birth based on the alleles it inherited This individualheterozygosity then informed the survival rate of that individualbased on the empirically estimated relationship between in-dividual heterozygosity and age‐specific survival rates Hetero-zygosity did not change throughout an individualrsquos lifetime butits effect on survival changed as individuals aged because theeffect of individual heterozygosity on survival rate differedamong age classes Survival rates of genetically sampled kittenswere biased high because some kittens were sampled later in lifewhen they had already survived for some time To correct forthis bias we adjusted kitten survival rates predicted by theheterozygosity model proportionally From the population tra-jectories we estimated the probability of quasi‐extinction (de-fined as the probability that the population will fall below 10 or30 individuals or that individuals of only 1 sex will remain)

Cost‐benefit analysismdashExperts estimated financial costs ofintrogression based on their experience with the geneticintrogression executed in 1995 and we adjusted estimates tocurrent costs to account for inflation (R T McBride RancherrsquosSupply Incorporated M Cunningham and D P OnoratoFlorida Fish and Wildlife Conservation Commission personalcommunication) Costs included capturing and transportingfemale pumas from the Texas source population caring for thepumas while they were in quarantine and post‐introgressionmonitoring To compare the financial costs of the differentscenarios we also calculated the average costs incurred over

Table 2 Sample size (n) number of alleles (Na) allelic richness (AR) observedheterozygosity (Ho) and expected heterozygosity (He) at 16 microsatellite loci inradio‐collared Florida panthers sampled from 1981 to 2013 in South FloridaUSA We calculated these values from a subset (n= 137) of the samples used inthe analyses and they include only adult and subadult panthers We excludedkittens because they can result in an overrepresentation of certain alleles

Locus n Na AR Ho He

FCA090 129 5 50 0333 0401FCA133 136 3 30 0618 0608FCA243 136 5 50 0419 0487F124 137 7 69 0708 0729F37 129 4 40 0395 0397FCA075 131 5 50 0534 0598FCA559 133 5 50 0496 0503FCA057 134 6 59 0679 0624FCA081 134 7 69 0209 0206FCA566 130 6 60 0462 0475F42 133 10 100 0737 0766FCA043 136 5 49 0596 0588FCA161 132 6 60 0500 0515FCA293 132 4 40 0614 0624FCA369 131 6 60 0573 0567FCA668 133 7 70 0406 0462Mean 1329 57 57 0517 0535

Figure 4 Proportional membership (q) of radio‐collared Florida panthers (n= 218) and pumas from Texas USA (n= 7) in 2 clusters identified by STRUCTUREEach individual animal is represented by a separate vertical bar Yellow indicates canonical panther ancestry and blue represents admixed ancestry The admixedancestry of the Texas females and F1 generation panthers that were radiocollared are highlighted red and green respectively to demarcate those groups The pre‐introgression period is inclusive of panthers radiocollared from 10 February 1981 to 6 March 1996 The post‐introgression era inclusive of the F1 panthers includesanimals radiocollared from 4 March 1997 to 18 February 2015 in South Florida USA

16 Wildlife Monographs bull 203

100 years assuming that the population would be managedaccording to a particular strategy Our goal with theseexploratory analyses was to evaluate the relative costs andbenefits of alternative management scenarios

RESULTS

Genetic VariationWe assessed genetic variation at 16 microsatellite loci for 137DNA samples collected from adult and subadult panthers(1981ndash2013) that were captured and subsequently radiocollaredThe number of alleles at these loci ranged from 3ndash10 and allelicrichness averaged 57 (Table 2) Average expected hetero-zygosity for this sample was 0535 and average observed het-erozygosity was 0517 (Table 2)We determined individual heterozygosity (1 minusHL) and an-

cestry for the complete sample of panther genotypes (n= 503)collected from 1981 to 2015 for use as covariates in subsequentanalyses Average individual heterozygosity was 0559 andranged from 0ndash0980 Our STRUCTURE analysis providedstrong support for K= 2 clusters based on the probability of thedata (log Pr(D)|K) and ΔK in STRUCTURE HARVESTERBased on this result we categorized the ancestry of each pantheras either canonical (n= 123) or admixed (n= 380) using valuesof proportional membership (q Fig 4) The average individualheterozygosity of canonical panthers was 0386plusmn 0012 (SE) foradmixed panthers it was 0615plusmn 0007

Survival and Cause‐Specific Mortality RatesOf the 395 kittens that were PIT‐tagged and released 55 werelater encountered alive 15 were recovered dead and 85 werepart of failed litters (Table 1) We radio‐tracked 209 subadult oradult panthers for a total of 244195 panther‐days and docu-mented 126 mortalities (Table 1)Survival depended strongly on age and kittens had the lowest

overall rate (032plusmn 009 Fig 5A Table 3) The model‐averagedkitten survival was 047plusmn 009 when we included only geneti-cally sampled individuals in the analysis (Table 3) as statedabove this estimate is biased high because many kittenswere genetically sampled when they were recaptured alive orrecovered dead at older ages We found no evidence for sex‐specific differences in kitten survival but females survived betterthan males in all older age classes For females subadults hadthe highest survival rates followed by prime adults and oldadults For males prime adults had the highest survival ratefollowed by subadults and old adults (Fig 5A and Table 3)For panthers of all ages there was substantial evidence that

ancestry affected survival with F1 admixed panthers of all ageshaving the highest rates and canonical individuals having thelowest (Fig 5B) The combined AIC weights of models thatincluded an ancestry covariate were 0880 for kittens and 0992for older panthers The survival rates of subadult prime‐adultand old‐adult panthers in the period immediately after the in-trogression (1995ndash2004) were similar to those in more recentyears (2005ndash2013 Fig 5C)After genetic introgression individual panther heterozygosity

increased (Fig 6) and varied among ancestry categories individualheterozygosity was the highest for F1 panthers (078plusmn 001)

A

B

C

Figure 5 Overall age‐ and sex‐specific survival rates (plusmnSE A) age‐ and sex‐specific survival rates (plusmnSE) for Florida panthers of canonical F1 admixed andother admixed ancestry (B) and age‐ and sex‐specific survival estimates (plusmnSE)for subadult and adult Florida panthers in the period immediately post‐introgression (1995ndash2004) and more recently (2005ndash2013 C) in SouthFlorida USA

Table 3 Model‐averaged estimates (plusmnSE) of demographic parameters for theFlorida panther obtained using data collected during 1981ndash2013 in SouthFlorida USA

Parameter Sex Age class Estimate

Survival Both Kitten 032plusmn 009a

Female Subadult 097plusmn 002Prime adult 086+ 003Old adult 078plusmn 009

Male Subadult 066plusmn 006Prime adult 077plusmn 005Old adult 065plusmn 010

Probability of reproduction Female Subadult 035plusmn 008Prime adult 050plusmn 005Old adult 025plusmn 006

Kittens produced annually Female Subadult 280plusmn 005Prime adult 267plusmn 001Old adult 228plusmn 013

a Overall kitten survival (based on all kittens in our data set including thosethat were not genetically sampled) Estimate of survival of geneticallysampled kittens was 047plusmn 009 this estimate is biased high because manykittens were genetically sampled when they were recaptured alive or re-covered dead at older ages

van de Kerk et al bull Florida Panther Population and Genetic Viability 17

followed by those for non‐F1 admixed (062plusmn 001) and canonical(039plusmn 001) panthers Individual heterozygosity positively affectedsurvival rates of kittens (slope parameter η= 1455 95CI= 0505ndash2405) Individual heterozygosity also positively af-fected survival of subadult and adult panthers but the evidence forthis effect was weaker because the confidence interval for the ha-zard ratio overlapped 10 (hazard ratio exp(ɩ)= 0311 95CI= 0092ndash1054 Table 4 and Fig 7)The MPC increased after genetic introgression (Fig 2) This

abundance index was also included in top‐ranking modelsindicating that population density affected survival (Table 4)Abundance reduced the survival of kittens (slope parameterη= minus0035 95 CI= minus0049 to minus0021) evidence was weak forthe effect of abundance on survival of subadult and adult panthers(hazard ratio exp(ɩ)= 1002 95 CI= 0995ndash1010 Fig 7) Re-gression coefficients associated with the effect of abundance andindividual heterozygosity on demographic parameters are presentedin Table B1 (available online in Supporting Information)The most frequently observed causes of death for radio‐

collared panthers were vehicle collision and intraspecific ag-gression Cause‐specific mortality analyses revealed thatmortality rates among possible causes were generally similar forthe 2 sexes but that males were more likely to die from in-traspecific aggression than were females (Fig 8A) Male mor-tality from intraspecific aggression was higher for subadult andold adult panthers than for prime adults (Fig 8B) For bothmales and females canonical panthers were more likely to diefrom intraspecific aggression than were admixed panthers(Fig 8C)

Reproductive ParametersOf 101 radio‐collared females 67 reproduced at least once duringour monitoring we used these individuals to assess the annualprobability of reproduction The top‐ranked model for the annual

probability of reproduction included age effect only suggesting thatadding ancestry or abundance did not improve model parsimony(Table 4) Model‐averaged annual probabilities of reproductiondiffered between female subadults (035plusmn 008) prime adults(050plusmn 005) and older adults (025plusmn 006)The most parsimonious model for the number of kittens

produced annually by females that produced at least 1 litter (ieconditional on reproduction) included effects of age ancestryand abundance a model that included only the effect of age wasalmost equally supported (Table 4) The model‐averagednumber of kittens produced annually conditional on re-production was 280plusmn 075 for subadults 267plusmn 043 for primeadults and 228plusmn 083 for old adults Confidence intervals forthe slope parameter (β) overlapped 0 for both abundance(β= 0010 95 CI= minus0001 to 0021) and ancestry (β= minus073795 CI= minus1637 to 0162)

Population Dynamics and PersistenceDeterministic and stochastic demographymdashThe deterministic (λ)

and stochastic (λs) annual population growth rate calculated usingthe matrix population model were 106 (5th and 95th percentiles099ndash114) and 104 (072ndash141) respectively The generation timewas 47 years A female starting in age class one (kitten) wasexpected to live another 58 years and produce 10 kittens onaverage during her lifetime Overall λs was proportionately(elasticity) most sensitive to changes in female prime adultsurvival on the absolute scale (sensitivity) however it was mostsensitive to changes in kitten survival (Fig 9) Populationtrajectories probabilities of quasi‐extinction and times untilquasi‐extinction estimated using the matrix population modeland IBM for comparable scenarios were nearly identical for acritical quasi‐extinction threshold of 10 panthers (Figs 10 and 11)and for a critical quasi‐extinction threshold of 30 panthers(Appendix E available online in Supporting Information)Minimum population count scenariomdashUnder the MPC scenario

(ie density dependence estimated using the minimumpopulation count data) with a critical threshold of 10panthers the population size reached 99 (72ndash104) and 100(77ndash105) at 200 years (Fig 10) as predicted by the IBM and thematrix model respectively The cumulative quasi‐extinctionprobabilities within 100 years based on the MPC scenario were15 (IBM 0ndash48) and 30 (matrix model 0ndash76) Theseprobabilities increased to 25 (IBM 0ndash114) and 38 (matrixmodel 0ndash154) by year 200 (Fig 10) The mean times untilquasi‐extinction were 15 years (IBM 0ndash67) and 15 years (matrixmodel 0ndash72) conditioned on going quasi‐extinct within 100years Expected times until quasi‐extinction conditioned onquasi‐extinction within 200 years were higher 34 years (IBM0ndash137) and 32 years (matrix model 0ndash134 Fig 10)Motor vehicle mortality scenariomdashUnder the MVM scenario

(ie density dependence estimated using estimates of pantherpopulation size from McClintock et al 2015) with a criticalthreshold of 10 panthers the projected population reached 188(142ndash217) and 190 (143ndash219) at 200 years as predicted by theIBM and the matrix model respectively (Fig 11) Thecumulative quasi‐extinction probabilities within 100 years were14 (IBM 0ndash08) and 13 (matrix model 0ndash06) Theseprobabilities increased to 20 (IBM 0ndash17) and 19 (matrix

000

025

050

075

100

1995 2000 2005 2010Year of birth

Indi

vidu

al h

eter

ozyg

osity

Figure 6 Temporal changes in the individual heterozygosity of Floridapanthers born during the period 1995ndash2013 in South Florida USA Pointsare measurements solid line is linear regression shaded area is 95 confidenceinterval of slope Slope= 0006plusmn 0001 R2= 009

18 Wildlife Monographs bull 203

model 0ndash16) within year 200 (Fig 11) The mean times until quasi‐extinction based on the MVM scenario were 5 years (IBM 0ndash61)and 6 years (matrix model 0ndash64) conditioned on going quasi‐extinctwithin 100 years Expected times until quasi‐extinction conditionedon quasi‐extinction within 200 years were 11 years (IBM 0ndash113)and 11 years (matrix model 0ndash112 Fig 11)

Sensitivity of extinction probability to demographic parametersmdashThe partial rank correlation coefficients between quasi‐extinctionprobabilities and demographic parameters were negative anincrease in any of the demographic parameters decreased thequasi‐extinction probability Probabilities of quasi‐extinction within200 years were most sensitive to changes in kitten survival for bothscenarios (Fig 12) followed by survival of subadult females in theMVM scenarios and survival of prime adult females in the MPCscenario The probability of reproduction of prime adult femaleshad the third‐largest partial rank correlation coefficient with quasi‐extinction probability for both scenarios

Benefits and Costs of Genetic ManagementMinimum population count scenariomdashThe MPC scenario with

a critical threshold of 10 panthers predicted that withoutmanagement intervention average individual heterozygositywould decrease reaching 057 (049ndash064) at 100 years and053 (042ndash062) at 200 years (Fig 13) The population woulddecrease slightly reaching an average of 79 (41ndash99) at 100 yearsand 76 (43ndash98) at 200 years (Fig 14) The probabilities of quasi‐extinction under the no‐intervention MPC scenario when weconsidered the effects of genetic erosion were 13 (0ndash99) at 100years and 23 (0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with

the translocation of 5 Texas females average individualheterozygosity under the MPC scenario increased and thenstabilized at 068 (064ndash072) at 100 years and remained atthat level at 200 years (Fig 13) The population reached anaverage of 88 (62ndash104) at 100 years and stayed the same at200 years (Fig 14) The probabilities of quasi‐extinctionunder this introgression strategy were 6 (0ndash53) within 100years and 7 (0ndash82) within 200 years (Fig 15) Results ofanalyses using a critical threshold of 30 panthers revealed asimilar pattern of relative benefits However the prob-abilities of quasi‐extinction were substantially higher (ge45)across all scenarios (Appendix E)

Motor vehicle mortality scenariomdashWithout managementintervention the average individual heterozygosity predictedby the MVM scenario and a critical threshold of 10 panthersdecreased to 060 (046ndash069) at 100 years and to 059 (039ndash070) at 200 years (Fig 13) The population increased slightlyand reached an equilibrium at 154 individuals (46ndash217) at 100years and 157 (57ndash218) at 200 years (Fig 14) The probabilitiesof quasi‐extinction were 17 (0ndash100) at 100 years and 22(0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with 5

female pumas from Texas average individual heterozygosityincreased slightly reaching 067 (060ndash073) at 100 years and068 (059ndash075) at 200 years (Fig 13) Under this strategy thepopulation reached an average of 164 individuals (44ndash226) at

Table 4 Model comparison results testing for the effects of several covariateson survival of all kittens survival of genetically sampled kittens survival ofsubadult and adult Florida panthers probability of reproduction and kittensproduced annually for Florida panthers sampled from 1981ndash2013 in SouthFlorida USA

Model Parameters ΔAICa Weight

Kitten survival (all data)Base+ F1b+ abundancec 10 000 0988Base+ abundance 9 1018 0006Base+ F1 9 1035 0005Base 8 2711 lt0001Base+ sexd 9 2912 lt0001Base+ litter sizee 9 2914 lt0001

Kitten survival (genetically sampled kittens only)Base+ F1+ abundance 10 000 0541Base+ F1CanAdmf+ abundance 11 099 0329Base+Hetg+ abundance 10 310 0115Base+CanAdmh+ abundance 10 791 0010Base+ abundance 9 944 0005Base+ F1 9 1638 lt0001Base+ F1CanAdm 10 1837 lt0001Base+Het 9 2872 lt0001Base 8 3296 lt0001Base+CanAdm 9 3348 lt0001

Subadult and adult survivalBase+ F1CanAdm+ abundance 7 000 0360Base+ F1 5 053 0276Base+ F1CanAdm 6 105 0213Base+ F1+ abundance 6 222 0119Base+CanAdm+ abundance 6 575 0020Base+CanAdm 5 908 0004Base+Het 5 964 0003Base+Het+ abundance 6 984 0003Base 4 1118 0001Base+ abundance 5 1278 0001

Probability of reproductionAge 3 000 0242Age+CanAdm 4 012 0228Age+ abundance 4 173 0102Age+ F1 4 190 0094Age+CanAdm+ abundance 5 216 0082Age+Het 4 219 0081Age+ F1CanAdm 5 232 0076Age+ F1+ abundance 5 372 0038Age+Het+ abundance 5 399 0033Age+ F1CanAdm+ abundance 6 444 0026

Kittens produced annuallyAge+ F1+ abundance 5 000 0194Age 3 060 0144Age+ abundance 4 061 0143Age+ F1 4 097 0120Age+CanAdm 4 139 0097Age+ F1CanAdm+ abundance 6 196 0073Age+ F1CanAdm 5 231 0061Age+CanAdm+ abundance 5 238 0059Age+Het+ abundance 5 250 0056Age+Het 4 259 0053

a Akaikersquos information criterion (AIC) for kitten survival models was adjustedfor small sample size and overdispersion (QAICc)

b F1 divides Florida panthers into 2 groups F1 admixed and all other panthersc Index of Florida panther abundanced Sex of an individual Florida panther (male or female)e Size of the litter in which a Florida panther kitten was bornf F1CanAdm divides panthers into 3 groups F1 admixed canonical andother admixed panthers

g Het is the individual heterozygosityh CanAdm divides panthers into 2 groups canonical and admixed (includingF1 admixed) panthers

van de Kerk et al bull Florida Panther Population and Genetic Viability 19

100 years and 170 (67ndash231) at 200 years based on the MVMscenario (Fig 14) The probabilities of quasi‐extinction underthis introgression strategy were 10 (0ndash99) at 100 years and12 (0ndash99) at 200 years (Fig 15)The projected population size and average individual hetero-

zygosity reached equilibrium levels with periodic fluctuations inthese values when introgression was implemented (Figs 16ndash17)Genetic introgression decreased the time until quasi‐extinctionacross all scenarios (Fig 18) Results of analyses using a criticalthreshold of 30 panthers revealed a similar pattern of relative ben-efits However the probabilities of quasi‐extinction were sub-stantially higher (ge45) across all scenarios (Appendix E)

Addition of pumas from Texas increased both the populationsize and average individual heterozygosity consequentlyintrogression caused periodic increases in average individualheterozygosity and population size at the interval at which theintrogression occurred Whereas average individual hetero-zygosity gradually decreased following introgression densitydependence caused the population size to fluctuate especiallywhen a larger number of pumas from Texas were released Thepositive effect of genetic introgression initiatives on quasi‐extinction probabilities decreased as the time interval betweenthese events increased More frequent genetic introgressions ledto greater increases in average individual heterozygosity and

Old adult

Prime adult

Subadult

Kitten

025 050 075

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Individual heterozygosity

Ann

ual s

urvi

val r

ate

Old adult

Prime adult

Subadult

Kitten

40 60 80 100 120

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Abundance index

Ann

ual s

urvi

val r

ate

Kittens Males Females

Figure 7 The effect of individual heterozygosity (left) and the abundance index (right) on Florida panther age‐ and sex‐specific survival estimates (shaded area isSE) in South Florida USA 1981ndash2013 Abundance index data are from McBride et al (2008) and R T McBride and C McBride (Rancherrsquos Supply Incorporatedunpublished report)

20 Wildlife Monographs bull 203

equilibrium population size while reducing the probability ofquasi‐extinction Comparatively performing genetic introgres-sion less often but with more individuals per release was lesseffective at improving the long‐term prospects for the pantherpopulation (Figs 13ndash15)

Financial cost and persistence benefitsmdashThe total estimated costof 1 genetic introgression was $40000 $80000 or $120000for releasing 5 10 or 15 pumas from Texas respectively(Table 5) The total cost incurred over 100 years for eachstrategy ranged from $50000 to $12 million (Table 6)The most expensive strategy involved the introduction of15 pumas every 10 years this strategy reduced the probability

of quasi‐extinction by 58 and 40 under the MPC and MVMscenarios respectively (Table 6) Introducing 5 pumas every80 years cost the least but improvements in populationpersistence under this strategy were insubstantial (Table 6)Releasing more panthers more frequently caused increasedpopulation fluctuations but did not necessarily reduce the quasi‐extinction probability (Figs 14 and 15 Table 6)

DISCUSSION

The duration (34 yr of data) and sample sizes associated with theFlorida Panther Project allowed us to obtain a comprehensiveunderstanding of genetic variation demographic parameters

A

B

C

Figure 8 Cause‐specific mortality rates (plusmnSE) for male and female Florida panthers (A) subadult prime‐adult and old‐adult Florida panthers of both sexes (B)and canonical and admixed Florida panthers of both sexes (C) in South Florida USA 1981ndash2013 Causes of mortality are vehicle collision intraspecific aggression(ISA) other causes (known causes such as diseases injuries and infections unrelated to the first 2 causes) and unknown cause (mortalities for which we could notassign a cause)

van de Kerk et al bull Florida Panther Population and Genetic Viability 21

probability of population persistence and the duration of thepositive impacts of genetic restoration on this endangered po-pulation The Florida panther population was in dire straits inthe early 1990s and our results clearly demonstrated that the

implementation of the introgression project in 1995 subse-quently accrued a series of genetic and demographic improve-ments that have led to the change from a declining population toone that is growing and expanding its current breeding range

Sensitivity Elasticity

0 1 2 3 00 02 04

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

Para

met

er

Figure 9 Sensitivity and elasticity (proportional sensitivity) of stochastic population growth rate (λs) of the Florida panther to vital demographic parameters in SouthFlorida USA 1981ndash2013 Horizontal lines represent 5th and 95th percentiles

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

70

80

90

100

110

0

5

10

15

0

50

100

Years

StatisticMeanMedian

Figure 10 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the minimum population count(MPC) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

125

150

175

200

00

05

10

15

20

0

30

60

90

Years

StatisticMeanMedian

Figure 11 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the motor vehicle mortality(MVM) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

22 Wildlife Monographs bull 203

MPC MVM

minus08 minus06 minus04 minus02 00 minus08 minus06 minus04 minus02 00

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Male subadult survival

Male prime adult survival

Male old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

PRCC

Para

met

er

Figure 12 Partial rank correlation coefficient (PRCC) between quasi‐extinction probabilities and the demographic parameters of the Florida panther for theminimum population count (MPC) and motor vehicle mortality (MVM) scenarios Error bars indicate standard deviation Based on data collected in South FloridaUSA 1981ndash2013

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 13 Average projected individual heterozygosities in the Florida panther population over 200 years without genetic management intervention and with eachof the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include theintroduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years Average individual heterozygosity peaks when pumas are added to the Floridapanther population Based on data collected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 23

Our research has further validated the success of genetic in-trogression by quantifying the reduction in the probability ofextinction of the panther population because of this manage-ment initiative These findings all bode well for continuedprogress toward recovery That said the Florida panther po-pulation remains small and isolated from other populations ofpuma from western North America thereby denoting that theeventual impact of genetic drift and inbreeding may negativelyaffect the population in the future Realizing this will continueto be an issue that managers will have to contemplate wemodeled the impact of varied genetic management scenarios todetermine the frequency of introgression along with the numberof panthers that should be released during each initiative tominimize cost and maximize benefits to the population Theseresults will prove invaluable to managers attempting to conservethe Florida panther

Genetic Variation Demographic Parametersand Persistence of Heterotic Benefits from GeneticIntrogressionOur measure of individual heterozygosity (1 minusHL) was higheron average for the admixed (0615) panthers than for those ofcanonical ancestry (0386) Levels of individual heterozygosityfor admixed panthers were comparable to levels observed in asample of pumas from west Texas (x plusmn SE= 0695plusmn 0019n= 41) which further demonstrates the improvement in levelsof genetic variation in the Florida population since 1995 The

correlation between HL and several demographic parametershas been noted in wild populations including cheetahs (Terrellet al 2016) and several species of birds such as European shags(Phalacrocorax aristotelis male and female survival probabilityVelando et al 2015) and Egyptian vulture (Neophron percnop-terus age at recruitment Agudo et al 2012) If we focus only onpanthers captured and radiocollared from 2012ndash2015 (FP211ndashFP240) all of which had estimated birth years ge2005 (ge10 yrpost‐introgression) those panthers had an average individualheterozygosity level of 0618plusmn 0029 highlighting the elevatedlevels of genetic variation that continue to persist in cohorts ofpanthers 5 generations after the release of female pumas fromTexasThe proportional membership values (q) from our

STRUCTURE analysis allowed us to delineate 2 clusters ofancestry for Florida panthers (Fig 4) During the pre‐in-trogression era the population was dominated by panthers thatretained a high percentage of the canonical ancestry which wasaffected by inbreeding depression The post‐introgression eraancestry values are comprised of many admixed individuals inessence demonstrating the success of the introgression in termsof increasing genetic variation in the population and mi-micking gene flow that historically occurred with panthers andother populations of pumas (Onorato et al 2010) A somewhatanalogous situation although abetted by natural immigrationwas evident in the ancestral variation in a small population ofpuma intersected by a major freeway in California USA In

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

75

80

85

90

95

100

130

150

170

190

Years

Popu

latio

n si

ze

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 14 Average projected population sizes in the Florida panther population over 200 years without intervention and with each of the genetic introgressionstrategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include the introduction of 5 10 or 15pumas from Texas USA every 10 20 40 or 80 years Peaks occur when pumas are added to the Florida panther population Based on data collected in SouthFlorida USA 1981ndash2013

24 Wildlife Monographs bull 203

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

25

50

75

100

0

25

50

75

100

Introgression interval (yrs)

PQ

E (

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 15 Average probability of quasi‐extinction (PQE) of the Florida panther population within 100 years and within 200 years without genetic managementintervention (N) and with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenariosIntrogression strategies include the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Errorbars indicate 5th and 95th percentiles Based on data collected in South Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

50

60

70

80

90

100

110

50

100

150

200

Years

Popu

latio

n si

ze

Figure 16 Average projected Florida panther population trajectories under a geneticintrogression strategy involving the release of 5 female pumas from Texas USA every20 years for the minimum population count (MPC) and motor vehicle mortality(MVM) scenarios Shaded area indicates 5th and 95th percentiles and isrepresentative of confidence intervals for other scenarios Based on data collected inSouth Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Figure 17 Average projected Florida panther population‐level heterozygositiesunder a genetic introgression strategy involving the release of 5 female pumas fromTexas USA every 20 years for the minimum population count (MPC) and motorvehicle mortality (MVM) scenarios Shaded area bounds the 5th and 95th percentilesand is representative of confidence intervals for other scenarios Based on datacollected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 25

this case the migration of just a single male across the freewayto the south resulted in an increase in the number of in-dividuals with admixed ancestry and substantially improvedgenetic diversity of the subpopulation south of the freeway(Riley et al 2014)Our estimate of overall kitten survival (0321plusmn 0056) was

similar to that reported by Hostetler et al (2010) who used13 years of post‐introgression data (0323plusmn 0071) These valuesare lower than kitten survival estimates derived from severalpuma populations in western North America (044ndash091 Loganand Sweanor 2001 Lambert et al 2006 Laundre et al 2007Robinson et al 2008 Ruth et al 2011) When we included onlydata for individuals that had been genetically sampled our es-timate was 15 higher The proportion of admixed kittens thatwere genetically sampled increased as the population expandedwhich could have resulted in higher estimates of kitten survivalbecause admixed kittens survive better than canonical kittensKitten survival was strongly influenced by genetic ancestry withsurvival rates of F1 kittens being almost twice that of canonicalkittens (Fig 5B) Consistent with the findings of Hostetler et al(2010) increases in heterozygosity coincided with increasedkitten survival whereas population density negatively affectedkitten survival Population density often has a strong negativeimpact on survival of young age classes due to infanticide andcannibalism which has been reported from several carnivore

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

50

100

150

0

50

100

150

Introgression interval (yrs)

TQE

(yrs

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 18 Average times until quasi‐extinction (TQE) for the Florida panther population within 100 years and within 200 years without genetic management interventionand with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategiesinclude the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Error bars indicate 5th and 95thpercentiles Based on data collected in South Florida USA 1981ndash2013

Table 5 Average cost (2017 US dollars) of introgression strategies employedover 100 years that involve the introduction of 5 10 or 15 pumas from TexasUSA into the Florida panther population in South Florida USA at an in-creasingly higher frequency Costs of capture (including transportation) quar-antine and post‐introgression monitoring were estimated by Roy McBrideMark Cunningham and Dave Onorato respectively

Frequency Component 5 pumas 10 pumas 15 pumas

Once Capture 25000 50000 75000

Quarantine 5000 10000 15000

Monitoring 10000 20000 30000

Total 40000 80000 120000

Every 80 years Capture 31250 62500 93750

Quarantine 6250 12500 18750

Monitoring 12500 25000 37500

Total 50000 100000 150000

Every 40 years Capture 62500 125000 187500

Quarantine 12500 25000 37500

Monitoring 25000 50000 75000

Total 100000 200000 300000

Every 20 years Capture 125000 250000 375000

Quarantine 25000 50000 75000

Monitoring 50000 100000 150000

Total 200000 400000 600000

Every 10 years Capture 250000 500000 750000

Quarantine 50000 100000 150000

Monitoring 100000 200000 300000

Total 400000 800000 1200000

26 Wildlife Monographs bull 203

species including polar bears (Ursus maritimus Derocher andWiig 1999) black bears (Ursus americanus Garrison et al 2007)and pumas (Logan and Sweanor 2001)Our estimates of sex‐ and age‐specific survival rates of subadult

and adult panthers (Table 3) and the effects of covariates on theserates were similar to those reported by Benson et al (2011) derivedfrom data collected through 2006 Our survival rates for males(065ndash077) were lower than females (078ndash097) for all 3 ageclasses (subadult prime adult and old adult) which is the typicaltrend observed in most puma populations (eg Ruth et al 19982011) However an unhunted puma population occupying afragmented urban landscape in southern California exhibited lowersurvival (0586 females 0525 males Vickers et al 2015) perhapsas a result of severe habitat fragmentation and the lack of extensiveswaths of public land protected in perpetuity Most populations ofpuma in western North America are typically subject to variedlevels of management via regulated hunting which often is biasedtoward the take of males (Cooley et al 2009 Ruth et al 2011)Consequently annual survival estimates from hunted populationsin northeast Washington (0679ndash0733 females 0341 malesRobinson et al 2008) and southeastern Arizona (0685 females0584 males Cunningham et al 2001) were generally lower thanthose we estimated for Florida panthersCause‐specific mortality analysis showed that male panthers

experienced a substantially greater mortality risk from in-traspecific aggression This differs from the pattern observedduring a long‐term study in Arizona where females were at ahigher risk of intraspecific mortality than males (46 of maleand 53 of female mortality Logan and Sweanor 2001)Mortality associated with intraspecific strife has been docu-mented in hunted and unhunted populations (this study Loganand Sweanor 2001 Stoner et al 2010) and its root cause hasbeen difficult to determine (eg population density and preypopulation trends) That being the case finding ways to reducewhat is often a major source of mortality for puma populationscontinues to pose a challenge to managersCanonical panthers were substantially more likely to die from

intraspecific aggression than were admixed panthers This mayat least partly explain the difference in survival between admixed

and canonical panthers Canonical panthers are known to bemore likely to suffer from varied correlates of inbreeding de-pression than admixed animals including atrial septal defects(Johnson et al 2010) and compromised immune systems(Roelke et al 1993) These factors have the potential to affectfitness and perhaps dominance of individuals when engaged inan intraspecific encounter A somewhat similar scenario wasobserved by Hogg et al (2006) in a population of bighorn sheepthat were part of an introgression project Admixed males in thispopulation had a greater probability of paternity than males thatwere less outbred This included coursing males with higherlevels of admixture being markedly more successful at fightingmales that were defending females in estrous (Hogg et al 2006)Heterosis resulting from genetic introgression is expected to be

the strongest in F1 individuals (Shull 1908 Crow 1948 Burkeand Arnold 2001 Waller 2015) and to progressively decline insubsequent generations (Pickup et al 2013 Frankham 2015)Given that admixed (especially F1) panthers survived better thancanonical panthers and that individual heterozygosity improvedsurvival of both sexes and all age classes our data provide evi-dence for heterotic advantage whereby individuals of mixedancestry had higher fitness (Shull 1908 Crow 1948) Our resultsare consistent with findings of other studies that involved in-tentional genetic introgression for conservation purposes Forexample Hogg et al (2006) detected substantial improvementsin survival and reproduction of introgressed bighorn sheep andMadsen et al (1999 2004) reported a dramatic increase in thepopulation of adders after genetic introgressionKnowledge of the persistence of benefits to a population ac-

crued via genetic introgression is essential for determiningwhen additional introgression may be warranted There is adepauperate amount of information regarding this topic and theidentification of factors that may influence persistence of thebenefits of genetic introgression (Tallmon et al 2004 Hedricket al 2014 Whiteley et al 2015) For example in minks(Neovison vison) Thirstrup et al (2014) found that outcrossingled to larger litters in F2 and F3 but this benefit disappeared insubsequent generations In a population of gray wolves Hedricket al (2014) suggested that benefits of genetic introgression may

Table 6 Costs and benefits accumulated over 100 years for genetic introgression at different intervals and with different numbers of pumas from Texas USAintroduced into the Florida panther population in South Florida USA Costs (2017 US dollars) include capture and transportation quarantine and post‐introgression monitoring Benefits are represented as average probability of quasi‐extinction (PQE and 5th and 95th percentiles) and changes (Δ) therein under thescenario using the minimum population count (McBride et al 2008 MPC) and the scenario using the motor vehicle mortality population estimates (McClintocket al 2015 MVM) The table is sorted by percent change in probability of quasi‐extinction under the MPC scenario

Interval (yr) Pumas Cost MPC PQE () ΔMPC PQE () MVM PQE () ΔMVM PQE ()

ndash 0 0 13 (0ndash99) 0 17 (0ndash100) 080 10 100000 12 (0ndash99) 10 (0ndash58) 16 (0ndash100) 5 (0ndash38)80 15 150000 12 (0ndash99) 13 (0ndash65) 15 (0ndash100) 8 (0ndash56)80 5 50000 12 (0ndash99) 14 (0ndash73) 15 (0ndash99) 9 (0ndash41)40 15 300000 10 (0ndash98) 26 (0ndash104) 14 (0ndash99) 13 (0ndash60)40 10 200000 9 (0ndash94) 35 (0ndash183) 13 (0ndash99) 21 (0ndash95)40 5 100000 8 (0ndash89) 39 (0ndash211) 12 (0ndash99) 26 (0ndash124)20 15 600000 8 (0ndash88) 42 (0ndash257) 13 (0ndash99) 24 (0ndash123)20 10 400000 6 (0ndash67) 54 (0ndash318) 10 (0ndash99) 37 (0ndash217)10 15 1200000 6 (0ndash53) 58 (0ndash372) 10 (0ndash99) 40 (0ndash208)20 5 200000 5 (0ndash50) 59 (0ndash338) 9 (0ndash99) 44 (0ndash352)10 10 800000 4 (0ndash16) 69 (0ndash489) 8 (0ndash92) 55 (0ndash401)10 5 400000 4 (0ndash7) 73 (0ndash502) 6 (0ndash71) 63 (0ndash503)

van de Kerk et al bull Florida Panther Population and Genetic Viability 27

wane after 2 or 3 generations The WrightndashFisher model ofgenetic drift predicts a geometric decline in expected hetero-zygosity at the rate of (1 minus 12Ne) per generation where Ne is theeffective population size (Hartl and Clark 1997 Frankham et al2014) Thus it seems logical to assume that the duration of thebenefits of introgression is limited especially in a species per-sisting in a small population such as Florida panthersWe expected Florida panther survival in the most recent years

(2005ndash2013) post‐introgression to differ from that observed in thedecade following the introgression in 1995 because pantherabundance and heterozygosity factors known to influence panthersurvival (Hostetler et al 2010 Benson et al 2011) have changed(Figs 2 and 6) We found that subadult and adult survival rates inrecent years (2005ndash2013) were similar to those in the period im-mediately following introgression (1995ndash2004 Fig 5C) overallkitten survival was similar to estimates of Hostetler et al (2010) Infact the pattern of covariate effects on Florida panther survivalreported by Benson et al (2011) and Hostetler et al (2010) hasbarely changed The negative effects of population density andpositive effects of heterozygosity may counterbalance survival ratesreducing population fluctuations Our result that benefits of geneticrestoration have remained in the population nearly for 5 genera-tions post‐intervention was surprising but also encouraging Pos-sible explanations for this observation may include 1) genetic ero-sion occurs at slower rate than previously thought 2) introducing 5migrants in 1 genetic intervention may have beneficial effects si-milar to those from 1 migrant per generation over multiple gen-erations or 3) other and as yet unknown factors may reduce therate of genetic erosion and subsequent demographic effects Adensity‐dependent effect on kitten survival could also explain whynon‐F1 admixed kittens did not survive substantially better thandid canonical kittens most kittens sampled from 2005ndash2013 wereadmixed and their survival was lower possibly because of a density‐dependent effect as the Florida panther population continuouslygrew during that period Trinkel et al (2010) also found thatpopulation density and inbreeding coefficient interacted to affectdemographic parameters and growth rate of an African lion po-pulation Likewise population density interacted with winterweather to affect the survival and population growth rates in Soaysheep (Ovis aries Milner et al 1999 Coulson et al 2001)

Population Dynamics and PersistenceThe finite deterministic and stochastic growth rates were gt10reflecting that much of the data used in this study were collectedfollowing genetic introgression in 1995 during which time thepopulation increased substantially Because our demographicparameter estimates did not change markedly in comparison tothose of Hostetler et al (2013) the similarity between thepopulation growth estimates from these studies was expectedBut these estimates reflect population growth during thedata‐collection period and do not predict future growth In factestimates of population size reported by McClintock et al(2015) suggest that population growth may already be slowing(Fig 2) A similar pattern was observed in an introduced po-pulation of African lion which increased steadily from 1995until 2001 then fluctuated around the presumed carrying ca-pacity thereafter (Trinkel et al 2010) Several other African lionpopulations that experienced various degrees of recovery

subsequently became relatively stable or exhibited decliningtrends (Bauer et al 2015)Our estimates of quasi‐extinction probabilities were lower than

those reported by Hostetler et al (2013) using a 2‐sex age‐structured matrix population model Lower probabilities ofquasi‐extinction than those reported by the earlier study forcomparable scenarios were likely a consequence of the fact thatthe estimates of abundance (McBride et al 2008) and thus thedensity‐dependent effects on survival of kittens and old panthers(gt10 yr) have changed in recent years Additionally these earlyestimates of the probability of quasi‐extinction and of time toquasi‐extinction did not consider genetic erosion that will in-evitably occur without management interventionUnder the MVM scenario the equilibrium population size was

twice that attained under the MPC scenario The MVM popula-tion estimates are approximately twice the MPCs (Fig 2) as aresult the simulated population under the MVM scenario achieveda higher equilibrium size Probability of quasi‐extinction under theMVM scenario was generally greater than that under the MPCscenario This result is a consequence of the fact that the estimateddensity dependence on kitten survival based on the MPC scenario isstronger than that for theMVM scenario (regression coefficients forabundance for the MPC scenario= minus0068 vs minus0002 for theMVM scenario Appendix B Table B1) Consequently simulatedpopulations under the MPC scenario have a stronger tendency tomove toward the equilibrium population size which inherentlyreduces the quasi‐extinction probability (eg Royama 1992)As expected for long‐lived mammals (Heppell et al 2000 Oli

and Dobson 2003) the population growth rate was proportio-nately most sensitive to changes in survival of prime adultsfollowed by that in younger age classes Global sensitivity ana-lysis in contrast showed that under all scenarios extinctionparameters were particularly sensitive to changes in survival ofFlorida panther kittens This result is a consequence of the highsensitivity of the population growth rate to kitten survival(Fig 9) and a larger standard error of kitten survival (Table 3)Results of our sensitivity analyses indicate that future researchshould focus on acquiring additional information on early lifestages to improve the accuracy and precision of estimates ofkitten survival Our results suggest that demographic variables(or their environmental drivers) that strongly influence extinc-tion risks are not necessarily those with the largest elasticityvalues A study of island fox (Bakker et al 2009) found thatextinction risk was strongly influenced by survival modifierparameters that had low elasticity values

Genetic Management When and How to InterveneThe use of genetic introgression as a management tool has al-ways been controversial and the probable effectiveness it mighthave in preventing the imminent demise of the panther popu-lation was initially debated (Creel 2006 Maehr et al 2006Pimm et al 2006) These debates notwithstanding the pantherpopulation had suffered from genetic morphological and bio-medical correlates of inbreeding before this management in-tervention (Johnson et al 2010 Onorato et al 2010) whereasthe post‐introgression period has been characterized by a sub-stantial reduction in the biomedical correlates of inbreeding andimprovements in demographic vigor and abundance (McBride

28 Wildlife Monographs bull 203

et al 2008 Hostetler et al 2010 2013 Johnson et al 2010Benson et al 2011 McClintock et al 2015) Nevertheless thepopulation remains small and isolated habitat is still being lostand there is no current possibility of natural gene flow betweenthis and other North American puma populations thus therecurrence of inbreeding‐related problems and subsequent po-pulation declines are inevitable The question therefore is notwhether genetic management of the Florida panther populationis needed but when and how it should be implementedWithout genetic introgression probabilities of quasi‐extinc-

tion were substantially greater than those from models thatincorporated periodic genetic introgression events (Fig 15)highlighting the importance of genetic management in smallisolated populations When 5 female pumas are added to theFlorida panther population every 10 years genetic variationincreased substantially under the MPC and MVM scenariosThe IBM predicted that the equillibrium population size wouldbe substantially larger when 5 pumas were released into thepopulation every 10 years than populations without intervention(ie no introgression) Releasing 15 Texas females did not af-ford substantially greater benefit to the equilibrium populationsize than did releasing only 5 Texas females Instead addingmore individuals caused increasingly large fluctuations in po-pulation size due to the numerical increases and density de-pendence (Fig 14) possibly diminishing the benefits associatedwith increased genetic variationCompared with the no‐intervention scenario (ie no genetic

introgression implemented) the introduction of 5 female pumasevery 10 years reduced quasi‐extinction probabilities from 13to 4 and from 17 to 6 for the MPC and MVM scenariosrespectively The benefit of genetic‐introgression strategies de-creased as the interval between successive management inter-ventions increased and no benefit was evident once the intervalreached 80 years (Fig 15) Introgression reduced the averagetime until quasi‐extinction (Fig 18) This somewhat counter‐intuitive result can be explained by the fact that repeated geneticintrogression makes the population more robust over timewhich will reduce the probability of extinction Consequentlyfew simulated population trajectories that fall below the criticalthreshold do so early reducing the mean time to extinctionAlso density dependence has a stabilizing effect on populations(Royama 1992) populations tend toward equilibrium popula-tion sizes which reduces the probability over time of popula-tions falling below quasi‐extinction threshold However timeuntil quasi‐extinction must be interpreted in conjunction withthe probability of quasi‐extinction which is very low for allscenarios with genetic introgression (Fig 15)Cost is a significant impediment to the implementation of a

genetic introgression program because captures relocations andmonitoring efforts before and after introgression are expensive(Edmands 2007 Frankham 2010b 2015 2016) Costs may beacceptable to society if the management actions result in sub-stantial fitness benefits especially if the species of concern ispopular and charismatic (Frankham 2015) We estimated the100‐year cost of introgression strategies modeled here to rangefrom $50000 to $12 million More expensive strategies gen-erally led to larger decreases in the probability of quasi‐extinc-tion but cheaper strategies also substantially improved

population viability (ie reduced quasi‐extinction probabilityTable 6) One of the cheaper strategies (5 pumas released every20 years) which cost an estimated $200000 over 100 yearsreduced the probability of quasi‐extinction by 59 and 44 forthe MPC and MVM scenarios respectively But these pre-liminary cost estimates are based on expenses incurred duringthe 1995 introgression and include costs of capture quarantinetransportation release and post‐release monitoring of femalesonly Although the cost for future genetic interventions wouldlikely be higher than those reported here the relative differencesin cost between alternative intervention strategies should remainapproximately the sameOur ability to simulate genetics and link it to the viability of

the Florida panther population is based on the empirically es-timated relationship between individual heterozygosity andsurvival Such heterozygosity and fitness correlations have beenstudied for decades (David 1998) but have only recently beenused to predict population viability (Benson et al 2016) noneto our knowledge have incorporated cost into their modelingefforts The mechanisms underlying heterozygosity and fitnesscorrelations remain unclear (David 1998 Szulkin et al 2010)and their significance as a proxy for inbreeding depression hasbeen debated (Balloux et al 2004 Miller and Coltman 2014)Nevertheless evidence continues to mount demonstratingpositive relationships between heterozygosity and survival(Coulson et al 1998ab Hansson et al 2004 Silva et al 2005Acevedo‐Whitehouse et al 2006 Jensen et al 2007 Velandoet al 2015) and between heterozygosity and fecundity (Seddonet al 2004 Charpentier et al 2005 Agudo et al 2012 Velandoet al 2015) Although the reported correlations are generallyweak their consequences can be substantial if population growthand persistence parameters are highly sensitive to the affectedvital rates (Velando et al 2015) Compared with scenarios thatignore genetics incorporating the effects of inbreeding on sur-vival increased quasi‐extinction probabilities within a 100‐yeartimeframe from 15 to 13 and from 14 to 17 for theMPC and MVM scenarios respectively These results high-light the importance of incorporating genetics when analyzingthe viability of small isolated populationsAlthough conservation biologists have recognized the im-

portance of genetics in population viability many still fail toincorporate genetic factors into PVA models (Reed et al 2002)Explicit consideration of genetics in PVAs requires long‐termgenetic and demographic data This permits the assessment ofthe functional relationship between genetic variability and de-mographic parameters Population viability analysis softwarepackages such as VORTEX (Lacy 1993 2000) offer a powerfulframework for individual‐based simulations but they often donot adequately capture a speciesrsquo life history or genetics Basedon a thorough review of the importance of genetic factors inwildlife conservation Frankham et al (2014) concluded thatgenetic factors can strongly influence the dynamics and persis-tence of small andor fragmented populations They furthernoted that ldquoMost PVA‐based risk assessments ignore or in-adequately model genetic factorsrdquo and recommended that ldquoPVAshould routinely include realistic inbreeding depressionhelliprdquo(Frankham et al 201457) Our custom‐built IBM permitted usto develop a model that adequately captured the Florida panther

van de Kerk et al bull Florida Panther Population and Genetic Viability 29

life history and we could parameterize the model with our long‐term genetic and demographic dataThe explicit consideration of the effects of inbreeding on

Florida panther population dynamics and persistence representsan important step forward Nonetheless our model remainsimperfect First we neglected the possible effects of habitat lossdetrimental anthropogenic activities climate change the prob-ability of immigration of pumas from western populationsdisease or catastrophes on demographic rates and populationviability Although immigration from puma populations outsideFlorida is possible its probability is very low given the extensivechallenges the anthropogenically altered landscape would posefor such a migrant to make it successfully to South Florida Weomitted mutations from our model because we have no in-formation on mutation rates though we expect that they are lowbased on values reported for other mammals (Kumar and Sub-ramanian 2002) Finally we only considered possible effects ofinbreeding on age‐specific survival rates because there was noevidence that heterozygosity affected female reproduction Lossof heterozygosity can negatively affect reproduction becauseinbred male panthers can suffer from cryptorchidism which canadversely affect their fecundity However given the polygynousmating system we expect the Florida panther population growthis primarily female‐limitedAlthough it is generally accepted that genetic introgression

only temporarily relieves a population from inbreeding depres-sion we know of no cases in which it has been implementedmore than once to conserve a threatened wildlife populationOur study provides an example of how demographic and mo-lecular data can be used within an IBM framework to evaluatethe efficacy of alternative genetic management strategies via theFlorida panther as a case study Our model can be adjusted forand applied to other species and offers great potential as a toolfor genetic management of small isolated wildlife populations

MANAGEMENT IMPLICATIONS

Our results and those of Hostetler et al (2013) and Johnsonet al (2010) provide persuasive evidence that the genetic in-tervention implemented in 1995 prevented the demise of theFlorida panther and restored demographic vigor to the popu-lation The Florida panther population remains small and iso-lated from other puma populations the closest puma populationis located gt1000 km away in Texas and the intervening land-scape is highly developed and fragmented with many interstate(and other high‐traffic) highways and major urban centersTheory suggests that 1 migrant per generation is sufficient tominimize the loss of genetic diversity (eg Mills and Allendorf1996) However the possibility of successful migration of apuma to South Florida followed by successful mating every ~5years is very low considering the distance between Texas andFlorida (Hedrick 1995) and the many risks and barriers thatdispersing pumas face (Beier 1995 Maehr et al 2002 Thatcheret al 2006) Consequently the pantherrsquos long‐term persistencewill likely depend on subsequent timely genetic introgressionsgiven the near‐impossibility of natural gene flow from otherpuma populations To that end our study offers considerableinsights into benefits of different genetic management strategiesand associated costs

Without genetic management the Florida panther populationfaces a substantial risk of quasi‐extinction within 100 years (10ndash46 depending on quasi‐extinction threshold and scenarios)Twenty‐three years have passed since genetic introgression wasimplemented and the population appears healthy as indicatedby measures of genetic variation increases in the populationsize and improvements in age‐specific survival rates Ourfindings suggest that releasing more pumas more frequently doesnot necessarily improve persistence probability because such astrategy would cause larger fluctuations in population size(Fig 14) which negatively affects persistence probability Thusextinction risks faced by inbred populations of large carnivorescan be substantially reduced by releasing approximately 5 im-migrants every 2 decades We suggest that such a managementstrategy be followed only in conjunction with continued long‐term monitoring of the population Our analyses demonstratethat population growth and persistence are highly sensitive tokitten and female (adult and subadult) survival Continuing tocollect data on these demographic variables along with bio-metric and genetic sampling will remain important to pantherrecovery and vigilance in identifying issues associated with in-breeding depression The collection of these monitoring datashould allow managers to detect any demographic or geneticchanges in a timely fashion and respond with genetic in-trogression or other management actions if warrantedRecovery objectives as defined by the USFWS in its current

recovery plan (USFWS 2008) delineate the need for additionalpopulations in the historical range to downlist or delist theFlorida panther If the only extant population of Floridapanthers continued to grow it could serve as a source ofpanthers to be translocated to suitable habitat to seed addi-tional populations Maintaining current information on thegenetic health of the population and precise estimates of itssize will be important in assessing whether it can withstand theremoval of a subset of animals for translocation Although the2017 documentation of a female panther with kittens north ofthe current breeding range is encouraging in terms of thepotential for population expansionmdashgiven that no female hadbeen recorded north of the Caloosahatchee River since 1972mdashthe re‐establishment of separate new populations will mostlikely require translocations by wildlife managers and a lengthysociopolitical processConservation of threatened or endangered species such as the

Florida panther is inherently challenging For panthers and otherlarge carnivores that require vast extents of natural habitat topersist habitat is typically the most limiting factor that will de-termine whether recovery efforts succeed Although geneticmanagement invariably plays a key role in the long‐term persis-tence of the panther population even a healthy population caneventually succumb to the impacts of habitat loss The humanpopulation of Florida continues to be one of the fastest‐growingin the nation the United States Census Bureau currently ranksFlorida as the third most populous Therefore habitat loss isgoing to continue to be a key issue for panthers Diligence towardpreserving panther habitat in its historical range via conservationeasements or other means and trying to keep such areas inter-connected via corridors is a goal stakeholders such as governmentagencies sportsmen non‐governmental organizations ranchers

30 Wildlife Monographs bull 203

and other private landowners must work on collaboratively toachieve

ACKNOWLEDGMENTS

We thank D Shindle D Jennings T OrsquoMeara D Land andC Belden for valuable advice throughout the project We thankS Bass M Criffield M Cunningham D Giardina D JansenA Johnson D Land M Lotz C McBride Rocky McBrideRowdy McBride Roy McBride L Oberhofer S Schulze staffsat Everglades National Park and Big Cypress National Preserveand others for assistance with fieldwork We also thank JAustin M Conroy J Hines J Laake S Mills J Nichols JHines M Sunquist J Fryxell and T Coulson for advice andassistance regarding data analysis and panther biology TheMuseum of Texas Tech University Natural Science ResearchLaboratory provided samples from pumas from west Texas forcomparative genetic analyses M Ben‐David D Land WMcCown E Hellgren and 4 anonymous reviewers providedmany helpful comments on the manuscript We thank NereaUbierna Lopez for completing the Spanish translation of theabstract We are grateful to the Department of ResearchComputing University of Florida for excellent high‐perfor-mance computing services We would like to thank US Fishand Wildlife Service Florida Fish and Wildlife ConservationCommission (FWC) US National Park Service QuantitativeSpatial Ecology Evolution and Environment (QSE3) In-tegrative Graduate Education and Research Traineeship(IGERT) program funded by the National Science Foundationand the University of Florida for financially supporting thisstudy Funding for panther research conducted by the FWC issupported predominantly via donations accrued with vehicleregistrations for ldquoProtect the Pantherrdquo license plates

LITERATURE CITEDAcevedo‐Whitehouse K T R Spraker E Lyons S R Melin F Gulland RL Delong and W Amos 2006 Contrasting effects of heterozygosity onsurvival and hookworm resistance in California sea lion pups MolecularEcology 151973ndash1982

Adams J R L M Vucetich P W Hedrick R O Peterson and J AVucetich 2011 Genomic sweep and potential genetic rescue during limitingenvironmental conditions in an isolated wolf population Proceedings of theRoyal Society B Biological Sciences 2783336ndash3344

Agresti A 2013 Categorical data analysis Third edition John Wiley amp SonsHoboken New Jersey USA

Agudo R M Carrete M Alcaide C Rico F Hiraldo and J A Donaacutezar2012 Genetic diversity at neutral and adaptive loci determines individualfitness in a long‐lived territorial bird Proceedings of the Royal Society BBiological Sciences 2793241ndash3249

Albeke S E N P Nibbelink and M Ben‐David 2015 Modeling behavior bycoastal river otter (Lontra canadensis) in response to prey availability in PrinceWilliam Sound Alaska a spatially‐explicit individual‐based approach PLOSOne 10e0126208

Alho J S K Vaumllimaumlki and J Merilauml 2010 Rhh an R extension for estimatingmultilocus heterozygosity and heterozygosity‐heterozygosity correlationMolecular Ecology Resources 10720ndash722

Alvarez K 1993 Twilight of the panther Myakka River Publishing SarasotaFlorida USA

Aparicio J M J Ortego and P J Cordero 2006 What should we weigh toestimate heterozygosity alleles or loci Molecular Ecology 154659ndash4665

Bakker V J D F Doak G W Roemer D K Garcelon T J Coonan S AMorrison C Lynch K Ralls and R Shaw 2009 Incorporating ecologicaldrivers and uncertainty into a demographic population viability analysis for theisland fox Ecological Monographs 7977ndash108

Balloux F W Amos and T Coulson 2004 Does heterozygosity estimateinbreeding in real populations Molecular Ecology 133021ndash3031

Barone M A M E Roelke J Howard J L Brown A E Anderson and DE Wildt 1994 Reproductive characteristics of male Florida panthersmdashcomparative studies from Florida Texas Colorado Latin‐America andNorth‐American Zoos Journal of Mammalogy 75150ndash162

Bateson Z W P O Dunn S D Hull A E Henschen J A Johnson and LA Whittingham 2014 Genetic restoration of a threatened population ofgreater prairie‐chickens Biological Conservation 17412ndash19

Bauer H G Chapron K Nowell P Henschel P Funston L T B HunterD W Macdonald and C Packer 2015 Lion (Panthera leo) populations aredeclining rapidly across Africa except in intensively managed areas Pro-ceedings of National Academy of Sciences of the United States of America11214894ndash14899

Beier P 1995 Dispersal of juvenile cougars in fragmented habitat Journal ofWildlife Management 59228ndash237

Benson J F J A Hostetler D P Onorato W E Johnson M E Roelke SJ OrsquoBrien D Jansen and M K Oli 2011 Intentional genetic introgressioninfluences survival of adults and subadults in a small inbred felid populationJournal of Animal Ecology 80958ndash967

Benson J F P J Mahoney J A Sikich L E K Serieys J P Pollinger H BErnest and S P D Riley 2016 Interactions between demography geneticsand landscape connectivity increase extinction probability for a small popu-lation of large carnivores in a major metropolitan area Proceedings of theRoyal Society B Biological Sciences 28320160957

Beverly F 1874 The Florida panther Forest and Stream 3290Boyce M S C V Haridas C T Lee and the NCEAS Stochastic Demo-graphy Working Group 2006 Demography in an increasingly variable worldTrends in Ecology amp Evolution 21141ndash148

Brook B W J J OrsquoGrady A P Chapman M A Burgman H R Akcakayaand R Frankham 2000 Predictive accuracy of population viability analysis inconservation biology Nature 404385ndash387

Brys R and H Jacquemyn 2016 Severe outbreeding and inbreeding de-pression maintain mating system differentiation in Epipactis (Orchidaceae)Journal of Evolutionary Biology 29352ndash359

Burke J M and M L Arnold 2001 Genetics and the fitness of hybridsAnnual Review of Genetics 3531ndash52

Burnham K P 1993 A theory for combined analysis of ring recovery andrecapture data Pages 199ndash213 in J‐D Lebreton and P M North editorsMarked individuals in the study of bird population Springer‐Verlag NewYork New York USA

Burnham K P and D R Anderson 2002 Model selection and multimodelinference a practical information‐theoretic approach Second edition SpringerVerlag New York New York USA

Caswell H 2001 Matrix population models construction analysis and in-terpretation Sinauer Associates Sunderland Massachusetts USA

Charpentier M J M Setchell F Prugnolle L A Knapp E J Wickings PPeignot and M Hossaert‐McKey 2005 Genetic diversity and reproductivesuccess in mandrills (Mandrillus sphinx) Proceedings of the NationalAcademy of Sciences of the United States of America 10216723ndash16728

Cooch E and G C White editors 2018 Program MARK a gentle in-troduction lthttpwwwphidotorgsoftwaremarkdocsbookgt Accessed 7Apr 2016

Cooley H S R B Wielgus G M Koehler H S Robinson and B TMaletzke 2009 Does hunting regulate cougar populations A test of thecompensatory mortality hypothesis Ecology 902913ndash2921

Coulson T E A Catchpole S D Albon B J Morgan J M Pemberton TH Clutton‐Brock M J Crawley and B T Grenfell 2001 Age sex densitywinter weather and population crashes in Soay sheep Science 2921528ndash1531

Coulson T J Pemberton S Albon M Beaumont T Marshall J Slate FGuinness and T Clutton‐Brock 1998a Microsatellites reveal heterosis in reddeer Proceedings of the Royal Society B Biological Sciences 265489ndash495

Coulson T N S D Albon J M Pemberton J Slate F E Guinness and TH Clutton‐Brock 1998b Genotype by environment interactions in wintersurvival in red deer Journal of Animal Ecology 67434ndash445

Cox D 1972 Regression models and life‐tables Journal of the Royal StatisticalSociety Series B Statistical Methodology 34187ndash220

Creel S 2006 Recovery of the Florida panthermdashgenetic rescue demographic rescueor both Response to Pimm et al (2006) Animal Conservation 9125ndash126

Crnokrak P and D A Roff 1999 Inbreeding depression in the wild Heredity83260ndash270

Crow J 1948 Alternative hypotheses of hybrid vigor Genetics 33477ndash487

van de Kerk et al bull Florida Panther Population and Genetic Viability 31

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 2: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

is not whether genetic management of the Florida panther population is needed but when and how it should beimplemented Thus we evaluated genetic and population consequences of alternative genetic introgressionstrategies to identify optimal management actions using individual‐based simulation models Releasing 5 pumasevery 20 years would cost much less ($200000 over 100 years) than releasing 15 pumas every 10 years($1200000 over 100 years) yet would reduce the risk of quasi‐extinction by comparable amount (44ndash59 vs40ndash58) Generally releasing more females per introgression attempt provided little added benefit The positiveeffects of the genetic introgression project persist in the panther population after 20 years We suggest thatmanagers contemplate repeating genetic introgression by releasing 5ndash10 individuals from other puma popula-tions every 20ndash40 years We also recommend that managers continue to collect data that will permit estimationand monitoring of kitten adult and subadult survival We identified these parameters via sensitivity analyses asmost critical in terms of their impact on the probability of and expected times to quasi‐extinction The con-tinuation of long‐term monitoring should permit the adaptation of genetic management strategies as necessarywhile collecting data that have proved essential in assessing the genetic and demographic health of the popu-lation The prospects for recovery of the panther will certainly be improved by following these guidelines copy2019 The Authors Wildlife Monographs published by Wiley Periodicals Inc on behalf of The Wildlife Society

KEY WORDS Florida panther demography genetic introgression genetic rescue heterosis inbreeding depressionindividual‐based population models population viability analysis Puma concolor coryi

Dinaacutemicas Persistencia y Manejo Geneacutetico de la Poblacioacutenen Peligro de Extincioacuten de Pantera de Florida

RESUMEN La poblacioacuten de pantera de Florida (Puma concolor coryi) mejoroacute tras la implementacioacuten en 1995 delproyecto de introgresioacuten geneacutetica en el sur de Florida USA como lo demuestran varias liacuteneas de evidencia Desdeentonces su diversidad geneacutetica ha mejorado la frecuencia de iacutendices morfoloacutegicos y biomeacutedicos correlacionadoscon depresioacuten endogaacutemica ha disminuido y el tamantildeo de la poblacioacuten ha aumentado Sin embargo la poblacioacuten depanteras permanece pequentildea aislada y se enfrenta a retos sustanciales producidos por fuerzas determiniacutesticas yestocaacutesticas Los objetivos de este estudio fueron 1) evaluar exhaustivamente la demografiacutea de la poblacioacuten depanteras de Florida usando datos de campo (del periodo 1981ndash2015) y modelos con el fin de calibrar en que medidapersisten los beneficios adquiridos a traveacutes de la introgresioacuten geneacutetica y 2) evaluar la efectividad de varias estrategiasde manejo geneacutetico La diversidad geneacutetica de la poblacioacuten mejoroacute sustancialmente con la introduccioacuten de pumashembra (Puma concolor stanleyana) procedentes de Texas USA En panteras canoacutenicas (no procedentes de in-trogresioacuten) el valor medio de heterocigosidad individual fue 0386plusmn 0012 (SE) y en panteras mezcladas0615plusmn 0007 En gran medida las tasas de supervivencia dependieron de la edad (los cachorros teniacutean las tasas desupervivencia maacutes bajas) estuvieron afectadas positivamente por la heterocigosidad individual y disminuyeroncuando la poblacioacuten aumentoacute La tasa de supervivencia total independientemente del sexo del cachorro fue de032plusmn 009 La tasa de supervivencia anual de panteras adultas y subadultas varioacute seguacuten el sexo independientementede la edad las hembras vivieron maacutes que los machos Las tasas anuales de supervivencia de hembras subadultasadultas y adultas mayores fueron 097plusmn 002 086plusmn 003 y 078plusmn 009 respectivamente Las tasas de supervivenciade machos subadultos adultos y adultos mayores fueron 066plusmn 006 077plusmn 005 y 065plusmn 010 respectivamenteLa ascendencia geneacutetica determinoacute en gran medida la tasa de supervivencia de panteras de cualquier edad siendomayor en la primera generacioacuten filiar (F1) de panteras mezcladas en todas las edades y menor en los individuoscanoacutenicos (sobre todo panteras pre‐introgresioacuten y sus descendientes post‐introgresioacuten) En panteras con collares deradio telemetriacutea las causas de mortalidad maacutes frecuentes fueron la agresioacuten intraespeciacutefica y la colisioacuten convehiacuteculos El anaacutelisis de las causas de mortalidad reveloacute que en las categoriacuteas colisioacuten con vehiacuteculos agresioacutenintraespeciacutefica otras causas y motivos desconocidos la tasa de mortalidad de machos y hembras era similar aunquelos machos teniacutean maacutes posibilidades de morir por agresioacuten intraespeciacutefica que las hembras Las probabilidades dereproduccioacuten y el nuacutemero anual de cachorros dependieron de la edad pero no de los ancestros o el tamantildeo de lapoblacioacuten Las probabilidades de reproduccioacuten de hembras subadultas adultas y adultas mayores se estimaron en035plusmn 008 050plusmn 005 y 025plusmn 006 respectivamente El nuacutemero de cachorros por antildeo y por pantera subadultaadulta y adulta mayor se estimoacute en 280plusmn 075 267plusmn 043 y 228plusmn 083 respectivamente Usando un modelodemograacutefico matricial se estimoacute la tasa anual de crecimiento estocaacutestico poblacional en 104 (95 CI= 072ndash141)Usando un modelo de poblacioacuten basado en el individuo e ignorando el impacto adverso de la erosioacuten geneacutetica seestimoacute la probabilidad de que la poblacioacuten disminuyese a menos de 10 panteras en 100 antildeos (cuasi‐extincioacuten) en14 (0ndash08) Sin embargo incluyendo el impacto de la erosioacuten geneacutetica la probabilidad de cuasi‐extincioacuten en 100antildeos aumentoacute al 17 (0ndash100) El plazo medio para la cuasi‐extincioacuten asumiendo que la cuasi‐extincioacuten ocurre en

4 Wildlife Monographs bull 203

100 antildeos e incluyendo el impacto de la erosioacuten geneacutetica fue de 22 (0ndash75) antildeos Anaacutelisis de sensibilidad demostraronque la probabilidad de cuasi‐extincioacuten y el plazo hasta alcanzarla dependiacutean de los valores utilizados para losparaacutemetros de supervivencia de cachorros Sin manejo geneacutetico la poblacioacuten de panteras de Florida se enfrentariacutea aun aumento sustancial del riesgo de cuasi‐extincioacuten Por lo tanto la pregunta no es si es necesario el manejo geneacuteticode la poblacioacuten de las panteras de Florida sino cuaacutendo y coacutemo implementarlo Usando modelos de simulacioacutenbasados en individuos evaluamos diferentes estrategias de introgresioacuten geneacutetica y sus posibles impactos en lapoblacioacuten y en su geneacutetica La reduccioacuten del riesgo de cuasi‐extincioacuten fue similar introduciendo 5 pumas cada 20antildeos o 15 pumas cada 10 antildeos (44ndash59 vs 40ndash58) pero la primera opcioacuten resulta maacutes econoacutemica ($200000 en100 antildeos) que la segunda ($12000000 en 100 antildeos) Introducir maacutes hembras en cada intento de introgresioacuten noprodujo beneficios adicionales El impacto positivo del proyecto de introgresioacuten geneacutetica persiste en la poblacioacuten depanteras veinte antildeos despueacutes de su implementacioacuten Recomendamos a los gestores que consideren repetir la in-trogresioacuten geneacutetica introduciendo 5ndash10 individuos de otras poblaciones de puma cada 20ndash40 antildeos Adicionalmenterecomendamos que se continuacutee con la recopilacioacuten de datos lo que es crucial para poder estimar y monitorizar lasupervivencia de cachorros adultos y subadultos Estos paraacutemetros son los maacutes criacuteticos para estimar la probabilidad yel periodo de cuasi‐extincioacuten seguacuten se demostroacute con el anaacutelisis de sensibilidad Se debe continuar con la mon-itorizacioacuten de la poblacioacuten a largo plazo lo que permitiraacute adaptar las estrategias de manejo seguacuten sea necesario a lavez que recopilar informacioacuten esencial para evaluar la salud demograacutefica de la poblacioacuten Siguiendo estas re-comendaciones las perspectivas de recuperacioacuten de las panteras mejoraraacuten

Contents

INTRODUCTION 5

STUDY AREA 9

METHODS 9

Field Methods 9

Genetic Analysis 10

Demographic Parameters 11

Survival of kittens 11

Subadult and adult survival 11

Probability of reproduction 11

Reproductive output 12

Covariates 12

Cause‐specific mortality 12Statistical inference 12

Matrix Population Models 13

IBMs and PVA 14

Sensitivity Analysis 15

Genetic Management 15

Introgression strategies 15

Cost‐benefit analysis 16RESULTS 17

Genetic Variation 17

Survival and Cause‐Specific Mortality Rates 17

Reproductive Parameters 18

Population Dynamics and Persistence 18

Deterministic and stochastic demography 18

Minimum population count scenario 18

Motor vehicle mortality scenario 18

Sensitivity of extinction probability to demographic parameters 19

Benefits and Costs of Genetic Management 19

Minimum population count scenario 19

Motor vehicle mortality scenario 19

Financial cost and persistence benefits 21

DISCUSSION 21

Genetic Variation Demographic Parameters

and Persistence of Heterotic Benefits from Genetic Introgression 24

Population Dynamics and Persistence 28

Genetic Management When and How to Intervene 28

MANAGEMENT IMPLICATIONS 30

ACKNOWLEDGMENTS 31

LITERATURE CITED 31

INTRODUCTION

Promoting genetic admixture between individuals from differentpopulations commonly referred to as genetic introgression hasthe potential to serve as a powerful management and con-servation tool (Keller and Waller 2002 Tallmon et al 2004Whiteley et al 2015) Genetic introgression can occur naturallythrough immigration of individuals into small or isolated po-pulations (Marr et al 2002 Vila et al 2003 Adams et al 2011)but it can also be implemented as a management strategy via theintentional translocation of individuals from large viable popu-lations to small endangered populations Intentional geneticintrogression of populations of wild animals has prevented thedemise of species such as adders (Vipera berus Madsen et al2004) prairie chickens (Tympanuchus cupido Bateson et al

2014) and bighorn sheep (Ovis canadensis Madsen et al 1999Hogg et al 2006) In most of these cases genetic introgressionhas been effective at improving demographic performance ofpopulations affected by inbreeding depressionInbreeding depression a reduction in fitness resulting from

the production of offspring by individuals related by ancestryis a conservation challenge common to small isolated popu-lations Inbreeding depression has historically been docu-mented in captive populations (Lacy et al 1993) The dele-terious impact of inbreeding depression on wild populationswas somewhat controversial (Frankham 2010a) until theseminal work of Crnokrak and Roff (1999) which verified thenegative effects of inbreeding on the fitness traits of severalwildlife populations including cheetahs (Acinonyx jubatus)

van de Kerk et al bull Florida Panther Population and Genetic Viability 5

black‐tailed prairie dogs (Cynomys ludovicianus) and white‐footed mice (Peromyscus leucopus) Subsequently the effect ofinbreeding depression has been extensively documented inseveral wild animal populations including mammals such aswolves (Canis lupus Liberg et al 2005) African lions (Pan-thera leo Trinkel et al 2010 2011) and Iberian lynx (Lynxpardinus Ruiz‐Loacutepez et al 2012) In these examples in-breeding depression had negative effects on a variety of fitnesstraits including survival of neonates susceptibility to diseaseand male fertility The demographic and genetic impacts ofinbreeding on population growth in carnivores is highlightedbest in one of the quintessential long‐term conservation stu-dies on the gray wolves of Isle Royale in northern MichiganUSA (Peterson 1977 Peterson and Page 1988 Peterson et al1998 Adams et al 2011) Research on this population hasproved informative for gaining a better understanding of po-pulation consequences of loss of genetic variation and impactsof intrinsic and extrinsic factors that have led to its decline toonly 2 individuals as of 2017 (Hedrick et al 2017)Conversely outbreeding depression is a scenario when mating

between members of genetically divergent populations might resultin a reduction in fitness correlated with the loss of local adaptationsor genetic incompatibilities (Moritz 1999) Outbreeding depressionhas been hypothesized to be a possible side effect of genetic in-trogression (Reinert 1991 Maehr and Caddick 1995) albeit in-frequently and with less and weaker empirical evidence for verte-brate species Negative fitness responses resulting from matingbetween individuals from highly divergent populations of copepodswere noted by Hwang et al (2012) Other examples of outbreedingdepression in varied species of plants are provided by Fenster andGallaway (2000) Waser et al (2000) Goto et al (2011) and Brysand Jacquemyn (2016) In general however outbreeding depres-sion is thought to be less of a concern when implementing con-servation measures such as genetic introgression (Miller et al 1999Whiteley et al 2015)In the last 3 decades conservation biologists have exponentially

increased their use of molecular techniques to assist with mon-itoring managing and conserving wildlife populations (DeYoungand Honeycutt 2005) Although early research relied on methodssuch as restriction fragment length polymorphisms allozymeelectrophoresis and minisatellites (Smith and Wayne 1996) thelast 20 years have been dominated by the implementation of im-proved automated sequencing technology to provide more in-formative and fine‐scale data from both nuclear and mitochondrialDNA sequences (eg mtDNA sequencing microsatellitesDeYoung and Honeycutt 2005) Microsatellites in particular havebecome a common tool for conservation projects to assess geneticparameters including inbreeding depression and the impacts ofgenetic introgression in imperiled populations They continue tobe widely used molecular markers in conservation researchThe field of molecular ecology is changing rapidly and new

technology is constantly being developed Currently there is in-terest in applying genomics or so‐called whole‐genome sequencingtechniques to assist with conserving imperiled species Two recentexamples of the application of whole‐genome sequencing to en-dangered species of carnivores include the work of Robinson et al(2016) on the Channel Island fox (Urocyon littoralis) and Murch-ison et al (2012) on the Tasmanian devil (Sarcophilus harrisii) For

instance Robinson et al (2016) reported a near absence ofgenomic variation in Channel Island foxes from the island of SanNicolas demonstrating an extreme reduction in heterozygositycompared to mainland populations Similarly Murchison et al(2012) applied whole genome sequencing to ascertain whether theTasmanian devil facial tumor disease originated from a male or afemale devil and further used whole genome sequencing techniquesto assess transmission of the disease across Tasmania The appli-cation of whole genome sequencing in conservation biology andwildlife management will continue to increase (Whiteley et al2015) Although costs associated with implementing wholegenome sequencing are steadily declining (Whiteley et al 2015)other factorsmdashsuch as sample preparation and the need forbioinformatic expertise to deal with voluminous amounts of datamdashare still an impediment to their widespread use in conservation andmanagement of wildlife populations Therefore long‐term datasets frommore traditional molecular markers such as microsatelliteswill continue to have utility for providing insight into the genetichealth of endangered species and for assessing impacts of man-agement and recovery initiativesThe identification and quantification of inbreeding depression

using molecular techniques (Kardos et al 2016) along with pro-tocols for implementing translocations to promote genetic in-trogression have been documented for varied endangered popu-lations (Whiteley et al 2015) Conversely studies that provide anassessment of the longevity of benefits accrued to wild populationsvia genetic introgression are uncommon Hedrick et al (2014)provided evidence for the declining effect of genetic introgressionon the Isle Royale population of wolves after only 2 or 3 genera-tions Whereas heterosis is expected to increase individual fitness itcan also result in outbreeding depression (Fenster and Gallaway2000 Waser et al 2000 Goto et al 2011 Waller 2015 Brys andJacquemyn 2016) Depending on the mechanisms outbreedingdepression may not become evident until generations subsequent tothe first filial (F1) generation (Pickup et al 2013 Waller 2015)For example ancestral chromosomes and coevolved genes remainintact in the F1 but recombination may break up favorable genecomplexes in F2 and later generations (Waser et al 2000 Waller2015 Frankham 2016) Investigating the timeframe over whichthe positive impacts of genetic introgression are sustained wouldlikely necessitate long‐term monitoring but for most studies of theprocess populations are typically monitored only for 1 or 2 gen-erations following genetic introgression (Willi et al 2007 Hedrickand Fredrickson 2009 Hedrick et al 2014 Whiteley et al 2015)Consequently data are rarely gathered that would allow us to es-timate the time over which populations continue to benefit fromgenetic introgression (Hedrick et al 2014 Whiteley et al 2015)The Florida panther (Puma concolor coryi) provides a textbook

example of how inbreeding depression in a small isolated po-pulation can result in low levels of genetic variation that areassociated with morphological and biomedical abnormalities andprecipitous population decline The timeline associated with thedecline of panthers parallels that of most large carnivores inNorth America (Onorato et al 2010) A robust population ofpanthers was thought to extend across Florida USA inpre‐Columbian times (Fig 1 Alvarez 1993) Subsequentlydocumented declines in the numbers of panthers were noted inthe nineteenth and twentieth centuries mainly as a result of

6 Wildlife Monographs bull 203

anthropogenic alteration of the landscape and unregulatedhunting As early as the 1870s panthers were rarely encounteredand considered mythical in some portions of their range (Beverly1874) The decline in the population of panthers eventuallycaptured the attention of lawmakers subsequently they receivedpartial legal protection as a game animal in 1950 and completelegal protection in Florida by 1958 (Onorato et al 2010) By the1960s the core of remaining wilderness within the Big Cypressregion of South Florida where the last group of breedingpanthers was thought to persist was further affected by theconstruction of Alligator Alley This roadway along with theTamiami Trail allowed access to these once inaccessible areasAll these factors ultimately led to the listing of the Floridapanther as endangered on the inaugural list of endangeredspecies on 11 March 1967 (US Federal Register 1967) andsubsequent protection under the Endangered Species Act of1973 (Public Law 93‐205) This designation invariably raisedawareness of the plight of the panther and served as a catalyst toinitiate research to avoid what appeared to be their imminentextinction (Onorato et al 2010)In 1973 the World Wildlife Fund initiated a survey that re-

sulted in the capture of a single female panther and doc-umentation of their sign in just a few locations in South Florida(Nowak and McBride 1974) The information accrued duringthis survey would ultimately lead to the commencement of along‐term research project on the Florida panther in 1981 by

what is now known as the Florida Fish and Wildlife Con-servation Commission (FWC) Over the next decade a multi-tude of studies were published that helped improve knowledgeof panther reproduction (Maehr et al 1989 Barone et al 1994)diet (Maehr 1990) and genetics (OrsquoBrien et al 1990 Roelkeet al 1993) among other topics Early research was essential foridentifying challenges faced by the remaining panthers mostimportantly the impacts of the continued loss of habitat isola-tion of the population and inbreeding depression (Onoratoet al 2010) In combination these factors would force wildlifemanagers to contemplate the implementation of genetic in-trogression to avert the extinction of the pantherBy the early 1990s the Florida panther population had declined

to a minimum of 20ndash30 individuals (McBride et al 2008) Severalstudies documented low levels of genetic diversity and expressedtraits thought to be correlated with inbreeding depression (Roelkeet al 1993 Johnson et al 2010 Onorato et al 2010) includingcowlicks (mid‐dorsal pelage whorls) and kinked tails (Wilkinset al 1997) These morphological traits probably had little directimpact on fitness of Florida panthers More concerning were theproportion of individuals in the population in the early 1990s thatwere unilaterally or bilaterally cryptorchid (Roelke et al 1993)possessed poor sperm quality (Barone et al 1994) sufferedcompromised immune systems (OrsquoBrien 1994) exhibited atrialseptal defects (Roelke et al 1993) and possessed some of thelowest levels of genetic variation observed in wild felids (Driscoll

Figure 1 Historical and current (2015) breeding range of the Florida panther in the southeastern United States Delineation of current breeding range is describedin Frakes et al (2015)

van de Kerk et al bull Florida Panther Population and Genetic Viability 7

et al 2002) Taken together these factors suggested that thelong‐term prospects for the persistence of the Florida pantherwere poorSubsequently the United States Fish and Wildlife Service

(USFWS) conducted an assessment that led to the delineation of 4possible management options 1) initiate a captive breeding pro-gram 2) introduce genetic material from captive stockpurported to contain both pure Florida panther and SouthAmerican puma lineages 3) translocate wild female pumas (Pumaconcolor stanleyana) from Texas USA into South Florida or 4)take no action at all (USFWS 1994) Collaborative discussionsinvolving academics non‐governmental organizations and stateand federal wildlife agencies eventually led to the selection of op-tion 3 and a plan for genetic introgression that involved the releaseof 8 female pumas from Texas into South Florida was im-plemented in 1995 (Seal 1994 Onorato et al 2010)Continued research conducted after the introgression interven-

tion provided clear evidence of the benefits accrued to the pantherpopulation including increased levels of genetic diversity declinesin the frequency of morphological and biomedical correlates ofinbreeding depression and the substantial increase of the popu-lation size (McBride et al 2008 Johnson et al 2010 McClintocket al 2015) Furthermore Hostetler et al (2013) showed that thepost‐introgression population growth was driven primarily byhigher survival rates associated with admixed panthers (offspringof pairings between the female Texas pumas and male Floridapanthers and subsequent generations of those progeny) followinggenetic introgression The apparent success of genetic introgres-sion in preventing the imminent extinction of the Florida pantherpopulation is encouraging but the population remains small andisolated and continues to face substantial challenges especially thelarge‐scale loss of habitatAlmost 5 panther generations (generation time= 45 years

Hostetler et al 2013) have passed since the implementation of thegenetic introgression program The panther population has beenmonitored continuously during this time period leading to asubstantial amount of additional data and biological insights sincethe demographic assessment made immediately following the in-trogression (Hostetler et al 2010 2012 2013 Benson et al 2011)Given that the effects of genetic introgression are not permanent itis important to discern whether the positive impacts of this man-agement initiative are continuing or waning Decline in the benefitsof the introgression could reduce the population growth rate andprobability of persistence At the same time panther habitat israpidly lost to a variety of human activities (eg land development)and environmental changes associated with invasive species andclimate change (Land et al 2008 Dorcas et al 2012 Frakes et al2015) When long‐term monitoring data are available it is prudentto periodically evaluate demographic rates population growth andpersistence parameters so that changes are detected and timelymanagement action can be implementedTwo population modeling approaches have become popular

among ecologists and wildlife managers for studying the dynamicsand persistence of age‐ or stage‐structured populations matrixpopulation models and individual‐based models (IBMs Brooket al 2000 Morris and Doak 2002 Morrison et al 2016) Matrixpopulation models are the generalized discrete time‐version of theLotka‐Euler equation which describes the dynamics of age‐ or

stage‐structured populations of plants animals and humans (Lotka1924 Caswell 2001) They have become standard tools for mod-eling human and wildlife populations (Tuljapurkar 1990 Caswell2001 Keyfitz and Caswell 2005) because 1) they are powerful andflexible permitting adequate representation of life history of a widevariety of organisms including those with complex life histories 2)parameters for these models can be empirically estimated usingstandard analytical methods such as multi‐state capture‐recapturemethods (eg Williams et al 2002) 3) it is relatively straight‐forward to include the effects of factors such as environmental anddemographic stochasticity and density dependence 4) importantdemographic quantities such as asymptotic population growth ratestable age‐ or stage distribution and generation time can be easilycalculated from these models 5) they permit prospective and ret-rospective perturbation analyses which have been proven useful inwildlife conservation and management 6) they are computer‐friendly do not require extensive programming experience andtheir implementation using software packages such as R (R CoreTeam 2016) and MATLAB (MathWorks 2014) is straight‐for-ward and 7) they have sound theoretical foundation offeringanalytical solutions to many aspects of population modeling Weused matrix population models for 1) calculating the aforemen-tioned demographic quantities 2) performing sensitivity analysesinvolving asymptotic population growth rate and 3) comparativepurposes to check the results obtained from equivalent simulation‐based modelsDespite many advantages offered by matrix population models it

is difficult to incorporate attributes of individuals such as geneticsor behavior within that framework Agent‐based or IBMs (Grimm1999 Grimm and Railsback 2005 McLane et al 2011 Railsbackand Grimm 2011 Albeke et al 2015) offer an alternative modelingframework with tremendous flexibility Individual‐based modelsare computer simulation models that rely on a bottom‐up approachthat begins by explicitly considering the components of a system(ie individuals) population‐level properties emerge from the be-havior of and interactions among discrete individuals UsingIBMs it is possible to explicitly represent attributes of individualssuch as movement mating or genetics (McLane et al 2011Railsback and Grimm 2011 Albeke et al 2015) We used in-dividual‐based population models as the primary populationmodeling approach because they offered a framework for an ex-plicit consideration of genetics both at individual and populationlevelsOur overall goal was to provide a comprehensive demographic

assessment of the sole extant population of the Florida pantherusing long‐term (1981ndash2015) field data and novel modelingtechniques Our objectives were 1) to estimate demographicparameters for the Florida panther and investigate whetherpositive impacts of the 1995 genetic introgression remained inthe population and if so to what extent 2) to estimate popu-lation growth and persistence parameters using an IBM tailoredto the Florida pantherrsquos life history and 3) to evaluate thesensitivity of population growth and persistence parameters tovital demographic ratesThe Florida panther population remains small (le230 adults and

subadults FWC 2017) with no realistic possibility of natural geneflow so the loss of genetic variation and concomitant increase ininbreeding‐related problems are inevitable Ensuring the long‐term

8 Wildlife Monographs bull 203

persistence of the population will likely necessitate periodic geneticmanagement interventions Therefore our final objective was to 4)evaluate genetic and population‐dynamic consequences of variedgenetic introgression strategies and from these identify manage-ment strategies that are affordable and effective at improvingprospects of Florida panther persistence To achieve this objectivewe extended the individual‐based population model to track in-dividual genetics based on empirical allele frequencies taking ad-vantage of the long‐term demographic and genetic data that havebeen collected on Florida panthers since 1981 We used individualheterozygosity to inform vital rates based on empirically estimatedrelationships between individual heterozygosity and demographicparameters We then simulated population trajectories trackedindividual heterozygosity over time and estimated quasi‐extinctiontimes and probabilities without introgression and with introgres-sion for a range of intervals and number of immigrants perintrogression event We compared the genetic and demographicbenefits of implementing alternative genetic introgression scenariosand ranked them by the level of improvement in population per-sistence parameters and estimated costs because funding is a per-sistent challenge to conservation planningManagement of imperiled species is a complex process invol-

ving many stakeholders and plagued by layers of uncertainties(Runge 2011) It would therefore seem that management ofthreatened and endangered species would be a perfect case foradaptive resource management approach (ARM) because ARMfocuses on structured decision making for recurrent decisionsmade under uncertainty (Williams et al 2002 Runge 2011)Because it combines structured decision making with learningprudent application of ARM reduces uncertainty over time andcan lead to optimal management actions (Williams and Brown2012 2016) Key ingredients of a successful adaptive manage-ment program include stakeholder engagement clearly definedand mutually agreed‐upon objectives alternative managementactions and objective decision rules (Nichols et al 2007Williams and Brown 2012 2016) Unfortunately Florida pan-ther stakeholders have divergent views about the state of thesystem (eg panther abundance) and it is a challenge to havemutually agreed‐upon objectives management actions or deci-sion rules acceptable to all or most stakeholders AlthoughFlorida panther conservation efforts would be best served withinthe ARM framework in the long run impediments to its im-plementation (Runge 2011 Williams and Brown 2012) areunlikely to be overcome in the near future

STUDY AREA

The breeding population of Florida panthers mainly persists as asingle population South of the Caloosahatchee River (approxi-mately 267133degN latitude 815566degW longitude see Fig 1)One exception is a female that was documented just north of theCaloosahatchee River in 2016 (and subsequently documentedwith kittens in 2017) by FWC the first validation of a femalepanther north of the River since 1973 (FWC unpublisheddata) The breeding range in South Florida is topographicallyflat has a subtropical climate and is characterized by permanentand ephemeral wetlands influenced by rains from May throughOctober (Duever et al 1986) The study area contains a variety

of wildland land cover types including hardwood hammockscypress forests pine flatwoods freshwater marshes prairies andgrasslands (Davis 1943) and land characterized by human ac-tivity including citrus groves croplands pastureland rockmining and residential developments (Onorato et al 2011)The area is intersected by a multitude of roads that fragmentpanther habitat Most notable among these is Interstate 75(Alligator Alley) which was expanded from 2 to 4 lanes in 1990when fencing and wildlife crossings were constructed with theintention of minimizing harmful effects of road expansion onlocal wildlife (Foster and Humphrey 1995) A large portion ofthe area that comprises South Florida is under public ownershipmainly as federal or state lands Everglades National Park BigCypress National Preserve Florida Panther National WildlifeRefuge Picayune Strand State Forest and Fakahatchee StrandPreserve State Park alone encompass 9745 km2 though not allof the area is considered suitable for panthers (eg large ex-panses of sawgrass marshes in the western Everglades) Frakeset al (2015) delineated 5579 km2 of adult panther habitat inSouth Florida 1399 km2 of which are in private ownership Amajority of these private lands are located in the northern extentof the breeding range Although some private lands may beprotected (eg conservation easements) other areas are sus-ceptible to incompatible land uses such as rock mining or re-sidential developments The breeding range in South Florida isbounded to the east and west by large metropolitan areas in-clusive of Miami‐Fort Lauderdale‐Pompano Beach and CapeCoral‐Fort Myers‐Naples‐Marco Island‐Immokalee respec-tively that are populated by gt6000000 people (httpswwwcensusgov accessed 24 Jan 2019) These characteristics of thestudy area highlight the recovery challenges faced by panthers

METHODS

Field MethodsSince 1981 Florida panthers have been captured radiocollared andtracked by the FWC and National Park Service staff (Table 1)Capture and handling protocols followed guidelines of theAmerican Society of Mammalogists (Sikes 2016) Livestock Pro-tection Company (Alpine TX USA) provided trained hounds andhoundsmen for captures Teams either treed or bayed panthers onthe ground and then darted them with a 3‐ml compressed‐air dartfired from a CO2‐powered rifle Immobilization drugs have variedduring the tenure of the project but most recently teamsimmobilized panthers with a combination of ketamine HCl (10mgkg Doc Lanersquos Veterinary Pharmacy Lexington KY USA) andxylazine HCl (1mgkg Doc Lanersquos Veterinary Pharmacy) Fol-lowing immobilization teams lowered treed panthers to the groundby a rope or caught panthers with a net in some cases they used aportable cushion (McCown et al 1990) to further mitigate theimpact of a fall Capture teams administered propofol (PropoFlotradeAbbott Laboratories Abbott Park IL USA) intravenously either asa bolus or continuous drip to maintain anesthesia They adminis-tered midazolam HCl (003mgkg) intramuscularly or intravenouslyto supplement anesthesia in some panthers Panthers recovered in ashaded area away from water In some cases capture teams reversedxylazine HCl with yohimbine HCl (Yobinereg Lloyd IncShenandoah IA USA) at 25 its recommended dose

van de Kerk et al bull Florida Panther Population and Genetic Viability 9

Capture teams determined sex and age of panthers marked themwith an ear tattoo and a subcutaneous passive integrated trans-ponder (PIT‐tag) and fitted them with a very high frequency(VHF) or global positioning system (GPS) radio‐collar equippedwith a mortality sensor The GPS collars were programmed toattempt to collect a position at a variety of time intervals(range= 1ndash12 hr) depending on objectives of concurrent studiesobservers routinely tracked panthers with VHF radiocollars from afixed‐wing aircraft 3 times per week When a radio‐collar trans-mitted a mortality signal or when a location did not change forseveral days observers investigated the site to establish the pan-therrsquos fate When observers found a dead panther they assessed thecause of death based on an examination of the carcass and sur-rounding area if possible They then transported carcasses to anexperienced veterinarian for necropsy and a histopathological ex-amination to attempt to assign the cause of death Starting in 1995observers continually assessed successive locations of radio‐collaredfemales to determine if they initiated denning behavior Three to 4fixes at the same location (VHF collars) or successive returns to thesame location over a weeklong period (GPS collars) served as anindicator of a possible den Observers then located dens moreprecisely via triangulation with ground telemetry They visited denswhen the dam left the site typically to make a kill Teams capturedkittens by hand captured kittens ranged in age from 4 to 35 dayspost‐partum Teams counted sexed weighed dewormed andpermanently marked kittens with a subcutaneous PIT‐tag

Genetic AnalysisTeams collected blood tissue or hair from the majority ofcaptured kittens subadult and adult Florida panthers for DNAsamples The United States Department of Agriculture ForestService National Genomics Center for Wildlife and FishConservation (Missoula MT USA) processed samplesTechnicians used Qiagen DNeasy Blood and Tissue Kits(Qiagen Valencia CA USA) to extract whole genomic DNAfrom blood and tissue samples and used a slight modification(overnight incubation of sample in lysis buffer and proteinase Kon a rocker at 60deg elution of DNA in 50 μl of buffer) of theQiagen DNA extraction protocol (Mills et al 2000) to extractDNA from hair samples Technicians used a panel of 16 mi-crosatellite loci (Fca090 Fca133 Fca243 F124 F37 Fca075Fca559 Fca057 Fca081 Fca566 F42 Fca043 Fca161

Fca293 Fca369 and Fca668) identified by Menotti‐Raymondet al (1999) which had been previously applied to Floridapanther samples (Johnson et al 2010) Technicians amplifiedDNA samples in 10‐microl reaction volumes that included 10 microl ofDNA 1times reaction buffer (Applied Biosystems Life Technolo-gies Corporation Grand Island NY USA) 20 mM of MgCl2200 microM of each dNTP 1 microM of reverse primer 1 microM of dye‐labeled forward primer 15 mgml of bovine serum albumin(BSA) and 1U Taq polymerase (Applied Biosystems) Thepolymerase chain reactions (PCRs) included a thermal profile of94degC for 5 minutes followed by 36 cycles of 94degC for 60 seconds55degC for 60 seconds and 72degC for 30 seconds Techniciansvisualized the resulting PCR products on a LI‐COR DNAanalyzer (LI‐COR Biotechnology Lincoln NE USA) Tech-nicians collected genotypes from hair samples (n= 16) using amultitube approach (Taberlet et al 1996) error‐checked geno-types using Program DROPOUT (McKelvey and Schwartz2005) and combined them with genotypes from tissue andblood samples (n= 487) We evaluated all 16 loci for variousdescriptive statistics for the radio‐collared sample of adult andsubadult panthers We excluded genotypes of kittens from thatanalysis because related individuals are known to result in theoverrepresentation of certain alleles (Marshall et al 1998) Wedetermined the number of alleles observed heterozygosity andexpected heterozygosity using GenAlex (Peakall and Smouse2006 2012) We assessed allelic richness at each locus in Fstat293 (Goudet 1995) We assessed conformance of genotypedata from these 16 loci to HardyndashWeinberg assumptionslinkage disequilibrium and null alleles (Appendix A availableonline in Supporting Information) The Genomics Center alsoprovided us with genotype data from the same 16 loci for 41pumas from western Texas the source population for thegenetic rescue programEvidence of inbreeding depression in wild populations is

commonly determined on the basis of heterozygosityndashfitnesscorrelations (Silva et al 2005 Ortego et al 2007 Mainguyet al 2009 Johnson et al 2011) so we also estimated theheterozygosity of each individual panther We calculatedhomozygosity by loci (HL) which varies between 0 (all lociheterozygous) and 1 (all loci homozygous Aparicio et al 2006)using the Rhh package (Alho et al 2010) in Program R (R CoreTeam 2016) A microsatellite locus has more weight in HL

Table 1 Number of Florida panthers by age sex and genetic ancestry categories used in subsequent demographic analyses during the study period (1981ndash2013) inSouth Florida USA

Kittensa Subadults and adultsb

Males Females Total Males Females Total

PIT‐taggedc 215 180 395 Radiocollaredc 108 101 209Recovered 11 4 15 Subadult 72 50 122Litter failed 47 38 85 Prime adult 65 95 160Recaptured 25 30 55 Old adult 13 20 33Canonical 27 20 47 Canonical 43 29 72F1 admixed 6 12 18 F1 admixed 2 6 8Other admixed 130 105 235 Other admixed 45 53 98

a Kitten samples included those that were PIT‐tagged recovered (ie panthers that were found dead) part of a failed litter or recaptured as subadults or adultsb Radiocollared subadult and adult panthers are presented as sample size within each age class and ancestry classc Combined number of panthers of different ancestries is less than the number radiocollared or tagged with a subcutaneous passive integrated transponder (PIT‐tagged) because we did not conduct genetic sampling on all panthers Combined number of panthers in different age classes is greater than the number ofradiocollared panthers because some individuals were counted in ge2 age classes as animals aged

10 Wildlife Monographs bull 203

when it is more informative (ie has more alleles) and has aneven distribution of allele frequencies For our demographicanalyses we calculated individual heterozygosity for each in-dividual panther as 1 minusHL As do most indices of individualheterozygosity HL takes into account the average allele fre-quency in the population (Aparicio et al 2006) Thus duringsimulations the allele frequency in the population changesthroughout an individualrsquos lifetime which causes its hetero-zygosity also to change This is not biologically plausible becausegenetic properties are retained throughout an individualrsquos life-time Therefore for use in model simulations we simply cal-culated heterozygosity for each individual as the proportion ofheterozygous lociWe also used genotype data to implement a Bayesian clus-

tering analysis in Program STRUCTURE (Pritchard et al2000) to infer ancestral clusters of panthers This method uses aMarkov chain Monte Carlo (MCMC) method to determine thenumber of genetic clusters (K) in the sample while also pro-viding an individualrsquos percentage of ancestry (q values) allocatedto each cluster Runs for this analysis included a 100000MCMC burn‐in period followed by 500000 MCMC iterationsfor 1ndash10 ancestral genetic clusters (K= 1 to 10) with 10 re-petitions for each K To determine the most likely number of Kgenetic clusters we used the logarithm of the probability of thedata (log Pr(D)∣K Pritchard et al 2000) and estimates of ΔK(Evanno et al 2005) in Program STRUCTURE HAR-VESTER (Earl and VonHoldt 2011) We then used q valuesprovided in runs for the best K to assign individual panthers aseither canonical (in most cases ge90 pre‐introgression ancestrysee Johnson et al 2010) or admixed (lt90 pre‐introgressionancestry) We recognized admixed panthers that were the off-spring of matings between a Texas female puma and a canonicalmale panther as F1 admixed panthers for analyses

Demographic ParametersSurvival of kittensmdashMost kittens PIT‐tagged at den sites were

never encountered again A small proportion of the PIT‐taggedkittens were recovered dead or were recaptured as subadults oradults and radiocollared so that their fate could be monitoredWe also identified instances of complete litter failure when afemale died within 9 months after giving birth (the minimumage of independence for kittens) or when a female wasdocumented to have another litter less than 12 months aftergiving birth (the minimum age of independence plus a 3‐monthgestation period Hostetler et al 2010) Thus our data setconsisted of 1) live recaptures and subsequent radio‐tracking ofpanthers of various ages that had been PIT‐tagged as kittens 2)recovery of dead panthers of various ages that had been PIT‐tagged as kitten 3) live captures and subsequent radio‐trackingof other panthers and 4) instances of complete litter failure(Table 1) We analyzed these data using Burnhamrsquos live‐recapture dead‐recovery modeling framework (Burnham 1993Williams et al 2002) to estimate survival of kittens (0ndash1‐yearold) and to test covariate effects on this parameter We includeddata on individuals encountered dead or alive at an older age inour analyses because their encounter histories also providedinformation on kitten survival We used a live‐dead data inputformat coded with a 1‐year time step (Cooch and White 2018)

All known litter failures occurred within the first year of thekittensrsquo lives so we treated all litter failures as recoveries withinthe first year We treated panthers that would have died if wehad not removed them to captivity as having died on the date ofremoval Data preparation and analytical approach are describedin more detail by Hostetler et al (2010)In addition to survival probability (S) the Burnham model

incorporates parameters for recapture probability (p) recoveryprobability (r) and site fidelity (f) We set r and p to 10 forradio‐tracked individuals because their status was known eachyear We also fixed f at 10 for all individuals because the re-capture and recovery areas were the same and encompassed theentire range of the Florida panther Based on findings of Hos-tetler et al (2010) we used a base model that 1) allowed survivalto differ between kittens and older panthers 2) allowed survivalto differ between sexes and among age classes (females ages 1and 2 ge3 males ages 1 2 and 3 ge4) 3) constrained recaptureprobability to be the same for all uncollared panthers and 4)allowed recovery rates to differ between kittens and uncollaredolder panthers Even with our additional data this model re-mained the best fit for a range of models for recapture recoveryand survival of kittens subadults and adults of various age‐classcategories (Hostetler et al 2010)Subadult and adult survivalmdashTo estimate survival for panthers

ge1 year of age we analyzed radio‐tracking data using a Coxproportional hazard‐modeling framework (Cox 1972 Therneauand Grambsch 2000) We followed procedures outlined byBenson et al (2011) for data preparation and analysis Brieflywe right‐censored panthers on the last day of observation beforetheir collar failed When an individual aged into a new age classwhile being tracked we created 2 data entries 1 ending the dayit entered the new age class the other starting on that day Weused the FlemingndashHarrington method to generate survivalestimates from the Cox analysis and the Huber sandwichmethod to estimate robust standard errors (Therneau andGrambsch 2000 Benson et al 2011)We considered 3 age classes subadults (1ndash25 yr for females and

1ndash35 yr for males) prime adults (25ndash10 yr for females and35ndash10 yr for males) and old adults (gt10 yr old for both sexes)after Benson et al (2011) We estimated overall survival using abase model that allowed survival to differ between subadultfemales subadult males prime adult females prime adult malesand old adults of either sex (old age was additive with sex other-wise age and sex were interactive) This model provided the best fitto the data relative to a range of models considering differences insurvival between prime‐adult and older‐adult age classes both ad-ditively and interactively with sex (Benson et al 2011)To examine whether survival rates observed in recent years

differed from those observed immediately following introgres-sion we also estimated age‐specific survival rates for 2 nearlyequal periods of time immediately post‐introgression (1995ndash2004) and the more recent period (2005ndash2013) For theseanalyses we separated the radio‐tracking data into the 2 periodsand estimated survival rates for each period separately using thebase model Similar analyses for kitten survival were not possiblebecause of data limitationsProbability of reproductionmdashWe used telemetry data obtained

from radio‐tracked females that reproduced at least once to

van de Kerk et al bull Florida Panther Population and Genetic Viability 11

estimate annual probability of reproduction of subadult prime‐adult and old‐adult females using complementary logndashlogregression (Agresti 2013) following methods described indetail by Hostetler et al (2012) Briefly we modeled theprobability of a female panther giving birth (b) as a function ofcovariates as

γ= minus [minus ( )]b z1 exp exp i i

where zi is a row vector of covariates (see below for covariateinformation) for female panther i and γ is a column vector ofmodel coefficients (Appendix B Table B2) To account fordifferences in observation time we included log(m) as an offsetin the model where m is the number of months a femalepanther was radio‐tracked (Hostetler et al 2012)

Reproductive outputmdashTo estimate the number of kittensproduced by reproductive females we used cumulative logitregression (Min and Agresti 2005 Agresti 2013) Weconsidered a model that includes J categories for the numberof offspring with j denoting the number of offspring ( j= 1 2hellip J and J= 6) provided that a female reproduces Weconsidered J= 6 which is greater than the largest litter sizeobserved in our study (4 kittens) because females can produce 2or more litters per year if litters fail We modeled the probabilitythat a female i produces a litter of size at most j (Yi) as

le⎧

⎨⎩

δ( ) = == hellip minus

=

( )+ θ βminus +Y jj J

j JPr

1 2 1

1i ij e

1

1 xj i

where θj is the intercept for the annual number of kittens pro-duced by a female i being at most j xi is a row vector of cov-ariates for female panther i (see below) and β is a column vectorof model coefficients The probability that Yi will be exactly j isgiven by

⎧⎨⎩

πδ

δ δ( = ) = =

=

minus = hellip( minus )Y j

jj J

Pr 1

2 i ij

i

ij i j

1

1

Finally the expected annual number of kittens produced byfemale i (νi) is given by

sumν π==

j i

j

J

ij

1

Additional details regarding the estimation and modelingof the annual number of kittens produced can be found inHostetler et al (2012)

CovariatesmdashIn addition to sex and age we tested for the effect ofabundance genetic ancestry and individual heterozygosity on theaforementioned demographic parameters (model coefficients forkitten and adult survival are given by η and ι respectively AppendixB Table B2) We used a minimum population count (MPC) basedon radio‐tracking data and field evidence of subadult and adult (ieindependent age) panthers as an index of abundance (McBride et al2008) For these analyses we did not use the population sizeestimates reported by McClintock et al (2015) because unlike theminimum counts they did not cover the entire study period

(McBride et al 2008 R T McBride and C McBride RancherrsquosSupply Incorporated unpublished report Fig 2)To test if demographic parameters differed depending on an-

cestry we considered 3 ancestry models dividing the panthers into1) F1 admixed and all other panthers 2) canonical and admixed(including F1 admixed) panthers and 3) F1 admixed canonicaland other admixed panthers We quantified genetic diversity usingindividual heterozygosity as described previously We tested for theeffect of genetic covariates using a subset of kittens that were ge-netically sampled We calculated model‐averaged estimates ofmodel parameters on the scale in which they were estimated(Appendix B available online in Supporting Information)

Cause‐specific mortalitymdashWe performed cause‐specificmortality analysis only on dead radio‐collared panthers becausedetection rates of uncollared panther mortalities are highlycorrelated with the cause of death (eg vehicle collision)Following Benson et al (2011) we attributed each mortality to1 of 4 causes 1) vehicle collision 2) intraspecific aggression 3)other causes (including known causes such as diseases injuriesand infections unrelated to the first 2 causes) and 4) unknown(ie mortalities for which we could not assign a cause) Wethen used the non‐parametric cumulative incidence functionestimator a generalization of the staggered‐entry KaplanndashMeiermethod of survival estimation to estimate cause‐specificmortality rates for male and female panthers in different ageclasses (Pollock et al 1989 Heisey and Patterson 2006 Bensonet al 2011) We compared cause‐specific mortality rates betweensexes age classes and ancestries

Statistical inferencemdashWe used an information‐theoreticapproach (Akaikersquos information criterion [AIC]) for modelselection and statistical inference (Burnham and Anderson 2002)For the analysis of kitten survival we adjusted the AIC foroverdispersion and small sample size (QAICc details in Hostetleret al 2010) We performed all statistical analyses in Program R (RCore Team 2016) We used Burnhamrsquos live recapturendashdead

0

100

200

300

1985 1990 1995 2000 2005 2010Year

Indi

vidu

als Method

MPC

MVM

Figure 2 The Florida panther minimum population count (MPC) and populationsize estimated using motor vehicle collision mortalities (MVM) during our studyperiod Minimum population counts are from McBride et al (2008) and R TMcBride and C McBride (Rancherrsquos Supply Incorporated unpublished report)Estimates based on MVM are from McClintock et al (2015) The solid vertical lineindicates the year of genetic introgression

12 Wildlife Monographs bull 203

recovery model (Burnham 1993) to estimate and model kittensurvival using the RMark package version 2113 (Laake 2013) asan interface for Program MARK (White and Burnham 1999)We implemented the Cox proportional hazard model for esti-

mation and modeling of survival of subadults and adults using the Rpackage SURVIVAL (Therneau and Grambsch 2000) We per-formed cause‐specific mortality analyses using an R version of S‐PLUS code provided by Heisey and Patterson (2006)To account for model selection uncertainty we calculated

model‐averaged parameter estimates across all models using theR package AICcmodavg (Mazerolle 2017) using AIC weights asmodel weights (Burnham and Anderson 2002) We used modelswithout the categorical ancestry variables for model averagingand estimated parameters based on the average value for thecontinuous covariates (individual heterozygosity and abundanceindex) Because only a subset of the kittens was geneticallysampled we estimated 2 kitten survival rates First we estimatedkitten survival based on the data set that included all kittens(overall kitten survival) Second we estimated kitten survivalusing a subset of the data that only included kittens that weregenetically sampled (survival of genetically sampled kittens)

Matrix Population ModelsWe investigated Florida panther population dynamics andpersistence using 2 complementary modeling frameworks ma-trix population models (Caswell 2001) and IBMs (Grimm andRailsback 2005 McLane et al 2011 Railsback and Grimm2011 Albeke et al 2015) Matrix population models offer aflexible and powerful framework for investigating the dynamicsand persistence of age‐ or stage‐structured populations (egCaswell 2001 Robinson et al 2008 Kohira et al 2009 Hunteret al 2010 Hostetler et al 2012) With a fully developed theorybehind them these models also serve as a check for resultsobtained from simulation‐based models such as IBMs We usedan age‐structured‐matrix population model for deterministic andstochastic demographic analyses and for preliminary investiga-tions of Florida panther population viability (Caswell 2001Morris and Doak 2002)For deterministic and stochastic demographic analyses we

parameterized a female‐only 19 times 19 age‐specific populationprojection matrix of the form

⎢⎢⎢⎢⎢

⋯ ⋯

⋯ ⋯

⋱ ⋱ ⋮

⋮ ⋱ ⋱ ⋱ ⋮

⎥⎥⎥⎥⎥

( )=t

F F F

P

P

P

A0 0

0

0 0 0

t t t

t

t

t

1 2 19

1

2

18

where Fi and Pi are the age‐specific fertility and survival ratesrespectively and ( )tA is a time‐varying (or stochastic) popula-tion projection matrix We assumed that the longevity and ageof last reproduction of female Florida panthers was 185 yearswe based this assumption on the observation that the oldestfemale in our study was 186 years old Florida panthers re-produce year‐round thus we used birth‐flow methods to esti-mate Fi and Pi values following Caswell (2001) and Hostetleret al (2013) We estimated Pi as

= +P S S i i i 1

where Si is the age‐specific survival probability We estimated Fi as

⎛⎝

⎞⎠

=+ +F S

m Pm

2i k

i i i 1

where Sk is the kitten survival probability and mi is the age‐specific fecundity rate which was estimated as

ν=m b f i i i k

where bi is the breeding probability for a female in age class iduring time step t νi is the annual number of kittens producedby a breeding female in age class i and fk is the proportionfemale kittens at birth (assumed to be 05)We estimated the finite deterministic (λ) and stochastic annual

population growth rate (λs) and stochastic sensitivities and elasti-cities following methods described in detail by Caswell (2001) Toestimate λs we assumed identically and independently distributedenvironments (Caswell 2001 Morris and Doak 2002 Haridas andTuljapurkar 2005) and used a parametric bootstrapping approachto obtain a sequence of demographic parameters for 50000 ma-trices We then estimated log(λs) from this sequence of matrices(Tuljapurkar 1990 Caswell 2001) and exponentiated it to obtain λsWe calculated the sensitivity and elasticity of λs to changes in lower‐level vital demographic rates as per Caswell (2001) and Haridas andTuljapurkar (2005)For population projection and population viability analysis (PVA)

simulations we expanded the population projection matrix into a34 times 34 2‐sex age‐structured matrix to include female and maleFlorida panthers (Caswell 2001 Hostetler et al 2013) Malescontribute to the population only through survival in this modelingframework We assumed maximum longevity for males to be 145years of age because the oldest male in our study was 144 years oldThe population projection matrix was of the following form

⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢

⋯ ⋯ ⋯ ⋯

⋯ ⋯ ⋯ ⋯

⋱ ⋱ ⋮ ⋮ ⋱ ⋱ ⋮

⋮ ⋱ ⋱ ⋱ ⋮ ⋮ ⋱ ⋱ ⋮

⋯ ⋯ ⋯

⋯ ⋯ ⋯ ⋯

⋯ ⋯ ⋱ ⋱ ⋮

⋮ ⋮ ⋱ ⋱ ⋮ ⋱ ⋱ ⋮

⋯ ⋯ ⋯

⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥

( )=

( ) ( ) ( )

( )

( )

( )

( ) ( ) ( )

( )

( )

t

F N F N F N

P N

P N

P N

M N M N M N

Q N

Q N

A n

0 0

0 0 0 0

0

0 0 0 0 0

0 0

0 0 0

00 0 0 0 0

t t t

t

t

t

t t t

t

t

1 2 19

1

2

18

1 2 19

1

14

where Pi and Qi are age‐specific survival rates of female and maleFlorida panthers respectively and Fi and Mi are the rates atwhich female and male kittens are produced by females of ageclass i respectively Here the population projection matrix isboth time‐varying and density‐dependent (Caswell 2001)We incorporated the influence of parameter uncertainty en-

vironmental stochasticity and density dependence on Floridapanther population dynamics and persistence following the ap-proach outlined by Hostetler et al (2013) we used the sameapproach in the IBM analysis which is described below UnlikeIBMs matrix models do not intrinsically include demographicstochasticity so we explicitly included the influence of

van de Kerk et al bull Florida Panther Population and Genetic Viability 13

demographic stochasticity by simulating the fates of individualsas described in Caswell (2001)Population projections using matrix models necessitate a

vector of initial sex‐ and age‐specific abundance Because suchdata are currently unavailable for the Florida panther we esti-mated it based on the stable sex and age distribution and MPCin 2013 Using point estimates of the demographic parameterswe estimated the stable sex and age distribution for the 2‐sexdeterministic and density‐independent population projectionmatrix We then multiplied the stable sex and age distributionby the MPC in 2013 to get the starting population vector n(0)We projected the population over time as n(t+ 1)=A(tn)n(t)where A(tn) indicates the time‐specific density‐dependentpopulation projection matrix (Caswell 2001)We had 2 indices of abundance available the MPC (McBride

et al 2008) and population size estimated using motor vehiclecollision mortalities (MVM McClintock et al 2015) These 2indices are substantially different because MPC does not ac-count for imperfect detection whereas MVM accounts forimperfect detection and provides an estimate of population size(Fig 2) Therefore we examined 2 scenarios one using densitydependence based on MPC and the other using density de-pendence based on MVM We ran 1000 bootstraps of para-meter values and 1000 simulations per bootstrap for each sce-nario We projected population trajectories over 200 years andthen estimated population growth rates probabilities ofquasi‐extinction and time until quasi‐extinction and comparedthese with the results obtained from the IBM (see below) Weran scenarios with a quasi‐extinction threshold of 10 and 30panthers We used the mean of probabilities of quasi‐extinction(PQE) across simulations as the point estimate for PQE and the5th and 95th percentiles across simulations as the 90 con-fidence interval Because the confidence intervals werequantile‐based but the point estimates were not it is possiblewith skewed distributions for the point estimates to be outsidethe confidence interval We implemented the matrix

population models in the R computing environment (R CoreTeam 2016)

IBMs and PVABecause of well‐developed theory ease of implementation andthe modeling flexibility that they offer matrix populationmodels have become the model of choice for investigating thedynamics and persistence of age‐ and stage‐structured popula-tions (Caswell 2001) However it is difficult to incorporateattributes of individuals such as genetics and behavior usingmatrix models Quantifying the effects of genetic erosion on theFlorida panther population dynamics and persistence andevaluating benefits and costs of alternative genetic managementscenarios were important goals of our study Because IBMs relyon a bottom‐up approach that begins by explicitly consideringindividuals (McLane et al 2011 Railsback and Grimm 2011Albeke et al 2015) it is possible to represent genetic attributesof each individual and to track genetic variation over time Weused IBMs as the primary modeling framework in this studybecause they allowed for explicit consideration of genetics bothat individual and population levels and population viabilityassessment under alternative genetic management scenariosWe developed an IBM (Grimm 1999 Grimm and Railsback

2005 McLane et al 2011 Railsback and Grimm 2011 Albekeet al 2015) tailored to the life history of the Florida panther(Fig 3) We simulated every individual from birth until deathbased on a set of rules describing fates (eg birth and death) andattributes (eg genetics) of individuals at each annual time stepAs in other IBMs population‐level processes (eg populationsize and genetic variation) emerged from fates of and interac-tions among individuals (Grimm 1999 Railsback and Grimm2011) We developed an overview design concepts and details(ODD) protocol (Grimm et al 2006 2010) to describe ourIBM (Appendix C available online in Supporting Information)and provide pseudocodes for individual‐based simulations(Appendix D available online in Supporting Information)

For each panther at time = t

Kitten (lt1 yr)

Reproduce

Survive

Kitten(s) born

Kitten

Sex and age class

Subadult male

Prime adult male

Old adult male

Subadult female

Prime adult female

Old adult female

Age one year

No Male

Female

Yes

No

Yes

t=t+1

Yes

lt35 yrs

35-10 yrs

gt10 yrs

lt25 yrs

25-10 yrs

gt10 yrs

Figure 3 Schematic representation of Florida panther life history We tailored the individual‐based population model to represent the life‐history patterndepicted here

14 Wildlife Monographs bull 203

Dynamics and persistence of populations are influenced bymany factors such as demographic and environmental sto-chasticity and density dependence these effects and parametricuncertainty must be considered in population models whenpossible (Caswell 2001 Boyce et al 2006 Bakker et al 2009Hostetler et al 2013) Because by definition IBMs simulate thelife history of individuals in a population they inherently in-corporate the effect of demographic stochasticity Our analysesrevealed that survival of kittens subadults and adults and theprobability of reproduction varied over time so we incorporatedenvironmental stochasticity in the model for these variablesfollowing Hostetler et al (2013) Likewise we found evidenceof density‐dependent effects on kitten survival Because theMPC might be an underestimate of population size (McBrideet al 2008) we performed simulations using both the MPC andMVM abundance indicesTo account for model selection and parametric uncertainty we

used a Monte Carlo simulation approach For each bootstrapsample we selected a model based on the AIC (or QAICc)weights We then sampled the intercept and slope for kittensurvival (on the logit scale) subadult and adult survival (on thelogit scale with log‐hazard effect sizes) and probability of re-production (on the complementary logndashlog scale Appendix B)from multivariate normal distributions with mean vectors equalto the estimated parameters and variance‐covariance matricesequal to the sampling variance‐covariance matrices We thentransformed the results to nominal scale and converted the es-timates to age‐specific survival and reproduction parameters asdescribed in Hostetler et al (2013)We ran 1000 simulations for 1000 parametric bootstraps for

200 years for the MPC and the MVM scenario for both thestochastic 2‐sex matrix‐population model and the IBM Wetracked the population size at each time step and estimatedquasi‐extinction probabilities and times based on the populationtrajectory We considered the population to be quasi‐extinct ifindividuals of only 1 sex remained or if the population size fellbelow critical thresholds set at 10 and 30 individuals We alsoestimated expected time to quasi‐extinction for both thresholdsdefined as the mean and median time until a population be-comes quasi‐extinct conditioned on quasi‐extinction We im-plemented the IBM using the RNetLogo package version 10ndash2(Thiele et al 2012 Thiele 2014) as an interface for the IBMplatform NetLogo (Wilensky 1999)

Sensitivity AnalysisAn important component of demographic analysis is sensitivityanalysis which is the quantification of the sensitivity of amodelrsquos outcome to changes in the input parameters (Caswell2001 Bakker et al 2009 Thiele et al 2014) Using an IBM weperformed global sensitivity analyses for 2 (ie MPC andMVM) density‐dependent scenarios to assess how sensitivequasi‐extinction times and probabilities were to changes (anduncertainties) in the estimated demographic rates We usedLatin hypercube sampling to sample parameters from the entireparameter space (Marino et al 2008 Thiele et al 2014) Wesampled intercept and slope for kitten survival (on logit scale)subadult and adult survival (on logit scale with log‐hazard effectsizes) and probability of reproduction (on complementary

logndashlog scale) from normal distributions We sampled theprobability distribution for the number of kittens producedannually (on the real scale) from a Dirichlet distribution themultivariate generalization of the beta distribution (Kotz et al2000) We sampled 1000 different parameter sets and ran 1000simulations for each scenario We calculated the probability ofquasi‐extinction within 200 years for each density‐dependencescenario (MPC or MVM) using a quasi‐extinction threshold of10 individuals and then calculated the partial rank correlationcoefficient between each of those and each parameter trans-formed to the real scale (Bakker et al 2009 Hostetler et al2013) Partial rank correlation coefficients quantify the sensi-tivity of the probability of quasi‐extinction to input variables(ie survival and reproductive parameters) with higher absolutevalues indicating stronger influence of that variable on quasi‐extinction probabilities

Genetic ManagementOne of our goals was to estimate the benefits (improvements ingenetic variation demographic parameters and populationgrowth and persistence parameters) and costs (financial costs ofcapture transportation quarantine release and monitoring ofintroduced panthers) of genetic management To achieve thisobjective we extended the IBM to include a genetic componentTo simulate individual‐ and population‐level heterozygosity overtime we assigned each panther alleles for 16 microsatellite locibased on the empirical allele frequency in the population (esti-mated from genetic samples collected during 2008ndash2015) whenwe initialized the model (Pierson et al 2015) At each time stepwe simulated Mendelian inheritance for kittens produced suchthat each kitten randomly inherited alleles from its mother andfrom a male panther randomly chosen to have putatively siredthe litter We did not explicitly force inbreeding to occur butthe probability of inbreeding naturally increased as the popula-tion size decreasedIntrogression strategiesmdashWe modeled genetic introgression as a

result of the release of 2‐year‐old female pumas from Texas intothe simulated Florida panther population For the geneticintrogression implemented in 1995 Texas females between 15and 25 years of age were released because females at that agehave lower mortality rates (Benson et al 2011) higherreproductive potential and also because it is much easier todetect with certainty the successful reproduction by females thanby males The ability to more closely monitor reproduction andthe birth of kittens is an important consideration in reducing theprobability of a genomic sweep of Texas alleles into the Floridapanther population Wildlife managers removed the last 2surviving Texas females from the wild population in Florida in2003 to reduce the possibility of genomic sweep To simulatethis strategy in the model we removed any Texas females thatwere still alive 5 years after each introgression event Weassigned Texas females alleles based on the allele frequency ofthe Texas population (based on genotypes from 48 samplescollected in west Texas that was inclusive of the pumas releasedin South Florida in 1995) when they entered the Florida pantherpopulation We could not estimate demographic rates separatelyfor the released Texas females because of small sample size (only8 females were released in 1995) therefore we assumed that

van de Kerk et al bull Florida Panther Population and Genetic Viability 15

survival and reproductive rates and the effects of covariates onthese rates for the released Texas females were identical to thoseof female Florida panthersThe impact of genetic introgression depends on the frequency

of introgression events and the number of individuals introducedat each attempt Thus we defined introgression strategies basedon the number of Texas females to be released (0 5 10 or 15)and the interval between genetic introgressions (10 20 40 and80 yr) for a total of 13 strategies We simulated each manage-ment strategy with density dependence estimated using theMPC and MVM abundance indices

We sampled 1000 parameter sets and for each of these setswe ran all introgression strategies 1000 times for 200 years Weapplied the same parametric uncertainty and environmentalstochasticity across all management scenarios to allow for bettercomparison of introgression strategies At each time step werecorded the projected population size and average individualheterozygosity We calculated the heterozygosity for each in-dividual at birth based on the alleles it inherited This individualheterozygosity then informed the survival rate of that individualbased on the empirically estimated relationship between in-dividual heterozygosity and age‐specific survival rates Hetero-zygosity did not change throughout an individualrsquos lifetime butits effect on survival changed as individuals aged because theeffect of individual heterozygosity on survival rate differedamong age classes Survival rates of genetically sampled kittenswere biased high because some kittens were sampled later in lifewhen they had already survived for some time To correct forthis bias we adjusted kitten survival rates predicted by theheterozygosity model proportionally From the population tra-jectories we estimated the probability of quasi‐extinction (de-fined as the probability that the population will fall below 10 or30 individuals or that individuals of only 1 sex will remain)

Cost‐benefit analysismdashExperts estimated financial costs ofintrogression based on their experience with the geneticintrogression executed in 1995 and we adjusted estimates tocurrent costs to account for inflation (R T McBride RancherrsquosSupply Incorporated M Cunningham and D P OnoratoFlorida Fish and Wildlife Conservation Commission personalcommunication) Costs included capturing and transportingfemale pumas from the Texas source population caring for thepumas while they were in quarantine and post‐introgressionmonitoring To compare the financial costs of the differentscenarios we also calculated the average costs incurred over

Table 2 Sample size (n) number of alleles (Na) allelic richness (AR) observedheterozygosity (Ho) and expected heterozygosity (He) at 16 microsatellite loci inradio‐collared Florida panthers sampled from 1981 to 2013 in South FloridaUSA We calculated these values from a subset (n= 137) of the samples used inthe analyses and they include only adult and subadult panthers We excludedkittens because they can result in an overrepresentation of certain alleles

Locus n Na AR Ho He

FCA090 129 5 50 0333 0401FCA133 136 3 30 0618 0608FCA243 136 5 50 0419 0487F124 137 7 69 0708 0729F37 129 4 40 0395 0397FCA075 131 5 50 0534 0598FCA559 133 5 50 0496 0503FCA057 134 6 59 0679 0624FCA081 134 7 69 0209 0206FCA566 130 6 60 0462 0475F42 133 10 100 0737 0766FCA043 136 5 49 0596 0588FCA161 132 6 60 0500 0515FCA293 132 4 40 0614 0624FCA369 131 6 60 0573 0567FCA668 133 7 70 0406 0462Mean 1329 57 57 0517 0535

Figure 4 Proportional membership (q) of radio‐collared Florida panthers (n= 218) and pumas from Texas USA (n= 7) in 2 clusters identified by STRUCTUREEach individual animal is represented by a separate vertical bar Yellow indicates canonical panther ancestry and blue represents admixed ancestry The admixedancestry of the Texas females and F1 generation panthers that were radiocollared are highlighted red and green respectively to demarcate those groups The pre‐introgression period is inclusive of panthers radiocollared from 10 February 1981 to 6 March 1996 The post‐introgression era inclusive of the F1 panthers includesanimals radiocollared from 4 March 1997 to 18 February 2015 in South Florida USA

16 Wildlife Monographs bull 203

100 years assuming that the population would be managedaccording to a particular strategy Our goal with theseexploratory analyses was to evaluate the relative costs andbenefits of alternative management scenarios

RESULTS

Genetic VariationWe assessed genetic variation at 16 microsatellite loci for 137DNA samples collected from adult and subadult panthers(1981ndash2013) that were captured and subsequently radiocollaredThe number of alleles at these loci ranged from 3ndash10 and allelicrichness averaged 57 (Table 2) Average expected hetero-zygosity for this sample was 0535 and average observed het-erozygosity was 0517 (Table 2)We determined individual heterozygosity (1 minusHL) and an-

cestry for the complete sample of panther genotypes (n= 503)collected from 1981 to 2015 for use as covariates in subsequentanalyses Average individual heterozygosity was 0559 andranged from 0ndash0980 Our STRUCTURE analysis providedstrong support for K= 2 clusters based on the probability of thedata (log Pr(D)|K) and ΔK in STRUCTURE HARVESTERBased on this result we categorized the ancestry of each pantheras either canonical (n= 123) or admixed (n= 380) using valuesof proportional membership (q Fig 4) The average individualheterozygosity of canonical panthers was 0386plusmn 0012 (SE) foradmixed panthers it was 0615plusmn 0007

Survival and Cause‐Specific Mortality RatesOf the 395 kittens that were PIT‐tagged and released 55 werelater encountered alive 15 were recovered dead and 85 werepart of failed litters (Table 1) We radio‐tracked 209 subadult oradult panthers for a total of 244195 panther‐days and docu-mented 126 mortalities (Table 1)Survival depended strongly on age and kittens had the lowest

overall rate (032plusmn 009 Fig 5A Table 3) The model‐averagedkitten survival was 047plusmn 009 when we included only geneti-cally sampled individuals in the analysis (Table 3) as statedabove this estimate is biased high because many kittenswere genetically sampled when they were recaptured alive orrecovered dead at older ages We found no evidence for sex‐specific differences in kitten survival but females survived betterthan males in all older age classes For females subadults hadthe highest survival rates followed by prime adults and oldadults For males prime adults had the highest survival ratefollowed by subadults and old adults (Fig 5A and Table 3)For panthers of all ages there was substantial evidence that

ancestry affected survival with F1 admixed panthers of all ageshaving the highest rates and canonical individuals having thelowest (Fig 5B) The combined AIC weights of models thatincluded an ancestry covariate were 0880 for kittens and 0992for older panthers The survival rates of subadult prime‐adultand old‐adult panthers in the period immediately after the in-trogression (1995ndash2004) were similar to those in more recentyears (2005ndash2013 Fig 5C)After genetic introgression individual panther heterozygosity

increased (Fig 6) and varied among ancestry categories individualheterozygosity was the highest for F1 panthers (078plusmn 001)

A

B

C

Figure 5 Overall age‐ and sex‐specific survival rates (plusmnSE A) age‐ and sex‐specific survival rates (plusmnSE) for Florida panthers of canonical F1 admixed andother admixed ancestry (B) and age‐ and sex‐specific survival estimates (plusmnSE)for subadult and adult Florida panthers in the period immediately post‐introgression (1995ndash2004) and more recently (2005ndash2013 C) in SouthFlorida USA

Table 3 Model‐averaged estimates (plusmnSE) of demographic parameters for theFlorida panther obtained using data collected during 1981ndash2013 in SouthFlorida USA

Parameter Sex Age class Estimate

Survival Both Kitten 032plusmn 009a

Female Subadult 097plusmn 002Prime adult 086+ 003Old adult 078plusmn 009

Male Subadult 066plusmn 006Prime adult 077plusmn 005Old adult 065plusmn 010

Probability of reproduction Female Subadult 035plusmn 008Prime adult 050plusmn 005Old adult 025plusmn 006

Kittens produced annually Female Subadult 280plusmn 005Prime adult 267plusmn 001Old adult 228plusmn 013

a Overall kitten survival (based on all kittens in our data set including thosethat were not genetically sampled) Estimate of survival of geneticallysampled kittens was 047plusmn 009 this estimate is biased high because manykittens were genetically sampled when they were recaptured alive or re-covered dead at older ages

van de Kerk et al bull Florida Panther Population and Genetic Viability 17

followed by those for non‐F1 admixed (062plusmn 001) and canonical(039plusmn 001) panthers Individual heterozygosity positively affectedsurvival rates of kittens (slope parameter η= 1455 95CI= 0505ndash2405) Individual heterozygosity also positively af-fected survival of subadult and adult panthers but the evidence forthis effect was weaker because the confidence interval for the ha-zard ratio overlapped 10 (hazard ratio exp(ɩ)= 0311 95CI= 0092ndash1054 Table 4 and Fig 7)The MPC increased after genetic introgression (Fig 2) This

abundance index was also included in top‐ranking modelsindicating that population density affected survival (Table 4)Abundance reduced the survival of kittens (slope parameterη= minus0035 95 CI= minus0049 to minus0021) evidence was weak forthe effect of abundance on survival of subadult and adult panthers(hazard ratio exp(ɩ)= 1002 95 CI= 0995ndash1010 Fig 7) Re-gression coefficients associated with the effect of abundance andindividual heterozygosity on demographic parameters are presentedin Table B1 (available online in Supporting Information)The most frequently observed causes of death for radio‐

collared panthers were vehicle collision and intraspecific ag-gression Cause‐specific mortality analyses revealed thatmortality rates among possible causes were generally similar forthe 2 sexes but that males were more likely to die from in-traspecific aggression than were females (Fig 8A) Male mor-tality from intraspecific aggression was higher for subadult andold adult panthers than for prime adults (Fig 8B) For bothmales and females canonical panthers were more likely to diefrom intraspecific aggression than were admixed panthers(Fig 8C)

Reproductive ParametersOf 101 radio‐collared females 67 reproduced at least once duringour monitoring we used these individuals to assess the annualprobability of reproduction The top‐ranked model for the annual

probability of reproduction included age effect only suggesting thatadding ancestry or abundance did not improve model parsimony(Table 4) Model‐averaged annual probabilities of reproductiondiffered between female subadults (035plusmn 008) prime adults(050plusmn 005) and older adults (025plusmn 006)The most parsimonious model for the number of kittens

produced annually by females that produced at least 1 litter (ieconditional on reproduction) included effects of age ancestryand abundance a model that included only the effect of age wasalmost equally supported (Table 4) The model‐averagednumber of kittens produced annually conditional on re-production was 280plusmn 075 for subadults 267plusmn 043 for primeadults and 228plusmn 083 for old adults Confidence intervals forthe slope parameter (β) overlapped 0 for both abundance(β= 0010 95 CI= minus0001 to 0021) and ancestry (β= minus073795 CI= minus1637 to 0162)

Population Dynamics and PersistenceDeterministic and stochastic demographymdashThe deterministic (λ)

and stochastic (λs) annual population growth rate calculated usingthe matrix population model were 106 (5th and 95th percentiles099ndash114) and 104 (072ndash141) respectively The generation timewas 47 years A female starting in age class one (kitten) wasexpected to live another 58 years and produce 10 kittens onaverage during her lifetime Overall λs was proportionately(elasticity) most sensitive to changes in female prime adultsurvival on the absolute scale (sensitivity) however it was mostsensitive to changes in kitten survival (Fig 9) Populationtrajectories probabilities of quasi‐extinction and times untilquasi‐extinction estimated using the matrix population modeland IBM for comparable scenarios were nearly identical for acritical quasi‐extinction threshold of 10 panthers (Figs 10 and 11)and for a critical quasi‐extinction threshold of 30 panthers(Appendix E available online in Supporting Information)Minimum population count scenariomdashUnder the MPC scenario

(ie density dependence estimated using the minimumpopulation count data) with a critical threshold of 10panthers the population size reached 99 (72ndash104) and 100(77ndash105) at 200 years (Fig 10) as predicted by the IBM and thematrix model respectively The cumulative quasi‐extinctionprobabilities within 100 years based on the MPC scenario were15 (IBM 0ndash48) and 30 (matrix model 0ndash76) Theseprobabilities increased to 25 (IBM 0ndash114) and 38 (matrixmodel 0ndash154) by year 200 (Fig 10) The mean times untilquasi‐extinction were 15 years (IBM 0ndash67) and 15 years (matrixmodel 0ndash72) conditioned on going quasi‐extinct within 100years Expected times until quasi‐extinction conditioned onquasi‐extinction within 200 years were higher 34 years (IBM0ndash137) and 32 years (matrix model 0ndash134 Fig 10)Motor vehicle mortality scenariomdashUnder the MVM scenario

(ie density dependence estimated using estimates of pantherpopulation size from McClintock et al 2015) with a criticalthreshold of 10 panthers the projected population reached 188(142ndash217) and 190 (143ndash219) at 200 years as predicted by theIBM and the matrix model respectively (Fig 11) Thecumulative quasi‐extinction probabilities within 100 years were14 (IBM 0ndash08) and 13 (matrix model 0ndash06) Theseprobabilities increased to 20 (IBM 0ndash17) and 19 (matrix

000

025

050

075

100

1995 2000 2005 2010Year of birth

Indi

vidu

al h

eter

ozyg

osity

Figure 6 Temporal changes in the individual heterozygosity of Floridapanthers born during the period 1995ndash2013 in South Florida USA Pointsare measurements solid line is linear regression shaded area is 95 confidenceinterval of slope Slope= 0006plusmn 0001 R2= 009

18 Wildlife Monographs bull 203

model 0ndash16) within year 200 (Fig 11) The mean times until quasi‐extinction based on the MVM scenario were 5 years (IBM 0ndash61)and 6 years (matrix model 0ndash64) conditioned on going quasi‐extinctwithin 100 years Expected times until quasi‐extinction conditionedon quasi‐extinction within 200 years were 11 years (IBM 0ndash113)and 11 years (matrix model 0ndash112 Fig 11)

Sensitivity of extinction probability to demographic parametersmdashThe partial rank correlation coefficients between quasi‐extinctionprobabilities and demographic parameters were negative anincrease in any of the demographic parameters decreased thequasi‐extinction probability Probabilities of quasi‐extinction within200 years were most sensitive to changes in kitten survival for bothscenarios (Fig 12) followed by survival of subadult females in theMVM scenarios and survival of prime adult females in the MPCscenario The probability of reproduction of prime adult femaleshad the third‐largest partial rank correlation coefficient with quasi‐extinction probability for both scenarios

Benefits and Costs of Genetic ManagementMinimum population count scenariomdashThe MPC scenario with

a critical threshold of 10 panthers predicted that withoutmanagement intervention average individual heterozygositywould decrease reaching 057 (049ndash064) at 100 years and053 (042ndash062) at 200 years (Fig 13) The population woulddecrease slightly reaching an average of 79 (41ndash99) at 100 yearsand 76 (43ndash98) at 200 years (Fig 14) The probabilities of quasi‐extinction under the no‐intervention MPC scenario when weconsidered the effects of genetic erosion were 13 (0ndash99) at 100years and 23 (0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with

the translocation of 5 Texas females average individualheterozygosity under the MPC scenario increased and thenstabilized at 068 (064ndash072) at 100 years and remained atthat level at 200 years (Fig 13) The population reached anaverage of 88 (62ndash104) at 100 years and stayed the same at200 years (Fig 14) The probabilities of quasi‐extinctionunder this introgression strategy were 6 (0ndash53) within 100years and 7 (0ndash82) within 200 years (Fig 15) Results ofanalyses using a critical threshold of 30 panthers revealed asimilar pattern of relative benefits However the prob-abilities of quasi‐extinction were substantially higher (ge45)across all scenarios (Appendix E)

Motor vehicle mortality scenariomdashWithout managementintervention the average individual heterozygosity predictedby the MVM scenario and a critical threshold of 10 panthersdecreased to 060 (046ndash069) at 100 years and to 059 (039ndash070) at 200 years (Fig 13) The population increased slightlyand reached an equilibrium at 154 individuals (46ndash217) at 100years and 157 (57ndash218) at 200 years (Fig 14) The probabilitiesof quasi‐extinction were 17 (0ndash100) at 100 years and 22(0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with 5

female pumas from Texas average individual heterozygosityincreased slightly reaching 067 (060ndash073) at 100 years and068 (059ndash075) at 200 years (Fig 13) Under this strategy thepopulation reached an average of 164 individuals (44ndash226) at

Table 4 Model comparison results testing for the effects of several covariateson survival of all kittens survival of genetically sampled kittens survival ofsubadult and adult Florida panthers probability of reproduction and kittensproduced annually for Florida panthers sampled from 1981ndash2013 in SouthFlorida USA

Model Parameters ΔAICa Weight

Kitten survival (all data)Base+ F1b+ abundancec 10 000 0988Base+ abundance 9 1018 0006Base+ F1 9 1035 0005Base 8 2711 lt0001Base+ sexd 9 2912 lt0001Base+ litter sizee 9 2914 lt0001

Kitten survival (genetically sampled kittens only)Base+ F1+ abundance 10 000 0541Base+ F1CanAdmf+ abundance 11 099 0329Base+Hetg+ abundance 10 310 0115Base+CanAdmh+ abundance 10 791 0010Base+ abundance 9 944 0005Base+ F1 9 1638 lt0001Base+ F1CanAdm 10 1837 lt0001Base+Het 9 2872 lt0001Base 8 3296 lt0001Base+CanAdm 9 3348 lt0001

Subadult and adult survivalBase+ F1CanAdm+ abundance 7 000 0360Base+ F1 5 053 0276Base+ F1CanAdm 6 105 0213Base+ F1+ abundance 6 222 0119Base+CanAdm+ abundance 6 575 0020Base+CanAdm 5 908 0004Base+Het 5 964 0003Base+Het+ abundance 6 984 0003Base 4 1118 0001Base+ abundance 5 1278 0001

Probability of reproductionAge 3 000 0242Age+CanAdm 4 012 0228Age+ abundance 4 173 0102Age+ F1 4 190 0094Age+CanAdm+ abundance 5 216 0082Age+Het 4 219 0081Age+ F1CanAdm 5 232 0076Age+ F1+ abundance 5 372 0038Age+Het+ abundance 5 399 0033Age+ F1CanAdm+ abundance 6 444 0026

Kittens produced annuallyAge+ F1+ abundance 5 000 0194Age 3 060 0144Age+ abundance 4 061 0143Age+ F1 4 097 0120Age+CanAdm 4 139 0097Age+ F1CanAdm+ abundance 6 196 0073Age+ F1CanAdm 5 231 0061Age+CanAdm+ abundance 5 238 0059Age+Het+ abundance 5 250 0056Age+Het 4 259 0053

a Akaikersquos information criterion (AIC) for kitten survival models was adjustedfor small sample size and overdispersion (QAICc)

b F1 divides Florida panthers into 2 groups F1 admixed and all other panthersc Index of Florida panther abundanced Sex of an individual Florida panther (male or female)e Size of the litter in which a Florida panther kitten was bornf F1CanAdm divides panthers into 3 groups F1 admixed canonical andother admixed panthers

g Het is the individual heterozygosityh CanAdm divides panthers into 2 groups canonical and admixed (includingF1 admixed) panthers

van de Kerk et al bull Florida Panther Population and Genetic Viability 19

100 years and 170 (67ndash231) at 200 years based on the MVMscenario (Fig 14) The probabilities of quasi‐extinction underthis introgression strategy were 10 (0ndash99) at 100 years and12 (0ndash99) at 200 years (Fig 15)The projected population size and average individual hetero-

zygosity reached equilibrium levels with periodic fluctuations inthese values when introgression was implemented (Figs 16ndash17)Genetic introgression decreased the time until quasi‐extinctionacross all scenarios (Fig 18) Results of analyses using a criticalthreshold of 30 panthers revealed a similar pattern of relative ben-efits However the probabilities of quasi‐extinction were sub-stantially higher (ge45) across all scenarios (Appendix E)

Addition of pumas from Texas increased both the populationsize and average individual heterozygosity consequentlyintrogression caused periodic increases in average individualheterozygosity and population size at the interval at which theintrogression occurred Whereas average individual hetero-zygosity gradually decreased following introgression densitydependence caused the population size to fluctuate especiallywhen a larger number of pumas from Texas were released Thepositive effect of genetic introgression initiatives on quasi‐extinction probabilities decreased as the time interval betweenthese events increased More frequent genetic introgressions ledto greater increases in average individual heterozygosity and

Old adult

Prime adult

Subadult

Kitten

025 050 075

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Individual heterozygosity

Ann

ual s

urvi

val r

ate

Old adult

Prime adult

Subadult

Kitten

40 60 80 100 120

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Abundance index

Ann

ual s

urvi

val r

ate

Kittens Males Females

Figure 7 The effect of individual heterozygosity (left) and the abundance index (right) on Florida panther age‐ and sex‐specific survival estimates (shaded area isSE) in South Florida USA 1981ndash2013 Abundance index data are from McBride et al (2008) and R T McBride and C McBride (Rancherrsquos Supply Incorporatedunpublished report)

20 Wildlife Monographs bull 203

equilibrium population size while reducing the probability ofquasi‐extinction Comparatively performing genetic introgres-sion less often but with more individuals per release was lesseffective at improving the long‐term prospects for the pantherpopulation (Figs 13ndash15)

Financial cost and persistence benefitsmdashThe total estimated costof 1 genetic introgression was $40000 $80000 or $120000for releasing 5 10 or 15 pumas from Texas respectively(Table 5) The total cost incurred over 100 years for eachstrategy ranged from $50000 to $12 million (Table 6)The most expensive strategy involved the introduction of15 pumas every 10 years this strategy reduced the probability

of quasi‐extinction by 58 and 40 under the MPC and MVMscenarios respectively (Table 6) Introducing 5 pumas every80 years cost the least but improvements in populationpersistence under this strategy were insubstantial (Table 6)Releasing more panthers more frequently caused increasedpopulation fluctuations but did not necessarily reduce the quasi‐extinction probability (Figs 14 and 15 Table 6)

DISCUSSION

The duration (34 yr of data) and sample sizes associated with theFlorida Panther Project allowed us to obtain a comprehensiveunderstanding of genetic variation demographic parameters

A

B

C

Figure 8 Cause‐specific mortality rates (plusmnSE) for male and female Florida panthers (A) subadult prime‐adult and old‐adult Florida panthers of both sexes (B)and canonical and admixed Florida panthers of both sexes (C) in South Florida USA 1981ndash2013 Causes of mortality are vehicle collision intraspecific aggression(ISA) other causes (known causes such as diseases injuries and infections unrelated to the first 2 causes) and unknown cause (mortalities for which we could notassign a cause)

van de Kerk et al bull Florida Panther Population and Genetic Viability 21

probability of population persistence and the duration of thepositive impacts of genetic restoration on this endangered po-pulation The Florida panther population was in dire straits inthe early 1990s and our results clearly demonstrated that the

implementation of the introgression project in 1995 subse-quently accrued a series of genetic and demographic improve-ments that have led to the change from a declining population toone that is growing and expanding its current breeding range

Sensitivity Elasticity

0 1 2 3 00 02 04

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

Para

met

er

Figure 9 Sensitivity and elasticity (proportional sensitivity) of stochastic population growth rate (λs) of the Florida panther to vital demographic parameters in SouthFlorida USA 1981ndash2013 Horizontal lines represent 5th and 95th percentiles

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

70

80

90

100

110

0

5

10

15

0

50

100

Years

StatisticMeanMedian

Figure 10 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the minimum population count(MPC) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

125

150

175

200

00

05

10

15

20

0

30

60

90

Years

StatisticMeanMedian

Figure 11 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the motor vehicle mortality(MVM) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

22 Wildlife Monographs bull 203

MPC MVM

minus08 minus06 minus04 minus02 00 minus08 minus06 minus04 minus02 00

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Male subadult survival

Male prime adult survival

Male old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

PRCC

Para

met

er

Figure 12 Partial rank correlation coefficient (PRCC) between quasi‐extinction probabilities and the demographic parameters of the Florida panther for theminimum population count (MPC) and motor vehicle mortality (MVM) scenarios Error bars indicate standard deviation Based on data collected in South FloridaUSA 1981ndash2013

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 13 Average projected individual heterozygosities in the Florida panther population over 200 years without genetic management intervention and with eachof the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include theintroduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years Average individual heterozygosity peaks when pumas are added to the Floridapanther population Based on data collected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 23

Our research has further validated the success of genetic in-trogression by quantifying the reduction in the probability ofextinction of the panther population because of this manage-ment initiative These findings all bode well for continuedprogress toward recovery That said the Florida panther po-pulation remains small and isolated from other populations ofpuma from western North America thereby denoting that theeventual impact of genetic drift and inbreeding may negativelyaffect the population in the future Realizing this will continueto be an issue that managers will have to contemplate wemodeled the impact of varied genetic management scenarios todetermine the frequency of introgression along with the numberof panthers that should be released during each initiative tominimize cost and maximize benefits to the population Theseresults will prove invaluable to managers attempting to conservethe Florida panther

Genetic Variation Demographic Parametersand Persistence of Heterotic Benefits from GeneticIntrogressionOur measure of individual heterozygosity (1 minusHL) was higheron average for the admixed (0615) panthers than for those ofcanonical ancestry (0386) Levels of individual heterozygosityfor admixed panthers were comparable to levels observed in asample of pumas from west Texas (x plusmn SE= 0695plusmn 0019n= 41) which further demonstrates the improvement in levelsof genetic variation in the Florida population since 1995 The

correlation between HL and several demographic parametershas been noted in wild populations including cheetahs (Terrellet al 2016) and several species of birds such as European shags(Phalacrocorax aristotelis male and female survival probabilityVelando et al 2015) and Egyptian vulture (Neophron percnop-terus age at recruitment Agudo et al 2012) If we focus only onpanthers captured and radiocollared from 2012ndash2015 (FP211ndashFP240) all of which had estimated birth years ge2005 (ge10 yrpost‐introgression) those panthers had an average individualheterozygosity level of 0618plusmn 0029 highlighting the elevatedlevels of genetic variation that continue to persist in cohorts ofpanthers 5 generations after the release of female pumas fromTexasThe proportional membership values (q) from our

STRUCTURE analysis allowed us to delineate 2 clusters ofancestry for Florida panthers (Fig 4) During the pre‐in-trogression era the population was dominated by panthers thatretained a high percentage of the canonical ancestry which wasaffected by inbreeding depression The post‐introgression eraancestry values are comprised of many admixed individuals inessence demonstrating the success of the introgression in termsof increasing genetic variation in the population and mi-micking gene flow that historically occurred with panthers andother populations of pumas (Onorato et al 2010) A somewhatanalogous situation although abetted by natural immigrationwas evident in the ancestral variation in a small population ofpuma intersected by a major freeway in California USA In

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

75

80

85

90

95

100

130

150

170

190

Years

Popu

latio

n si

ze

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 14 Average projected population sizes in the Florida panther population over 200 years without intervention and with each of the genetic introgressionstrategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include the introduction of 5 10 or 15pumas from Texas USA every 10 20 40 or 80 years Peaks occur when pumas are added to the Florida panther population Based on data collected in SouthFlorida USA 1981ndash2013

24 Wildlife Monographs bull 203

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

25

50

75

100

0

25

50

75

100

Introgression interval (yrs)

PQ

E (

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 15 Average probability of quasi‐extinction (PQE) of the Florida panther population within 100 years and within 200 years without genetic managementintervention (N) and with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenariosIntrogression strategies include the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Errorbars indicate 5th and 95th percentiles Based on data collected in South Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

50

60

70

80

90

100

110

50

100

150

200

Years

Popu

latio

n si

ze

Figure 16 Average projected Florida panther population trajectories under a geneticintrogression strategy involving the release of 5 female pumas from Texas USA every20 years for the minimum population count (MPC) and motor vehicle mortality(MVM) scenarios Shaded area indicates 5th and 95th percentiles and isrepresentative of confidence intervals for other scenarios Based on data collected inSouth Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Figure 17 Average projected Florida panther population‐level heterozygositiesunder a genetic introgression strategy involving the release of 5 female pumas fromTexas USA every 20 years for the minimum population count (MPC) and motorvehicle mortality (MVM) scenarios Shaded area bounds the 5th and 95th percentilesand is representative of confidence intervals for other scenarios Based on datacollected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 25

this case the migration of just a single male across the freewayto the south resulted in an increase in the number of in-dividuals with admixed ancestry and substantially improvedgenetic diversity of the subpopulation south of the freeway(Riley et al 2014)Our estimate of overall kitten survival (0321plusmn 0056) was

similar to that reported by Hostetler et al (2010) who used13 years of post‐introgression data (0323plusmn 0071) These valuesare lower than kitten survival estimates derived from severalpuma populations in western North America (044ndash091 Loganand Sweanor 2001 Lambert et al 2006 Laundre et al 2007Robinson et al 2008 Ruth et al 2011) When we included onlydata for individuals that had been genetically sampled our es-timate was 15 higher The proportion of admixed kittens thatwere genetically sampled increased as the population expandedwhich could have resulted in higher estimates of kitten survivalbecause admixed kittens survive better than canonical kittensKitten survival was strongly influenced by genetic ancestry withsurvival rates of F1 kittens being almost twice that of canonicalkittens (Fig 5B) Consistent with the findings of Hostetler et al(2010) increases in heterozygosity coincided with increasedkitten survival whereas population density negatively affectedkitten survival Population density often has a strong negativeimpact on survival of young age classes due to infanticide andcannibalism which has been reported from several carnivore

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

50

100

150

0

50

100

150

Introgression interval (yrs)

TQE

(yrs

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 18 Average times until quasi‐extinction (TQE) for the Florida panther population within 100 years and within 200 years without genetic management interventionand with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategiesinclude the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Error bars indicate 5th and 95thpercentiles Based on data collected in South Florida USA 1981ndash2013

Table 5 Average cost (2017 US dollars) of introgression strategies employedover 100 years that involve the introduction of 5 10 or 15 pumas from TexasUSA into the Florida panther population in South Florida USA at an in-creasingly higher frequency Costs of capture (including transportation) quar-antine and post‐introgression monitoring were estimated by Roy McBrideMark Cunningham and Dave Onorato respectively

Frequency Component 5 pumas 10 pumas 15 pumas

Once Capture 25000 50000 75000

Quarantine 5000 10000 15000

Monitoring 10000 20000 30000

Total 40000 80000 120000

Every 80 years Capture 31250 62500 93750

Quarantine 6250 12500 18750

Monitoring 12500 25000 37500

Total 50000 100000 150000

Every 40 years Capture 62500 125000 187500

Quarantine 12500 25000 37500

Monitoring 25000 50000 75000

Total 100000 200000 300000

Every 20 years Capture 125000 250000 375000

Quarantine 25000 50000 75000

Monitoring 50000 100000 150000

Total 200000 400000 600000

Every 10 years Capture 250000 500000 750000

Quarantine 50000 100000 150000

Monitoring 100000 200000 300000

Total 400000 800000 1200000

26 Wildlife Monographs bull 203

species including polar bears (Ursus maritimus Derocher andWiig 1999) black bears (Ursus americanus Garrison et al 2007)and pumas (Logan and Sweanor 2001)Our estimates of sex‐ and age‐specific survival rates of subadult

and adult panthers (Table 3) and the effects of covariates on theserates were similar to those reported by Benson et al (2011) derivedfrom data collected through 2006 Our survival rates for males(065ndash077) were lower than females (078ndash097) for all 3 ageclasses (subadult prime adult and old adult) which is the typicaltrend observed in most puma populations (eg Ruth et al 19982011) However an unhunted puma population occupying afragmented urban landscape in southern California exhibited lowersurvival (0586 females 0525 males Vickers et al 2015) perhapsas a result of severe habitat fragmentation and the lack of extensiveswaths of public land protected in perpetuity Most populations ofpuma in western North America are typically subject to variedlevels of management via regulated hunting which often is biasedtoward the take of males (Cooley et al 2009 Ruth et al 2011)Consequently annual survival estimates from hunted populationsin northeast Washington (0679ndash0733 females 0341 malesRobinson et al 2008) and southeastern Arizona (0685 females0584 males Cunningham et al 2001) were generally lower thanthose we estimated for Florida panthersCause‐specific mortality analysis showed that male panthers

experienced a substantially greater mortality risk from in-traspecific aggression This differs from the pattern observedduring a long‐term study in Arizona where females were at ahigher risk of intraspecific mortality than males (46 of maleand 53 of female mortality Logan and Sweanor 2001)Mortality associated with intraspecific strife has been docu-mented in hunted and unhunted populations (this study Loganand Sweanor 2001 Stoner et al 2010) and its root cause hasbeen difficult to determine (eg population density and preypopulation trends) That being the case finding ways to reducewhat is often a major source of mortality for puma populationscontinues to pose a challenge to managersCanonical panthers were substantially more likely to die from

intraspecific aggression than were admixed panthers This mayat least partly explain the difference in survival between admixed

and canonical panthers Canonical panthers are known to bemore likely to suffer from varied correlates of inbreeding de-pression than admixed animals including atrial septal defects(Johnson et al 2010) and compromised immune systems(Roelke et al 1993) These factors have the potential to affectfitness and perhaps dominance of individuals when engaged inan intraspecific encounter A somewhat similar scenario wasobserved by Hogg et al (2006) in a population of bighorn sheepthat were part of an introgression project Admixed males in thispopulation had a greater probability of paternity than males thatwere less outbred This included coursing males with higherlevels of admixture being markedly more successful at fightingmales that were defending females in estrous (Hogg et al 2006)Heterosis resulting from genetic introgression is expected to be

the strongest in F1 individuals (Shull 1908 Crow 1948 Burkeand Arnold 2001 Waller 2015) and to progressively decline insubsequent generations (Pickup et al 2013 Frankham 2015)Given that admixed (especially F1) panthers survived better thancanonical panthers and that individual heterozygosity improvedsurvival of both sexes and all age classes our data provide evi-dence for heterotic advantage whereby individuals of mixedancestry had higher fitness (Shull 1908 Crow 1948) Our resultsare consistent with findings of other studies that involved in-tentional genetic introgression for conservation purposes Forexample Hogg et al (2006) detected substantial improvementsin survival and reproduction of introgressed bighorn sheep andMadsen et al (1999 2004) reported a dramatic increase in thepopulation of adders after genetic introgressionKnowledge of the persistence of benefits to a population ac-

crued via genetic introgression is essential for determiningwhen additional introgression may be warranted There is adepauperate amount of information regarding this topic and theidentification of factors that may influence persistence of thebenefits of genetic introgression (Tallmon et al 2004 Hedricket al 2014 Whiteley et al 2015) For example in minks(Neovison vison) Thirstrup et al (2014) found that outcrossingled to larger litters in F2 and F3 but this benefit disappeared insubsequent generations In a population of gray wolves Hedricket al (2014) suggested that benefits of genetic introgression may

Table 6 Costs and benefits accumulated over 100 years for genetic introgression at different intervals and with different numbers of pumas from Texas USAintroduced into the Florida panther population in South Florida USA Costs (2017 US dollars) include capture and transportation quarantine and post‐introgression monitoring Benefits are represented as average probability of quasi‐extinction (PQE and 5th and 95th percentiles) and changes (Δ) therein under thescenario using the minimum population count (McBride et al 2008 MPC) and the scenario using the motor vehicle mortality population estimates (McClintocket al 2015 MVM) The table is sorted by percent change in probability of quasi‐extinction under the MPC scenario

Interval (yr) Pumas Cost MPC PQE () ΔMPC PQE () MVM PQE () ΔMVM PQE ()

ndash 0 0 13 (0ndash99) 0 17 (0ndash100) 080 10 100000 12 (0ndash99) 10 (0ndash58) 16 (0ndash100) 5 (0ndash38)80 15 150000 12 (0ndash99) 13 (0ndash65) 15 (0ndash100) 8 (0ndash56)80 5 50000 12 (0ndash99) 14 (0ndash73) 15 (0ndash99) 9 (0ndash41)40 15 300000 10 (0ndash98) 26 (0ndash104) 14 (0ndash99) 13 (0ndash60)40 10 200000 9 (0ndash94) 35 (0ndash183) 13 (0ndash99) 21 (0ndash95)40 5 100000 8 (0ndash89) 39 (0ndash211) 12 (0ndash99) 26 (0ndash124)20 15 600000 8 (0ndash88) 42 (0ndash257) 13 (0ndash99) 24 (0ndash123)20 10 400000 6 (0ndash67) 54 (0ndash318) 10 (0ndash99) 37 (0ndash217)10 15 1200000 6 (0ndash53) 58 (0ndash372) 10 (0ndash99) 40 (0ndash208)20 5 200000 5 (0ndash50) 59 (0ndash338) 9 (0ndash99) 44 (0ndash352)10 10 800000 4 (0ndash16) 69 (0ndash489) 8 (0ndash92) 55 (0ndash401)10 5 400000 4 (0ndash7) 73 (0ndash502) 6 (0ndash71) 63 (0ndash503)

van de Kerk et al bull Florida Panther Population and Genetic Viability 27

wane after 2 or 3 generations The WrightndashFisher model ofgenetic drift predicts a geometric decline in expected hetero-zygosity at the rate of (1 minus 12Ne) per generation where Ne is theeffective population size (Hartl and Clark 1997 Frankham et al2014) Thus it seems logical to assume that the duration of thebenefits of introgression is limited especially in a species per-sisting in a small population such as Florida panthersWe expected Florida panther survival in the most recent years

(2005ndash2013) post‐introgression to differ from that observed in thedecade following the introgression in 1995 because pantherabundance and heterozygosity factors known to influence panthersurvival (Hostetler et al 2010 Benson et al 2011) have changed(Figs 2 and 6) We found that subadult and adult survival rates inrecent years (2005ndash2013) were similar to those in the period im-mediately following introgression (1995ndash2004 Fig 5C) overallkitten survival was similar to estimates of Hostetler et al (2010) Infact the pattern of covariate effects on Florida panther survivalreported by Benson et al (2011) and Hostetler et al (2010) hasbarely changed The negative effects of population density andpositive effects of heterozygosity may counterbalance survival ratesreducing population fluctuations Our result that benefits of geneticrestoration have remained in the population nearly for 5 genera-tions post‐intervention was surprising but also encouraging Pos-sible explanations for this observation may include 1) genetic ero-sion occurs at slower rate than previously thought 2) introducing 5migrants in 1 genetic intervention may have beneficial effects si-milar to those from 1 migrant per generation over multiple gen-erations or 3) other and as yet unknown factors may reduce therate of genetic erosion and subsequent demographic effects Adensity‐dependent effect on kitten survival could also explain whynon‐F1 admixed kittens did not survive substantially better thandid canonical kittens most kittens sampled from 2005ndash2013 wereadmixed and their survival was lower possibly because of a density‐dependent effect as the Florida panther population continuouslygrew during that period Trinkel et al (2010) also found thatpopulation density and inbreeding coefficient interacted to affectdemographic parameters and growth rate of an African lion po-pulation Likewise population density interacted with winterweather to affect the survival and population growth rates in Soaysheep (Ovis aries Milner et al 1999 Coulson et al 2001)

Population Dynamics and PersistenceThe finite deterministic and stochastic growth rates were gt10reflecting that much of the data used in this study were collectedfollowing genetic introgression in 1995 during which time thepopulation increased substantially Because our demographicparameter estimates did not change markedly in comparison tothose of Hostetler et al (2013) the similarity between thepopulation growth estimates from these studies was expectedBut these estimates reflect population growth during thedata‐collection period and do not predict future growth In factestimates of population size reported by McClintock et al(2015) suggest that population growth may already be slowing(Fig 2) A similar pattern was observed in an introduced po-pulation of African lion which increased steadily from 1995until 2001 then fluctuated around the presumed carrying ca-pacity thereafter (Trinkel et al 2010) Several other African lionpopulations that experienced various degrees of recovery

subsequently became relatively stable or exhibited decliningtrends (Bauer et al 2015)Our estimates of quasi‐extinction probabilities were lower than

those reported by Hostetler et al (2013) using a 2‐sex age‐structured matrix population model Lower probabilities ofquasi‐extinction than those reported by the earlier study forcomparable scenarios were likely a consequence of the fact thatthe estimates of abundance (McBride et al 2008) and thus thedensity‐dependent effects on survival of kittens and old panthers(gt10 yr) have changed in recent years Additionally these earlyestimates of the probability of quasi‐extinction and of time toquasi‐extinction did not consider genetic erosion that will in-evitably occur without management interventionUnder the MVM scenario the equilibrium population size was

twice that attained under the MPC scenario The MVM popula-tion estimates are approximately twice the MPCs (Fig 2) as aresult the simulated population under the MVM scenario achieveda higher equilibrium size Probability of quasi‐extinction under theMVM scenario was generally greater than that under the MPCscenario This result is a consequence of the fact that the estimateddensity dependence on kitten survival based on the MPC scenario isstronger than that for theMVM scenario (regression coefficients forabundance for the MPC scenario= minus0068 vs minus0002 for theMVM scenario Appendix B Table B1) Consequently simulatedpopulations under the MPC scenario have a stronger tendency tomove toward the equilibrium population size which inherentlyreduces the quasi‐extinction probability (eg Royama 1992)As expected for long‐lived mammals (Heppell et al 2000 Oli

and Dobson 2003) the population growth rate was proportio-nately most sensitive to changes in survival of prime adultsfollowed by that in younger age classes Global sensitivity ana-lysis in contrast showed that under all scenarios extinctionparameters were particularly sensitive to changes in survival ofFlorida panther kittens This result is a consequence of the highsensitivity of the population growth rate to kitten survival(Fig 9) and a larger standard error of kitten survival (Table 3)Results of our sensitivity analyses indicate that future researchshould focus on acquiring additional information on early lifestages to improve the accuracy and precision of estimates ofkitten survival Our results suggest that demographic variables(or their environmental drivers) that strongly influence extinc-tion risks are not necessarily those with the largest elasticityvalues A study of island fox (Bakker et al 2009) found thatextinction risk was strongly influenced by survival modifierparameters that had low elasticity values

Genetic Management When and How to InterveneThe use of genetic introgression as a management tool has al-ways been controversial and the probable effectiveness it mighthave in preventing the imminent demise of the panther popu-lation was initially debated (Creel 2006 Maehr et al 2006Pimm et al 2006) These debates notwithstanding the pantherpopulation had suffered from genetic morphological and bio-medical correlates of inbreeding before this management in-tervention (Johnson et al 2010 Onorato et al 2010) whereasthe post‐introgression period has been characterized by a sub-stantial reduction in the biomedical correlates of inbreeding andimprovements in demographic vigor and abundance (McBride

28 Wildlife Monographs bull 203

et al 2008 Hostetler et al 2010 2013 Johnson et al 2010Benson et al 2011 McClintock et al 2015) Nevertheless thepopulation remains small and isolated habitat is still being lostand there is no current possibility of natural gene flow betweenthis and other North American puma populations thus therecurrence of inbreeding‐related problems and subsequent po-pulation declines are inevitable The question therefore is notwhether genetic management of the Florida panther populationis needed but when and how it should be implementedWithout genetic introgression probabilities of quasi‐extinc-

tion were substantially greater than those from models thatincorporated periodic genetic introgression events (Fig 15)highlighting the importance of genetic management in smallisolated populations When 5 female pumas are added to theFlorida panther population every 10 years genetic variationincreased substantially under the MPC and MVM scenariosThe IBM predicted that the equillibrium population size wouldbe substantially larger when 5 pumas were released into thepopulation every 10 years than populations without intervention(ie no introgression) Releasing 15 Texas females did not af-ford substantially greater benefit to the equilibrium populationsize than did releasing only 5 Texas females Instead addingmore individuals caused increasingly large fluctuations in po-pulation size due to the numerical increases and density de-pendence (Fig 14) possibly diminishing the benefits associatedwith increased genetic variationCompared with the no‐intervention scenario (ie no genetic

introgression implemented) the introduction of 5 female pumasevery 10 years reduced quasi‐extinction probabilities from 13to 4 and from 17 to 6 for the MPC and MVM scenariosrespectively The benefit of genetic‐introgression strategies de-creased as the interval between successive management inter-ventions increased and no benefit was evident once the intervalreached 80 years (Fig 15) Introgression reduced the averagetime until quasi‐extinction (Fig 18) This somewhat counter‐intuitive result can be explained by the fact that repeated geneticintrogression makes the population more robust over timewhich will reduce the probability of extinction Consequentlyfew simulated population trajectories that fall below the criticalthreshold do so early reducing the mean time to extinctionAlso density dependence has a stabilizing effect on populations(Royama 1992) populations tend toward equilibrium popula-tion sizes which reduces the probability over time of popula-tions falling below quasi‐extinction threshold However timeuntil quasi‐extinction must be interpreted in conjunction withthe probability of quasi‐extinction which is very low for allscenarios with genetic introgression (Fig 15)Cost is a significant impediment to the implementation of a

genetic introgression program because captures relocations andmonitoring efforts before and after introgression are expensive(Edmands 2007 Frankham 2010b 2015 2016) Costs may beacceptable to society if the management actions result in sub-stantial fitness benefits especially if the species of concern ispopular and charismatic (Frankham 2015) We estimated the100‐year cost of introgression strategies modeled here to rangefrom $50000 to $12 million More expensive strategies gen-erally led to larger decreases in the probability of quasi‐extinc-tion but cheaper strategies also substantially improved

population viability (ie reduced quasi‐extinction probabilityTable 6) One of the cheaper strategies (5 pumas released every20 years) which cost an estimated $200000 over 100 yearsreduced the probability of quasi‐extinction by 59 and 44 forthe MPC and MVM scenarios respectively But these pre-liminary cost estimates are based on expenses incurred duringthe 1995 introgression and include costs of capture quarantinetransportation release and post‐release monitoring of femalesonly Although the cost for future genetic interventions wouldlikely be higher than those reported here the relative differencesin cost between alternative intervention strategies should remainapproximately the sameOur ability to simulate genetics and link it to the viability of

the Florida panther population is based on the empirically es-timated relationship between individual heterozygosity andsurvival Such heterozygosity and fitness correlations have beenstudied for decades (David 1998) but have only recently beenused to predict population viability (Benson et al 2016) noneto our knowledge have incorporated cost into their modelingefforts The mechanisms underlying heterozygosity and fitnesscorrelations remain unclear (David 1998 Szulkin et al 2010)and their significance as a proxy for inbreeding depression hasbeen debated (Balloux et al 2004 Miller and Coltman 2014)Nevertheless evidence continues to mount demonstratingpositive relationships between heterozygosity and survival(Coulson et al 1998ab Hansson et al 2004 Silva et al 2005Acevedo‐Whitehouse et al 2006 Jensen et al 2007 Velandoet al 2015) and between heterozygosity and fecundity (Seddonet al 2004 Charpentier et al 2005 Agudo et al 2012 Velandoet al 2015) Although the reported correlations are generallyweak their consequences can be substantial if population growthand persistence parameters are highly sensitive to the affectedvital rates (Velando et al 2015) Compared with scenarios thatignore genetics incorporating the effects of inbreeding on sur-vival increased quasi‐extinction probabilities within a 100‐yeartimeframe from 15 to 13 and from 14 to 17 for theMPC and MVM scenarios respectively These results high-light the importance of incorporating genetics when analyzingthe viability of small isolated populationsAlthough conservation biologists have recognized the im-

portance of genetics in population viability many still fail toincorporate genetic factors into PVA models (Reed et al 2002)Explicit consideration of genetics in PVAs requires long‐termgenetic and demographic data This permits the assessment ofthe functional relationship between genetic variability and de-mographic parameters Population viability analysis softwarepackages such as VORTEX (Lacy 1993 2000) offer a powerfulframework for individual‐based simulations but they often donot adequately capture a speciesrsquo life history or genetics Basedon a thorough review of the importance of genetic factors inwildlife conservation Frankham et al (2014) concluded thatgenetic factors can strongly influence the dynamics and persis-tence of small andor fragmented populations They furthernoted that ldquoMost PVA‐based risk assessments ignore or in-adequately model genetic factorsrdquo and recommended that ldquoPVAshould routinely include realistic inbreeding depressionhelliprdquo(Frankham et al 201457) Our custom‐built IBM permitted usto develop a model that adequately captured the Florida panther

van de Kerk et al bull Florida Panther Population and Genetic Viability 29

life history and we could parameterize the model with our long‐term genetic and demographic dataThe explicit consideration of the effects of inbreeding on

Florida panther population dynamics and persistence representsan important step forward Nonetheless our model remainsimperfect First we neglected the possible effects of habitat lossdetrimental anthropogenic activities climate change the prob-ability of immigration of pumas from western populationsdisease or catastrophes on demographic rates and populationviability Although immigration from puma populations outsideFlorida is possible its probability is very low given the extensivechallenges the anthropogenically altered landscape would posefor such a migrant to make it successfully to South Florida Weomitted mutations from our model because we have no in-formation on mutation rates though we expect that they are lowbased on values reported for other mammals (Kumar and Sub-ramanian 2002) Finally we only considered possible effects ofinbreeding on age‐specific survival rates because there was noevidence that heterozygosity affected female reproduction Lossof heterozygosity can negatively affect reproduction becauseinbred male panthers can suffer from cryptorchidism which canadversely affect their fecundity However given the polygynousmating system we expect the Florida panther population growthis primarily female‐limitedAlthough it is generally accepted that genetic introgression

only temporarily relieves a population from inbreeding depres-sion we know of no cases in which it has been implementedmore than once to conserve a threatened wildlife populationOur study provides an example of how demographic and mo-lecular data can be used within an IBM framework to evaluatethe efficacy of alternative genetic management strategies via theFlorida panther as a case study Our model can be adjusted forand applied to other species and offers great potential as a toolfor genetic management of small isolated wildlife populations

MANAGEMENT IMPLICATIONS

Our results and those of Hostetler et al (2013) and Johnsonet al (2010) provide persuasive evidence that the genetic in-tervention implemented in 1995 prevented the demise of theFlorida panther and restored demographic vigor to the popu-lation The Florida panther population remains small and iso-lated from other puma populations the closest puma populationis located gt1000 km away in Texas and the intervening land-scape is highly developed and fragmented with many interstate(and other high‐traffic) highways and major urban centersTheory suggests that 1 migrant per generation is sufficient tominimize the loss of genetic diversity (eg Mills and Allendorf1996) However the possibility of successful migration of apuma to South Florida followed by successful mating every ~5years is very low considering the distance between Texas andFlorida (Hedrick 1995) and the many risks and barriers thatdispersing pumas face (Beier 1995 Maehr et al 2002 Thatcheret al 2006) Consequently the pantherrsquos long‐term persistencewill likely depend on subsequent timely genetic introgressionsgiven the near‐impossibility of natural gene flow from otherpuma populations To that end our study offers considerableinsights into benefits of different genetic management strategiesand associated costs

Without genetic management the Florida panther populationfaces a substantial risk of quasi‐extinction within 100 years (10ndash46 depending on quasi‐extinction threshold and scenarios)Twenty‐three years have passed since genetic introgression wasimplemented and the population appears healthy as indicatedby measures of genetic variation increases in the populationsize and improvements in age‐specific survival rates Ourfindings suggest that releasing more pumas more frequently doesnot necessarily improve persistence probability because such astrategy would cause larger fluctuations in population size(Fig 14) which negatively affects persistence probability Thusextinction risks faced by inbred populations of large carnivorescan be substantially reduced by releasing approximately 5 im-migrants every 2 decades We suggest that such a managementstrategy be followed only in conjunction with continued long‐term monitoring of the population Our analyses demonstratethat population growth and persistence are highly sensitive tokitten and female (adult and subadult) survival Continuing tocollect data on these demographic variables along with bio-metric and genetic sampling will remain important to pantherrecovery and vigilance in identifying issues associated with in-breeding depression The collection of these monitoring datashould allow managers to detect any demographic or geneticchanges in a timely fashion and respond with genetic in-trogression or other management actions if warrantedRecovery objectives as defined by the USFWS in its current

recovery plan (USFWS 2008) delineate the need for additionalpopulations in the historical range to downlist or delist theFlorida panther If the only extant population of Floridapanthers continued to grow it could serve as a source ofpanthers to be translocated to suitable habitat to seed addi-tional populations Maintaining current information on thegenetic health of the population and precise estimates of itssize will be important in assessing whether it can withstand theremoval of a subset of animals for translocation Although the2017 documentation of a female panther with kittens north ofthe current breeding range is encouraging in terms of thepotential for population expansionmdashgiven that no female hadbeen recorded north of the Caloosahatchee River since 1972mdashthe re‐establishment of separate new populations will mostlikely require translocations by wildlife managers and a lengthysociopolitical processConservation of threatened or endangered species such as the

Florida panther is inherently challenging For panthers and otherlarge carnivores that require vast extents of natural habitat topersist habitat is typically the most limiting factor that will de-termine whether recovery efforts succeed Although geneticmanagement invariably plays a key role in the long‐term persis-tence of the panther population even a healthy population caneventually succumb to the impacts of habitat loss The humanpopulation of Florida continues to be one of the fastest‐growingin the nation the United States Census Bureau currently ranksFlorida as the third most populous Therefore habitat loss isgoing to continue to be a key issue for panthers Diligence towardpreserving panther habitat in its historical range via conservationeasements or other means and trying to keep such areas inter-connected via corridors is a goal stakeholders such as governmentagencies sportsmen non‐governmental organizations ranchers

30 Wildlife Monographs bull 203

and other private landowners must work on collaboratively toachieve

ACKNOWLEDGMENTS

We thank D Shindle D Jennings T OrsquoMeara D Land andC Belden for valuable advice throughout the project We thankS Bass M Criffield M Cunningham D Giardina D JansenA Johnson D Land M Lotz C McBride Rocky McBrideRowdy McBride Roy McBride L Oberhofer S Schulze staffsat Everglades National Park and Big Cypress National Preserveand others for assistance with fieldwork We also thank JAustin M Conroy J Hines J Laake S Mills J Nichols JHines M Sunquist J Fryxell and T Coulson for advice andassistance regarding data analysis and panther biology TheMuseum of Texas Tech University Natural Science ResearchLaboratory provided samples from pumas from west Texas forcomparative genetic analyses M Ben‐David D Land WMcCown E Hellgren and 4 anonymous reviewers providedmany helpful comments on the manuscript We thank NereaUbierna Lopez for completing the Spanish translation of theabstract We are grateful to the Department of ResearchComputing University of Florida for excellent high‐perfor-mance computing services We would like to thank US Fishand Wildlife Service Florida Fish and Wildlife ConservationCommission (FWC) US National Park Service QuantitativeSpatial Ecology Evolution and Environment (QSE3) In-tegrative Graduate Education and Research Traineeship(IGERT) program funded by the National Science Foundationand the University of Florida for financially supporting thisstudy Funding for panther research conducted by the FWC issupported predominantly via donations accrued with vehicleregistrations for ldquoProtect the Pantherrdquo license plates

LITERATURE CITEDAcevedo‐Whitehouse K T R Spraker E Lyons S R Melin F Gulland RL Delong and W Amos 2006 Contrasting effects of heterozygosity onsurvival and hookworm resistance in California sea lion pups MolecularEcology 151973ndash1982

Adams J R L M Vucetich P W Hedrick R O Peterson and J AVucetich 2011 Genomic sweep and potential genetic rescue during limitingenvironmental conditions in an isolated wolf population Proceedings of theRoyal Society B Biological Sciences 2783336ndash3344

Agresti A 2013 Categorical data analysis Third edition John Wiley amp SonsHoboken New Jersey USA

Agudo R M Carrete M Alcaide C Rico F Hiraldo and J A Donaacutezar2012 Genetic diversity at neutral and adaptive loci determines individualfitness in a long‐lived territorial bird Proceedings of the Royal Society BBiological Sciences 2793241ndash3249

Albeke S E N P Nibbelink and M Ben‐David 2015 Modeling behavior bycoastal river otter (Lontra canadensis) in response to prey availability in PrinceWilliam Sound Alaska a spatially‐explicit individual‐based approach PLOSOne 10e0126208

Alho J S K Vaumllimaumlki and J Merilauml 2010 Rhh an R extension for estimatingmultilocus heterozygosity and heterozygosity‐heterozygosity correlationMolecular Ecology Resources 10720ndash722

Alvarez K 1993 Twilight of the panther Myakka River Publishing SarasotaFlorida USA

Aparicio J M J Ortego and P J Cordero 2006 What should we weigh toestimate heterozygosity alleles or loci Molecular Ecology 154659ndash4665

Bakker V J D F Doak G W Roemer D K Garcelon T J Coonan S AMorrison C Lynch K Ralls and R Shaw 2009 Incorporating ecologicaldrivers and uncertainty into a demographic population viability analysis for theisland fox Ecological Monographs 7977ndash108

Balloux F W Amos and T Coulson 2004 Does heterozygosity estimateinbreeding in real populations Molecular Ecology 133021ndash3031

Barone M A M E Roelke J Howard J L Brown A E Anderson and DE Wildt 1994 Reproductive characteristics of male Florida panthersmdashcomparative studies from Florida Texas Colorado Latin‐America andNorth‐American Zoos Journal of Mammalogy 75150ndash162

Bateson Z W P O Dunn S D Hull A E Henschen J A Johnson and LA Whittingham 2014 Genetic restoration of a threatened population ofgreater prairie‐chickens Biological Conservation 17412ndash19

Bauer H G Chapron K Nowell P Henschel P Funston L T B HunterD W Macdonald and C Packer 2015 Lion (Panthera leo) populations aredeclining rapidly across Africa except in intensively managed areas Pro-ceedings of National Academy of Sciences of the United States of America11214894ndash14899

Beier P 1995 Dispersal of juvenile cougars in fragmented habitat Journal ofWildlife Management 59228ndash237

Benson J F J A Hostetler D P Onorato W E Johnson M E Roelke SJ OrsquoBrien D Jansen and M K Oli 2011 Intentional genetic introgressioninfluences survival of adults and subadults in a small inbred felid populationJournal of Animal Ecology 80958ndash967

Benson J F P J Mahoney J A Sikich L E K Serieys J P Pollinger H BErnest and S P D Riley 2016 Interactions between demography geneticsand landscape connectivity increase extinction probability for a small popu-lation of large carnivores in a major metropolitan area Proceedings of theRoyal Society B Biological Sciences 28320160957

Beverly F 1874 The Florida panther Forest and Stream 3290Boyce M S C V Haridas C T Lee and the NCEAS Stochastic Demo-graphy Working Group 2006 Demography in an increasingly variable worldTrends in Ecology amp Evolution 21141ndash148

Brook B W J J OrsquoGrady A P Chapman M A Burgman H R Akcakayaand R Frankham 2000 Predictive accuracy of population viability analysis inconservation biology Nature 404385ndash387

Brys R and H Jacquemyn 2016 Severe outbreeding and inbreeding de-pression maintain mating system differentiation in Epipactis (Orchidaceae)Journal of Evolutionary Biology 29352ndash359

Burke J M and M L Arnold 2001 Genetics and the fitness of hybridsAnnual Review of Genetics 3531ndash52

Burnham K P 1993 A theory for combined analysis of ring recovery andrecapture data Pages 199ndash213 in J‐D Lebreton and P M North editorsMarked individuals in the study of bird population Springer‐Verlag NewYork New York USA

Burnham K P and D R Anderson 2002 Model selection and multimodelinference a practical information‐theoretic approach Second edition SpringerVerlag New York New York USA

Caswell H 2001 Matrix population models construction analysis and in-terpretation Sinauer Associates Sunderland Massachusetts USA

Charpentier M J M Setchell F Prugnolle L A Knapp E J Wickings PPeignot and M Hossaert‐McKey 2005 Genetic diversity and reproductivesuccess in mandrills (Mandrillus sphinx) Proceedings of the NationalAcademy of Sciences of the United States of America 10216723ndash16728

Cooch E and G C White editors 2018 Program MARK a gentle in-troduction lthttpwwwphidotorgsoftwaremarkdocsbookgt Accessed 7Apr 2016

Cooley H S R B Wielgus G M Koehler H S Robinson and B TMaletzke 2009 Does hunting regulate cougar populations A test of thecompensatory mortality hypothesis Ecology 902913ndash2921

Coulson T E A Catchpole S D Albon B J Morgan J M Pemberton TH Clutton‐Brock M J Crawley and B T Grenfell 2001 Age sex densitywinter weather and population crashes in Soay sheep Science 2921528ndash1531

Coulson T J Pemberton S Albon M Beaumont T Marshall J Slate FGuinness and T Clutton‐Brock 1998a Microsatellites reveal heterosis in reddeer Proceedings of the Royal Society B Biological Sciences 265489ndash495

Coulson T N S D Albon J M Pemberton J Slate F E Guinness and TH Clutton‐Brock 1998b Genotype by environment interactions in wintersurvival in red deer Journal of Animal Ecology 67434ndash445

Cox D 1972 Regression models and life‐tables Journal of the Royal StatisticalSociety Series B Statistical Methodology 34187ndash220

Creel S 2006 Recovery of the Florida panthermdashgenetic rescue demographic rescueor both Response to Pimm et al (2006) Animal Conservation 9125ndash126

Crnokrak P and D A Roff 1999 Inbreeding depression in the wild Heredity83260ndash270

Crow J 1948 Alternative hypotheses of hybrid vigor Genetics 33477ndash487

van de Kerk et al bull Florida Panther Population and Genetic Viability 31

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 3: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

100 antildeos e incluyendo el impacto de la erosioacuten geneacutetica fue de 22 (0ndash75) antildeos Anaacutelisis de sensibilidad demostraronque la probabilidad de cuasi‐extincioacuten y el plazo hasta alcanzarla dependiacutean de los valores utilizados para losparaacutemetros de supervivencia de cachorros Sin manejo geneacutetico la poblacioacuten de panteras de Florida se enfrentariacutea aun aumento sustancial del riesgo de cuasi‐extincioacuten Por lo tanto la pregunta no es si es necesario el manejo geneacuteticode la poblacioacuten de las panteras de Florida sino cuaacutendo y coacutemo implementarlo Usando modelos de simulacioacutenbasados en individuos evaluamos diferentes estrategias de introgresioacuten geneacutetica y sus posibles impactos en lapoblacioacuten y en su geneacutetica La reduccioacuten del riesgo de cuasi‐extincioacuten fue similar introduciendo 5 pumas cada 20antildeos o 15 pumas cada 10 antildeos (44ndash59 vs 40ndash58) pero la primera opcioacuten resulta maacutes econoacutemica ($200000 en100 antildeos) que la segunda ($12000000 en 100 antildeos) Introducir maacutes hembras en cada intento de introgresioacuten noprodujo beneficios adicionales El impacto positivo del proyecto de introgresioacuten geneacutetica persiste en la poblacioacuten depanteras veinte antildeos despueacutes de su implementacioacuten Recomendamos a los gestores que consideren repetir la in-trogresioacuten geneacutetica introduciendo 5ndash10 individuos de otras poblaciones de puma cada 20ndash40 antildeos Adicionalmenterecomendamos que se continuacutee con la recopilacioacuten de datos lo que es crucial para poder estimar y monitorizar lasupervivencia de cachorros adultos y subadultos Estos paraacutemetros son los maacutes criacuteticos para estimar la probabilidad yel periodo de cuasi‐extincioacuten seguacuten se demostroacute con el anaacutelisis de sensibilidad Se debe continuar con la mon-itorizacioacuten de la poblacioacuten a largo plazo lo que permitiraacute adaptar las estrategias de manejo seguacuten sea necesario a lavez que recopilar informacioacuten esencial para evaluar la salud demograacutefica de la poblacioacuten Siguiendo estas re-comendaciones las perspectivas de recuperacioacuten de las panteras mejoraraacuten

Contents

INTRODUCTION 5

STUDY AREA 9

METHODS 9

Field Methods 9

Genetic Analysis 10

Demographic Parameters 11

Survival of kittens 11

Subadult and adult survival 11

Probability of reproduction 11

Reproductive output 12

Covariates 12

Cause‐specific mortality 12Statistical inference 12

Matrix Population Models 13

IBMs and PVA 14

Sensitivity Analysis 15

Genetic Management 15

Introgression strategies 15

Cost‐benefit analysis 16RESULTS 17

Genetic Variation 17

Survival and Cause‐Specific Mortality Rates 17

Reproductive Parameters 18

Population Dynamics and Persistence 18

Deterministic and stochastic demography 18

Minimum population count scenario 18

Motor vehicle mortality scenario 18

Sensitivity of extinction probability to demographic parameters 19

Benefits and Costs of Genetic Management 19

Minimum population count scenario 19

Motor vehicle mortality scenario 19

Financial cost and persistence benefits 21

DISCUSSION 21

Genetic Variation Demographic Parameters

and Persistence of Heterotic Benefits from Genetic Introgression 24

Population Dynamics and Persistence 28

Genetic Management When and How to Intervene 28

MANAGEMENT IMPLICATIONS 30

ACKNOWLEDGMENTS 31

LITERATURE CITED 31

INTRODUCTION

Promoting genetic admixture between individuals from differentpopulations commonly referred to as genetic introgression hasthe potential to serve as a powerful management and con-servation tool (Keller and Waller 2002 Tallmon et al 2004Whiteley et al 2015) Genetic introgression can occur naturallythrough immigration of individuals into small or isolated po-pulations (Marr et al 2002 Vila et al 2003 Adams et al 2011)but it can also be implemented as a management strategy via theintentional translocation of individuals from large viable popu-lations to small endangered populations Intentional geneticintrogression of populations of wild animals has prevented thedemise of species such as adders (Vipera berus Madsen et al2004) prairie chickens (Tympanuchus cupido Bateson et al

2014) and bighorn sheep (Ovis canadensis Madsen et al 1999Hogg et al 2006) In most of these cases genetic introgressionhas been effective at improving demographic performance ofpopulations affected by inbreeding depressionInbreeding depression a reduction in fitness resulting from

the production of offspring by individuals related by ancestryis a conservation challenge common to small isolated popu-lations Inbreeding depression has historically been docu-mented in captive populations (Lacy et al 1993) The dele-terious impact of inbreeding depression on wild populationswas somewhat controversial (Frankham 2010a) until theseminal work of Crnokrak and Roff (1999) which verified thenegative effects of inbreeding on the fitness traits of severalwildlife populations including cheetahs (Acinonyx jubatus)

van de Kerk et al bull Florida Panther Population and Genetic Viability 5

black‐tailed prairie dogs (Cynomys ludovicianus) and white‐footed mice (Peromyscus leucopus) Subsequently the effect ofinbreeding depression has been extensively documented inseveral wild animal populations including mammals such aswolves (Canis lupus Liberg et al 2005) African lions (Pan-thera leo Trinkel et al 2010 2011) and Iberian lynx (Lynxpardinus Ruiz‐Loacutepez et al 2012) In these examples in-breeding depression had negative effects on a variety of fitnesstraits including survival of neonates susceptibility to diseaseand male fertility The demographic and genetic impacts ofinbreeding on population growth in carnivores is highlightedbest in one of the quintessential long‐term conservation stu-dies on the gray wolves of Isle Royale in northern MichiganUSA (Peterson 1977 Peterson and Page 1988 Peterson et al1998 Adams et al 2011) Research on this population hasproved informative for gaining a better understanding of po-pulation consequences of loss of genetic variation and impactsof intrinsic and extrinsic factors that have led to its decline toonly 2 individuals as of 2017 (Hedrick et al 2017)Conversely outbreeding depression is a scenario when mating

between members of genetically divergent populations might resultin a reduction in fitness correlated with the loss of local adaptationsor genetic incompatibilities (Moritz 1999) Outbreeding depressionhas been hypothesized to be a possible side effect of genetic in-trogression (Reinert 1991 Maehr and Caddick 1995) albeit in-frequently and with less and weaker empirical evidence for verte-brate species Negative fitness responses resulting from matingbetween individuals from highly divergent populations of copepodswere noted by Hwang et al (2012) Other examples of outbreedingdepression in varied species of plants are provided by Fenster andGallaway (2000) Waser et al (2000) Goto et al (2011) and Brysand Jacquemyn (2016) In general however outbreeding depres-sion is thought to be less of a concern when implementing con-servation measures such as genetic introgression (Miller et al 1999Whiteley et al 2015)In the last 3 decades conservation biologists have exponentially

increased their use of molecular techniques to assist with mon-itoring managing and conserving wildlife populations (DeYoungand Honeycutt 2005) Although early research relied on methodssuch as restriction fragment length polymorphisms allozymeelectrophoresis and minisatellites (Smith and Wayne 1996) thelast 20 years have been dominated by the implementation of im-proved automated sequencing technology to provide more in-formative and fine‐scale data from both nuclear and mitochondrialDNA sequences (eg mtDNA sequencing microsatellitesDeYoung and Honeycutt 2005) Microsatellites in particular havebecome a common tool for conservation projects to assess geneticparameters including inbreeding depression and the impacts ofgenetic introgression in imperiled populations They continue tobe widely used molecular markers in conservation researchThe field of molecular ecology is changing rapidly and new

technology is constantly being developed Currently there is in-terest in applying genomics or so‐called whole‐genome sequencingtechniques to assist with conserving imperiled species Two recentexamples of the application of whole‐genome sequencing to en-dangered species of carnivores include the work of Robinson et al(2016) on the Channel Island fox (Urocyon littoralis) and Murch-ison et al (2012) on the Tasmanian devil (Sarcophilus harrisii) For

instance Robinson et al (2016) reported a near absence ofgenomic variation in Channel Island foxes from the island of SanNicolas demonstrating an extreme reduction in heterozygositycompared to mainland populations Similarly Murchison et al(2012) applied whole genome sequencing to ascertain whether theTasmanian devil facial tumor disease originated from a male or afemale devil and further used whole genome sequencing techniquesto assess transmission of the disease across Tasmania The appli-cation of whole genome sequencing in conservation biology andwildlife management will continue to increase (Whiteley et al2015) Although costs associated with implementing wholegenome sequencing are steadily declining (Whiteley et al 2015)other factorsmdashsuch as sample preparation and the need forbioinformatic expertise to deal with voluminous amounts of datamdashare still an impediment to their widespread use in conservation andmanagement of wildlife populations Therefore long‐term datasets frommore traditional molecular markers such as microsatelliteswill continue to have utility for providing insight into the genetichealth of endangered species and for assessing impacts of man-agement and recovery initiativesThe identification and quantification of inbreeding depression

using molecular techniques (Kardos et al 2016) along with pro-tocols for implementing translocations to promote genetic in-trogression have been documented for varied endangered popu-lations (Whiteley et al 2015) Conversely studies that provide anassessment of the longevity of benefits accrued to wild populationsvia genetic introgression are uncommon Hedrick et al (2014)provided evidence for the declining effect of genetic introgressionon the Isle Royale population of wolves after only 2 or 3 genera-tions Whereas heterosis is expected to increase individual fitness itcan also result in outbreeding depression (Fenster and Gallaway2000 Waser et al 2000 Goto et al 2011 Waller 2015 Brys andJacquemyn 2016) Depending on the mechanisms outbreedingdepression may not become evident until generations subsequent tothe first filial (F1) generation (Pickup et al 2013 Waller 2015)For example ancestral chromosomes and coevolved genes remainintact in the F1 but recombination may break up favorable genecomplexes in F2 and later generations (Waser et al 2000 Waller2015 Frankham 2016) Investigating the timeframe over whichthe positive impacts of genetic introgression are sustained wouldlikely necessitate long‐term monitoring but for most studies of theprocess populations are typically monitored only for 1 or 2 gen-erations following genetic introgression (Willi et al 2007 Hedrickand Fredrickson 2009 Hedrick et al 2014 Whiteley et al 2015)Consequently data are rarely gathered that would allow us to es-timate the time over which populations continue to benefit fromgenetic introgression (Hedrick et al 2014 Whiteley et al 2015)The Florida panther (Puma concolor coryi) provides a textbook

example of how inbreeding depression in a small isolated po-pulation can result in low levels of genetic variation that areassociated with morphological and biomedical abnormalities andprecipitous population decline The timeline associated with thedecline of panthers parallels that of most large carnivores inNorth America (Onorato et al 2010) A robust population ofpanthers was thought to extend across Florida USA inpre‐Columbian times (Fig 1 Alvarez 1993) Subsequentlydocumented declines in the numbers of panthers were noted inthe nineteenth and twentieth centuries mainly as a result of

6 Wildlife Monographs bull 203

anthropogenic alteration of the landscape and unregulatedhunting As early as the 1870s panthers were rarely encounteredand considered mythical in some portions of their range (Beverly1874) The decline in the population of panthers eventuallycaptured the attention of lawmakers subsequently they receivedpartial legal protection as a game animal in 1950 and completelegal protection in Florida by 1958 (Onorato et al 2010) By the1960s the core of remaining wilderness within the Big Cypressregion of South Florida where the last group of breedingpanthers was thought to persist was further affected by theconstruction of Alligator Alley This roadway along with theTamiami Trail allowed access to these once inaccessible areasAll these factors ultimately led to the listing of the Floridapanther as endangered on the inaugural list of endangeredspecies on 11 March 1967 (US Federal Register 1967) andsubsequent protection under the Endangered Species Act of1973 (Public Law 93‐205) This designation invariably raisedawareness of the plight of the panther and served as a catalyst toinitiate research to avoid what appeared to be their imminentextinction (Onorato et al 2010)In 1973 the World Wildlife Fund initiated a survey that re-

sulted in the capture of a single female panther and doc-umentation of their sign in just a few locations in South Florida(Nowak and McBride 1974) The information accrued duringthis survey would ultimately lead to the commencement of along‐term research project on the Florida panther in 1981 by

what is now known as the Florida Fish and Wildlife Con-servation Commission (FWC) Over the next decade a multi-tude of studies were published that helped improve knowledgeof panther reproduction (Maehr et al 1989 Barone et al 1994)diet (Maehr 1990) and genetics (OrsquoBrien et al 1990 Roelkeet al 1993) among other topics Early research was essential foridentifying challenges faced by the remaining panthers mostimportantly the impacts of the continued loss of habitat isola-tion of the population and inbreeding depression (Onoratoet al 2010) In combination these factors would force wildlifemanagers to contemplate the implementation of genetic in-trogression to avert the extinction of the pantherBy the early 1990s the Florida panther population had declined

to a minimum of 20ndash30 individuals (McBride et al 2008) Severalstudies documented low levels of genetic diversity and expressedtraits thought to be correlated with inbreeding depression (Roelkeet al 1993 Johnson et al 2010 Onorato et al 2010) includingcowlicks (mid‐dorsal pelage whorls) and kinked tails (Wilkinset al 1997) These morphological traits probably had little directimpact on fitness of Florida panthers More concerning were theproportion of individuals in the population in the early 1990s thatwere unilaterally or bilaterally cryptorchid (Roelke et al 1993)possessed poor sperm quality (Barone et al 1994) sufferedcompromised immune systems (OrsquoBrien 1994) exhibited atrialseptal defects (Roelke et al 1993) and possessed some of thelowest levels of genetic variation observed in wild felids (Driscoll

Figure 1 Historical and current (2015) breeding range of the Florida panther in the southeastern United States Delineation of current breeding range is describedin Frakes et al (2015)

van de Kerk et al bull Florida Panther Population and Genetic Viability 7

et al 2002) Taken together these factors suggested that thelong‐term prospects for the persistence of the Florida pantherwere poorSubsequently the United States Fish and Wildlife Service

(USFWS) conducted an assessment that led to the delineation of 4possible management options 1) initiate a captive breeding pro-gram 2) introduce genetic material from captive stockpurported to contain both pure Florida panther and SouthAmerican puma lineages 3) translocate wild female pumas (Pumaconcolor stanleyana) from Texas USA into South Florida or 4)take no action at all (USFWS 1994) Collaborative discussionsinvolving academics non‐governmental organizations and stateand federal wildlife agencies eventually led to the selection of op-tion 3 and a plan for genetic introgression that involved the releaseof 8 female pumas from Texas into South Florida was im-plemented in 1995 (Seal 1994 Onorato et al 2010)Continued research conducted after the introgression interven-

tion provided clear evidence of the benefits accrued to the pantherpopulation including increased levels of genetic diversity declinesin the frequency of morphological and biomedical correlates ofinbreeding depression and the substantial increase of the popu-lation size (McBride et al 2008 Johnson et al 2010 McClintocket al 2015) Furthermore Hostetler et al (2013) showed that thepost‐introgression population growth was driven primarily byhigher survival rates associated with admixed panthers (offspringof pairings between the female Texas pumas and male Floridapanthers and subsequent generations of those progeny) followinggenetic introgression The apparent success of genetic introgres-sion in preventing the imminent extinction of the Florida pantherpopulation is encouraging but the population remains small andisolated and continues to face substantial challenges especially thelarge‐scale loss of habitatAlmost 5 panther generations (generation time= 45 years

Hostetler et al 2013) have passed since the implementation of thegenetic introgression program The panther population has beenmonitored continuously during this time period leading to asubstantial amount of additional data and biological insights sincethe demographic assessment made immediately following the in-trogression (Hostetler et al 2010 2012 2013 Benson et al 2011)Given that the effects of genetic introgression are not permanent itis important to discern whether the positive impacts of this man-agement initiative are continuing or waning Decline in the benefitsof the introgression could reduce the population growth rate andprobability of persistence At the same time panther habitat israpidly lost to a variety of human activities (eg land development)and environmental changes associated with invasive species andclimate change (Land et al 2008 Dorcas et al 2012 Frakes et al2015) When long‐term monitoring data are available it is prudentto periodically evaluate demographic rates population growth andpersistence parameters so that changes are detected and timelymanagement action can be implementedTwo population modeling approaches have become popular

among ecologists and wildlife managers for studying the dynamicsand persistence of age‐ or stage‐structured populations matrixpopulation models and individual‐based models (IBMs Brooket al 2000 Morris and Doak 2002 Morrison et al 2016) Matrixpopulation models are the generalized discrete time‐version of theLotka‐Euler equation which describes the dynamics of age‐ or

stage‐structured populations of plants animals and humans (Lotka1924 Caswell 2001) They have become standard tools for mod-eling human and wildlife populations (Tuljapurkar 1990 Caswell2001 Keyfitz and Caswell 2005) because 1) they are powerful andflexible permitting adequate representation of life history of a widevariety of organisms including those with complex life histories 2)parameters for these models can be empirically estimated usingstandard analytical methods such as multi‐state capture‐recapturemethods (eg Williams et al 2002) 3) it is relatively straight‐forward to include the effects of factors such as environmental anddemographic stochasticity and density dependence 4) importantdemographic quantities such as asymptotic population growth ratestable age‐ or stage distribution and generation time can be easilycalculated from these models 5) they permit prospective and ret-rospective perturbation analyses which have been proven useful inwildlife conservation and management 6) they are computer‐friendly do not require extensive programming experience andtheir implementation using software packages such as R (R CoreTeam 2016) and MATLAB (MathWorks 2014) is straight‐for-ward and 7) they have sound theoretical foundation offeringanalytical solutions to many aspects of population modeling Weused matrix population models for 1) calculating the aforemen-tioned demographic quantities 2) performing sensitivity analysesinvolving asymptotic population growth rate and 3) comparativepurposes to check the results obtained from equivalent simulation‐based modelsDespite many advantages offered by matrix population models it

is difficult to incorporate attributes of individuals such as geneticsor behavior within that framework Agent‐based or IBMs (Grimm1999 Grimm and Railsback 2005 McLane et al 2011 Railsbackand Grimm 2011 Albeke et al 2015) offer an alternative modelingframework with tremendous flexibility Individual‐based modelsare computer simulation models that rely on a bottom‐up approachthat begins by explicitly considering the components of a system(ie individuals) population‐level properties emerge from the be-havior of and interactions among discrete individuals UsingIBMs it is possible to explicitly represent attributes of individualssuch as movement mating or genetics (McLane et al 2011Railsback and Grimm 2011 Albeke et al 2015) We used in-dividual‐based population models as the primary populationmodeling approach because they offered a framework for an ex-plicit consideration of genetics both at individual and populationlevelsOur overall goal was to provide a comprehensive demographic

assessment of the sole extant population of the Florida pantherusing long‐term (1981ndash2015) field data and novel modelingtechniques Our objectives were 1) to estimate demographicparameters for the Florida panther and investigate whetherpositive impacts of the 1995 genetic introgression remained inthe population and if so to what extent 2) to estimate popu-lation growth and persistence parameters using an IBM tailoredto the Florida pantherrsquos life history and 3) to evaluate thesensitivity of population growth and persistence parameters tovital demographic ratesThe Florida panther population remains small (le230 adults and

subadults FWC 2017) with no realistic possibility of natural geneflow so the loss of genetic variation and concomitant increase ininbreeding‐related problems are inevitable Ensuring the long‐term

8 Wildlife Monographs bull 203

persistence of the population will likely necessitate periodic geneticmanagement interventions Therefore our final objective was to 4)evaluate genetic and population‐dynamic consequences of variedgenetic introgression strategies and from these identify manage-ment strategies that are affordable and effective at improvingprospects of Florida panther persistence To achieve this objectivewe extended the individual‐based population model to track in-dividual genetics based on empirical allele frequencies taking ad-vantage of the long‐term demographic and genetic data that havebeen collected on Florida panthers since 1981 We used individualheterozygosity to inform vital rates based on empirically estimatedrelationships between individual heterozygosity and demographicparameters We then simulated population trajectories trackedindividual heterozygosity over time and estimated quasi‐extinctiontimes and probabilities without introgression and with introgres-sion for a range of intervals and number of immigrants perintrogression event We compared the genetic and demographicbenefits of implementing alternative genetic introgression scenariosand ranked them by the level of improvement in population per-sistence parameters and estimated costs because funding is a per-sistent challenge to conservation planningManagement of imperiled species is a complex process invol-

ving many stakeholders and plagued by layers of uncertainties(Runge 2011) It would therefore seem that management ofthreatened and endangered species would be a perfect case foradaptive resource management approach (ARM) because ARMfocuses on structured decision making for recurrent decisionsmade under uncertainty (Williams et al 2002 Runge 2011)Because it combines structured decision making with learningprudent application of ARM reduces uncertainty over time andcan lead to optimal management actions (Williams and Brown2012 2016) Key ingredients of a successful adaptive manage-ment program include stakeholder engagement clearly definedand mutually agreed‐upon objectives alternative managementactions and objective decision rules (Nichols et al 2007Williams and Brown 2012 2016) Unfortunately Florida pan-ther stakeholders have divergent views about the state of thesystem (eg panther abundance) and it is a challenge to havemutually agreed‐upon objectives management actions or deci-sion rules acceptable to all or most stakeholders AlthoughFlorida panther conservation efforts would be best served withinthe ARM framework in the long run impediments to its im-plementation (Runge 2011 Williams and Brown 2012) areunlikely to be overcome in the near future

STUDY AREA

The breeding population of Florida panthers mainly persists as asingle population South of the Caloosahatchee River (approxi-mately 267133degN latitude 815566degW longitude see Fig 1)One exception is a female that was documented just north of theCaloosahatchee River in 2016 (and subsequently documentedwith kittens in 2017) by FWC the first validation of a femalepanther north of the River since 1973 (FWC unpublisheddata) The breeding range in South Florida is topographicallyflat has a subtropical climate and is characterized by permanentand ephemeral wetlands influenced by rains from May throughOctober (Duever et al 1986) The study area contains a variety

of wildland land cover types including hardwood hammockscypress forests pine flatwoods freshwater marshes prairies andgrasslands (Davis 1943) and land characterized by human ac-tivity including citrus groves croplands pastureland rockmining and residential developments (Onorato et al 2011)The area is intersected by a multitude of roads that fragmentpanther habitat Most notable among these is Interstate 75(Alligator Alley) which was expanded from 2 to 4 lanes in 1990when fencing and wildlife crossings were constructed with theintention of minimizing harmful effects of road expansion onlocal wildlife (Foster and Humphrey 1995) A large portion ofthe area that comprises South Florida is under public ownershipmainly as federal or state lands Everglades National Park BigCypress National Preserve Florida Panther National WildlifeRefuge Picayune Strand State Forest and Fakahatchee StrandPreserve State Park alone encompass 9745 km2 though not allof the area is considered suitable for panthers (eg large ex-panses of sawgrass marshes in the western Everglades) Frakeset al (2015) delineated 5579 km2 of adult panther habitat inSouth Florida 1399 km2 of which are in private ownership Amajority of these private lands are located in the northern extentof the breeding range Although some private lands may beprotected (eg conservation easements) other areas are sus-ceptible to incompatible land uses such as rock mining or re-sidential developments The breeding range in South Florida isbounded to the east and west by large metropolitan areas in-clusive of Miami‐Fort Lauderdale‐Pompano Beach and CapeCoral‐Fort Myers‐Naples‐Marco Island‐Immokalee respec-tively that are populated by gt6000000 people (httpswwwcensusgov accessed 24 Jan 2019) These characteristics of thestudy area highlight the recovery challenges faced by panthers

METHODS

Field MethodsSince 1981 Florida panthers have been captured radiocollared andtracked by the FWC and National Park Service staff (Table 1)Capture and handling protocols followed guidelines of theAmerican Society of Mammalogists (Sikes 2016) Livestock Pro-tection Company (Alpine TX USA) provided trained hounds andhoundsmen for captures Teams either treed or bayed panthers onthe ground and then darted them with a 3‐ml compressed‐air dartfired from a CO2‐powered rifle Immobilization drugs have variedduring the tenure of the project but most recently teamsimmobilized panthers with a combination of ketamine HCl (10mgkg Doc Lanersquos Veterinary Pharmacy Lexington KY USA) andxylazine HCl (1mgkg Doc Lanersquos Veterinary Pharmacy) Fol-lowing immobilization teams lowered treed panthers to the groundby a rope or caught panthers with a net in some cases they used aportable cushion (McCown et al 1990) to further mitigate theimpact of a fall Capture teams administered propofol (PropoFlotradeAbbott Laboratories Abbott Park IL USA) intravenously either asa bolus or continuous drip to maintain anesthesia They adminis-tered midazolam HCl (003mgkg) intramuscularly or intravenouslyto supplement anesthesia in some panthers Panthers recovered in ashaded area away from water In some cases capture teams reversedxylazine HCl with yohimbine HCl (Yobinereg Lloyd IncShenandoah IA USA) at 25 its recommended dose

van de Kerk et al bull Florida Panther Population and Genetic Viability 9

Capture teams determined sex and age of panthers marked themwith an ear tattoo and a subcutaneous passive integrated trans-ponder (PIT‐tag) and fitted them with a very high frequency(VHF) or global positioning system (GPS) radio‐collar equippedwith a mortality sensor The GPS collars were programmed toattempt to collect a position at a variety of time intervals(range= 1ndash12 hr) depending on objectives of concurrent studiesobservers routinely tracked panthers with VHF radiocollars from afixed‐wing aircraft 3 times per week When a radio‐collar trans-mitted a mortality signal or when a location did not change forseveral days observers investigated the site to establish the pan-therrsquos fate When observers found a dead panther they assessed thecause of death based on an examination of the carcass and sur-rounding area if possible They then transported carcasses to anexperienced veterinarian for necropsy and a histopathological ex-amination to attempt to assign the cause of death Starting in 1995observers continually assessed successive locations of radio‐collaredfemales to determine if they initiated denning behavior Three to 4fixes at the same location (VHF collars) or successive returns to thesame location over a weeklong period (GPS collars) served as anindicator of a possible den Observers then located dens moreprecisely via triangulation with ground telemetry They visited denswhen the dam left the site typically to make a kill Teams capturedkittens by hand captured kittens ranged in age from 4 to 35 dayspost‐partum Teams counted sexed weighed dewormed andpermanently marked kittens with a subcutaneous PIT‐tag

Genetic AnalysisTeams collected blood tissue or hair from the majority ofcaptured kittens subadult and adult Florida panthers for DNAsamples The United States Department of Agriculture ForestService National Genomics Center for Wildlife and FishConservation (Missoula MT USA) processed samplesTechnicians used Qiagen DNeasy Blood and Tissue Kits(Qiagen Valencia CA USA) to extract whole genomic DNAfrom blood and tissue samples and used a slight modification(overnight incubation of sample in lysis buffer and proteinase Kon a rocker at 60deg elution of DNA in 50 μl of buffer) of theQiagen DNA extraction protocol (Mills et al 2000) to extractDNA from hair samples Technicians used a panel of 16 mi-crosatellite loci (Fca090 Fca133 Fca243 F124 F37 Fca075Fca559 Fca057 Fca081 Fca566 F42 Fca043 Fca161

Fca293 Fca369 and Fca668) identified by Menotti‐Raymondet al (1999) which had been previously applied to Floridapanther samples (Johnson et al 2010) Technicians amplifiedDNA samples in 10‐microl reaction volumes that included 10 microl ofDNA 1times reaction buffer (Applied Biosystems Life Technolo-gies Corporation Grand Island NY USA) 20 mM of MgCl2200 microM of each dNTP 1 microM of reverse primer 1 microM of dye‐labeled forward primer 15 mgml of bovine serum albumin(BSA) and 1U Taq polymerase (Applied Biosystems) Thepolymerase chain reactions (PCRs) included a thermal profile of94degC for 5 minutes followed by 36 cycles of 94degC for 60 seconds55degC for 60 seconds and 72degC for 30 seconds Techniciansvisualized the resulting PCR products on a LI‐COR DNAanalyzer (LI‐COR Biotechnology Lincoln NE USA) Tech-nicians collected genotypes from hair samples (n= 16) using amultitube approach (Taberlet et al 1996) error‐checked geno-types using Program DROPOUT (McKelvey and Schwartz2005) and combined them with genotypes from tissue andblood samples (n= 487) We evaluated all 16 loci for variousdescriptive statistics for the radio‐collared sample of adult andsubadult panthers We excluded genotypes of kittens from thatanalysis because related individuals are known to result in theoverrepresentation of certain alleles (Marshall et al 1998) Wedetermined the number of alleles observed heterozygosity andexpected heterozygosity using GenAlex (Peakall and Smouse2006 2012) We assessed allelic richness at each locus in Fstat293 (Goudet 1995) We assessed conformance of genotypedata from these 16 loci to HardyndashWeinberg assumptionslinkage disequilibrium and null alleles (Appendix A availableonline in Supporting Information) The Genomics Center alsoprovided us with genotype data from the same 16 loci for 41pumas from western Texas the source population for thegenetic rescue programEvidence of inbreeding depression in wild populations is

commonly determined on the basis of heterozygosityndashfitnesscorrelations (Silva et al 2005 Ortego et al 2007 Mainguyet al 2009 Johnson et al 2011) so we also estimated theheterozygosity of each individual panther We calculatedhomozygosity by loci (HL) which varies between 0 (all lociheterozygous) and 1 (all loci homozygous Aparicio et al 2006)using the Rhh package (Alho et al 2010) in Program R (R CoreTeam 2016) A microsatellite locus has more weight in HL

Table 1 Number of Florida panthers by age sex and genetic ancestry categories used in subsequent demographic analyses during the study period (1981ndash2013) inSouth Florida USA

Kittensa Subadults and adultsb

Males Females Total Males Females Total

PIT‐taggedc 215 180 395 Radiocollaredc 108 101 209Recovered 11 4 15 Subadult 72 50 122Litter failed 47 38 85 Prime adult 65 95 160Recaptured 25 30 55 Old adult 13 20 33Canonical 27 20 47 Canonical 43 29 72F1 admixed 6 12 18 F1 admixed 2 6 8Other admixed 130 105 235 Other admixed 45 53 98

a Kitten samples included those that were PIT‐tagged recovered (ie panthers that were found dead) part of a failed litter or recaptured as subadults or adultsb Radiocollared subadult and adult panthers are presented as sample size within each age class and ancestry classc Combined number of panthers of different ancestries is less than the number radiocollared or tagged with a subcutaneous passive integrated transponder (PIT‐tagged) because we did not conduct genetic sampling on all panthers Combined number of panthers in different age classes is greater than the number ofradiocollared panthers because some individuals were counted in ge2 age classes as animals aged

10 Wildlife Monographs bull 203

when it is more informative (ie has more alleles) and has aneven distribution of allele frequencies For our demographicanalyses we calculated individual heterozygosity for each in-dividual panther as 1 minusHL As do most indices of individualheterozygosity HL takes into account the average allele fre-quency in the population (Aparicio et al 2006) Thus duringsimulations the allele frequency in the population changesthroughout an individualrsquos lifetime which causes its hetero-zygosity also to change This is not biologically plausible becausegenetic properties are retained throughout an individualrsquos life-time Therefore for use in model simulations we simply cal-culated heterozygosity for each individual as the proportion ofheterozygous lociWe also used genotype data to implement a Bayesian clus-

tering analysis in Program STRUCTURE (Pritchard et al2000) to infer ancestral clusters of panthers This method uses aMarkov chain Monte Carlo (MCMC) method to determine thenumber of genetic clusters (K) in the sample while also pro-viding an individualrsquos percentage of ancestry (q values) allocatedto each cluster Runs for this analysis included a 100000MCMC burn‐in period followed by 500000 MCMC iterationsfor 1ndash10 ancestral genetic clusters (K= 1 to 10) with 10 re-petitions for each K To determine the most likely number of Kgenetic clusters we used the logarithm of the probability of thedata (log Pr(D)∣K Pritchard et al 2000) and estimates of ΔK(Evanno et al 2005) in Program STRUCTURE HAR-VESTER (Earl and VonHoldt 2011) We then used q valuesprovided in runs for the best K to assign individual panthers aseither canonical (in most cases ge90 pre‐introgression ancestrysee Johnson et al 2010) or admixed (lt90 pre‐introgressionancestry) We recognized admixed panthers that were the off-spring of matings between a Texas female puma and a canonicalmale panther as F1 admixed panthers for analyses

Demographic ParametersSurvival of kittensmdashMost kittens PIT‐tagged at den sites were

never encountered again A small proportion of the PIT‐taggedkittens were recovered dead or were recaptured as subadults oradults and radiocollared so that their fate could be monitoredWe also identified instances of complete litter failure when afemale died within 9 months after giving birth (the minimumage of independence for kittens) or when a female wasdocumented to have another litter less than 12 months aftergiving birth (the minimum age of independence plus a 3‐monthgestation period Hostetler et al 2010) Thus our data setconsisted of 1) live recaptures and subsequent radio‐tracking ofpanthers of various ages that had been PIT‐tagged as kittens 2)recovery of dead panthers of various ages that had been PIT‐tagged as kitten 3) live captures and subsequent radio‐trackingof other panthers and 4) instances of complete litter failure(Table 1) We analyzed these data using Burnhamrsquos live‐recapture dead‐recovery modeling framework (Burnham 1993Williams et al 2002) to estimate survival of kittens (0ndash1‐yearold) and to test covariate effects on this parameter We includeddata on individuals encountered dead or alive at an older age inour analyses because their encounter histories also providedinformation on kitten survival We used a live‐dead data inputformat coded with a 1‐year time step (Cooch and White 2018)

All known litter failures occurred within the first year of thekittensrsquo lives so we treated all litter failures as recoveries withinthe first year We treated panthers that would have died if wehad not removed them to captivity as having died on the date ofremoval Data preparation and analytical approach are describedin more detail by Hostetler et al (2010)In addition to survival probability (S) the Burnham model

incorporates parameters for recapture probability (p) recoveryprobability (r) and site fidelity (f) We set r and p to 10 forradio‐tracked individuals because their status was known eachyear We also fixed f at 10 for all individuals because the re-capture and recovery areas were the same and encompassed theentire range of the Florida panther Based on findings of Hos-tetler et al (2010) we used a base model that 1) allowed survivalto differ between kittens and older panthers 2) allowed survivalto differ between sexes and among age classes (females ages 1and 2 ge3 males ages 1 2 and 3 ge4) 3) constrained recaptureprobability to be the same for all uncollared panthers and 4)allowed recovery rates to differ between kittens and uncollaredolder panthers Even with our additional data this model re-mained the best fit for a range of models for recapture recoveryand survival of kittens subadults and adults of various age‐classcategories (Hostetler et al 2010)Subadult and adult survivalmdashTo estimate survival for panthers

ge1 year of age we analyzed radio‐tracking data using a Coxproportional hazard‐modeling framework (Cox 1972 Therneauand Grambsch 2000) We followed procedures outlined byBenson et al (2011) for data preparation and analysis Brieflywe right‐censored panthers on the last day of observation beforetheir collar failed When an individual aged into a new age classwhile being tracked we created 2 data entries 1 ending the dayit entered the new age class the other starting on that day Weused the FlemingndashHarrington method to generate survivalestimates from the Cox analysis and the Huber sandwichmethod to estimate robust standard errors (Therneau andGrambsch 2000 Benson et al 2011)We considered 3 age classes subadults (1ndash25 yr for females and

1ndash35 yr for males) prime adults (25ndash10 yr for females and35ndash10 yr for males) and old adults (gt10 yr old for both sexes)after Benson et al (2011) We estimated overall survival using abase model that allowed survival to differ between subadultfemales subadult males prime adult females prime adult malesand old adults of either sex (old age was additive with sex other-wise age and sex were interactive) This model provided the best fitto the data relative to a range of models considering differences insurvival between prime‐adult and older‐adult age classes both ad-ditively and interactively with sex (Benson et al 2011)To examine whether survival rates observed in recent years

differed from those observed immediately following introgres-sion we also estimated age‐specific survival rates for 2 nearlyequal periods of time immediately post‐introgression (1995ndash2004) and the more recent period (2005ndash2013) For theseanalyses we separated the radio‐tracking data into the 2 periodsand estimated survival rates for each period separately using thebase model Similar analyses for kitten survival were not possiblebecause of data limitationsProbability of reproductionmdashWe used telemetry data obtained

from radio‐tracked females that reproduced at least once to

van de Kerk et al bull Florida Panther Population and Genetic Viability 11

estimate annual probability of reproduction of subadult prime‐adult and old‐adult females using complementary logndashlogregression (Agresti 2013) following methods described indetail by Hostetler et al (2012) Briefly we modeled theprobability of a female panther giving birth (b) as a function ofcovariates as

γ= minus [minus ( )]b z1 exp exp i i

where zi is a row vector of covariates (see below for covariateinformation) for female panther i and γ is a column vector ofmodel coefficients (Appendix B Table B2) To account fordifferences in observation time we included log(m) as an offsetin the model where m is the number of months a femalepanther was radio‐tracked (Hostetler et al 2012)

Reproductive outputmdashTo estimate the number of kittensproduced by reproductive females we used cumulative logitregression (Min and Agresti 2005 Agresti 2013) Weconsidered a model that includes J categories for the numberof offspring with j denoting the number of offspring ( j= 1 2hellip J and J= 6) provided that a female reproduces Weconsidered J= 6 which is greater than the largest litter sizeobserved in our study (4 kittens) because females can produce 2or more litters per year if litters fail We modeled the probabilitythat a female i produces a litter of size at most j (Yi) as

le⎧

⎨⎩

δ( ) = == hellip minus

=

( )+ θ βminus +Y jj J

j JPr

1 2 1

1i ij e

1

1 xj i

where θj is the intercept for the annual number of kittens pro-duced by a female i being at most j xi is a row vector of cov-ariates for female panther i (see below) and β is a column vectorof model coefficients The probability that Yi will be exactly j isgiven by

⎧⎨⎩

πδ

δ δ( = ) = =

=

minus = hellip( minus )Y j

jj J

Pr 1

2 i ij

i

ij i j

1

1

Finally the expected annual number of kittens produced byfemale i (νi) is given by

sumν π==

j i

j

J

ij

1

Additional details regarding the estimation and modelingof the annual number of kittens produced can be found inHostetler et al (2012)

CovariatesmdashIn addition to sex and age we tested for the effect ofabundance genetic ancestry and individual heterozygosity on theaforementioned demographic parameters (model coefficients forkitten and adult survival are given by η and ι respectively AppendixB Table B2) We used a minimum population count (MPC) basedon radio‐tracking data and field evidence of subadult and adult (ieindependent age) panthers as an index of abundance (McBride et al2008) For these analyses we did not use the population sizeestimates reported by McClintock et al (2015) because unlike theminimum counts they did not cover the entire study period

(McBride et al 2008 R T McBride and C McBride RancherrsquosSupply Incorporated unpublished report Fig 2)To test if demographic parameters differed depending on an-

cestry we considered 3 ancestry models dividing the panthers into1) F1 admixed and all other panthers 2) canonical and admixed(including F1 admixed) panthers and 3) F1 admixed canonicaland other admixed panthers We quantified genetic diversity usingindividual heterozygosity as described previously We tested for theeffect of genetic covariates using a subset of kittens that were ge-netically sampled We calculated model‐averaged estimates ofmodel parameters on the scale in which they were estimated(Appendix B available online in Supporting Information)

Cause‐specific mortalitymdashWe performed cause‐specificmortality analysis only on dead radio‐collared panthers becausedetection rates of uncollared panther mortalities are highlycorrelated with the cause of death (eg vehicle collision)Following Benson et al (2011) we attributed each mortality to1 of 4 causes 1) vehicle collision 2) intraspecific aggression 3)other causes (including known causes such as diseases injuriesand infections unrelated to the first 2 causes) and 4) unknown(ie mortalities for which we could not assign a cause) Wethen used the non‐parametric cumulative incidence functionestimator a generalization of the staggered‐entry KaplanndashMeiermethod of survival estimation to estimate cause‐specificmortality rates for male and female panthers in different ageclasses (Pollock et al 1989 Heisey and Patterson 2006 Bensonet al 2011) We compared cause‐specific mortality rates betweensexes age classes and ancestries

Statistical inferencemdashWe used an information‐theoreticapproach (Akaikersquos information criterion [AIC]) for modelselection and statistical inference (Burnham and Anderson 2002)For the analysis of kitten survival we adjusted the AIC foroverdispersion and small sample size (QAICc details in Hostetleret al 2010) We performed all statistical analyses in Program R (RCore Team 2016) We used Burnhamrsquos live recapturendashdead

0

100

200

300

1985 1990 1995 2000 2005 2010Year

Indi

vidu

als Method

MPC

MVM

Figure 2 The Florida panther minimum population count (MPC) and populationsize estimated using motor vehicle collision mortalities (MVM) during our studyperiod Minimum population counts are from McBride et al (2008) and R TMcBride and C McBride (Rancherrsquos Supply Incorporated unpublished report)Estimates based on MVM are from McClintock et al (2015) The solid vertical lineindicates the year of genetic introgression

12 Wildlife Monographs bull 203

recovery model (Burnham 1993) to estimate and model kittensurvival using the RMark package version 2113 (Laake 2013) asan interface for Program MARK (White and Burnham 1999)We implemented the Cox proportional hazard model for esti-

mation and modeling of survival of subadults and adults using the Rpackage SURVIVAL (Therneau and Grambsch 2000) We per-formed cause‐specific mortality analyses using an R version of S‐PLUS code provided by Heisey and Patterson (2006)To account for model selection uncertainty we calculated

model‐averaged parameter estimates across all models using theR package AICcmodavg (Mazerolle 2017) using AIC weights asmodel weights (Burnham and Anderson 2002) We used modelswithout the categorical ancestry variables for model averagingand estimated parameters based on the average value for thecontinuous covariates (individual heterozygosity and abundanceindex) Because only a subset of the kittens was geneticallysampled we estimated 2 kitten survival rates First we estimatedkitten survival based on the data set that included all kittens(overall kitten survival) Second we estimated kitten survivalusing a subset of the data that only included kittens that weregenetically sampled (survival of genetically sampled kittens)

Matrix Population ModelsWe investigated Florida panther population dynamics andpersistence using 2 complementary modeling frameworks ma-trix population models (Caswell 2001) and IBMs (Grimm andRailsback 2005 McLane et al 2011 Railsback and Grimm2011 Albeke et al 2015) Matrix population models offer aflexible and powerful framework for investigating the dynamicsand persistence of age‐ or stage‐structured populations (egCaswell 2001 Robinson et al 2008 Kohira et al 2009 Hunteret al 2010 Hostetler et al 2012) With a fully developed theorybehind them these models also serve as a check for resultsobtained from simulation‐based models such as IBMs We usedan age‐structured‐matrix population model for deterministic andstochastic demographic analyses and for preliminary investiga-tions of Florida panther population viability (Caswell 2001Morris and Doak 2002)For deterministic and stochastic demographic analyses we

parameterized a female‐only 19 times 19 age‐specific populationprojection matrix of the form

⎢⎢⎢⎢⎢

⋯ ⋯

⋯ ⋯

⋱ ⋱ ⋮

⋮ ⋱ ⋱ ⋱ ⋮

⎥⎥⎥⎥⎥

( )=t

F F F

P

P

P

A0 0

0

0 0 0

t t t

t

t

t

1 2 19

1

2

18

where Fi and Pi are the age‐specific fertility and survival ratesrespectively and ( )tA is a time‐varying (or stochastic) popula-tion projection matrix We assumed that the longevity and ageof last reproduction of female Florida panthers was 185 yearswe based this assumption on the observation that the oldestfemale in our study was 186 years old Florida panthers re-produce year‐round thus we used birth‐flow methods to esti-mate Fi and Pi values following Caswell (2001) and Hostetleret al (2013) We estimated Pi as

= +P S S i i i 1

where Si is the age‐specific survival probability We estimated Fi as

⎛⎝

⎞⎠

=+ +F S

m Pm

2i k

i i i 1

where Sk is the kitten survival probability and mi is the age‐specific fecundity rate which was estimated as

ν=m b f i i i k

where bi is the breeding probability for a female in age class iduring time step t νi is the annual number of kittens producedby a breeding female in age class i and fk is the proportionfemale kittens at birth (assumed to be 05)We estimated the finite deterministic (λ) and stochastic annual

population growth rate (λs) and stochastic sensitivities and elasti-cities following methods described in detail by Caswell (2001) Toestimate λs we assumed identically and independently distributedenvironments (Caswell 2001 Morris and Doak 2002 Haridas andTuljapurkar 2005) and used a parametric bootstrapping approachto obtain a sequence of demographic parameters for 50000 ma-trices We then estimated log(λs) from this sequence of matrices(Tuljapurkar 1990 Caswell 2001) and exponentiated it to obtain λsWe calculated the sensitivity and elasticity of λs to changes in lower‐level vital demographic rates as per Caswell (2001) and Haridas andTuljapurkar (2005)For population projection and population viability analysis (PVA)

simulations we expanded the population projection matrix into a34 times 34 2‐sex age‐structured matrix to include female and maleFlorida panthers (Caswell 2001 Hostetler et al 2013) Malescontribute to the population only through survival in this modelingframework We assumed maximum longevity for males to be 145years of age because the oldest male in our study was 144 years oldThe population projection matrix was of the following form

⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢

⋯ ⋯ ⋯ ⋯

⋯ ⋯ ⋯ ⋯

⋱ ⋱ ⋮ ⋮ ⋱ ⋱ ⋮

⋮ ⋱ ⋱ ⋱ ⋮ ⋮ ⋱ ⋱ ⋮

⋯ ⋯ ⋯

⋯ ⋯ ⋯ ⋯

⋯ ⋯ ⋱ ⋱ ⋮

⋮ ⋮ ⋱ ⋱ ⋮ ⋱ ⋱ ⋮

⋯ ⋯ ⋯

⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥

( )=

( ) ( ) ( )

( )

( )

( )

( ) ( ) ( )

( )

( )

t

F N F N F N

P N

P N

P N

M N M N M N

Q N

Q N

A n

0 0

0 0 0 0

0

0 0 0 0 0

0 0

0 0 0

00 0 0 0 0

t t t

t

t

t

t t t

t

t

1 2 19

1

2

18

1 2 19

1

14

where Pi and Qi are age‐specific survival rates of female and maleFlorida panthers respectively and Fi and Mi are the rates atwhich female and male kittens are produced by females of ageclass i respectively Here the population projection matrix isboth time‐varying and density‐dependent (Caswell 2001)We incorporated the influence of parameter uncertainty en-

vironmental stochasticity and density dependence on Floridapanther population dynamics and persistence following the ap-proach outlined by Hostetler et al (2013) we used the sameapproach in the IBM analysis which is described below UnlikeIBMs matrix models do not intrinsically include demographicstochasticity so we explicitly included the influence of

van de Kerk et al bull Florida Panther Population and Genetic Viability 13

demographic stochasticity by simulating the fates of individualsas described in Caswell (2001)Population projections using matrix models necessitate a

vector of initial sex‐ and age‐specific abundance Because suchdata are currently unavailable for the Florida panther we esti-mated it based on the stable sex and age distribution and MPCin 2013 Using point estimates of the demographic parameterswe estimated the stable sex and age distribution for the 2‐sexdeterministic and density‐independent population projectionmatrix We then multiplied the stable sex and age distributionby the MPC in 2013 to get the starting population vector n(0)We projected the population over time as n(t+ 1)=A(tn)n(t)where A(tn) indicates the time‐specific density‐dependentpopulation projection matrix (Caswell 2001)We had 2 indices of abundance available the MPC (McBride

et al 2008) and population size estimated using motor vehiclecollision mortalities (MVM McClintock et al 2015) These 2indices are substantially different because MPC does not ac-count for imperfect detection whereas MVM accounts forimperfect detection and provides an estimate of population size(Fig 2) Therefore we examined 2 scenarios one using densitydependence based on MPC and the other using density de-pendence based on MVM We ran 1000 bootstraps of para-meter values and 1000 simulations per bootstrap for each sce-nario We projected population trajectories over 200 years andthen estimated population growth rates probabilities ofquasi‐extinction and time until quasi‐extinction and comparedthese with the results obtained from the IBM (see below) Weran scenarios with a quasi‐extinction threshold of 10 and 30panthers We used the mean of probabilities of quasi‐extinction(PQE) across simulations as the point estimate for PQE and the5th and 95th percentiles across simulations as the 90 con-fidence interval Because the confidence intervals werequantile‐based but the point estimates were not it is possiblewith skewed distributions for the point estimates to be outsidethe confidence interval We implemented the matrix

population models in the R computing environment (R CoreTeam 2016)

IBMs and PVABecause of well‐developed theory ease of implementation andthe modeling flexibility that they offer matrix populationmodels have become the model of choice for investigating thedynamics and persistence of age‐ and stage‐structured popula-tions (Caswell 2001) However it is difficult to incorporateattributes of individuals such as genetics and behavior usingmatrix models Quantifying the effects of genetic erosion on theFlorida panther population dynamics and persistence andevaluating benefits and costs of alternative genetic managementscenarios were important goals of our study Because IBMs relyon a bottom‐up approach that begins by explicitly consideringindividuals (McLane et al 2011 Railsback and Grimm 2011Albeke et al 2015) it is possible to represent genetic attributesof each individual and to track genetic variation over time Weused IBMs as the primary modeling framework in this studybecause they allowed for explicit consideration of genetics bothat individual and population levels and population viabilityassessment under alternative genetic management scenariosWe developed an IBM (Grimm 1999 Grimm and Railsback

2005 McLane et al 2011 Railsback and Grimm 2011 Albekeet al 2015) tailored to the life history of the Florida panther(Fig 3) We simulated every individual from birth until deathbased on a set of rules describing fates (eg birth and death) andattributes (eg genetics) of individuals at each annual time stepAs in other IBMs population‐level processes (eg populationsize and genetic variation) emerged from fates of and interac-tions among individuals (Grimm 1999 Railsback and Grimm2011) We developed an overview design concepts and details(ODD) protocol (Grimm et al 2006 2010) to describe ourIBM (Appendix C available online in Supporting Information)and provide pseudocodes for individual‐based simulations(Appendix D available online in Supporting Information)

For each panther at time = t

Kitten (lt1 yr)

Reproduce

Survive

Kitten(s) born

Kitten

Sex and age class

Subadult male

Prime adult male

Old adult male

Subadult female

Prime adult female

Old adult female

Age one year

No Male

Female

Yes

No

Yes

t=t+1

Yes

lt35 yrs

35-10 yrs

gt10 yrs

lt25 yrs

25-10 yrs

gt10 yrs

Figure 3 Schematic representation of Florida panther life history We tailored the individual‐based population model to represent the life‐history patterndepicted here

14 Wildlife Monographs bull 203

Dynamics and persistence of populations are influenced bymany factors such as demographic and environmental sto-chasticity and density dependence these effects and parametricuncertainty must be considered in population models whenpossible (Caswell 2001 Boyce et al 2006 Bakker et al 2009Hostetler et al 2013) Because by definition IBMs simulate thelife history of individuals in a population they inherently in-corporate the effect of demographic stochasticity Our analysesrevealed that survival of kittens subadults and adults and theprobability of reproduction varied over time so we incorporatedenvironmental stochasticity in the model for these variablesfollowing Hostetler et al (2013) Likewise we found evidenceof density‐dependent effects on kitten survival Because theMPC might be an underestimate of population size (McBrideet al 2008) we performed simulations using both the MPC andMVM abundance indicesTo account for model selection and parametric uncertainty we

used a Monte Carlo simulation approach For each bootstrapsample we selected a model based on the AIC (or QAICc)weights We then sampled the intercept and slope for kittensurvival (on the logit scale) subadult and adult survival (on thelogit scale with log‐hazard effect sizes) and probability of re-production (on the complementary logndashlog scale Appendix B)from multivariate normal distributions with mean vectors equalto the estimated parameters and variance‐covariance matricesequal to the sampling variance‐covariance matrices We thentransformed the results to nominal scale and converted the es-timates to age‐specific survival and reproduction parameters asdescribed in Hostetler et al (2013)We ran 1000 simulations for 1000 parametric bootstraps for

200 years for the MPC and the MVM scenario for both thestochastic 2‐sex matrix‐population model and the IBM Wetracked the population size at each time step and estimatedquasi‐extinction probabilities and times based on the populationtrajectory We considered the population to be quasi‐extinct ifindividuals of only 1 sex remained or if the population size fellbelow critical thresholds set at 10 and 30 individuals We alsoestimated expected time to quasi‐extinction for both thresholdsdefined as the mean and median time until a population be-comes quasi‐extinct conditioned on quasi‐extinction We im-plemented the IBM using the RNetLogo package version 10ndash2(Thiele et al 2012 Thiele 2014) as an interface for the IBMplatform NetLogo (Wilensky 1999)

Sensitivity AnalysisAn important component of demographic analysis is sensitivityanalysis which is the quantification of the sensitivity of amodelrsquos outcome to changes in the input parameters (Caswell2001 Bakker et al 2009 Thiele et al 2014) Using an IBM weperformed global sensitivity analyses for 2 (ie MPC andMVM) density‐dependent scenarios to assess how sensitivequasi‐extinction times and probabilities were to changes (anduncertainties) in the estimated demographic rates We usedLatin hypercube sampling to sample parameters from the entireparameter space (Marino et al 2008 Thiele et al 2014) Wesampled intercept and slope for kitten survival (on logit scale)subadult and adult survival (on logit scale with log‐hazard effectsizes) and probability of reproduction (on complementary

logndashlog scale) from normal distributions We sampled theprobability distribution for the number of kittens producedannually (on the real scale) from a Dirichlet distribution themultivariate generalization of the beta distribution (Kotz et al2000) We sampled 1000 different parameter sets and ran 1000simulations for each scenario We calculated the probability ofquasi‐extinction within 200 years for each density‐dependencescenario (MPC or MVM) using a quasi‐extinction threshold of10 individuals and then calculated the partial rank correlationcoefficient between each of those and each parameter trans-formed to the real scale (Bakker et al 2009 Hostetler et al2013) Partial rank correlation coefficients quantify the sensi-tivity of the probability of quasi‐extinction to input variables(ie survival and reproductive parameters) with higher absolutevalues indicating stronger influence of that variable on quasi‐extinction probabilities

Genetic ManagementOne of our goals was to estimate the benefits (improvements ingenetic variation demographic parameters and populationgrowth and persistence parameters) and costs (financial costs ofcapture transportation quarantine release and monitoring ofintroduced panthers) of genetic management To achieve thisobjective we extended the IBM to include a genetic componentTo simulate individual‐ and population‐level heterozygosity overtime we assigned each panther alleles for 16 microsatellite locibased on the empirical allele frequency in the population (esti-mated from genetic samples collected during 2008ndash2015) whenwe initialized the model (Pierson et al 2015) At each time stepwe simulated Mendelian inheritance for kittens produced suchthat each kitten randomly inherited alleles from its mother andfrom a male panther randomly chosen to have putatively siredthe litter We did not explicitly force inbreeding to occur butthe probability of inbreeding naturally increased as the popula-tion size decreasedIntrogression strategiesmdashWe modeled genetic introgression as a

result of the release of 2‐year‐old female pumas from Texas intothe simulated Florida panther population For the geneticintrogression implemented in 1995 Texas females between 15and 25 years of age were released because females at that agehave lower mortality rates (Benson et al 2011) higherreproductive potential and also because it is much easier todetect with certainty the successful reproduction by females thanby males The ability to more closely monitor reproduction andthe birth of kittens is an important consideration in reducing theprobability of a genomic sweep of Texas alleles into the Floridapanther population Wildlife managers removed the last 2surviving Texas females from the wild population in Florida in2003 to reduce the possibility of genomic sweep To simulatethis strategy in the model we removed any Texas females thatwere still alive 5 years after each introgression event Weassigned Texas females alleles based on the allele frequency ofthe Texas population (based on genotypes from 48 samplescollected in west Texas that was inclusive of the pumas releasedin South Florida in 1995) when they entered the Florida pantherpopulation We could not estimate demographic rates separatelyfor the released Texas females because of small sample size (only8 females were released in 1995) therefore we assumed that

van de Kerk et al bull Florida Panther Population and Genetic Viability 15

survival and reproductive rates and the effects of covariates onthese rates for the released Texas females were identical to thoseof female Florida panthersThe impact of genetic introgression depends on the frequency

of introgression events and the number of individuals introducedat each attempt Thus we defined introgression strategies basedon the number of Texas females to be released (0 5 10 or 15)and the interval between genetic introgressions (10 20 40 and80 yr) for a total of 13 strategies We simulated each manage-ment strategy with density dependence estimated using theMPC and MVM abundance indices

We sampled 1000 parameter sets and for each of these setswe ran all introgression strategies 1000 times for 200 years Weapplied the same parametric uncertainty and environmentalstochasticity across all management scenarios to allow for bettercomparison of introgression strategies At each time step werecorded the projected population size and average individualheterozygosity We calculated the heterozygosity for each in-dividual at birth based on the alleles it inherited This individualheterozygosity then informed the survival rate of that individualbased on the empirically estimated relationship between in-dividual heterozygosity and age‐specific survival rates Hetero-zygosity did not change throughout an individualrsquos lifetime butits effect on survival changed as individuals aged because theeffect of individual heterozygosity on survival rate differedamong age classes Survival rates of genetically sampled kittenswere biased high because some kittens were sampled later in lifewhen they had already survived for some time To correct forthis bias we adjusted kitten survival rates predicted by theheterozygosity model proportionally From the population tra-jectories we estimated the probability of quasi‐extinction (de-fined as the probability that the population will fall below 10 or30 individuals or that individuals of only 1 sex will remain)

Cost‐benefit analysismdashExperts estimated financial costs ofintrogression based on their experience with the geneticintrogression executed in 1995 and we adjusted estimates tocurrent costs to account for inflation (R T McBride RancherrsquosSupply Incorporated M Cunningham and D P OnoratoFlorida Fish and Wildlife Conservation Commission personalcommunication) Costs included capturing and transportingfemale pumas from the Texas source population caring for thepumas while they were in quarantine and post‐introgressionmonitoring To compare the financial costs of the differentscenarios we also calculated the average costs incurred over

Table 2 Sample size (n) number of alleles (Na) allelic richness (AR) observedheterozygosity (Ho) and expected heterozygosity (He) at 16 microsatellite loci inradio‐collared Florida panthers sampled from 1981 to 2013 in South FloridaUSA We calculated these values from a subset (n= 137) of the samples used inthe analyses and they include only adult and subadult panthers We excludedkittens because they can result in an overrepresentation of certain alleles

Locus n Na AR Ho He

FCA090 129 5 50 0333 0401FCA133 136 3 30 0618 0608FCA243 136 5 50 0419 0487F124 137 7 69 0708 0729F37 129 4 40 0395 0397FCA075 131 5 50 0534 0598FCA559 133 5 50 0496 0503FCA057 134 6 59 0679 0624FCA081 134 7 69 0209 0206FCA566 130 6 60 0462 0475F42 133 10 100 0737 0766FCA043 136 5 49 0596 0588FCA161 132 6 60 0500 0515FCA293 132 4 40 0614 0624FCA369 131 6 60 0573 0567FCA668 133 7 70 0406 0462Mean 1329 57 57 0517 0535

Figure 4 Proportional membership (q) of radio‐collared Florida panthers (n= 218) and pumas from Texas USA (n= 7) in 2 clusters identified by STRUCTUREEach individual animal is represented by a separate vertical bar Yellow indicates canonical panther ancestry and blue represents admixed ancestry The admixedancestry of the Texas females and F1 generation panthers that were radiocollared are highlighted red and green respectively to demarcate those groups The pre‐introgression period is inclusive of panthers radiocollared from 10 February 1981 to 6 March 1996 The post‐introgression era inclusive of the F1 panthers includesanimals radiocollared from 4 March 1997 to 18 February 2015 in South Florida USA

16 Wildlife Monographs bull 203

100 years assuming that the population would be managedaccording to a particular strategy Our goal with theseexploratory analyses was to evaluate the relative costs andbenefits of alternative management scenarios

RESULTS

Genetic VariationWe assessed genetic variation at 16 microsatellite loci for 137DNA samples collected from adult and subadult panthers(1981ndash2013) that were captured and subsequently radiocollaredThe number of alleles at these loci ranged from 3ndash10 and allelicrichness averaged 57 (Table 2) Average expected hetero-zygosity for this sample was 0535 and average observed het-erozygosity was 0517 (Table 2)We determined individual heterozygosity (1 minusHL) and an-

cestry for the complete sample of panther genotypes (n= 503)collected from 1981 to 2015 for use as covariates in subsequentanalyses Average individual heterozygosity was 0559 andranged from 0ndash0980 Our STRUCTURE analysis providedstrong support for K= 2 clusters based on the probability of thedata (log Pr(D)|K) and ΔK in STRUCTURE HARVESTERBased on this result we categorized the ancestry of each pantheras either canonical (n= 123) or admixed (n= 380) using valuesof proportional membership (q Fig 4) The average individualheterozygosity of canonical panthers was 0386plusmn 0012 (SE) foradmixed panthers it was 0615plusmn 0007

Survival and Cause‐Specific Mortality RatesOf the 395 kittens that were PIT‐tagged and released 55 werelater encountered alive 15 were recovered dead and 85 werepart of failed litters (Table 1) We radio‐tracked 209 subadult oradult panthers for a total of 244195 panther‐days and docu-mented 126 mortalities (Table 1)Survival depended strongly on age and kittens had the lowest

overall rate (032plusmn 009 Fig 5A Table 3) The model‐averagedkitten survival was 047plusmn 009 when we included only geneti-cally sampled individuals in the analysis (Table 3) as statedabove this estimate is biased high because many kittenswere genetically sampled when they were recaptured alive orrecovered dead at older ages We found no evidence for sex‐specific differences in kitten survival but females survived betterthan males in all older age classes For females subadults hadthe highest survival rates followed by prime adults and oldadults For males prime adults had the highest survival ratefollowed by subadults and old adults (Fig 5A and Table 3)For panthers of all ages there was substantial evidence that

ancestry affected survival with F1 admixed panthers of all ageshaving the highest rates and canonical individuals having thelowest (Fig 5B) The combined AIC weights of models thatincluded an ancestry covariate were 0880 for kittens and 0992for older panthers The survival rates of subadult prime‐adultand old‐adult panthers in the period immediately after the in-trogression (1995ndash2004) were similar to those in more recentyears (2005ndash2013 Fig 5C)After genetic introgression individual panther heterozygosity

increased (Fig 6) and varied among ancestry categories individualheterozygosity was the highest for F1 panthers (078plusmn 001)

A

B

C

Figure 5 Overall age‐ and sex‐specific survival rates (plusmnSE A) age‐ and sex‐specific survival rates (plusmnSE) for Florida panthers of canonical F1 admixed andother admixed ancestry (B) and age‐ and sex‐specific survival estimates (plusmnSE)for subadult and adult Florida panthers in the period immediately post‐introgression (1995ndash2004) and more recently (2005ndash2013 C) in SouthFlorida USA

Table 3 Model‐averaged estimates (plusmnSE) of demographic parameters for theFlorida panther obtained using data collected during 1981ndash2013 in SouthFlorida USA

Parameter Sex Age class Estimate

Survival Both Kitten 032plusmn 009a

Female Subadult 097plusmn 002Prime adult 086+ 003Old adult 078plusmn 009

Male Subadult 066plusmn 006Prime adult 077plusmn 005Old adult 065plusmn 010

Probability of reproduction Female Subadult 035plusmn 008Prime adult 050plusmn 005Old adult 025plusmn 006

Kittens produced annually Female Subadult 280plusmn 005Prime adult 267plusmn 001Old adult 228plusmn 013

a Overall kitten survival (based on all kittens in our data set including thosethat were not genetically sampled) Estimate of survival of geneticallysampled kittens was 047plusmn 009 this estimate is biased high because manykittens were genetically sampled when they were recaptured alive or re-covered dead at older ages

van de Kerk et al bull Florida Panther Population and Genetic Viability 17

followed by those for non‐F1 admixed (062plusmn 001) and canonical(039plusmn 001) panthers Individual heterozygosity positively affectedsurvival rates of kittens (slope parameter η= 1455 95CI= 0505ndash2405) Individual heterozygosity also positively af-fected survival of subadult and adult panthers but the evidence forthis effect was weaker because the confidence interval for the ha-zard ratio overlapped 10 (hazard ratio exp(ɩ)= 0311 95CI= 0092ndash1054 Table 4 and Fig 7)The MPC increased after genetic introgression (Fig 2) This

abundance index was also included in top‐ranking modelsindicating that population density affected survival (Table 4)Abundance reduced the survival of kittens (slope parameterη= minus0035 95 CI= minus0049 to minus0021) evidence was weak forthe effect of abundance on survival of subadult and adult panthers(hazard ratio exp(ɩ)= 1002 95 CI= 0995ndash1010 Fig 7) Re-gression coefficients associated with the effect of abundance andindividual heterozygosity on demographic parameters are presentedin Table B1 (available online in Supporting Information)The most frequently observed causes of death for radio‐

collared panthers were vehicle collision and intraspecific ag-gression Cause‐specific mortality analyses revealed thatmortality rates among possible causes were generally similar forthe 2 sexes but that males were more likely to die from in-traspecific aggression than were females (Fig 8A) Male mor-tality from intraspecific aggression was higher for subadult andold adult panthers than for prime adults (Fig 8B) For bothmales and females canonical panthers were more likely to diefrom intraspecific aggression than were admixed panthers(Fig 8C)

Reproductive ParametersOf 101 radio‐collared females 67 reproduced at least once duringour monitoring we used these individuals to assess the annualprobability of reproduction The top‐ranked model for the annual

probability of reproduction included age effect only suggesting thatadding ancestry or abundance did not improve model parsimony(Table 4) Model‐averaged annual probabilities of reproductiondiffered between female subadults (035plusmn 008) prime adults(050plusmn 005) and older adults (025plusmn 006)The most parsimonious model for the number of kittens

produced annually by females that produced at least 1 litter (ieconditional on reproduction) included effects of age ancestryand abundance a model that included only the effect of age wasalmost equally supported (Table 4) The model‐averagednumber of kittens produced annually conditional on re-production was 280plusmn 075 for subadults 267plusmn 043 for primeadults and 228plusmn 083 for old adults Confidence intervals forthe slope parameter (β) overlapped 0 for both abundance(β= 0010 95 CI= minus0001 to 0021) and ancestry (β= minus073795 CI= minus1637 to 0162)

Population Dynamics and PersistenceDeterministic and stochastic demographymdashThe deterministic (λ)

and stochastic (λs) annual population growth rate calculated usingthe matrix population model were 106 (5th and 95th percentiles099ndash114) and 104 (072ndash141) respectively The generation timewas 47 years A female starting in age class one (kitten) wasexpected to live another 58 years and produce 10 kittens onaverage during her lifetime Overall λs was proportionately(elasticity) most sensitive to changes in female prime adultsurvival on the absolute scale (sensitivity) however it was mostsensitive to changes in kitten survival (Fig 9) Populationtrajectories probabilities of quasi‐extinction and times untilquasi‐extinction estimated using the matrix population modeland IBM for comparable scenarios were nearly identical for acritical quasi‐extinction threshold of 10 panthers (Figs 10 and 11)and for a critical quasi‐extinction threshold of 30 panthers(Appendix E available online in Supporting Information)Minimum population count scenariomdashUnder the MPC scenario

(ie density dependence estimated using the minimumpopulation count data) with a critical threshold of 10panthers the population size reached 99 (72ndash104) and 100(77ndash105) at 200 years (Fig 10) as predicted by the IBM and thematrix model respectively The cumulative quasi‐extinctionprobabilities within 100 years based on the MPC scenario were15 (IBM 0ndash48) and 30 (matrix model 0ndash76) Theseprobabilities increased to 25 (IBM 0ndash114) and 38 (matrixmodel 0ndash154) by year 200 (Fig 10) The mean times untilquasi‐extinction were 15 years (IBM 0ndash67) and 15 years (matrixmodel 0ndash72) conditioned on going quasi‐extinct within 100years Expected times until quasi‐extinction conditioned onquasi‐extinction within 200 years were higher 34 years (IBM0ndash137) and 32 years (matrix model 0ndash134 Fig 10)Motor vehicle mortality scenariomdashUnder the MVM scenario

(ie density dependence estimated using estimates of pantherpopulation size from McClintock et al 2015) with a criticalthreshold of 10 panthers the projected population reached 188(142ndash217) and 190 (143ndash219) at 200 years as predicted by theIBM and the matrix model respectively (Fig 11) Thecumulative quasi‐extinction probabilities within 100 years were14 (IBM 0ndash08) and 13 (matrix model 0ndash06) Theseprobabilities increased to 20 (IBM 0ndash17) and 19 (matrix

000

025

050

075

100

1995 2000 2005 2010Year of birth

Indi

vidu

al h

eter

ozyg

osity

Figure 6 Temporal changes in the individual heterozygosity of Floridapanthers born during the period 1995ndash2013 in South Florida USA Pointsare measurements solid line is linear regression shaded area is 95 confidenceinterval of slope Slope= 0006plusmn 0001 R2= 009

18 Wildlife Monographs bull 203

model 0ndash16) within year 200 (Fig 11) The mean times until quasi‐extinction based on the MVM scenario were 5 years (IBM 0ndash61)and 6 years (matrix model 0ndash64) conditioned on going quasi‐extinctwithin 100 years Expected times until quasi‐extinction conditionedon quasi‐extinction within 200 years were 11 years (IBM 0ndash113)and 11 years (matrix model 0ndash112 Fig 11)

Sensitivity of extinction probability to demographic parametersmdashThe partial rank correlation coefficients between quasi‐extinctionprobabilities and demographic parameters were negative anincrease in any of the demographic parameters decreased thequasi‐extinction probability Probabilities of quasi‐extinction within200 years were most sensitive to changes in kitten survival for bothscenarios (Fig 12) followed by survival of subadult females in theMVM scenarios and survival of prime adult females in the MPCscenario The probability of reproduction of prime adult femaleshad the third‐largest partial rank correlation coefficient with quasi‐extinction probability for both scenarios

Benefits and Costs of Genetic ManagementMinimum population count scenariomdashThe MPC scenario with

a critical threshold of 10 panthers predicted that withoutmanagement intervention average individual heterozygositywould decrease reaching 057 (049ndash064) at 100 years and053 (042ndash062) at 200 years (Fig 13) The population woulddecrease slightly reaching an average of 79 (41ndash99) at 100 yearsand 76 (43ndash98) at 200 years (Fig 14) The probabilities of quasi‐extinction under the no‐intervention MPC scenario when weconsidered the effects of genetic erosion were 13 (0ndash99) at 100years and 23 (0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with

the translocation of 5 Texas females average individualheterozygosity under the MPC scenario increased and thenstabilized at 068 (064ndash072) at 100 years and remained atthat level at 200 years (Fig 13) The population reached anaverage of 88 (62ndash104) at 100 years and stayed the same at200 years (Fig 14) The probabilities of quasi‐extinctionunder this introgression strategy were 6 (0ndash53) within 100years and 7 (0ndash82) within 200 years (Fig 15) Results ofanalyses using a critical threshold of 30 panthers revealed asimilar pattern of relative benefits However the prob-abilities of quasi‐extinction were substantially higher (ge45)across all scenarios (Appendix E)

Motor vehicle mortality scenariomdashWithout managementintervention the average individual heterozygosity predictedby the MVM scenario and a critical threshold of 10 panthersdecreased to 060 (046ndash069) at 100 years and to 059 (039ndash070) at 200 years (Fig 13) The population increased slightlyand reached an equilibrium at 154 individuals (46ndash217) at 100years and 157 (57ndash218) at 200 years (Fig 14) The probabilitiesof quasi‐extinction were 17 (0ndash100) at 100 years and 22(0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with 5

female pumas from Texas average individual heterozygosityincreased slightly reaching 067 (060ndash073) at 100 years and068 (059ndash075) at 200 years (Fig 13) Under this strategy thepopulation reached an average of 164 individuals (44ndash226) at

Table 4 Model comparison results testing for the effects of several covariateson survival of all kittens survival of genetically sampled kittens survival ofsubadult and adult Florida panthers probability of reproduction and kittensproduced annually for Florida panthers sampled from 1981ndash2013 in SouthFlorida USA

Model Parameters ΔAICa Weight

Kitten survival (all data)Base+ F1b+ abundancec 10 000 0988Base+ abundance 9 1018 0006Base+ F1 9 1035 0005Base 8 2711 lt0001Base+ sexd 9 2912 lt0001Base+ litter sizee 9 2914 lt0001

Kitten survival (genetically sampled kittens only)Base+ F1+ abundance 10 000 0541Base+ F1CanAdmf+ abundance 11 099 0329Base+Hetg+ abundance 10 310 0115Base+CanAdmh+ abundance 10 791 0010Base+ abundance 9 944 0005Base+ F1 9 1638 lt0001Base+ F1CanAdm 10 1837 lt0001Base+Het 9 2872 lt0001Base 8 3296 lt0001Base+CanAdm 9 3348 lt0001

Subadult and adult survivalBase+ F1CanAdm+ abundance 7 000 0360Base+ F1 5 053 0276Base+ F1CanAdm 6 105 0213Base+ F1+ abundance 6 222 0119Base+CanAdm+ abundance 6 575 0020Base+CanAdm 5 908 0004Base+Het 5 964 0003Base+Het+ abundance 6 984 0003Base 4 1118 0001Base+ abundance 5 1278 0001

Probability of reproductionAge 3 000 0242Age+CanAdm 4 012 0228Age+ abundance 4 173 0102Age+ F1 4 190 0094Age+CanAdm+ abundance 5 216 0082Age+Het 4 219 0081Age+ F1CanAdm 5 232 0076Age+ F1+ abundance 5 372 0038Age+Het+ abundance 5 399 0033Age+ F1CanAdm+ abundance 6 444 0026

Kittens produced annuallyAge+ F1+ abundance 5 000 0194Age 3 060 0144Age+ abundance 4 061 0143Age+ F1 4 097 0120Age+CanAdm 4 139 0097Age+ F1CanAdm+ abundance 6 196 0073Age+ F1CanAdm 5 231 0061Age+CanAdm+ abundance 5 238 0059Age+Het+ abundance 5 250 0056Age+Het 4 259 0053

a Akaikersquos information criterion (AIC) for kitten survival models was adjustedfor small sample size and overdispersion (QAICc)

b F1 divides Florida panthers into 2 groups F1 admixed and all other panthersc Index of Florida panther abundanced Sex of an individual Florida panther (male or female)e Size of the litter in which a Florida panther kitten was bornf F1CanAdm divides panthers into 3 groups F1 admixed canonical andother admixed panthers

g Het is the individual heterozygosityh CanAdm divides panthers into 2 groups canonical and admixed (includingF1 admixed) panthers

van de Kerk et al bull Florida Panther Population and Genetic Viability 19

100 years and 170 (67ndash231) at 200 years based on the MVMscenario (Fig 14) The probabilities of quasi‐extinction underthis introgression strategy were 10 (0ndash99) at 100 years and12 (0ndash99) at 200 years (Fig 15)The projected population size and average individual hetero-

zygosity reached equilibrium levels with periodic fluctuations inthese values when introgression was implemented (Figs 16ndash17)Genetic introgression decreased the time until quasi‐extinctionacross all scenarios (Fig 18) Results of analyses using a criticalthreshold of 30 panthers revealed a similar pattern of relative ben-efits However the probabilities of quasi‐extinction were sub-stantially higher (ge45) across all scenarios (Appendix E)

Addition of pumas from Texas increased both the populationsize and average individual heterozygosity consequentlyintrogression caused periodic increases in average individualheterozygosity and population size at the interval at which theintrogression occurred Whereas average individual hetero-zygosity gradually decreased following introgression densitydependence caused the population size to fluctuate especiallywhen a larger number of pumas from Texas were released Thepositive effect of genetic introgression initiatives on quasi‐extinction probabilities decreased as the time interval betweenthese events increased More frequent genetic introgressions ledto greater increases in average individual heterozygosity and

Old adult

Prime adult

Subadult

Kitten

025 050 075

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Individual heterozygosity

Ann

ual s

urvi

val r

ate

Old adult

Prime adult

Subadult

Kitten

40 60 80 100 120

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Abundance index

Ann

ual s

urvi

val r

ate

Kittens Males Females

Figure 7 The effect of individual heterozygosity (left) and the abundance index (right) on Florida panther age‐ and sex‐specific survival estimates (shaded area isSE) in South Florida USA 1981ndash2013 Abundance index data are from McBride et al (2008) and R T McBride and C McBride (Rancherrsquos Supply Incorporatedunpublished report)

20 Wildlife Monographs bull 203

equilibrium population size while reducing the probability ofquasi‐extinction Comparatively performing genetic introgres-sion less often but with more individuals per release was lesseffective at improving the long‐term prospects for the pantherpopulation (Figs 13ndash15)

Financial cost and persistence benefitsmdashThe total estimated costof 1 genetic introgression was $40000 $80000 or $120000for releasing 5 10 or 15 pumas from Texas respectively(Table 5) The total cost incurred over 100 years for eachstrategy ranged from $50000 to $12 million (Table 6)The most expensive strategy involved the introduction of15 pumas every 10 years this strategy reduced the probability

of quasi‐extinction by 58 and 40 under the MPC and MVMscenarios respectively (Table 6) Introducing 5 pumas every80 years cost the least but improvements in populationpersistence under this strategy were insubstantial (Table 6)Releasing more panthers more frequently caused increasedpopulation fluctuations but did not necessarily reduce the quasi‐extinction probability (Figs 14 and 15 Table 6)

DISCUSSION

The duration (34 yr of data) and sample sizes associated with theFlorida Panther Project allowed us to obtain a comprehensiveunderstanding of genetic variation demographic parameters

A

B

C

Figure 8 Cause‐specific mortality rates (plusmnSE) for male and female Florida panthers (A) subadult prime‐adult and old‐adult Florida panthers of both sexes (B)and canonical and admixed Florida panthers of both sexes (C) in South Florida USA 1981ndash2013 Causes of mortality are vehicle collision intraspecific aggression(ISA) other causes (known causes such as diseases injuries and infections unrelated to the first 2 causes) and unknown cause (mortalities for which we could notassign a cause)

van de Kerk et al bull Florida Panther Population and Genetic Viability 21

probability of population persistence and the duration of thepositive impacts of genetic restoration on this endangered po-pulation The Florida panther population was in dire straits inthe early 1990s and our results clearly demonstrated that the

implementation of the introgression project in 1995 subse-quently accrued a series of genetic and demographic improve-ments that have led to the change from a declining population toone that is growing and expanding its current breeding range

Sensitivity Elasticity

0 1 2 3 00 02 04

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

Para

met

er

Figure 9 Sensitivity and elasticity (proportional sensitivity) of stochastic population growth rate (λs) of the Florida panther to vital demographic parameters in SouthFlorida USA 1981ndash2013 Horizontal lines represent 5th and 95th percentiles

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

70

80

90

100

110

0

5

10

15

0

50

100

Years

StatisticMeanMedian

Figure 10 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the minimum population count(MPC) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

125

150

175

200

00

05

10

15

20

0

30

60

90

Years

StatisticMeanMedian

Figure 11 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the motor vehicle mortality(MVM) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

22 Wildlife Monographs bull 203

MPC MVM

minus08 minus06 minus04 minus02 00 minus08 minus06 minus04 minus02 00

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Male subadult survival

Male prime adult survival

Male old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

PRCC

Para

met

er

Figure 12 Partial rank correlation coefficient (PRCC) between quasi‐extinction probabilities and the demographic parameters of the Florida panther for theminimum population count (MPC) and motor vehicle mortality (MVM) scenarios Error bars indicate standard deviation Based on data collected in South FloridaUSA 1981ndash2013

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 13 Average projected individual heterozygosities in the Florida panther population over 200 years without genetic management intervention and with eachof the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include theintroduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years Average individual heterozygosity peaks when pumas are added to the Floridapanther population Based on data collected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 23

Our research has further validated the success of genetic in-trogression by quantifying the reduction in the probability ofextinction of the panther population because of this manage-ment initiative These findings all bode well for continuedprogress toward recovery That said the Florida panther po-pulation remains small and isolated from other populations ofpuma from western North America thereby denoting that theeventual impact of genetic drift and inbreeding may negativelyaffect the population in the future Realizing this will continueto be an issue that managers will have to contemplate wemodeled the impact of varied genetic management scenarios todetermine the frequency of introgression along with the numberof panthers that should be released during each initiative tominimize cost and maximize benefits to the population Theseresults will prove invaluable to managers attempting to conservethe Florida panther

Genetic Variation Demographic Parametersand Persistence of Heterotic Benefits from GeneticIntrogressionOur measure of individual heterozygosity (1 minusHL) was higheron average for the admixed (0615) panthers than for those ofcanonical ancestry (0386) Levels of individual heterozygosityfor admixed panthers were comparable to levels observed in asample of pumas from west Texas (x plusmn SE= 0695plusmn 0019n= 41) which further demonstrates the improvement in levelsof genetic variation in the Florida population since 1995 The

correlation between HL and several demographic parametershas been noted in wild populations including cheetahs (Terrellet al 2016) and several species of birds such as European shags(Phalacrocorax aristotelis male and female survival probabilityVelando et al 2015) and Egyptian vulture (Neophron percnop-terus age at recruitment Agudo et al 2012) If we focus only onpanthers captured and radiocollared from 2012ndash2015 (FP211ndashFP240) all of which had estimated birth years ge2005 (ge10 yrpost‐introgression) those panthers had an average individualheterozygosity level of 0618plusmn 0029 highlighting the elevatedlevels of genetic variation that continue to persist in cohorts ofpanthers 5 generations after the release of female pumas fromTexasThe proportional membership values (q) from our

STRUCTURE analysis allowed us to delineate 2 clusters ofancestry for Florida panthers (Fig 4) During the pre‐in-trogression era the population was dominated by panthers thatretained a high percentage of the canonical ancestry which wasaffected by inbreeding depression The post‐introgression eraancestry values are comprised of many admixed individuals inessence demonstrating the success of the introgression in termsof increasing genetic variation in the population and mi-micking gene flow that historically occurred with panthers andother populations of pumas (Onorato et al 2010) A somewhatanalogous situation although abetted by natural immigrationwas evident in the ancestral variation in a small population ofpuma intersected by a major freeway in California USA In

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

75

80

85

90

95

100

130

150

170

190

Years

Popu

latio

n si

ze

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 14 Average projected population sizes in the Florida panther population over 200 years without intervention and with each of the genetic introgressionstrategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include the introduction of 5 10 or 15pumas from Texas USA every 10 20 40 or 80 years Peaks occur when pumas are added to the Florida panther population Based on data collected in SouthFlorida USA 1981ndash2013

24 Wildlife Monographs bull 203

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

25

50

75

100

0

25

50

75

100

Introgression interval (yrs)

PQ

E (

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 15 Average probability of quasi‐extinction (PQE) of the Florida panther population within 100 years and within 200 years without genetic managementintervention (N) and with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenariosIntrogression strategies include the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Errorbars indicate 5th and 95th percentiles Based on data collected in South Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

50

60

70

80

90

100

110

50

100

150

200

Years

Popu

latio

n si

ze

Figure 16 Average projected Florida panther population trajectories under a geneticintrogression strategy involving the release of 5 female pumas from Texas USA every20 years for the minimum population count (MPC) and motor vehicle mortality(MVM) scenarios Shaded area indicates 5th and 95th percentiles and isrepresentative of confidence intervals for other scenarios Based on data collected inSouth Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Figure 17 Average projected Florida panther population‐level heterozygositiesunder a genetic introgression strategy involving the release of 5 female pumas fromTexas USA every 20 years for the minimum population count (MPC) and motorvehicle mortality (MVM) scenarios Shaded area bounds the 5th and 95th percentilesand is representative of confidence intervals for other scenarios Based on datacollected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 25

this case the migration of just a single male across the freewayto the south resulted in an increase in the number of in-dividuals with admixed ancestry and substantially improvedgenetic diversity of the subpopulation south of the freeway(Riley et al 2014)Our estimate of overall kitten survival (0321plusmn 0056) was

similar to that reported by Hostetler et al (2010) who used13 years of post‐introgression data (0323plusmn 0071) These valuesare lower than kitten survival estimates derived from severalpuma populations in western North America (044ndash091 Loganand Sweanor 2001 Lambert et al 2006 Laundre et al 2007Robinson et al 2008 Ruth et al 2011) When we included onlydata for individuals that had been genetically sampled our es-timate was 15 higher The proportion of admixed kittens thatwere genetically sampled increased as the population expandedwhich could have resulted in higher estimates of kitten survivalbecause admixed kittens survive better than canonical kittensKitten survival was strongly influenced by genetic ancestry withsurvival rates of F1 kittens being almost twice that of canonicalkittens (Fig 5B) Consistent with the findings of Hostetler et al(2010) increases in heterozygosity coincided with increasedkitten survival whereas population density negatively affectedkitten survival Population density often has a strong negativeimpact on survival of young age classes due to infanticide andcannibalism which has been reported from several carnivore

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

50

100

150

0

50

100

150

Introgression interval (yrs)

TQE

(yrs

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 18 Average times until quasi‐extinction (TQE) for the Florida panther population within 100 years and within 200 years without genetic management interventionand with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategiesinclude the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Error bars indicate 5th and 95thpercentiles Based on data collected in South Florida USA 1981ndash2013

Table 5 Average cost (2017 US dollars) of introgression strategies employedover 100 years that involve the introduction of 5 10 or 15 pumas from TexasUSA into the Florida panther population in South Florida USA at an in-creasingly higher frequency Costs of capture (including transportation) quar-antine and post‐introgression monitoring were estimated by Roy McBrideMark Cunningham and Dave Onorato respectively

Frequency Component 5 pumas 10 pumas 15 pumas

Once Capture 25000 50000 75000

Quarantine 5000 10000 15000

Monitoring 10000 20000 30000

Total 40000 80000 120000

Every 80 years Capture 31250 62500 93750

Quarantine 6250 12500 18750

Monitoring 12500 25000 37500

Total 50000 100000 150000

Every 40 years Capture 62500 125000 187500

Quarantine 12500 25000 37500

Monitoring 25000 50000 75000

Total 100000 200000 300000

Every 20 years Capture 125000 250000 375000

Quarantine 25000 50000 75000

Monitoring 50000 100000 150000

Total 200000 400000 600000

Every 10 years Capture 250000 500000 750000

Quarantine 50000 100000 150000

Monitoring 100000 200000 300000

Total 400000 800000 1200000

26 Wildlife Monographs bull 203

species including polar bears (Ursus maritimus Derocher andWiig 1999) black bears (Ursus americanus Garrison et al 2007)and pumas (Logan and Sweanor 2001)Our estimates of sex‐ and age‐specific survival rates of subadult

and adult panthers (Table 3) and the effects of covariates on theserates were similar to those reported by Benson et al (2011) derivedfrom data collected through 2006 Our survival rates for males(065ndash077) were lower than females (078ndash097) for all 3 ageclasses (subadult prime adult and old adult) which is the typicaltrend observed in most puma populations (eg Ruth et al 19982011) However an unhunted puma population occupying afragmented urban landscape in southern California exhibited lowersurvival (0586 females 0525 males Vickers et al 2015) perhapsas a result of severe habitat fragmentation and the lack of extensiveswaths of public land protected in perpetuity Most populations ofpuma in western North America are typically subject to variedlevels of management via regulated hunting which often is biasedtoward the take of males (Cooley et al 2009 Ruth et al 2011)Consequently annual survival estimates from hunted populationsin northeast Washington (0679ndash0733 females 0341 malesRobinson et al 2008) and southeastern Arizona (0685 females0584 males Cunningham et al 2001) were generally lower thanthose we estimated for Florida panthersCause‐specific mortality analysis showed that male panthers

experienced a substantially greater mortality risk from in-traspecific aggression This differs from the pattern observedduring a long‐term study in Arizona where females were at ahigher risk of intraspecific mortality than males (46 of maleand 53 of female mortality Logan and Sweanor 2001)Mortality associated with intraspecific strife has been docu-mented in hunted and unhunted populations (this study Loganand Sweanor 2001 Stoner et al 2010) and its root cause hasbeen difficult to determine (eg population density and preypopulation trends) That being the case finding ways to reducewhat is often a major source of mortality for puma populationscontinues to pose a challenge to managersCanonical panthers were substantially more likely to die from

intraspecific aggression than were admixed panthers This mayat least partly explain the difference in survival between admixed

and canonical panthers Canonical panthers are known to bemore likely to suffer from varied correlates of inbreeding de-pression than admixed animals including atrial septal defects(Johnson et al 2010) and compromised immune systems(Roelke et al 1993) These factors have the potential to affectfitness and perhaps dominance of individuals when engaged inan intraspecific encounter A somewhat similar scenario wasobserved by Hogg et al (2006) in a population of bighorn sheepthat were part of an introgression project Admixed males in thispopulation had a greater probability of paternity than males thatwere less outbred This included coursing males with higherlevels of admixture being markedly more successful at fightingmales that were defending females in estrous (Hogg et al 2006)Heterosis resulting from genetic introgression is expected to be

the strongest in F1 individuals (Shull 1908 Crow 1948 Burkeand Arnold 2001 Waller 2015) and to progressively decline insubsequent generations (Pickup et al 2013 Frankham 2015)Given that admixed (especially F1) panthers survived better thancanonical panthers and that individual heterozygosity improvedsurvival of both sexes and all age classes our data provide evi-dence for heterotic advantage whereby individuals of mixedancestry had higher fitness (Shull 1908 Crow 1948) Our resultsare consistent with findings of other studies that involved in-tentional genetic introgression for conservation purposes Forexample Hogg et al (2006) detected substantial improvementsin survival and reproduction of introgressed bighorn sheep andMadsen et al (1999 2004) reported a dramatic increase in thepopulation of adders after genetic introgressionKnowledge of the persistence of benefits to a population ac-

crued via genetic introgression is essential for determiningwhen additional introgression may be warranted There is adepauperate amount of information regarding this topic and theidentification of factors that may influence persistence of thebenefits of genetic introgression (Tallmon et al 2004 Hedricket al 2014 Whiteley et al 2015) For example in minks(Neovison vison) Thirstrup et al (2014) found that outcrossingled to larger litters in F2 and F3 but this benefit disappeared insubsequent generations In a population of gray wolves Hedricket al (2014) suggested that benefits of genetic introgression may

Table 6 Costs and benefits accumulated over 100 years for genetic introgression at different intervals and with different numbers of pumas from Texas USAintroduced into the Florida panther population in South Florida USA Costs (2017 US dollars) include capture and transportation quarantine and post‐introgression monitoring Benefits are represented as average probability of quasi‐extinction (PQE and 5th and 95th percentiles) and changes (Δ) therein under thescenario using the minimum population count (McBride et al 2008 MPC) and the scenario using the motor vehicle mortality population estimates (McClintocket al 2015 MVM) The table is sorted by percent change in probability of quasi‐extinction under the MPC scenario

Interval (yr) Pumas Cost MPC PQE () ΔMPC PQE () MVM PQE () ΔMVM PQE ()

ndash 0 0 13 (0ndash99) 0 17 (0ndash100) 080 10 100000 12 (0ndash99) 10 (0ndash58) 16 (0ndash100) 5 (0ndash38)80 15 150000 12 (0ndash99) 13 (0ndash65) 15 (0ndash100) 8 (0ndash56)80 5 50000 12 (0ndash99) 14 (0ndash73) 15 (0ndash99) 9 (0ndash41)40 15 300000 10 (0ndash98) 26 (0ndash104) 14 (0ndash99) 13 (0ndash60)40 10 200000 9 (0ndash94) 35 (0ndash183) 13 (0ndash99) 21 (0ndash95)40 5 100000 8 (0ndash89) 39 (0ndash211) 12 (0ndash99) 26 (0ndash124)20 15 600000 8 (0ndash88) 42 (0ndash257) 13 (0ndash99) 24 (0ndash123)20 10 400000 6 (0ndash67) 54 (0ndash318) 10 (0ndash99) 37 (0ndash217)10 15 1200000 6 (0ndash53) 58 (0ndash372) 10 (0ndash99) 40 (0ndash208)20 5 200000 5 (0ndash50) 59 (0ndash338) 9 (0ndash99) 44 (0ndash352)10 10 800000 4 (0ndash16) 69 (0ndash489) 8 (0ndash92) 55 (0ndash401)10 5 400000 4 (0ndash7) 73 (0ndash502) 6 (0ndash71) 63 (0ndash503)

van de Kerk et al bull Florida Panther Population and Genetic Viability 27

wane after 2 or 3 generations The WrightndashFisher model ofgenetic drift predicts a geometric decline in expected hetero-zygosity at the rate of (1 minus 12Ne) per generation where Ne is theeffective population size (Hartl and Clark 1997 Frankham et al2014) Thus it seems logical to assume that the duration of thebenefits of introgression is limited especially in a species per-sisting in a small population such as Florida panthersWe expected Florida panther survival in the most recent years

(2005ndash2013) post‐introgression to differ from that observed in thedecade following the introgression in 1995 because pantherabundance and heterozygosity factors known to influence panthersurvival (Hostetler et al 2010 Benson et al 2011) have changed(Figs 2 and 6) We found that subadult and adult survival rates inrecent years (2005ndash2013) were similar to those in the period im-mediately following introgression (1995ndash2004 Fig 5C) overallkitten survival was similar to estimates of Hostetler et al (2010) Infact the pattern of covariate effects on Florida panther survivalreported by Benson et al (2011) and Hostetler et al (2010) hasbarely changed The negative effects of population density andpositive effects of heterozygosity may counterbalance survival ratesreducing population fluctuations Our result that benefits of geneticrestoration have remained in the population nearly for 5 genera-tions post‐intervention was surprising but also encouraging Pos-sible explanations for this observation may include 1) genetic ero-sion occurs at slower rate than previously thought 2) introducing 5migrants in 1 genetic intervention may have beneficial effects si-milar to those from 1 migrant per generation over multiple gen-erations or 3) other and as yet unknown factors may reduce therate of genetic erosion and subsequent demographic effects Adensity‐dependent effect on kitten survival could also explain whynon‐F1 admixed kittens did not survive substantially better thandid canonical kittens most kittens sampled from 2005ndash2013 wereadmixed and their survival was lower possibly because of a density‐dependent effect as the Florida panther population continuouslygrew during that period Trinkel et al (2010) also found thatpopulation density and inbreeding coefficient interacted to affectdemographic parameters and growth rate of an African lion po-pulation Likewise population density interacted with winterweather to affect the survival and population growth rates in Soaysheep (Ovis aries Milner et al 1999 Coulson et al 2001)

Population Dynamics and PersistenceThe finite deterministic and stochastic growth rates were gt10reflecting that much of the data used in this study were collectedfollowing genetic introgression in 1995 during which time thepopulation increased substantially Because our demographicparameter estimates did not change markedly in comparison tothose of Hostetler et al (2013) the similarity between thepopulation growth estimates from these studies was expectedBut these estimates reflect population growth during thedata‐collection period and do not predict future growth In factestimates of population size reported by McClintock et al(2015) suggest that population growth may already be slowing(Fig 2) A similar pattern was observed in an introduced po-pulation of African lion which increased steadily from 1995until 2001 then fluctuated around the presumed carrying ca-pacity thereafter (Trinkel et al 2010) Several other African lionpopulations that experienced various degrees of recovery

subsequently became relatively stable or exhibited decliningtrends (Bauer et al 2015)Our estimates of quasi‐extinction probabilities were lower than

those reported by Hostetler et al (2013) using a 2‐sex age‐structured matrix population model Lower probabilities ofquasi‐extinction than those reported by the earlier study forcomparable scenarios were likely a consequence of the fact thatthe estimates of abundance (McBride et al 2008) and thus thedensity‐dependent effects on survival of kittens and old panthers(gt10 yr) have changed in recent years Additionally these earlyestimates of the probability of quasi‐extinction and of time toquasi‐extinction did not consider genetic erosion that will in-evitably occur without management interventionUnder the MVM scenario the equilibrium population size was

twice that attained under the MPC scenario The MVM popula-tion estimates are approximately twice the MPCs (Fig 2) as aresult the simulated population under the MVM scenario achieveda higher equilibrium size Probability of quasi‐extinction under theMVM scenario was generally greater than that under the MPCscenario This result is a consequence of the fact that the estimateddensity dependence on kitten survival based on the MPC scenario isstronger than that for theMVM scenario (regression coefficients forabundance for the MPC scenario= minus0068 vs minus0002 for theMVM scenario Appendix B Table B1) Consequently simulatedpopulations under the MPC scenario have a stronger tendency tomove toward the equilibrium population size which inherentlyreduces the quasi‐extinction probability (eg Royama 1992)As expected for long‐lived mammals (Heppell et al 2000 Oli

and Dobson 2003) the population growth rate was proportio-nately most sensitive to changes in survival of prime adultsfollowed by that in younger age classes Global sensitivity ana-lysis in contrast showed that under all scenarios extinctionparameters were particularly sensitive to changes in survival ofFlorida panther kittens This result is a consequence of the highsensitivity of the population growth rate to kitten survival(Fig 9) and a larger standard error of kitten survival (Table 3)Results of our sensitivity analyses indicate that future researchshould focus on acquiring additional information on early lifestages to improve the accuracy and precision of estimates ofkitten survival Our results suggest that demographic variables(or their environmental drivers) that strongly influence extinc-tion risks are not necessarily those with the largest elasticityvalues A study of island fox (Bakker et al 2009) found thatextinction risk was strongly influenced by survival modifierparameters that had low elasticity values

Genetic Management When and How to InterveneThe use of genetic introgression as a management tool has al-ways been controversial and the probable effectiveness it mighthave in preventing the imminent demise of the panther popu-lation was initially debated (Creel 2006 Maehr et al 2006Pimm et al 2006) These debates notwithstanding the pantherpopulation had suffered from genetic morphological and bio-medical correlates of inbreeding before this management in-tervention (Johnson et al 2010 Onorato et al 2010) whereasthe post‐introgression period has been characterized by a sub-stantial reduction in the biomedical correlates of inbreeding andimprovements in demographic vigor and abundance (McBride

28 Wildlife Monographs bull 203

et al 2008 Hostetler et al 2010 2013 Johnson et al 2010Benson et al 2011 McClintock et al 2015) Nevertheless thepopulation remains small and isolated habitat is still being lostand there is no current possibility of natural gene flow betweenthis and other North American puma populations thus therecurrence of inbreeding‐related problems and subsequent po-pulation declines are inevitable The question therefore is notwhether genetic management of the Florida panther populationis needed but when and how it should be implementedWithout genetic introgression probabilities of quasi‐extinc-

tion were substantially greater than those from models thatincorporated periodic genetic introgression events (Fig 15)highlighting the importance of genetic management in smallisolated populations When 5 female pumas are added to theFlorida panther population every 10 years genetic variationincreased substantially under the MPC and MVM scenariosThe IBM predicted that the equillibrium population size wouldbe substantially larger when 5 pumas were released into thepopulation every 10 years than populations without intervention(ie no introgression) Releasing 15 Texas females did not af-ford substantially greater benefit to the equilibrium populationsize than did releasing only 5 Texas females Instead addingmore individuals caused increasingly large fluctuations in po-pulation size due to the numerical increases and density de-pendence (Fig 14) possibly diminishing the benefits associatedwith increased genetic variationCompared with the no‐intervention scenario (ie no genetic

introgression implemented) the introduction of 5 female pumasevery 10 years reduced quasi‐extinction probabilities from 13to 4 and from 17 to 6 for the MPC and MVM scenariosrespectively The benefit of genetic‐introgression strategies de-creased as the interval between successive management inter-ventions increased and no benefit was evident once the intervalreached 80 years (Fig 15) Introgression reduced the averagetime until quasi‐extinction (Fig 18) This somewhat counter‐intuitive result can be explained by the fact that repeated geneticintrogression makes the population more robust over timewhich will reduce the probability of extinction Consequentlyfew simulated population trajectories that fall below the criticalthreshold do so early reducing the mean time to extinctionAlso density dependence has a stabilizing effect on populations(Royama 1992) populations tend toward equilibrium popula-tion sizes which reduces the probability over time of popula-tions falling below quasi‐extinction threshold However timeuntil quasi‐extinction must be interpreted in conjunction withthe probability of quasi‐extinction which is very low for allscenarios with genetic introgression (Fig 15)Cost is a significant impediment to the implementation of a

genetic introgression program because captures relocations andmonitoring efforts before and after introgression are expensive(Edmands 2007 Frankham 2010b 2015 2016) Costs may beacceptable to society if the management actions result in sub-stantial fitness benefits especially if the species of concern ispopular and charismatic (Frankham 2015) We estimated the100‐year cost of introgression strategies modeled here to rangefrom $50000 to $12 million More expensive strategies gen-erally led to larger decreases in the probability of quasi‐extinc-tion but cheaper strategies also substantially improved

population viability (ie reduced quasi‐extinction probabilityTable 6) One of the cheaper strategies (5 pumas released every20 years) which cost an estimated $200000 over 100 yearsreduced the probability of quasi‐extinction by 59 and 44 forthe MPC and MVM scenarios respectively But these pre-liminary cost estimates are based on expenses incurred duringthe 1995 introgression and include costs of capture quarantinetransportation release and post‐release monitoring of femalesonly Although the cost for future genetic interventions wouldlikely be higher than those reported here the relative differencesin cost between alternative intervention strategies should remainapproximately the sameOur ability to simulate genetics and link it to the viability of

the Florida panther population is based on the empirically es-timated relationship between individual heterozygosity andsurvival Such heterozygosity and fitness correlations have beenstudied for decades (David 1998) but have only recently beenused to predict population viability (Benson et al 2016) noneto our knowledge have incorporated cost into their modelingefforts The mechanisms underlying heterozygosity and fitnesscorrelations remain unclear (David 1998 Szulkin et al 2010)and their significance as a proxy for inbreeding depression hasbeen debated (Balloux et al 2004 Miller and Coltman 2014)Nevertheless evidence continues to mount demonstratingpositive relationships between heterozygosity and survival(Coulson et al 1998ab Hansson et al 2004 Silva et al 2005Acevedo‐Whitehouse et al 2006 Jensen et al 2007 Velandoet al 2015) and between heterozygosity and fecundity (Seddonet al 2004 Charpentier et al 2005 Agudo et al 2012 Velandoet al 2015) Although the reported correlations are generallyweak their consequences can be substantial if population growthand persistence parameters are highly sensitive to the affectedvital rates (Velando et al 2015) Compared with scenarios thatignore genetics incorporating the effects of inbreeding on sur-vival increased quasi‐extinction probabilities within a 100‐yeartimeframe from 15 to 13 and from 14 to 17 for theMPC and MVM scenarios respectively These results high-light the importance of incorporating genetics when analyzingthe viability of small isolated populationsAlthough conservation biologists have recognized the im-

portance of genetics in population viability many still fail toincorporate genetic factors into PVA models (Reed et al 2002)Explicit consideration of genetics in PVAs requires long‐termgenetic and demographic data This permits the assessment ofthe functional relationship between genetic variability and de-mographic parameters Population viability analysis softwarepackages such as VORTEX (Lacy 1993 2000) offer a powerfulframework for individual‐based simulations but they often donot adequately capture a speciesrsquo life history or genetics Basedon a thorough review of the importance of genetic factors inwildlife conservation Frankham et al (2014) concluded thatgenetic factors can strongly influence the dynamics and persis-tence of small andor fragmented populations They furthernoted that ldquoMost PVA‐based risk assessments ignore or in-adequately model genetic factorsrdquo and recommended that ldquoPVAshould routinely include realistic inbreeding depressionhelliprdquo(Frankham et al 201457) Our custom‐built IBM permitted usto develop a model that adequately captured the Florida panther

van de Kerk et al bull Florida Panther Population and Genetic Viability 29

life history and we could parameterize the model with our long‐term genetic and demographic dataThe explicit consideration of the effects of inbreeding on

Florida panther population dynamics and persistence representsan important step forward Nonetheless our model remainsimperfect First we neglected the possible effects of habitat lossdetrimental anthropogenic activities climate change the prob-ability of immigration of pumas from western populationsdisease or catastrophes on demographic rates and populationviability Although immigration from puma populations outsideFlorida is possible its probability is very low given the extensivechallenges the anthropogenically altered landscape would posefor such a migrant to make it successfully to South Florida Weomitted mutations from our model because we have no in-formation on mutation rates though we expect that they are lowbased on values reported for other mammals (Kumar and Sub-ramanian 2002) Finally we only considered possible effects ofinbreeding on age‐specific survival rates because there was noevidence that heterozygosity affected female reproduction Lossof heterozygosity can negatively affect reproduction becauseinbred male panthers can suffer from cryptorchidism which canadversely affect their fecundity However given the polygynousmating system we expect the Florida panther population growthis primarily female‐limitedAlthough it is generally accepted that genetic introgression

only temporarily relieves a population from inbreeding depres-sion we know of no cases in which it has been implementedmore than once to conserve a threatened wildlife populationOur study provides an example of how demographic and mo-lecular data can be used within an IBM framework to evaluatethe efficacy of alternative genetic management strategies via theFlorida panther as a case study Our model can be adjusted forand applied to other species and offers great potential as a toolfor genetic management of small isolated wildlife populations

MANAGEMENT IMPLICATIONS

Our results and those of Hostetler et al (2013) and Johnsonet al (2010) provide persuasive evidence that the genetic in-tervention implemented in 1995 prevented the demise of theFlorida panther and restored demographic vigor to the popu-lation The Florida panther population remains small and iso-lated from other puma populations the closest puma populationis located gt1000 km away in Texas and the intervening land-scape is highly developed and fragmented with many interstate(and other high‐traffic) highways and major urban centersTheory suggests that 1 migrant per generation is sufficient tominimize the loss of genetic diversity (eg Mills and Allendorf1996) However the possibility of successful migration of apuma to South Florida followed by successful mating every ~5years is very low considering the distance between Texas andFlorida (Hedrick 1995) and the many risks and barriers thatdispersing pumas face (Beier 1995 Maehr et al 2002 Thatcheret al 2006) Consequently the pantherrsquos long‐term persistencewill likely depend on subsequent timely genetic introgressionsgiven the near‐impossibility of natural gene flow from otherpuma populations To that end our study offers considerableinsights into benefits of different genetic management strategiesand associated costs

Without genetic management the Florida panther populationfaces a substantial risk of quasi‐extinction within 100 years (10ndash46 depending on quasi‐extinction threshold and scenarios)Twenty‐three years have passed since genetic introgression wasimplemented and the population appears healthy as indicatedby measures of genetic variation increases in the populationsize and improvements in age‐specific survival rates Ourfindings suggest that releasing more pumas more frequently doesnot necessarily improve persistence probability because such astrategy would cause larger fluctuations in population size(Fig 14) which negatively affects persistence probability Thusextinction risks faced by inbred populations of large carnivorescan be substantially reduced by releasing approximately 5 im-migrants every 2 decades We suggest that such a managementstrategy be followed only in conjunction with continued long‐term monitoring of the population Our analyses demonstratethat population growth and persistence are highly sensitive tokitten and female (adult and subadult) survival Continuing tocollect data on these demographic variables along with bio-metric and genetic sampling will remain important to pantherrecovery and vigilance in identifying issues associated with in-breeding depression The collection of these monitoring datashould allow managers to detect any demographic or geneticchanges in a timely fashion and respond with genetic in-trogression or other management actions if warrantedRecovery objectives as defined by the USFWS in its current

recovery plan (USFWS 2008) delineate the need for additionalpopulations in the historical range to downlist or delist theFlorida panther If the only extant population of Floridapanthers continued to grow it could serve as a source ofpanthers to be translocated to suitable habitat to seed addi-tional populations Maintaining current information on thegenetic health of the population and precise estimates of itssize will be important in assessing whether it can withstand theremoval of a subset of animals for translocation Although the2017 documentation of a female panther with kittens north ofthe current breeding range is encouraging in terms of thepotential for population expansionmdashgiven that no female hadbeen recorded north of the Caloosahatchee River since 1972mdashthe re‐establishment of separate new populations will mostlikely require translocations by wildlife managers and a lengthysociopolitical processConservation of threatened or endangered species such as the

Florida panther is inherently challenging For panthers and otherlarge carnivores that require vast extents of natural habitat topersist habitat is typically the most limiting factor that will de-termine whether recovery efforts succeed Although geneticmanagement invariably plays a key role in the long‐term persis-tence of the panther population even a healthy population caneventually succumb to the impacts of habitat loss The humanpopulation of Florida continues to be one of the fastest‐growingin the nation the United States Census Bureau currently ranksFlorida as the third most populous Therefore habitat loss isgoing to continue to be a key issue for panthers Diligence towardpreserving panther habitat in its historical range via conservationeasements or other means and trying to keep such areas inter-connected via corridors is a goal stakeholders such as governmentagencies sportsmen non‐governmental organizations ranchers

30 Wildlife Monographs bull 203

and other private landowners must work on collaboratively toachieve

ACKNOWLEDGMENTS

We thank D Shindle D Jennings T OrsquoMeara D Land andC Belden for valuable advice throughout the project We thankS Bass M Criffield M Cunningham D Giardina D JansenA Johnson D Land M Lotz C McBride Rocky McBrideRowdy McBride Roy McBride L Oberhofer S Schulze staffsat Everglades National Park and Big Cypress National Preserveand others for assistance with fieldwork We also thank JAustin M Conroy J Hines J Laake S Mills J Nichols JHines M Sunquist J Fryxell and T Coulson for advice andassistance regarding data analysis and panther biology TheMuseum of Texas Tech University Natural Science ResearchLaboratory provided samples from pumas from west Texas forcomparative genetic analyses M Ben‐David D Land WMcCown E Hellgren and 4 anonymous reviewers providedmany helpful comments on the manuscript We thank NereaUbierna Lopez for completing the Spanish translation of theabstract We are grateful to the Department of ResearchComputing University of Florida for excellent high‐perfor-mance computing services We would like to thank US Fishand Wildlife Service Florida Fish and Wildlife ConservationCommission (FWC) US National Park Service QuantitativeSpatial Ecology Evolution and Environment (QSE3) In-tegrative Graduate Education and Research Traineeship(IGERT) program funded by the National Science Foundationand the University of Florida for financially supporting thisstudy Funding for panther research conducted by the FWC issupported predominantly via donations accrued with vehicleregistrations for ldquoProtect the Pantherrdquo license plates

LITERATURE CITEDAcevedo‐Whitehouse K T R Spraker E Lyons S R Melin F Gulland RL Delong and W Amos 2006 Contrasting effects of heterozygosity onsurvival and hookworm resistance in California sea lion pups MolecularEcology 151973ndash1982

Adams J R L M Vucetich P W Hedrick R O Peterson and J AVucetich 2011 Genomic sweep and potential genetic rescue during limitingenvironmental conditions in an isolated wolf population Proceedings of theRoyal Society B Biological Sciences 2783336ndash3344

Agresti A 2013 Categorical data analysis Third edition John Wiley amp SonsHoboken New Jersey USA

Agudo R M Carrete M Alcaide C Rico F Hiraldo and J A Donaacutezar2012 Genetic diversity at neutral and adaptive loci determines individualfitness in a long‐lived territorial bird Proceedings of the Royal Society BBiological Sciences 2793241ndash3249

Albeke S E N P Nibbelink and M Ben‐David 2015 Modeling behavior bycoastal river otter (Lontra canadensis) in response to prey availability in PrinceWilliam Sound Alaska a spatially‐explicit individual‐based approach PLOSOne 10e0126208

Alho J S K Vaumllimaumlki and J Merilauml 2010 Rhh an R extension for estimatingmultilocus heterozygosity and heterozygosity‐heterozygosity correlationMolecular Ecology Resources 10720ndash722

Alvarez K 1993 Twilight of the panther Myakka River Publishing SarasotaFlorida USA

Aparicio J M J Ortego and P J Cordero 2006 What should we weigh toestimate heterozygosity alleles or loci Molecular Ecology 154659ndash4665

Bakker V J D F Doak G W Roemer D K Garcelon T J Coonan S AMorrison C Lynch K Ralls and R Shaw 2009 Incorporating ecologicaldrivers and uncertainty into a demographic population viability analysis for theisland fox Ecological Monographs 7977ndash108

Balloux F W Amos and T Coulson 2004 Does heterozygosity estimateinbreeding in real populations Molecular Ecology 133021ndash3031

Barone M A M E Roelke J Howard J L Brown A E Anderson and DE Wildt 1994 Reproductive characteristics of male Florida panthersmdashcomparative studies from Florida Texas Colorado Latin‐America andNorth‐American Zoos Journal of Mammalogy 75150ndash162

Bateson Z W P O Dunn S D Hull A E Henschen J A Johnson and LA Whittingham 2014 Genetic restoration of a threatened population ofgreater prairie‐chickens Biological Conservation 17412ndash19

Bauer H G Chapron K Nowell P Henschel P Funston L T B HunterD W Macdonald and C Packer 2015 Lion (Panthera leo) populations aredeclining rapidly across Africa except in intensively managed areas Pro-ceedings of National Academy of Sciences of the United States of America11214894ndash14899

Beier P 1995 Dispersal of juvenile cougars in fragmented habitat Journal ofWildlife Management 59228ndash237

Benson J F J A Hostetler D P Onorato W E Johnson M E Roelke SJ OrsquoBrien D Jansen and M K Oli 2011 Intentional genetic introgressioninfluences survival of adults and subadults in a small inbred felid populationJournal of Animal Ecology 80958ndash967

Benson J F P J Mahoney J A Sikich L E K Serieys J P Pollinger H BErnest and S P D Riley 2016 Interactions between demography geneticsand landscape connectivity increase extinction probability for a small popu-lation of large carnivores in a major metropolitan area Proceedings of theRoyal Society B Biological Sciences 28320160957

Beverly F 1874 The Florida panther Forest and Stream 3290Boyce M S C V Haridas C T Lee and the NCEAS Stochastic Demo-graphy Working Group 2006 Demography in an increasingly variable worldTrends in Ecology amp Evolution 21141ndash148

Brook B W J J OrsquoGrady A P Chapman M A Burgman H R Akcakayaand R Frankham 2000 Predictive accuracy of population viability analysis inconservation biology Nature 404385ndash387

Brys R and H Jacquemyn 2016 Severe outbreeding and inbreeding de-pression maintain mating system differentiation in Epipactis (Orchidaceae)Journal of Evolutionary Biology 29352ndash359

Burke J M and M L Arnold 2001 Genetics and the fitness of hybridsAnnual Review of Genetics 3531ndash52

Burnham K P 1993 A theory for combined analysis of ring recovery andrecapture data Pages 199ndash213 in J‐D Lebreton and P M North editorsMarked individuals in the study of bird population Springer‐Verlag NewYork New York USA

Burnham K P and D R Anderson 2002 Model selection and multimodelinference a practical information‐theoretic approach Second edition SpringerVerlag New York New York USA

Caswell H 2001 Matrix population models construction analysis and in-terpretation Sinauer Associates Sunderland Massachusetts USA

Charpentier M J M Setchell F Prugnolle L A Knapp E J Wickings PPeignot and M Hossaert‐McKey 2005 Genetic diversity and reproductivesuccess in mandrills (Mandrillus sphinx) Proceedings of the NationalAcademy of Sciences of the United States of America 10216723ndash16728

Cooch E and G C White editors 2018 Program MARK a gentle in-troduction lthttpwwwphidotorgsoftwaremarkdocsbookgt Accessed 7Apr 2016

Cooley H S R B Wielgus G M Koehler H S Robinson and B TMaletzke 2009 Does hunting regulate cougar populations A test of thecompensatory mortality hypothesis Ecology 902913ndash2921

Coulson T E A Catchpole S D Albon B J Morgan J M Pemberton TH Clutton‐Brock M J Crawley and B T Grenfell 2001 Age sex densitywinter weather and population crashes in Soay sheep Science 2921528ndash1531

Coulson T J Pemberton S Albon M Beaumont T Marshall J Slate FGuinness and T Clutton‐Brock 1998a Microsatellites reveal heterosis in reddeer Proceedings of the Royal Society B Biological Sciences 265489ndash495

Coulson T N S D Albon J M Pemberton J Slate F E Guinness and TH Clutton‐Brock 1998b Genotype by environment interactions in wintersurvival in red deer Journal of Animal Ecology 67434ndash445

Cox D 1972 Regression models and life‐tables Journal of the Royal StatisticalSociety Series B Statistical Methodology 34187ndash220

Creel S 2006 Recovery of the Florida panthermdashgenetic rescue demographic rescueor both Response to Pimm et al (2006) Animal Conservation 9125ndash126

Crnokrak P and D A Roff 1999 Inbreeding depression in the wild Heredity83260ndash270

Crow J 1948 Alternative hypotheses of hybrid vigor Genetics 33477ndash487

van de Kerk et al bull Florida Panther Population and Genetic Viability 31

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 4: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

black‐tailed prairie dogs (Cynomys ludovicianus) and white‐footed mice (Peromyscus leucopus) Subsequently the effect ofinbreeding depression has been extensively documented inseveral wild animal populations including mammals such aswolves (Canis lupus Liberg et al 2005) African lions (Pan-thera leo Trinkel et al 2010 2011) and Iberian lynx (Lynxpardinus Ruiz‐Loacutepez et al 2012) In these examples in-breeding depression had negative effects on a variety of fitnesstraits including survival of neonates susceptibility to diseaseand male fertility The demographic and genetic impacts ofinbreeding on population growth in carnivores is highlightedbest in one of the quintessential long‐term conservation stu-dies on the gray wolves of Isle Royale in northern MichiganUSA (Peterson 1977 Peterson and Page 1988 Peterson et al1998 Adams et al 2011) Research on this population hasproved informative for gaining a better understanding of po-pulation consequences of loss of genetic variation and impactsof intrinsic and extrinsic factors that have led to its decline toonly 2 individuals as of 2017 (Hedrick et al 2017)Conversely outbreeding depression is a scenario when mating

between members of genetically divergent populations might resultin a reduction in fitness correlated with the loss of local adaptationsor genetic incompatibilities (Moritz 1999) Outbreeding depressionhas been hypothesized to be a possible side effect of genetic in-trogression (Reinert 1991 Maehr and Caddick 1995) albeit in-frequently and with less and weaker empirical evidence for verte-brate species Negative fitness responses resulting from matingbetween individuals from highly divergent populations of copepodswere noted by Hwang et al (2012) Other examples of outbreedingdepression in varied species of plants are provided by Fenster andGallaway (2000) Waser et al (2000) Goto et al (2011) and Brysand Jacquemyn (2016) In general however outbreeding depres-sion is thought to be less of a concern when implementing con-servation measures such as genetic introgression (Miller et al 1999Whiteley et al 2015)In the last 3 decades conservation biologists have exponentially

increased their use of molecular techniques to assist with mon-itoring managing and conserving wildlife populations (DeYoungand Honeycutt 2005) Although early research relied on methodssuch as restriction fragment length polymorphisms allozymeelectrophoresis and minisatellites (Smith and Wayne 1996) thelast 20 years have been dominated by the implementation of im-proved automated sequencing technology to provide more in-formative and fine‐scale data from both nuclear and mitochondrialDNA sequences (eg mtDNA sequencing microsatellitesDeYoung and Honeycutt 2005) Microsatellites in particular havebecome a common tool for conservation projects to assess geneticparameters including inbreeding depression and the impacts ofgenetic introgression in imperiled populations They continue tobe widely used molecular markers in conservation researchThe field of molecular ecology is changing rapidly and new

technology is constantly being developed Currently there is in-terest in applying genomics or so‐called whole‐genome sequencingtechniques to assist with conserving imperiled species Two recentexamples of the application of whole‐genome sequencing to en-dangered species of carnivores include the work of Robinson et al(2016) on the Channel Island fox (Urocyon littoralis) and Murch-ison et al (2012) on the Tasmanian devil (Sarcophilus harrisii) For

instance Robinson et al (2016) reported a near absence ofgenomic variation in Channel Island foxes from the island of SanNicolas demonstrating an extreme reduction in heterozygositycompared to mainland populations Similarly Murchison et al(2012) applied whole genome sequencing to ascertain whether theTasmanian devil facial tumor disease originated from a male or afemale devil and further used whole genome sequencing techniquesto assess transmission of the disease across Tasmania The appli-cation of whole genome sequencing in conservation biology andwildlife management will continue to increase (Whiteley et al2015) Although costs associated with implementing wholegenome sequencing are steadily declining (Whiteley et al 2015)other factorsmdashsuch as sample preparation and the need forbioinformatic expertise to deal with voluminous amounts of datamdashare still an impediment to their widespread use in conservation andmanagement of wildlife populations Therefore long‐term datasets frommore traditional molecular markers such as microsatelliteswill continue to have utility for providing insight into the genetichealth of endangered species and for assessing impacts of man-agement and recovery initiativesThe identification and quantification of inbreeding depression

using molecular techniques (Kardos et al 2016) along with pro-tocols for implementing translocations to promote genetic in-trogression have been documented for varied endangered popu-lations (Whiteley et al 2015) Conversely studies that provide anassessment of the longevity of benefits accrued to wild populationsvia genetic introgression are uncommon Hedrick et al (2014)provided evidence for the declining effect of genetic introgressionon the Isle Royale population of wolves after only 2 or 3 genera-tions Whereas heterosis is expected to increase individual fitness itcan also result in outbreeding depression (Fenster and Gallaway2000 Waser et al 2000 Goto et al 2011 Waller 2015 Brys andJacquemyn 2016) Depending on the mechanisms outbreedingdepression may not become evident until generations subsequent tothe first filial (F1) generation (Pickup et al 2013 Waller 2015)For example ancestral chromosomes and coevolved genes remainintact in the F1 but recombination may break up favorable genecomplexes in F2 and later generations (Waser et al 2000 Waller2015 Frankham 2016) Investigating the timeframe over whichthe positive impacts of genetic introgression are sustained wouldlikely necessitate long‐term monitoring but for most studies of theprocess populations are typically monitored only for 1 or 2 gen-erations following genetic introgression (Willi et al 2007 Hedrickand Fredrickson 2009 Hedrick et al 2014 Whiteley et al 2015)Consequently data are rarely gathered that would allow us to es-timate the time over which populations continue to benefit fromgenetic introgression (Hedrick et al 2014 Whiteley et al 2015)The Florida panther (Puma concolor coryi) provides a textbook

example of how inbreeding depression in a small isolated po-pulation can result in low levels of genetic variation that areassociated with morphological and biomedical abnormalities andprecipitous population decline The timeline associated with thedecline of panthers parallels that of most large carnivores inNorth America (Onorato et al 2010) A robust population ofpanthers was thought to extend across Florida USA inpre‐Columbian times (Fig 1 Alvarez 1993) Subsequentlydocumented declines in the numbers of panthers were noted inthe nineteenth and twentieth centuries mainly as a result of

6 Wildlife Monographs bull 203

anthropogenic alteration of the landscape and unregulatedhunting As early as the 1870s panthers were rarely encounteredand considered mythical in some portions of their range (Beverly1874) The decline in the population of panthers eventuallycaptured the attention of lawmakers subsequently they receivedpartial legal protection as a game animal in 1950 and completelegal protection in Florida by 1958 (Onorato et al 2010) By the1960s the core of remaining wilderness within the Big Cypressregion of South Florida where the last group of breedingpanthers was thought to persist was further affected by theconstruction of Alligator Alley This roadway along with theTamiami Trail allowed access to these once inaccessible areasAll these factors ultimately led to the listing of the Floridapanther as endangered on the inaugural list of endangeredspecies on 11 March 1967 (US Federal Register 1967) andsubsequent protection under the Endangered Species Act of1973 (Public Law 93‐205) This designation invariably raisedawareness of the plight of the panther and served as a catalyst toinitiate research to avoid what appeared to be their imminentextinction (Onorato et al 2010)In 1973 the World Wildlife Fund initiated a survey that re-

sulted in the capture of a single female panther and doc-umentation of their sign in just a few locations in South Florida(Nowak and McBride 1974) The information accrued duringthis survey would ultimately lead to the commencement of along‐term research project on the Florida panther in 1981 by

what is now known as the Florida Fish and Wildlife Con-servation Commission (FWC) Over the next decade a multi-tude of studies were published that helped improve knowledgeof panther reproduction (Maehr et al 1989 Barone et al 1994)diet (Maehr 1990) and genetics (OrsquoBrien et al 1990 Roelkeet al 1993) among other topics Early research was essential foridentifying challenges faced by the remaining panthers mostimportantly the impacts of the continued loss of habitat isola-tion of the population and inbreeding depression (Onoratoet al 2010) In combination these factors would force wildlifemanagers to contemplate the implementation of genetic in-trogression to avert the extinction of the pantherBy the early 1990s the Florida panther population had declined

to a minimum of 20ndash30 individuals (McBride et al 2008) Severalstudies documented low levels of genetic diversity and expressedtraits thought to be correlated with inbreeding depression (Roelkeet al 1993 Johnson et al 2010 Onorato et al 2010) includingcowlicks (mid‐dorsal pelage whorls) and kinked tails (Wilkinset al 1997) These morphological traits probably had little directimpact on fitness of Florida panthers More concerning were theproportion of individuals in the population in the early 1990s thatwere unilaterally or bilaterally cryptorchid (Roelke et al 1993)possessed poor sperm quality (Barone et al 1994) sufferedcompromised immune systems (OrsquoBrien 1994) exhibited atrialseptal defects (Roelke et al 1993) and possessed some of thelowest levels of genetic variation observed in wild felids (Driscoll

Figure 1 Historical and current (2015) breeding range of the Florida panther in the southeastern United States Delineation of current breeding range is describedin Frakes et al (2015)

van de Kerk et al bull Florida Panther Population and Genetic Viability 7

et al 2002) Taken together these factors suggested that thelong‐term prospects for the persistence of the Florida pantherwere poorSubsequently the United States Fish and Wildlife Service

(USFWS) conducted an assessment that led to the delineation of 4possible management options 1) initiate a captive breeding pro-gram 2) introduce genetic material from captive stockpurported to contain both pure Florida panther and SouthAmerican puma lineages 3) translocate wild female pumas (Pumaconcolor stanleyana) from Texas USA into South Florida or 4)take no action at all (USFWS 1994) Collaborative discussionsinvolving academics non‐governmental organizations and stateand federal wildlife agencies eventually led to the selection of op-tion 3 and a plan for genetic introgression that involved the releaseof 8 female pumas from Texas into South Florida was im-plemented in 1995 (Seal 1994 Onorato et al 2010)Continued research conducted after the introgression interven-

tion provided clear evidence of the benefits accrued to the pantherpopulation including increased levels of genetic diversity declinesin the frequency of morphological and biomedical correlates ofinbreeding depression and the substantial increase of the popu-lation size (McBride et al 2008 Johnson et al 2010 McClintocket al 2015) Furthermore Hostetler et al (2013) showed that thepost‐introgression population growth was driven primarily byhigher survival rates associated with admixed panthers (offspringof pairings between the female Texas pumas and male Floridapanthers and subsequent generations of those progeny) followinggenetic introgression The apparent success of genetic introgres-sion in preventing the imminent extinction of the Florida pantherpopulation is encouraging but the population remains small andisolated and continues to face substantial challenges especially thelarge‐scale loss of habitatAlmost 5 panther generations (generation time= 45 years

Hostetler et al 2013) have passed since the implementation of thegenetic introgression program The panther population has beenmonitored continuously during this time period leading to asubstantial amount of additional data and biological insights sincethe demographic assessment made immediately following the in-trogression (Hostetler et al 2010 2012 2013 Benson et al 2011)Given that the effects of genetic introgression are not permanent itis important to discern whether the positive impacts of this man-agement initiative are continuing or waning Decline in the benefitsof the introgression could reduce the population growth rate andprobability of persistence At the same time panther habitat israpidly lost to a variety of human activities (eg land development)and environmental changes associated with invasive species andclimate change (Land et al 2008 Dorcas et al 2012 Frakes et al2015) When long‐term monitoring data are available it is prudentto periodically evaluate demographic rates population growth andpersistence parameters so that changes are detected and timelymanagement action can be implementedTwo population modeling approaches have become popular

among ecologists and wildlife managers for studying the dynamicsand persistence of age‐ or stage‐structured populations matrixpopulation models and individual‐based models (IBMs Brooket al 2000 Morris and Doak 2002 Morrison et al 2016) Matrixpopulation models are the generalized discrete time‐version of theLotka‐Euler equation which describes the dynamics of age‐ or

stage‐structured populations of plants animals and humans (Lotka1924 Caswell 2001) They have become standard tools for mod-eling human and wildlife populations (Tuljapurkar 1990 Caswell2001 Keyfitz and Caswell 2005) because 1) they are powerful andflexible permitting adequate representation of life history of a widevariety of organisms including those with complex life histories 2)parameters for these models can be empirically estimated usingstandard analytical methods such as multi‐state capture‐recapturemethods (eg Williams et al 2002) 3) it is relatively straight‐forward to include the effects of factors such as environmental anddemographic stochasticity and density dependence 4) importantdemographic quantities such as asymptotic population growth ratestable age‐ or stage distribution and generation time can be easilycalculated from these models 5) they permit prospective and ret-rospective perturbation analyses which have been proven useful inwildlife conservation and management 6) they are computer‐friendly do not require extensive programming experience andtheir implementation using software packages such as R (R CoreTeam 2016) and MATLAB (MathWorks 2014) is straight‐for-ward and 7) they have sound theoretical foundation offeringanalytical solutions to many aspects of population modeling Weused matrix population models for 1) calculating the aforemen-tioned demographic quantities 2) performing sensitivity analysesinvolving asymptotic population growth rate and 3) comparativepurposes to check the results obtained from equivalent simulation‐based modelsDespite many advantages offered by matrix population models it

is difficult to incorporate attributes of individuals such as geneticsor behavior within that framework Agent‐based or IBMs (Grimm1999 Grimm and Railsback 2005 McLane et al 2011 Railsbackand Grimm 2011 Albeke et al 2015) offer an alternative modelingframework with tremendous flexibility Individual‐based modelsare computer simulation models that rely on a bottom‐up approachthat begins by explicitly considering the components of a system(ie individuals) population‐level properties emerge from the be-havior of and interactions among discrete individuals UsingIBMs it is possible to explicitly represent attributes of individualssuch as movement mating or genetics (McLane et al 2011Railsback and Grimm 2011 Albeke et al 2015) We used in-dividual‐based population models as the primary populationmodeling approach because they offered a framework for an ex-plicit consideration of genetics both at individual and populationlevelsOur overall goal was to provide a comprehensive demographic

assessment of the sole extant population of the Florida pantherusing long‐term (1981ndash2015) field data and novel modelingtechniques Our objectives were 1) to estimate demographicparameters for the Florida panther and investigate whetherpositive impacts of the 1995 genetic introgression remained inthe population and if so to what extent 2) to estimate popu-lation growth and persistence parameters using an IBM tailoredto the Florida pantherrsquos life history and 3) to evaluate thesensitivity of population growth and persistence parameters tovital demographic ratesThe Florida panther population remains small (le230 adults and

subadults FWC 2017) with no realistic possibility of natural geneflow so the loss of genetic variation and concomitant increase ininbreeding‐related problems are inevitable Ensuring the long‐term

8 Wildlife Monographs bull 203

persistence of the population will likely necessitate periodic geneticmanagement interventions Therefore our final objective was to 4)evaluate genetic and population‐dynamic consequences of variedgenetic introgression strategies and from these identify manage-ment strategies that are affordable and effective at improvingprospects of Florida panther persistence To achieve this objectivewe extended the individual‐based population model to track in-dividual genetics based on empirical allele frequencies taking ad-vantage of the long‐term demographic and genetic data that havebeen collected on Florida panthers since 1981 We used individualheterozygosity to inform vital rates based on empirically estimatedrelationships between individual heterozygosity and demographicparameters We then simulated population trajectories trackedindividual heterozygosity over time and estimated quasi‐extinctiontimes and probabilities without introgression and with introgres-sion for a range of intervals and number of immigrants perintrogression event We compared the genetic and demographicbenefits of implementing alternative genetic introgression scenariosand ranked them by the level of improvement in population per-sistence parameters and estimated costs because funding is a per-sistent challenge to conservation planningManagement of imperiled species is a complex process invol-

ving many stakeholders and plagued by layers of uncertainties(Runge 2011) It would therefore seem that management ofthreatened and endangered species would be a perfect case foradaptive resource management approach (ARM) because ARMfocuses on structured decision making for recurrent decisionsmade under uncertainty (Williams et al 2002 Runge 2011)Because it combines structured decision making with learningprudent application of ARM reduces uncertainty over time andcan lead to optimal management actions (Williams and Brown2012 2016) Key ingredients of a successful adaptive manage-ment program include stakeholder engagement clearly definedand mutually agreed‐upon objectives alternative managementactions and objective decision rules (Nichols et al 2007Williams and Brown 2012 2016) Unfortunately Florida pan-ther stakeholders have divergent views about the state of thesystem (eg panther abundance) and it is a challenge to havemutually agreed‐upon objectives management actions or deci-sion rules acceptable to all or most stakeholders AlthoughFlorida panther conservation efforts would be best served withinthe ARM framework in the long run impediments to its im-plementation (Runge 2011 Williams and Brown 2012) areunlikely to be overcome in the near future

STUDY AREA

The breeding population of Florida panthers mainly persists as asingle population South of the Caloosahatchee River (approxi-mately 267133degN latitude 815566degW longitude see Fig 1)One exception is a female that was documented just north of theCaloosahatchee River in 2016 (and subsequently documentedwith kittens in 2017) by FWC the first validation of a femalepanther north of the River since 1973 (FWC unpublisheddata) The breeding range in South Florida is topographicallyflat has a subtropical climate and is characterized by permanentand ephemeral wetlands influenced by rains from May throughOctober (Duever et al 1986) The study area contains a variety

of wildland land cover types including hardwood hammockscypress forests pine flatwoods freshwater marshes prairies andgrasslands (Davis 1943) and land characterized by human ac-tivity including citrus groves croplands pastureland rockmining and residential developments (Onorato et al 2011)The area is intersected by a multitude of roads that fragmentpanther habitat Most notable among these is Interstate 75(Alligator Alley) which was expanded from 2 to 4 lanes in 1990when fencing and wildlife crossings were constructed with theintention of minimizing harmful effects of road expansion onlocal wildlife (Foster and Humphrey 1995) A large portion ofthe area that comprises South Florida is under public ownershipmainly as federal or state lands Everglades National Park BigCypress National Preserve Florida Panther National WildlifeRefuge Picayune Strand State Forest and Fakahatchee StrandPreserve State Park alone encompass 9745 km2 though not allof the area is considered suitable for panthers (eg large ex-panses of sawgrass marshes in the western Everglades) Frakeset al (2015) delineated 5579 km2 of adult panther habitat inSouth Florida 1399 km2 of which are in private ownership Amajority of these private lands are located in the northern extentof the breeding range Although some private lands may beprotected (eg conservation easements) other areas are sus-ceptible to incompatible land uses such as rock mining or re-sidential developments The breeding range in South Florida isbounded to the east and west by large metropolitan areas in-clusive of Miami‐Fort Lauderdale‐Pompano Beach and CapeCoral‐Fort Myers‐Naples‐Marco Island‐Immokalee respec-tively that are populated by gt6000000 people (httpswwwcensusgov accessed 24 Jan 2019) These characteristics of thestudy area highlight the recovery challenges faced by panthers

METHODS

Field MethodsSince 1981 Florida panthers have been captured radiocollared andtracked by the FWC and National Park Service staff (Table 1)Capture and handling protocols followed guidelines of theAmerican Society of Mammalogists (Sikes 2016) Livestock Pro-tection Company (Alpine TX USA) provided trained hounds andhoundsmen for captures Teams either treed or bayed panthers onthe ground and then darted them with a 3‐ml compressed‐air dartfired from a CO2‐powered rifle Immobilization drugs have variedduring the tenure of the project but most recently teamsimmobilized panthers with a combination of ketamine HCl (10mgkg Doc Lanersquos Veterinary Pharmacy Lexington KY USA) andxylazine HCl (1mgkg Doc Lanersquos Veterinary Pharmacy) Fol-lowing immobilization teams lowered treed panthers to the groundby a rope or caught panthers with a net in some cases they used aportable cushion (McCown et al 1990) to further mitigate theimpact of a fall Capture teams administered propofol (PropoFlotradeAbbott Laboratories Abbott Park IL USA) intravenously either asa bolus or continuous drip to maintain anesthesia They adminis-tered midazolam HCl (003mgkg) intramuscularly or intravenouslyto supplement anesthesia in some panthers Panthers recovered in ashaded area away from water In some cases capture teams reversedxylazine HCl with yohimbine HCl (Yobinereg Lloyd IncShenandoah IA USA) at 25 its recommended dose

van de Kerk et al bull Florida Panther Population and Genetic Viability 9

Capture teams determined sex and age of panthers marked themwith an ear tattoo and a subcutaneous passive integrated trans-ponder (PIT‐tag) and fitted them with a very high frequency(VHF) or global positioning system (GPS) radio‐collar equippedwith a mortality sensor The GPS collars were programmed toattempt to collect a position at a variety of time intervals(range= 1ndash12 hr) depending on objectives of concurrent studiesobservers routinely tracked panthers with VHF radiocollars from afixed‐wing aircraft 3 times per week When a radio‐collar trans-mitted a mortality signal or when a location did not change forseveral days observers investigated the site to establish the pan-therrsquos fate When observers found a dead panther they assessed thecause of death based on an examination of the carcass and sur-rounding area if possible They then transported carcasses to anexperienced veterinarian for necropsy and a histopathological ex-amination to attempt to assign the cause of death Starting in 1995observers continually assessed successive locations of radio‐collaredfemales to determine if they initiated denning behavior Three to 4fixes at the same location (VHF collars) or successive returns to thesame location over a weeklong period (GPS collars) served as anindicator of a possible den Observers then located dens moreprecisely via triangulation with ground telemetry They visited denswhen the dam left the site typically to make a kill Teams capturedkittens by hand captured kittens ranged in age from 4 to 35 dayspost‐partum Teams counted sexed weighed dewormed andpermanently marked kittens with a subcutaneous PIT‐tag

Genetic AnalysisTeams collected blood tissue or hair from the majority ofcaptured kittens subadult and adult Florida panthers for DNAsamples The United States Department of Agriculture ForestService National Genomics Center for Wildlife and FishConservation (Missoula MT USA) processed samplesTechnicians used Qiagen DNeasy Blood and Tissue Kits(Qiagen Valencia CA USA) to extract whole genomic DNAfrom blood and tissue samples and used a slight modification(overnight incubation of sample in lysis buffer and proteinase Kon a rocker at 60deg elution of DNA in 50 μl of buffer) of theQiagen DNA extraction protocol (Mills et al 2000) to extractDNA from hair samples Technicians used a panel of 16 mi-crosatellite loci (Fca090 Fca133 Fca243 F124 F37 Fca075Fca559 Fca057 Fca081 Fca566 F42 Fca043 Fca161

Fca293 Fca369 and Fca668) identified by Menotti‐Raymondet al (1999) which had been previously applied to Floridapanther samples (Johnson et al 2010) Technicians amplifiedDNA samples in 10‐microl reaction volumes that included 10 microl ofDNA 1times reaction buffer (Applied Biosystems Life Technolo-gies Corporation Grand Island NY USA) 20 mM of MgCl2200 microM of each dNTP 1 microM of reverse primer 1 microM of dye‐labeled forward primer 15 mgml of bovine serum albumin(BSA) and 1U Taq polymerase (Applied Biosystems) Thepolymerase chain reactions (PCRs) included a thermal profile of94degC for 5 minutes followed by 36 cycles of 94degC for 60 seconds55degC for 60 seconds and 72degC for 30 seconds Techniciansvisualized the resulting PCR products on a LI‐COR DNAanalyzer (LI‐COR Biotechnology Lincoln NE USA) Tech-nicians collected genotypes from hair samples (n= 16) using amultitube approach (Taberlet et al 1996) error‐checked geno-types using Program DROPOUT (McKelvey and Schwartz2005) and combined them with genotypes from tissue andblood samples (n= 487) We evaluated all 16 loci for variousdescriptive statistics for the radio‐collared sample of adult andsubadult panthers We excluded genotypes of kittens from thatanalysis because related individuals are known to result in theoverrepresentation of certain alleles (Marshall et al 1998) Wedetermined the number of alleles observed heterozygosity andexpected heterozygosity using GenAlex (Peakall and Smouse2006 2012) We assessed allelic richness at each locus in Fstat293 (Goudet 1995) We assessed conformance of genotypedata from these 16 loci to HardyndashWeinberg assumptionslinkage disequilibrium and null alleles (Appendix A availableonline in Supporting Information) The Genomics Center alsoprovided us with genotype data from the same 16 loci for 41pumas from western Texas the source population for thegenetic rescue programEvidence of inbreeding depression in wild populations is

commonly determined on the basis of heterozygosityndashfitnesscorrelations (Silva et al 2005 Ortego et al 2007 Mainguyet al 2009 Johnson et al 2011) so we also estimated theheterozygosity of each individual panther We calculatedhomozygosity by loci (HL) which varies between 0 (all lociheterozygous) and 1 (all loci homozygous Aparicio et al 2006)using the Rhh package (Alho et al 2010) in Program R (R CoreTeam 2016) A microsatellite locus has more weight in HL

Table 1 Number of Florida panthers by age sex and genetic ancestry categories used in subsequent demographic analyses during the study period (1981ndash2013) inSouth Florida USA

Kittensa Subadults and adultsb

Males Females Total Males Females Total

PIT‐taggedc 215 180 395 Radiocollaredc 108 101 209Recovered 11 4 15 Subadult 72 50 122Litter failed 47 38 85 Prime adult 65 95 160Recaptured 25 30 55 Old adult 13 20 33Canonical 27 20 47 Canonical 43 29 72F1 admixed 6 12 18 F1 admixed 2 6 8Other admixed 130 105 235 Other admixed 45 53 98

a Kitten samples included those that were PIT‐tagged recovered (ie panthers that were found dead) part of a failed litter or recaptured as subadults or adultsb Radiocollared subadult and adult panthers are presented as sample size within each age class and ancestry classc Combined number of panthers of different ancestries is less than the number radiocollared or tagged with a subcutaneous passive integrated transponder (PIT‐tagged) because we did not conduct genetic sampling on all panthers Combined number of panthers in different age classes is greater than the number ofradiocollared panthers because some individuals were counted in ge2 age classes as animals aged

10 Wildlife Monographs bull 203

when it is more informative (ie has more alleles) and has aneven distribution of allele frequencies For our demographicanalyses we calculated individual heterozygosity for each in-dividual panther as 1 minusHL As do most indices of individualheterozygosity HL takes into account the average allele fre-quency in the population (Aparicio et al 2006) Thus duringsimulations the allele frequency in the population changesthroughout an individualrsquos lifetime which causes its hetero-zygosity also to change This is not biologically plausible becausegenetic properties are retained throughout an individualrsquos life-time Therefore for use in model simulations we simply cal-culated heterozygosity for each individual as the proportion ofheterozygous lociWe also used genotype data to implement a Bayesian clus-

tering analysis in Program STRUCTURE (Pritchard et al2000) to infer ancestral clusters of panthers This method uses aMarkov chain Monte Carlo (MCMC) method to determine thenumber of genetic clusters (K) in the sample while also pro-viding an individualrsquos percentage of ancestry (q values) allocatedto each cluster Runs for this analysis included a 100000MCMC burn‐in period followed by 500000 MCMC iterationsfor 1ndash10 ancestral genetic clusters (K= 1 to 10) with 10 re-petitions for each K To determine the most likely number of Kgenetic clusters we used the logarithm of the probability of thedata (log Pr(D)∣K Pritchard et al 2000) and estimates of ΔK(Evanno et al 2005) in Program STRUCTURE HAR-VESTER (Earl and VonHoldt 2011) We then used q valuesprovided in runs for the best K to assign individual panthers aseither canonical (in most cases ge90 pre‐introgression ancestrysee Johnson et al 2010) or admixed (lt90 pre‐introgressionancestry) We recognized admixed panthers that were the off-spring of matings between a Texas female puma and a canonicalmale panther as F1 admixed panthers for analyses

Demographic ParametersSurvival of kittensmdashMost kittens PIT‐tagged at den sites were

never encountered again A small proportion of the PIT‐taggedkittens were recovered dead or were recaptured as subadults oradults and radiocollared so that their fate could be monitoredWe also identified instances of complete litter failure when afemale died within 9 months after giving birth (the minimumage of independence for kittens) or when a female wasdocumented to have another litter less than 12 months aftergiving birth (the minimum age of independence plus a 3‐monthgestation period Hostetler et al 2010) Thus our data setconsisted of 1) live recaptures and subsequent radio‐tracking ofpanthers of various ages that had been PIT‐tagged as kittens 2)recovery of dead panthers of various ages that had been PIT‐tagged as kitten 3) live captures and subsequent radio‐trackingof other panthers and 4) instances of complete litter failure(Table 1) We analyzed these data using Burnhamrsquos live‐recapture dead‐recovery modeling framework (Burnham 1993Williams et al 2002) to estimate survival of kittens (0ndash1‐yearold) and to test covariate effects on this parameter We includeddata on individuals encountered dead or alive at an older age inour analyses because their encounter histories also providedinformation on kitten survival We used a live‐dead data inputformat coded with a 1‐year time step (Cooch and White 2018)

All known litter failures occurred within the first year of thekittensrsquo lives so we treated all litter failures as recoveries withinthe first year We treated panthers that would have died if wehad not removed them to captivity as having died on the date ofremoval Data preparation and analytical approach are describedin more detail by Hostetler et al (2010)In addition to survival probability (S) the Burnham model

incorporates parameters for recapture probability (p) recoveryprobability (r) and site fidelity (f) We set r and p to 10 forradio‐tracked individuals because their status was known eachyear We also fixed f at 10 for all individuals because the re-capture and recovery areas were the same and encompassed theentire range of the Florida panther Based on findings of Hos-tetler et al (2010) we used a base model that 1) allowed survivalto differ between kittens and older panthers 2) allowed survivalto differ between sexes and among age classes (females ages 1and 2 ge3 males ages 1 2 and 3 ge4) 3) constrained recaptureprobability to be the same for all uncollared panthers and 4)allowed recovery rates to differ between kittens and uncollaredolder panthers Even with our additional data this model re-mained the best fit for a range of models for recapture recoveryand survival of kittens subadults and adults of various age‐classcategories (Hostetler et al 2010)Subadult and adult survivalmdashTo estimate survival for panthers

ge1 year of age we analyzed radio‐tracking data using a Coxproportional hazard‐modeling framework (Cox 1972 Therneauand Grambsch 2000) We followed procedures outlined byBenson et al (2011) for data preparation and analysis Brieflywe right‐censored panthers on the last day of observation beforetheir collar failed When an individual aged into a new age classwhile being tracked we created 2 data entries 1 ending the dayit entered the new age class the other starting on that day Weused the FlemingndashHarrington method to generate survivalestimates from the Cox analysis and the Huber sandwichmethod to estimate robust standard errors (Therneau andGrambsch 2000 Benson et al 2011)We considered 3 age classes subadults (1ndash25 yr for females and

1ndash35 yr for males) prime adults (25ndash10 yr for females and35ndash10 yr for males) and old adults (gt10 yr old for both sexes)after Benson et al (2011) We estimated overall survival using abase model that allowed survival to differ between subadultfemales subadult males prime adult females prime adult malesand old adults of either sex (old age was additive with sex other-wise age and sex were interactive) This model provided the best fitto the data relative to a range of models considering differences insurvival between prime‐adult and older‐adult age classes both ad-ditively and interactively with sex (Benson et al 2011)To examine whether survival rates observed in recent years

differed from those observed immediately following introgres-sion we also estimated age‐specific survival rates for 2 nearlyequal periods of time immediately post‐introgression (1995ndash2004) and the more recent period (2005ndash2013) For theseanalyses we separated the radio‐tracking data into the 2 periodsand estimated survival rates for each period separately using thebase model Similar analyses for kitten survival were not possiblebecause of data limitationsProbability of reproductionmdashWe used telemetry data obtained

from radio‐tracked females that reproduced at least once to

van de Kerk et al bull Florida Panther Population and Genetic Viability 11

estimate annual probability of reproduction of subadult prime‐adult and old‐adult females using complementary logndashlogregression (Agresti 2013) following methods described indetail by Hostetler et al (2012) Briefly we modeled theprobability of a female panther giving birth (b) as a function ofcovariates as

γ= minus [minus ( )]b z1 exp exp i i

where zi is a row vector of covariates (see below for covariateinformation) for female panther i and γ is a column vector ofmodel coefficients (Appendix B Table B2) To account fordifferences in observation time we included log(m) as an offsetin the model where m is the number of months a femalepanther was radio‐tracked (Hostetler et al 2012)

Reproductive outputmdashTo estimate the number of kittensproduced by reproductive females we used cumulative logitregression (Min and Agresti 2005 Agresti 2013) Weconsidered a model that includes J categories for the numberof offspring with j denoting the number of offspring ( j= 1 2hellip J and J= 6) provided that a female reproduces Weconsidered J= 6 which is greater than the largest litter sizeobserved in our study (4 kittens) because females can produce 2or more litters per year if litters fail We modeled the probabilitythat a female i produces a litter of size at most j (Yi) as

le⎧

⎨⎩

δ( ) = == hellip minus

=

( )+ θ βminus +Y jj J

j JPr

1 2 1

1i ij e

1

1 xj i

where θj is the intercept for the annual number of kittens pro-duced by a female i being at most j xi is a row vector of cov-ariates for female panther i (see below) and β is a column vectorof model coefficients The probability that Yi will be exactly j isgiven by

⎧⎨⎩

πδ

δ δ( = ) = =

=

minus = hellip( minus )Y j

jj J

Pr 1

2 i ij

i

ij i j

1

1

Finally the expected annual number of kittens produced byfemale i (νi) is given by

sumν π==

j i

j

J

ij

1

Additional details regarding the estimation and modelingof the annual number of kittens produced can be found inHostetler et al (2012)

CovariatesmdashIn addition to sex and age we tested for the effect ofabundance genetic ancestry and individual heterozygosity on theaforementioned demographic parameters (model coefficients forkitten and adult survival are given by η and ι respectively AppendixB Table B2) We used a minimum population count (MPC) basedon radio‐tracking data and field evidence of subadult and adult (ieindependent age) panthers as an index of abundance (McBride et al2008) For these analyses we did not use the population sizeestimates reported by McClintock et al (2015) because unlike theminimum counts they did not cover the entire study period

(McBride et al 2008 R T McBride and C McBride RancherrsquosSupply Incorporated unpublished report Fig 2)To test if demographic parameters differed depending on an-

cestry we considered 3 ancestry models dividing the panthers into1) F1 admixed and all other panthers 2) canonical and admixed(including F1 admixed) panthers and 3) F1 admixed canonicaland other admixed panthers We quantified genetic diversity usingindividual heterozygosity as described previously We tested for theeffect of genetic covariates using a subset of kittens that were ge-netically sampled We calculated model‐averaged estimates ofmodel parameters on the scale in which they were estimated(Appendix B available online in Supporting Information)

Cause‐specific mortalitymdashWe performed cause‐specificmortality analysis only on dead radio‐collared panthers becausedetection rates of uncollared panther mortalities are highlycorrelated with the cause of death (eg vehicle collision)Following Benson et al (2011) we attributed each mortality to1 of 4 causes 1) vehicle collision 2) intraspecific aggression 3)other causes (including known causes such as diseases injuriesand infections unrelated to the first 2 causes) and 4) unknown(ie mortalities for which we could not assign a cause) Wethen used the non‐parametric cumulative incidence functionestimator a generalization of the staggered‐entry KaplanndashMeiermethod of survival estimation to estimate cause‐specificmortality rates for male and female panthers in different ageclasses (Pollock et al 1989 Heisey and Patterson 2006 Bensonet al 2011) We compared cause‐specific mortality rates betweensexes age classes and ancestries

Statistical inferencemdashWe used an information‐theoreticapproach (Akaikersquos information criterion [AIC]) for modelselection and statistical inference (Burnham and Anderson 2002)For the analysis of kitten survival we adjusted the AIC foroverdispersion and small sample size (QAICc details in Hostetleret al 2010) We performed all statistical analyses in Program R (RCore Team 2016) We used Burnhamrsquos live recapturendashdead

0

100

200

300

1985 1990 1995 2000 2005 2010Year

Indi

vidu

als Method

MPC

MVM

Figure 2 The Florida panther minimum population count (MPC) and populationsize estimated using motor vehicle collision mortalities (MVM) during our studyperiod Minimum population counts are from McBride et al (2008) and R TMcBride and C McBride (Rancherrsquos Supply Incorporated unpublished report)Estimates based on MVM are from McClintock et al (2015) The solid vertical lineindicates the year of genetic introgression

12 Wildlife Monographs bull 203

recovery model (Burnham 1993) to estimate and model kittensurvival using the RMark package version 2113 (Laake 2013) asan interface for Program MARK (White and Burnham 1999)We implemented the Cox proportional hazard model for esti-

mation and modeling of survival of subadults and adults using the Rpackage SURVIVAL (Therneau and Grambsch 2000) We per-formed cause‐specific mortality analyses using an R version of S‐PLUS code provided by Heisey and Patterson (2006)To account for model selection uncertainty we calculated

model‐averaged parameter estimates across all models using theR package AICcmodavg (Mazerolle 2017) using AIC weights asmodel weights (Burnham and Anderson 2002) We used modelswithout the categorical ancestry variables for model averagingand estimated parameters based on the average value for thecontinuous covariates (individual heterozygosity and abundanceindex) Because only a subset of the kittens was geneticallysampled we estimated 2 kitten survival rates First we estimatedkitten survival based on the data set that included all kittens(overall kitten survival) Second we estimated kitten survivalusing a subset of the data that only included kittens that weregenetically sampled (survival of genetically sampled kittens)

Matrix Population ModelsWe investigated Florida panther population dynamics andpersistence using 2 complementary modeling frameworks ma-trix population models (Caswell 2001) and IBMs (Grimm andRailsback 2005 McLane et al 2011 Railsback and Grimm2011 Albeke et al 2015) Matrix population models offer aflexible and powerful framework for investigating the dynamicsand persistence of age‐ or stage‐structured populations (egCaswell 2001 Robinson et al 2008 Kohira et al 2009 Hunteret al 2010 Hostetler et al 2012) With a fully developed theorybehind them these models also serve as a check for resultsobtained from simulation‐based models such as IBMs We usedan age‐structured‐matrix population model for deterministic andstochastic demographic analyses and for preliminary investiga-tions of Florida panther population viability (Caswell 2001Morris and Doak 2002)For deterministic and stochastic demographic analyses we

parameterized a female‐only 19 times 19 age‐specific populationprojection matrix of the form

⎢⎢⎢⎢⎢

⋯ ⋯

⋯ ⋯

⋱ ⋱ ⋮

⋮ ⋱ ⋱ ⋱ ⋮

⎥⎥⎥⎥⎥

( )=t

F F F

P

P

P

A0 0

0

0 0 0

t t t

t

t

t

1 2 19

1

2

18

where Fi and Pi are the age‐specific fertility and survival ratesrespectively and ( )tA is a time‐varying (or stochastic) popula-tion projection matrix We assumed that the longevity and ageof last reproduction of female Florida panthers was 185 yearswe based this assumption on the observation that the oldestfemale in our study was 186 years old Florida panthers re-produce year‐round thus we used birth‐flow methods to esti-mate Fi and Pi values following Caswell (2001) and Hostetleret al (2013) We estimated Pi as

= +P S S i i i 1

where Si is the age‐specific survival probability We estimated Fi as

⎛⎝

⎞⎠

=+ +F S

m Pm

2i k

i i i 1

where Sk is the kitten survival probability and mi is the age‐specific fecundity rate which was estimated as

ν=m b f i i i k

where bi is the breeding probability for a female in age class iduring time step t νi is the annual number of kittens producedby a breeding female in age class i and fk is the proportionfemale kittens at birth (assumed to be 05)We estimated the finite deterministic (λ) and stochastic annual

population growth rate (λs) and stochastic sensitivities and elasti-cities following methods described in detail by Caswell (2001) Toestimate λs we assumed identically and independently distributedenvironments (Caswell 2001 Morris and Doak 2002 Haridas andTuljapurkar 2005) and used a parametric bootstrapping approachto obtain a sequence of demographic parameters for 50000 ma-trices We then estimated log(λs) from this sequence of matrices(Tuljapurkar 1990 Caswell 2001) and exponentiated it to obtain λsWe calculated the sensitivity and elasticity of λs to changes in lower‐level vital demographic rates as per Caswell (2001) and Haridas andTuljapurkar (2005)For population projection and population viability analysis (PVA)

simulations we expanded the population projection matrix into a34 times 34 2‐sex age‐structured matrix to include female and maleFlorida panthers (Caswell 2001 Hostetler et al 2013) Malescontribute to the population only through survival in this modelingframework We assumed maximum longevity for males to be 145years of age because the oldest male in our study was 144 years oldThe population projection matrix was of the following form

⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢

⋯ ⋯ ⋯ ⋯

⋯ ⋯ ⋯ ⋯

⋱ ⋱ ⋮ ⋮ ⋱ ⋱ ⋮

⋮ ⋱ ⋱ ⋱ ⋮ ⋮ ⋱ ⋱ ⋮

⋯ ⋯ ⋯

⋯ ⋯ ⋯ ⋯

⋯ ⋯ ⋱ ⋱ ⋮

⋮ ⋮ ⋱ ⋱ ⋮ ⋱ ⋱ ⋮

⋯ ⋯ ⋯

⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥

( )=

( ) ( ) ( )

( )

( )

( )

( ) ( ) ( )

( )

( )

t

F N F N F N

P N

P N

P N

M N M N M N

Q N

Q N

A n

0 0

0 0 0 0

0

0 0 0 0 0

0 0

0 0 0

00 0 0 0 0

t t t

t

t

t

t t t

t

t

1 2 19

1

2

18

1 2 19

1

14

where Pi and Qi are age‐specific survival rates of female and maleFlorida panthers respectively and Fi and Mi are the rates atwhich female and male kittens are produced by females of ageclass i respectively Here the population projection matrix isboth time‐varying and density‐dependent (Caswell 2001)We incorporated the influence of parameter uncertainty en-

vironmental stochasticity and density dependence on Floridapanther population dynamics and persistence following the ap-proach outlined by Hostetler et al (2013) we used the sameapproach in the IBM analysis which is described below UnlikeIBMs matrix models do not intrinsically include demographicstochasticity so we explicitly included the influence of

van de Kerk et al bull Florida Panther Population and Genetic Viability 13

demographic stochasticity by simulating the fates of individualsas described in Caswell (2001)Population projections using matrix models necessitate a

vector of initial sex‐ and age‐specific abundance Because suchdata are currently unavailable for the Florida panther we esti-mated it based on the stable sex and age distribution and MPCin 2013 Using point estimates of the demographic parameterswe estimated the stable sex and age distribution for the 2‐sexdeterministic and density‐independent population projectionmatrix We then multiplied the stable sex and age distributionby the MPC in 2013 to get the starting population vector n(0)We projected the population over time as n(t+ 1)=A(tn)n(t)where A(tn) indicates the time‐specific density‐dependentpopulation projection matrix (Caswell 2001)We had 2 indices of abundance available the MPC (McBride

et al 2008) and population size estimated using motor vehiclecollision mortalities (MVM McClintock et al 2015) These 2indices are substantially different because MPC does not ac-count for imperfect detection whereas MVM accounts forimperfect detection and provides an estimate of population size(Fig 2) Therefore we examined 2 scenarios one using densitydependence based on MPC and the other using density de-pendence based on MVM We ran 1000 bootstraps of para-meter values and 1000 simulations per bootstrap for each sce-nario We projected population trajectories over 200 years andthen estimated population growth rates probabilities ofquasi‐extinction and time until quasi‐extinction and comparedthese with the results obtained from the IBM (see below) Weran scenarios with a quasi‐extinction threshold of 10 and 30panthers We used the mean of probabilities of quasi‐extinction(PQE) across simulations as the point estimate for PQE and the5th and 95th percentiles across simulations as the 90 con-fidence interval Because the confidence intervals werequantile‐based but the point estimates were not it is possiblewith skewed distributions for the point estimates to be outsidethe confidence interval We implemented the matrix

population models in the R computing environment (R CoreTeam 2016)

IBMs and PVABecause of well‐developed theory ease of implementation andthe modeling flexibility that they offer matrix populationmodels have become the model of choice for investigating thedynamics and persistence of age‐ and stage‐structured popula-tions (Caswell 2001) However it is difficult to incorporateattributes of individuals such as genetics and behavior usingmatrix models Quantifying the effects of genetic erosion on theFlorida panther population dynamics and persistence andevaluating benefits and costs of alternative genetic managementscenarios were important goals of our study Because IBMs relyon a bottom‐up approach that begins by explicitly consideringindividuals (McLane et al 2011 Railsback and Grimm 2011Albeke et al 2015) it is possible to represent genetic attributesof each individual and to track genetic variation over time Weused IBMs as the primary modeling framework in this studybecause they allowed for explicit consideration of genetics bothat individual and population levels and population viabilityassessment under alternative genetic management scenariosWe developed an IBM (Grimm 1999 Grimm and Railsback

2005 McLane et al 2011 Railsback and Grimm 2011 Albekeet al 2015) tailored to the life history of the Florida panther(Fig 3) We simulated every individual from birth until deathbased on a set of rules describing fates (eg birth and death) andattributes (eg genetics) of individuals at each annual time stepAs in other IBMs population‐level processes (eg populationsize and genetic variation) emerged from fates of and interac-tions among individuals (Grimm 1999 Railsback and Grimm2011) We developed an overview design concepts and details(ODD) protocol (Grimm et al 2006 2010) to describe ourIBM (Appendix C available online in Supporting Information)and provide pseudocodes for individual‐based simulations(Appendix D available online in Supporting Information)

For each panther at time = t

Kitten (lt1 yr)

Reproduce

Survive

Kitten(s) born

Kitten

Sex and age class

Subadult male

Prime adult male

Old adult male

Subadult female

Prime adult female

Old adult female

Age one year

No Male

Female

Yes

No

Yes

t=t+1

Yes

lt35 yrs

35-10 yrs

gt10 yrs

lt25 yrs

25-10 yrs

gt10 yrs

Figure 3 Schematic representation of Florida panther life history We tailored the individual‐based population model to represent the life‐history patterndepicted here

14 Wildlife Monographs bull 203

Dynamics and persistence of populations are influenced bymany factors such as demographic and environmental sto-chasticity and density dependence these effects and parametricuncertainty must be considered in population models whenpossible (Caswell 2001 Boyce et al 2006 Bakker et al 2009Hostetler et al 2013) Because by definition IBMs simulate thelife history of individuals in a population they inherently in-corporate the effect of demographic stochasticity Our analysesrevealed that survival of kittens subadults and adults and theprobability of reproduction varied over time so we incorporatedenvironmental stochasticity in the model for these variablesfollowing Hostetler et al (2013) Likewise we found evidenceof density‐dependent effects on kitten survival Because theMPC might be an underestimate of population size (McBrideet al 2008) we performed simulations using both the MPC andMVM abundance indicesTo account for model selection and parametric uncertainty we

used a Monte Carlo simulation approach For each bootstrapsample we selected a model based on the AIC (or QAICc)weights We then sampled the intercept and slope for kittensurvival (on the logit scale) subadult and adult survival (on thelogit scale with log‐hazard effect sizes) and probability of re-production (on the complementary logndashlog scale Appendix B)from multivariate normal distributions with mean vectors equalto the estimated parameters and variance‐covariance matricesequal to the sampling variance‐covariance matrices We thentransformed the results to nominal scale and converted the es-timates to age‐specific survival and reproduction parameters asdescribed in Hostetler et al (2013)We ran 1000 simulations for 1000 parametric bootstraps for

200 years for the MPC and the MVM scenario for both thestochastic 2‐sex matrix‐population model and the IBM Wetracked the population size at each time step and estimatedquasi‐extinction probabilities and times based on the populationtrajectory We considered the population to be quasi‐extinct ifindividuals of only 1 sex remained or if the population size fellbelow critical thresholds set at 10 and 30 individuals We alsoestimated expected time to quasi‐extinction for both thresholdsdefined as the mean and median time until a population be-comes quasi‐extinct conditioned on quasi‐extinction We im-plemented the IBM using the RNetLogo package version 10ndash2(Thiele et al 2012 Thiele 2014) as an interface for the IBMplatform NetLogo (Wilensky 1999)

Sensitivity AnalysisAn important component of demographic analysis is sensitivityanalysis which is the quantification of the sensitivity of amodelrsquos outcome to changes in the input parameters (Caswell2001 Bakker et al 2009 Thiele et al 2014) Using an IBM weperformed global sensitivity analyses for 2 (ie MPC andMVM) density‐dependent scenarios to assess how sensitivequasi‐extinction times and probabilities were to changes (anduncertainties) in the estimated demographic rates We usedLatin hypercube sampling to sample parameters from the entireparameter space (Marino et al 2008 Thiele et al 2014) Wesampled intercept and slope for kitten survival (on logit scale)subadult and adult survival (on logit scale with log‐hazard effectsizes) and probability of reproduction (on complementary

logndashlog scale) from normal distributions We sampled theprobability distribution for the number of kittens producedannually (on the real scale) from a Dirichlet distribution themultivariate generalization of the beta distribution (Kotz et al2000) We sampled 1000 different parameter sets and ran 1000simulations for each scenario We calculated the probability ofquasi‐extinction within 200 years for each density‐dependencescenario (MPC or MVM) using a quasi‐extinction threshold of10 individuals and then calculated the partial rank correlationcoefficient between each of those and each parameter trans-formed to the real scale (Bakker et al 2009 Hostetler et al2013) Partial rank correlation coefficients quantify the sensi-tivity of the probability of quasi‐extinction to input variables(ie survival and reproductive parameters) with higher absolutevalues indicating stronger influence of that variable on quasi‐extinction probabilities

Genetic ManagementOne of our goals was to estimate the benefits (improvements ingenetic variation demographic parameters and populationgrowth and persistence parameters) and costs (financial costs ofcapture transportation quarantine release and monitoring ofintroduced panthers) of genetic management To achieve thisobjective we extended the IBM to include a genetic componentTo simulate individual‐ and population‐level heterozygosity overtime we assigned each panther alleles for 16 microsatellite locibased on the empirical allele frequency in the population (esti-mated from genetic samples collected during 2008ndash2015) whenwe initialized the model (Pierson et al 2015) At each time stepwe simulated Mendelian inheritance for kittens produced suchthat each kitten randomly inherited alleles from its mother andfrom a male panther randomly chosen to have putatively siredthe litter We did not explicitly force inbreeding to occur butthe probability of inbreeding naturally increased as the popula-tion size decreasedIntrogression strategiesmdashWe modeled genetic introgression as a

result of the release of 2‐year‐old female pumas from Texas intothe simulated Florida panther population For the geneticintrogression implemented in 1995 Texas females between 15and 25 years of age were released because females at that agehave lower mortality rates (Benson et al 2011) higherreproductive potential and also because it is much easier todetect with certainty the successful reproduction by females thanby males The ability to more closely monitor reproduction andthe birth of kittens is an important consideration in reducing theprobability of a genomic sweep of Texas alleles into the Floridapanther population Wildlife managers removed the last 2surviving Texas females from the wild population in Florida in2003 to reduce the possibility of genomic sweep To simulatethis strategy in the model we removed any Texas females thatwere still alive 5 years after each introgression event Weassigned Texas females alleles based on the allele frequency ofthe Texas population (based on genotypes from 48 samplescollected in west Texas that was inclusive of the pumas releasedin South Florida in 1995) when they entered the Florida pantherpopulation We could not estimate demographic rates separatelyfor the released Texas females because of small sample size (only8 females were released in 1995) therefore we assumed that

van de Kerk et al bull Florida Panther Population and Genetic Viability 15

survival and reproductive rates and the effects of covariates onthese rates for the released Texas females were identical to thoseof female Florida panthersThe impact of genetic introgression depends on the frequency

of introgression events and the number of individuals introducedat each attempt Thus we defined introgression strategies basedon the number of Texas females to be released (0 5 10 or 15)and the interval between genetic introgressions (10 20 40 and80 yr) for a total of 13 strategies We simulated each manage-ment strategy with density dependence estimated using theMPC and MVM abundance indices

We sampled 1000 parameter sets and for each of these setswe ran all introgression strategies 1000 times for 200 years Weapplied the same parametric uncertainty and environmentalstochasticity across all management scenarios to allow for bettercomparison of introgression strategies At each time step werecorded the projected population size and average individualheterozygosity We calculated the heterozygosity for each in-dividual at birth based on the alleles it inherited This individualheterozygosity then informed the survival rate of that individualbased on the empirically estimated relationship between in-dividual heterozygosity and age‐specific survival rates Hetero-zygosity did not change throughout an individualrsquos lifetime butits effect on survival changed as individuals aged because theeffect of individual heterozygosity on survival rate differedamong age classes Survival rates of genetically sampled kittenswere biased high because some kittens were sampled later in lifewhen they had already survived for some time To correct forthis bias we adjusted kitten survival rates predicted by theheterozygosity model proportionally From the population tra-jectories we estimated the probability of quasi‐extinction (de-fined as the probability that the population will fall below 10 or30 individuals or that individuals of only 1 sex will remain)

Cost‐benefit analysismdashExperts estimated financial costs ofintrogression based on their experience with the geneticintrogression executed in 1995 and we adjusted estimates tocurrent costs to account for inflation (R T McBride RancherrsquosSupply Incorporated M Cunningham and D P OnoratoFlorida Fish and Wildlife Conservation Commission personalcommunication) Costs included capturing and transportingfemale pumas from the Texas source population caring for thepumas while they were in quarantine and post‐introgressionmonitoring To compare the financial costs of the differentscenarios we also calculated the average costs incurred over

Table 2 Sample size (n) number of alleles (Na) allelic richness (AR) observedheterozygosity (Ho) and expected heterozygosity (He) at 16 microsatellite loci inradio‐collared Florida panthers sampled from 1981 to 2013 in South FloridaUSA We calculated these values from a subset (n= 137) of the samples used inthe analyses and they include only adult and subadult panthers We excludedkittens because they can result in an overrepresentation of certain alleles

Locus n Na AR Ho He

FCA090 129 5 50 0333 0401FCA133 136 3 30 0618 0608FCA243 136 5 50 0419 0487F124 137 7 69 0708 0729F37 129 4 40 0395 0397FCA075 131 5 50 0534 0598FCA559 133 5 50 0496 0503FCA057 134 6 59 0679 0624FCA081 134 7 69 0209 0206FCA566 130 6 60 0462 0475F42 133 10 100 0737 0766FCA043 136 5 49 0596 0588FCA161 132 6 60 0500 0515FCA293 132 4 40 0614 0624FCA369 131 6 60 0573 0567FCA668 133 7 70 0406 0462Mean 1329 57 57 0517 0535

Figure 4 Proportional membership (q) of radio‐collared Florida panthers (n= 218) and pumas from Texas USA (n= 7) in 2 clusters identified by STRUCTUREEach individual animal is represented by a separate vertical bar Yellow indicates canonical panther ancestry and blue represents admixed ancestry The admixedancestry of the Texas females and F1 generation panthers that were radiocollared are highlighted red and green respectively to demarcate those groups The pre‐introgression period is inclusive of panthers radiocollared from 10 February 1981 to 6 March 1996 The post‐introgression era inclusive of the F1 panthers includesanimals radiocollared from 4 March 1997 to 18 February 2015 in South Florida USA

16 Wildlife Monographs bull 203

100 years assuming that the population would be managedaccording to a particular strategy Our goal with theseexploratory analyses was to evaluate the relative costs andbenefits of alternative management scenarios

RESULTS

Genetic VariationWe assessed genetic variation at 16 microsatellite loci for 137DNA samples collected from adult and subadult panthers(1981ndash2013) that were captured and subsequently radiocollaredThe number of alleles at these loci ranged from 3ndash10 and allelicrichness averaged 57 (Table 2) Average expected hetero-zygosity for this sample was 0535 and average observed het-erozygosity was 0517 (Table 2)We determined individual heterozygosity (1 minusHL) and an-

cestry for the complete sample of panther genotypes (n= 503)collected from 1981 to 2015 for use as covariates in subsequentanalyses Average individual heterozygosity was 0559 andranged from 0ndash0980 Our STRUCTURE analysis providedstrong support for K= 2 clusters based on the probability of thedata (log Pr(D)|K) and ΔK in STRUCTURE HARVESTERBased on this result we categorized the ancestry of each pantheras either canonical (n= 123) or admixed (n= 380) using valuesof proportional membership (q Fig 4) The average individualheterozygosity of canonical panthers was 0386plusmn 0012 (SE) foradmixed panthers it was 0615plusmn 0007

Survival and Cause‐Specific Mortality RatesOf the 395 kittens that were PIT‐tagged and released 55 werelater encountered alive 15 were recovered dead and 85 werepart of failed litters (Table 1) We radio‐tracked 209 subadult oradult panthers for a total of 244195 panther‐days and docu-mented 126 mortalities (Table 1)Survival depended strongly on age and kittens had the lowest

overall rate (032plusmn 009 Fig 5A Table 3) The model‐averagedkitten survival was 047plusmn 009 when we included only geneti-cally sampled individuals in the analysis (Table 3) as statedabove this estimate is biased high because many kittenswere genetically sampled when they were recaptured alive orrecovered dead at older ages We found no evidence for sex‐specific differences in kitten survival but females survived betterthan males in all older age classes For females subadults hadthe highest survival rates followed by prime adults and oldadults For males prime adults had the highest survival ratefollowed by subadults and old adults (Fig 5A and Table 3)For panthers of all ages there was substantial evidence that

ancestry affected survival with F1 admixed panthers of all ageshaving the highest rates and canonical individuals having thelowest (Fig 5B) The combined AIC weights of models thatincluded an ancestry covariate were 0880 for kittens and 0992for older panthers The survival rates of subadult prime‐adultand old‐adult panthers in the period immediately after the in-trogression (1995ndash2004) were similar to those in more recentyears (2005ndash2013 Fig 5C)After genetic introgression individual panther heterozygosity

increased (Fig 6) and varied among ancestry categories individualheterozygosity was the highest for F1 panthers (078plusmn 001)

A

B

C

Figure 5 Overall age‐ and sex‐specific survival rates (plusmnSE A) age‐ and sex‐specific survival rates (plusmnSE) for Florida panthers of canonical F1 admixed andother admixed ancestry (B) and age‐ and sex‐specific survival estimates (plusmnSE)for subadult and adult Florida panthers in the period immediately post‐introgression (1995ndash2004) and more recently (2005ndash2013 C) in SouthFlorida USA

Table 3 Model‐averaged estimates (plusmnSE) of demographic parameters for theFlorida panther obtained using data collected during 1981ndash2013 in SouthFlorida USA

Parameter Sex Age class Estimate

Survival Both Kitten 032plusmn 009a

Female Subadult 097plusmn 002Prime adult 086+ 003Old adult 078plusmn 009

Male Subadult 066plusmn 006Prime adult 077plusmn 005Old adult 065plusmn 010

Probability of reproduction Female Subadult 035plusmn 008Prime adult 050plusmn 005Old adult 025plusmn 006

Kittens produced annually Female Subadult 280plusmn 005Prime adult 267plusmn 001Old adult 228plusmn 013

a Overall kitten survival (based on all kittens in our data set including thosethat were not genetically sampled) Estimate of survival of geneticallysampled kittens was 047plusmn 009 this estimate is biased high because manykittens were genetically sampled when they were recaptured alive or re-covered dead at older ages

van de Kerk et al bull Florida Panther Population and Genetic Viability 17

followed by those for non‐F1 admixed (062plusmn 001) and canonical(039plusmn 001) panthers Individual heterozygosity positively affectedsurvival rates of kittens (slope parameter η= 1455 95CI= 0505ndash2405) Individual heterozygosity also positively af-fected survival of subadult and adult panthers but the evidence forthis effect was weaker because the confidence interval for the ha-zard ratio overlapped 10 (hazard ratio exp(ɩ)= 0311 95CI= 0092ndash1054 Table 4 and Fig 7)The MPC increased after genetic introgression (Fig 2) This

abundance index was also included in top‐ranking modelsindicating that population density affected survival (Table 4)Abundance reduced the survival of kittens (slope parameterη= minus0035 95 CI= minus0049 to minus0021) evidence was weak forthe effect of abundance on survival of subadult and adult panthers(hazard ratio exp(ɩ)= 1002 95 CI= 0995ndash1010 Fig 7) Re-gression coefficients associated with the effect of abundance andindividual heterozygosity on demographic parameters are presentedin Table B1 (available online in Supporting Information)The most frequently observed causes of death for radio‐

collared panthers were vehicle collision and intraspecific ag-gression Cause‐specific mortality analyses revealed thatmortality rates among possible causes were generally similar forthe 2 sexes but that males were more likely to die from in-traspecific aggression than were females (Fig 8A) Male mor-tality from intraspecific aggression was higher for subadult andold adult panthers than for prime adults (Fig 8B) For bothmales and females canonical panthers were more likely to diefrom intraspecific aggression than were admixed panthers(Fig 8C)

Reproductive ParametersOf 101 radio‐collared females 67 reproduced at least once duringour monitoring we used these individuals to assess the annualprobability of reproduction The top‐ranked model for the annual

probability of reproduction included age effect only suggesting thatadding ancestry or abundance did not improve model parsimony(Table 4) Model‐averaged annual probabilities of reproductiondiffered between female subadults (035plusmn 008) prime adults(050plusmn 005) and older adults (025plusmn 006)The most parsimonious model for the number of kittens

produced annually by females that produced at least 1 litter (ieconditional on reproduction) included effects of age ancestryand abundance a model that included only the effect of age wasalmost equally supported (Table 4) The model‐averagednumber of kittens produced annually conditional on re-production was 280plusmn 075 for subadults 267plusmn 043 for primeadults and 228plusmn 083 for old adults Confidence intervals forthe slope parameter (β) overlapped 0 for both abundance(β= 0010 95 CI= minus0001 to 0021) and ancestry (β= minus073795 CI= minus1637 to 0162)

Population Dynamics and PersistenceDeterministic and stochastic demographymdashThe deterministic (λ)

and stochastic (λs) annual population growth rate calculated usingthe matrix population model were 106 (5th and 95th percentiles099ndash114) and 104 (072ndash141) respectively The generation timewas 47 years A female starting in age class one (kitten) wasexpected to live another 58 years and produce 10 kittens onaverage during her lifetime Overall λs was proportionately(elasticity) most sensitive to changes in female prime adultsurvival on the absolute scale (sensitivity) however it was mostsensitive to changes in kitten survival (Fig 9) Populationtrajectories probabilities of quasi‐extinction and times untilquasi‐extinction estimated using the matrix population modeland IBM for comparable scenarios were nearly identical for acritical quasi‐extinction threshold of 10 panthers (Figs 10 and 11)and for a critical quasi‐extinction threshold of 30 panthers(Appendix E available online in Supporting Information)Minimum population count scenariomdashUnder the MPC scenario

(ie density dependence estimated using the minimumpopulation count data) with a critical threshold of 10panthers the population size reached 99 (72ndash104) and 100(77ndash105) at 200 years (Fig 10) as predicted by the IBM and thematrix model respectively The cumulative quasi‐extinctionprobabilities within 100 years based on the MPC scenario were15 (IBM 0ndash48) and 30 (matrix model 0ndash76) Theseprobabilities increased to 25 (IBM 0ndash114) and 38 (matrixmodel 0ndash154) by year 200 (Fig 10) The mean times untilquasi‐extinction were 15 years (IBM 0ndash67) and 15 years (matrixmodel 0ndash72) conditioned on going quasi‐extinct within 100years Expected times until quasi‐extinction conditioned onquasi‐extinction within 200 years were higher 34 years (IBM0ndash137) and 32 years (matrix model 0ndash134 Fig 10)Motor vehicle mortality scenariomdashUnder the MVM scenario

(ie density dependence estimated using estimates of pantherpopulation size from McClintock et al 2015) with a criticalthreshold of 10 panthers the projected population reached 188(142ndash217) and 190 (143ndash219) at 200 years as predicted by theIBM and the matrix model respectively (Fig 11) Thecumulative quasi‐extinction probabilities within 100 years were14 (IBM 0ndash08) and 13 (matrix model 0ndash06) Theseprobabilities increased to 20 (IBM 0ndash17) and 19 (matrix

000

025

050

075

100

1995 2000 2005 2010Year of birth

Indi

vidu

al h

eter

ozyg

osity

Figure 6 Temporal changes in the individual heterozygosity of Floridapanthers born during the period 1995ndash2013 in South Florida USA Pointsare measurements solid line is linear regression shaded area is 95 confidenceinterval of slope Slope= 0006plusmn 0001 R2= 009

18 Wildlife Monographs bull 203

model 0ndash16) within year 200 (Fig 11) The mean times until quasi‐extinction based on the MVM scenario were 5 years (IBM 0ndash61)and 6 years (matrix model 0ndash64) conditioned on going quasi‐extinctwithin 100 years Expected times until quasi‐extinction conditionedon quasi‐extinction within 200 years were 11 years (IBM 0ndash113)and 11 years (matrix model 0ndash112 Fig 11)

Sensitivity of extinction probability to demographic parametersmdashThe partial rank correlation coefficients between quasi‐extinctionprobabilities and demographic parameters were negative anincrease in any of the demographic parameters decreased thequasi‐extinction probability Probabilities of quasi‐extinction within200 years were most sensitive to changes in kitten survival for bothscenarios (Fig 12) followed by survival of subadult females in theMVM scenarios and survival of prime adult females in the MPCscenario The probability of reproduction of prime adult femaleshad the third‐largest partial rank correlation coefficient with quasi‐extinction probability for both scenarios

Benefits and Costs of Genetic ManagementMinimum population count scenariomdashThe MPC scenario with

a critical threshold of 10 panthers predicted that withoutmanagement intervention average individual heterozygositywould decrease reaching 057 (049ndash064) at 100 years and053 (042ndash062) at 200 years (Fig 13) The population woulddecrease slightly reaching an average of 79 (41ndash99) at 100 yearsand 76 (43ndash98) at 200 years (Fig 14) The probabilities of quasi‐extinction under the no‐intervention MPC scenario when weconsidered the effects of genetic erosion were 13 (0ndash99) at 100years and 23 (0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with

the translocation of 5 Texas females average individualheterozygosity under the MPC scenario increased and thenstabilized at 068 (064ndash072) at 100 years and remained atthat level at 200 years (Fig 13) The population reached anaverage of 88 (62ndash104) at 100 years and stayed the same at200 years (Fig 14) The probabilities of quasi‐extinctionunder this introgression strategy were 6 (0ndash53) within 100years and 7 (0ndash82) within 200 years (Fig 15) Results ofanalyses using a critical threshold of 30 panthers revealed asimilar pattern of relative benefits However the prob-abilities of quasi‐extinction were substantially higher (ge45)across all scenarios (Appendix E)

Motor vehicle mortality scenariomdashWithout managementintervention the average individual heterozygosity predictedby the MVM scenario and a critical threshold of 10 panthersdecreased to 060 (046ndash069) at 100 years and to 059 (039ndash070) at 200 years (Fig 13) The population increased slightlyand reached an equilibrium at 154 individuals (46ndash217) at 100years and 157 (57ndash218) at 200 years (Fig 14) The probabilitiesof quasi‐extinction were 17 (0ndash100) at 100 years and 22(0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with 5

female pumas from Texas average individual heterozygosityincreased slightly reaching 067 (060ndash073) at 100 years and068 (059ndash075) at 200 years (Fig 13) Under this strategy thepopulation reached an average of 164 individuals (44ndash226) at

Table 4 Model comparison results testing for the effects of several covariateson survival of all kittens survival of genetically sampled kittens survival ofsubadult and adult Florida panthers probability of reproduction and kittensproduced annually for Florida panthers sampled from 1981ndash2013 in SouthFlorida USA

Model Parameters ΔAICa Weight

Kitten survival (all data)Base+ F1b+ abundancec 10 000 0988Base+ abundance 9 1018 0006Base+ F1 9 1035 0005Base 8 2711 lt0001Base+ sexd 9 2912 lt0001Base+ litter sizee 9 2914 lt0001

Kitten survival (genetically sampled kittens only)Base+ F1+ abundance 10 000 0541Base+ F1CanAdmf+ abundance 11 099 0329Base+Hetg+ abundance 10 310 0115Base+CanAdmh+ abundance 10 791 0010Base+ abundance 9 944 0005Base+ F1 9 1638 lt0001Base+ F1CanAdm 10 1837 lt0001Base+Het 9 2872 lt0001Base 8 3296 lt0001Base+CanAdm 9 3348 lt0001

Subadult and adult survivalBase+ F1CanAdm+ abundance 7 000 0360Base+ F1 5 053 0276Base+ F1CanAdm 6 105 0213Base+ F1+ abundance 6 222 0119Base+CanAdm+ abundance 6 575 0020Base+CanAdm 5 908 0004Base+Het 5 964 0003Base+Het+ abundance 6 984 0003Base 4 1118 0001Base+ abundance 5 1278 0001

Probability of reproductionAge 3 000 0242Age+CanAdm 4 012 0228Age+ abundance 4 173 0102Age+ F1 4 190 0094Age+CanAdm+ abundance 5 216 0082Age+Het 4 219 0081Age+ F1CanAdm 5 232 0076Age+ F1+ abundance 5 372 0038Age+Het+ abundance 5 399 0033Age+ F1CanAdm+ abundance 6 444 0026

Kittens produced annuallyAge+ F1+ abundance 5 000 0194Age 3 060 0144Age+ abundance 4 061 0143Age+ F1 4 097 0120Age+CanAdm 4 139 0097Age+ F1CanAdm+ abundance 6 196 0073Age+ F1CanAdm 5 231 0061Age+CanAdm+ abundance 5 238 0059Age+Het+ abundance 5 250 0056Age+Het 4 259 0053

a Akaikersquos information criterion (AIC) for kitten survival models was adjustedfor small sample size and overdispersion (QAICc)

b F1 divides Florida panthers into 2 groups F1 admixed and all other panthersc Index of Florida panther abundanced Sex of an individual Florida panther (male or female)e Size of the litter in which a Florida panther kitten was bornf F1CanAdm divides panthers into 3 groups F1 admixed canonical andother admixed panthers

g Het is the individual heterozygosityh CanAdm divides panthers into 2 groups canonical and admixed (includingF1 admixed) panthers

van de Kerk et al bull Florida Panther Population and Genetic Viability 19

100 years and 170 (67ndash231) at 200 years based on the MVMscenario (Fig 14) The probabilities of quasi‐extinction underthis introgression strategy were 10 (0ndash99) at 100 years and12 (0ndash99) at 200 years (Fig 15)The projected population size and average individual hetero-

zygosity reached equilibrium levels with periodic fluctuations inthese values when introgression was implemented (Figs 16ndash17)Genetic introgression decreased the time until quasi‐extinctionacross all scenarios (Fig 18) Results of analyses using a criticalthreshold of 30 panthers revealed a similar pattern of relative ben-efits However the probabilities of quasi‐extinction were sub-stantially higher (ge45) across all scenarios (Appendix E)

Addition of pumas from Texas increased both the populationsize and average individual heterozygosity consequentlyintrogression caused periodic increases in average individualheterozygosity and population size at the interval at which theintrogression occurred Whereas average individual hetero-zygosity gradually decreased following introgression densitydependence caused the population size to fluctuate especiallywhen a larger number of pumas from Texas were released Thepositive effect of genetic introgression initiatives on quasi‐extinction probabilities decreased as the time interval betweenthese events increased More frequent genetic introgressions ledto greater increases in average individual heterozygosity and

Old adult

Prime adult

Subadult

Kitten

025 050 075

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Individual heterozygosity

Ann

ual s

urvi

val r

ate

Old adult

Prime adult

Subadult

Kitten

40 60 80 100 120

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Abundance index

Ann

ual s

urvi

val r

ate

Kittens Males Females

Figure 7 The effect of individual heterozygosity (left) and the abundance index (right) on Florida panther age‐ and sex‐specific survival estimates (shaded area isSE) in South Florida USA 1981ndash2013 Abundance index data are from McBride et al (2008) and R T McBride and C McBride (Rancherrsquos Supply Incorporatedunpublished report)

20 Wildlife Monographs bull 203

equilibrium population size while reducing the probability ofquasi‐extinction Comparatively performing genetic introgres-sion less often but with more individuals per release was lesseffective at improving the long‐term prospects for the pantherpopulation (Figs 13ndash15)

Financial cost and persistence benefitsmdashThe total estimated costof 1 genetic introgression was $40000 $80000 or $120000for releasing 5 10 or 15 pumas from Texas respectively(Table 5) The total cost incurred over 100 years for eachstrategy ranged from $50000 to $12 million (Table 6)The most expensive strategy involved the introduction of15 pumas every 10 years this strategy reduced the probability

of quasi‐extinction by 58 and 40 under the MPC and MVMscenarios respectively (Table 6) Introducing 5 pumas every80 years cost the least but improvements in populationpersistence under this strategy were insubstantial (Table 6)Releasing more panthers more frequently caused increasedpopulation fluctuations but did not necessarily reduce the quasi‐extinction probability (Figs 14 and 15 Table 6)

DISCUSSION

The duration (34 yr of data) and sample sizes associated with theFlorida Panther Project allowed us to obtain a comprehensiveunderstanding of genetic variation demographic parameters

A

B

C

Figure 8 Cause‐specific mortality rates (plusmnSE) for male and female Florida panthers (A) subadult prime‐adult and old‐adult Florida panthers of both sexes (B)and canonical and admixed Florida panthers of both sexes (C) in South Florida USA 1981ndash2013 Causes of mortality are vehicle collision intraspecific aggression(ISA) other causes (known causes such as diseases injuries and infections unrelated to the first 2 causes) and unknown cause (mortalities for which we could notassign a cause)

van de Kerk et al bull Florida Panther Population and Genetic Viability 21

probability of population persistence and the duration of thepositive impacts of genetic restoration on this endangered po-pulation The Florida panther population was in dire straits inthe early 1990s and our results clearly demonstrated that the

implementation of the introgression project in 1995 subse-quently accrued a series of genetic and demographic improve-ments that have led to the change from a declining population toone that is growing and expanding its current breeding range

Sensitivity Elasticity

0 1 2 3 00 02 04

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

Para

met

er

Figure 9 Sensitivity and elasticity (proportional sensitivity) of stochastic population growth rate (λs) of the Florida panther to vital demographic parameters in SouthFlorida USA 1981ndash2013 Horizontal lines represent 5th and 95th percentiles

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

70

80

90

100

110

0

5

10

15

0

50

100

Years

StatisticMeanMedian

Figure 10 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the minimum population count(MPC) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

125

150

175

200

00

05

10

15

20

0

30

60

90

Years

StatisticMeanMedian

Figure 11 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the motor vehicle mortality(MVM) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

22 Wildlife Monographs bull 203

MPC MVM

minus08 minus06 minus04 minus02 00 minus08 minus06 minus04 minus02 00

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Male subadult survival

Male prime adult survival

Male old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

PRCC

Para

met

er

Figure 12 Partial rank correlation coefficient (PRCC) between quasi‐extinction probabilities and the demographic parameters of the Florida panther for theminimum population count (MPC) and motor vehicle mortality (MVM) scenarios Error bars indicate standard deviation Based on data collected in South FloridaUSA 1981ndash2013

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 13 Average projected individual heterozygosities in the Florida panther population over 200 years without genetic management intervention and with eachof the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include theintroduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years Average individual heterozygosity peaks when pumas are added to the Floridapanther population Based on data collected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 23

Our research has further validated the success of genetic in-trogression by quantifying the reduction in the probability ofextinction of the panther population because of this manage-ment initiative These findings all bode well for continuedprogress toward recovery That said the Florida panther po-pulation remains small and isolated from other populations ofpuma from western North America thereby denoting that theeventual impact of genetic drift and inbreeding may negativelyaffect the population in the future Realizing this will continueto be an issue that managers will have to contemplate wemodeled the impact of varied genetic management scenarios todetermine the frequency of introgression along with the numberof panthers that should be released during each initiative tominimize cost and maximize benefits to the population Theseresults will prove invaluable to managers attempting to conservethe Florida panther

Genetic Variation Demographic Parametersand Persistence of Heterotic Benefits from GeneticIntrogressionOur measure of individual heterozygosity (1 minusHL) was higheron average for the admixed (0615) panthers than for those ofcanonical ancestry (0386) Levels of individual heterozygosityfor admixed panthers were comparable to levels observed in asample of pumas from west Texas (x plusmn SE= 0695plusmn 0019n= 41) which further demonstrates the improvement in levelsof genetic variation in the Florida population since 1995 The

correlation between HL and several demographic parametershas been noted in wild populations including cheetahs (Terrellet al 2016) and several species of birds such as European shags(Phalacrocorax aristotelis male and female survival probabilityVelando et al 2015) and Egyptian vulture (Neophron percnop-terus age at recruitment Agudo et al 2012) If we focus only onpanthers captured and radiocollared from 2012ndash2015 (FP211ndashFP240) all of which had estimated birth years ge2005 (ge10 yrpost‐introgression) those panthers had an average individualheterozygosity level of 0618plusmn 0029 highlighting the elevatedlevels of genetic variation that continue to persist in cohorts ofpanthers 5 generations after the release of female pumas fromTexasThe proportional membership values (q) from our

STRUCTURE analysis allowed us to delineate 2 clusters ofancestry for Florida panthers (Fig 4) During the pre‐in-trogression era the population was dominated by panthers thatretained a high percentage of the canonical ancestry which wasaffected by inbreeding depression The post‐introgression eraancestry values are comprised of many admixed individuals inessence demonstrating the success of the introgression in termsof increasing genetic variation in the population and mi-micking gene flow that historically occurred with panthers andother populations of pumas (Onorato et al 2010) A somewhatanalogous situation although abetted by natural immigrationwas evident in the ancestral variation in a small population ofpuma intersected by a major freeway in California USA In

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

75

80

85

90

95

100

130

150

170

190

Years

Popu

latio

n si

ze

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 14 Average projected population sizes in the Florida panther population over 200 years without intervention and with each of the genetic introgressionstrategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include the introduction of 5 10 or 15pumas from Texas USA every 10 20 40 or 80 years Peaks occur when pumas are added to the Florida panther population Based on data collected in SouthFlorida USA 1981ndash2013

24 Wildlife Monographs bull 203

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

25

50

75

100

0

25

50

75

100

Introgression interval (yrs)

PQ

E (

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 15 Average probability of quasi‐extinction (PQE) of the Florida panther population within 100 years and within 200 years without genetic managementintervention (N) and with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenariosIntrogression strategies include the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Errorbars indicate 5th and 95th percentiles Based on data collected in South Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

50

60

70

80

90

100

110

50

100

150

200

Years

Popu

latio

n si

ze

Figure 16 Average projected Florida panther population trajectories under a geneticintrogression strategy involving the release of 5 female pumas from Texas USA every20 years for the minimum population count (MPC) and motor vehicle mortality(MVM) scenarios Shaded area indicates 5th and 95th percentiles and isrepresentative of confidence intervals for other scenarios Based on data collected inSouth Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Figure 17 Average projected Florida panther population‐level heterozygositiesunder a genetic introgression strategy involving the release of 5 female pumas fromTexas USA every 20 years for the minimum population count (MPC) and motorvehicle mortality (MVM) scenarios Shaded area bounds the 5th and 95th percentilesand is representative of confidence intervals for other scenarios Based on datacollected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 25

this case the migration of just a single male across the freewayto the south resulted in an increase in the number of in-dividuals with admixed ancestry and substantially improvedgenetic diversity of the subpopulation south of the freeway(Riley et al 2014)Our estimate of overall kitten survival (0321plusmn 0056) was

similar to that reported by Hostetler et al (2010) who used13 years of post‐introgression data (0323plusmn 0071) These valuesare lower than kitten survival estimates derived from severalpuma populations in western North America (044ndash091 Loganand Sweanor 2001 Lambert et al 2006 Laundre et al 2007Robinson et al 2008 Ruth et al 2011) When we included onlydata for individuals that had been genetically sampled our es-timate was 15 higher The proportion of admixed kittens thatwere genetically sampled increased as the population expandedwhich could have resulted in higher estimates of kitten survivalbecause admixed kittens survive better than canonical kittensKitten survival was strongly influenced by genetic ancestry withsurvival rates of F1 kittens being almost twice that of canonicalkittens (Fig 5B) Consistent with the findings of Hostetler et al(2010) increases in heterozygosity coincided with increasedkitten survival whereas population density negatively affectedkitten survival Population density often has a strong negativeimpact on survival of young age classes due to infanticide andcannibalism which has been reported from several carnivore

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

50

100

150

0

50

100

150

Introgression interval (yrs)

TQE

(yrs

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 18 Average times until quasi‐extinction (TQE) for the Florida panther population within 100 years and within 200 years without genetic management interventionand with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategiesinclude the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Error bars indicate 5th and 95thpercentiles Based on data collected in South Florida USA 1981ndash2013

Table 5 Average cost (2017 US dollars) of introgression strategies employedover 100 years that involve the introduction of 5 10 or 15 pumas from TexasUSA into the Florida panther population in South Florida USA at an in-creasingly higher frequency Costs of capture (including transportation) quar-antine and post‐introgression monitoring were estimated by Roy McBrideMark Cunningham and Dave Onorato respectively

Frequency Component 5 pumas 10 pumas 15 pumas

Once Capture 25000 50000 75000

Quarantine 5000 10000 15000

Monitoring 10000 20000 30000

Total 40000 80000 120000

Every 80 years Capture 31250 62500 93750

Quarantine 6250 12500 18750

Monitoring 12500 25000 37500

Total 50000 100000 150000

Every 40 years Capture 62500 125000 187500

Quarantine 12500 25000 37500

Monitoring 25000 50000 75000

Total 100000 200000 300000

Every 20 years Capture 125000 250000 375000

Quarantine 25000 50000 75000

Monitoring 50000 100000 150000

Total 200000 400000 600000

Every 10 years Capture 250000 500000 750000

Quarantine 50000 100000 150000

Monitoring 100000 200000 300000

Total 400000 800000 1200000

26 Wildlife Monographs bull 203

species including polar bears (Ursus maritimus Derocher andWiig 1999) black bears (Ursus americanus Garrison et al 2007)and pumas (Logan and Sweanor 2001)Our estimates of sex‐ and age‐specific survival rates of subadult

and adult panthers (Table 3) and the effects of covariates on theserates were similar to those reported by Benson et al (2011) derivedfrom data collected through 2006 Our survival rates for males(065ndash077) were lower than females (078ndash097) for all 3 ageclasses (subadult prime adult and old adult) which is the typicaltrend observed in most puma populations (eg Ruth et al 19982011) However an unhunted puma population occupying afragmented urban landscape in southern California exhibited lowersurvival (0586 females 0525 males Vickers et al 2015) perhapsas a result of severe habitat fragmentation and the lack of extensiveswaths of public land protected in perpetuity Most populations ofpuma in western North America are typically subject to variedlevels of management via regulated hunting which often is biasedtoward the take of males (Cooley et al 2009 Ruth et al 2011)Consequently annual survival estimates from hunted populationsin northeast Washington (0679ndash0733 females 0341 malesRobinson et al 2008) and southeastern Arizona (0685 females0584 males Cunningham et al 2001) were generally lower thanthose we estimated for Florida panthersCause‐specific mortality analysis showed that male panthers

experienced a substantially greater mortality risk from in-traspecific aggression This differs from the pattern observedduring a long‐term study in Arizona where females were at ahigher risk of intraspecific mortality than males (46 of maleand 53 of female mortality Logan and Sweanor 2001)Mortality associated with intraspecific strife has been docu-mented in hunted and unhunted populations (this study Loganand Sweanor 2001 Stoner et al 2010) and its root cause hasbeen difficult to determine (eg population density and preypopulation trends) That being the case finding ways to reducewhat is often a major source of mortality for puma populationscontinues to pose a challenge to managersCanonical panthers were substantially more likely to die from

intraspecific aggression than were admixed panthers This mayat least partly explain the difference in survival between admixed

and canonical panthers Canonical panthers are known to bemore likely to suffer from varied correlates of inbreeding de-pression than admixed animals including atrial septal defects(Johnson et al 2010) and compromised immune systems(Roelke et al 1993) These factors have the potential to affectfitness and perhaps dominance of individuals when engaged inan intraspecific encounter A somewhat similar scenario wasobserved by Hogg et al (2006) in a population of bighorn sheepthat were part of an introgression project Admixed males in thispopulation had a greater probability of paternity than males thatwere less outbred This included coursing males with higherlevels of admixture being markedly more successful at fightingmales that were defending females in estrous (Hogg et al 2006)Heterosis resulting from genetic introgression is expected to be

the strongest in F1 individuals (Shull 1908 Crow 1948 Burkeand Arnold 2001 Waller 2015) and to progressively decline insubsequent generations (Pickup et al 2013 Frankham 2015)Given that admixed (especially F1) panthers survived better thancanonical panthers and that individual heterozygosity improvedsurvival of both sexes and all age classes our data provide evi-dence for heterotic advantage whereby individuals of mixedancestry had higher fitness (Shull 1908 Crow 1948) Our resultsare consistent with findings of other studies that involved in-tentional genetic introgression for conservation purposes Forexample Hogg et al (2006) detected substantial improvementsin survival and reproduction of introgressed bighorn sheep andMadsen et al (1999 2004) reported a dramatic increase in thepopulation of adders after genetic introgressionKnowledge of the persistence of benefits to a population ac-

crued via genetic introgression is essential for determiningwhen additional introgression may be warranted There is adepauperate amount of information regarding this topic and theidentification of factors that may influence persistence of thebenefits of genetic introgression (Tallmon et al 2004 Hedricket al 2014 Whiteley et al 2015) For example in minks(Neovison vison) Thirstrup et al (2014) found that outcrossingled to larger litters in F2 and F3 but this benefit disappeared insubsequent generations In a population of gray wolves Hedricket al (2014) suggested that benefits of genetic introgression may

Table 6 Costs and benefits accumulated over 100 years for genetic introgression at different intervals and with different numbers of pumas from Texas USAintroduced into the Florida panther population in South Florida USA Costs (2017 US dollars) include capture and transportation quarantine and post‐introgression monitoring Benefits are represented as average probability of quasi‐extinction (PQE and 5th and 95th percentiles) and changes (Δ) therein under thescenario using the minimum population count (McBride et al 2008 MPC) and the scenario using the motor vehicle mortality population estimates (McClintocket al 2015 MVM) The table is sorted by percent change in probability of quasi‐extinction under the MPC scenario

Interval (yr) Pumas Cost MPC PQE () ΔMPC PQE () MVM PQE () ΔMVM PQE ()

ndash 0 0 13 (0ndash99) 0 17 (0ndash100) 080 10 100000 12 (0ndash99) 10 (0ndash58) 16 (0ndash100) 5 (0ndash38)80 15 150000 12 (0ndash99) 13 (0ndash65) 15 (0ndash100) 8 (0ndash56)80 5 50000 12 (0ndash99) 14 (0ndash73) 15 (0ndash99) 9 (0ndash41)40 15 300000 10 (0ndash98) 26 (0ndash104) 14 (0ndash99) 13 (0ndash60)40 10 200000 9 (0ndash94) 35 (0ndash183) 13 (0ndash99) 21 (0ndash95)40 5 100000 8 (0ndash89) 39 (0ndash211) 12 (0ndash99) 26 (0ndash124)20 15 600000 8 (0ndash88) 42 (0ndash257) 13 (0ndash99) 24 (0ndash123)20 10 400000 6 (0ndash67) 54 (0ndash318) 10 (0ndash99) 37 (0ndash217)10 15 1200000 6 (0ndash53) 58 (0ndash372) 10 (0ndash99) 40 (0ndash208)20 5 200000 5 (0ndash50) 59 (0ndash338) 9 (0ndash99) 44 (0ndash352)10 10 800000 4 (0ndash16) 69 (0ndash489) 8 (0ndash92) 55 (0ndash401)10 5 400000 4 (0ndash7) 73 (0ndash502) 6 (0ndash71) 63 (0ndash503)

van de Kerk et al bull Florida Panther Population and Genetic Viability 27

wane after 2 or 3 generations The WrightndashFisher model ofgenetic drift predicts a geometric decline in expected hetero-zygosity at the rate of (1 minus 12Ne) per generation where Ne is theeffective population size (Hartl and Clark 1997 Frankham et al2014) Thus it seems logical to assume that the duration of thebenefits of introgression is limited especially in a species per-sisting in a small population such as Florida panthersWe expected Florida panther survival in the most recent years

(2005ndash2013) post‐introgression to differ from that observed in thedecade following the introgression in 1995 because pantherabundance and heterozygosity factors known to influence panthersurvival (Hostetler et al 2010 Benson et al 2011) have changed(Figs 2 and 6) We found that subadult and adult survival rates inrecent years (2005ndash2013) were similar to those in the period im-mediately following introgression (1995ndash2004 Fig 5C) overallkitten survival was similar to estimates of Hostetler et al (2010) Infact the pattern of covariate effects on Florida panther survivalreported by Benson et al (2011) and Hostetler et al (2010) hasbarely changed The negative effects of population density andpositive effects of heterozygosity may counterbalance survival ratesreducing population fluctuations Our result that benefits of geneticrestoration have remained in the population nearly for 5 genera-tions post‐intervention was surprising but also encouraging Pos-sible explanations for this observation may include 1) genetic ero-sion occurs at slower rate than previously thought 2) introducing 5migrants in 1 genetic intervention may have beneficial effects si-milar to those from 1 migrant per generation over multiple gen-erations or 3) other and as yet unknown factors may reduce therate of genetic erosion and subsequent demographic effects Adensity‐dependent effect on kitten survival could also explain whynon‐F1 admixed kittens did not survive substantially better thandid canonical kittens most kittens sampled from 2005ndash2013 wereadmixed and their survival was lower possibly because of a density‐dependent effect as the Florida panther population continuouslygrew during that period Trinkel et al (2010) also found thatpopulation density and inbreeding coefficient interacted to affectdemographic parameters and growth rate of an African lion po-pulation Likewise population density interacted with winterweather to affect the survival and population growth rates in Soaysheep (Ovis aries Milner et al 1999 Coulson et al 2001)

Population Dynamics and PersistenceThe finite deterministic and stochastic growth rates were gt10reflecting that much of the data used in this study were collectedfollowing genetic introgression in 1995 during which time thepopulation increased substantially Because our demographicparameter estimates did not change markedly in comparison tothose of Hostetler et al (2013) the similarity between thepopulation growth estimates from these studies was expectedBut these estimates reflect population growth during thedata‐collection period and do not predict future growth In factestimates of population size reported by McClintock et al(2015) suggest that population growth may already be slowing(Fig 2) A similar pattern was observed in an introduced po-pulation of African lion which increased steadily from 1995until 2001 then fluctuated around the presumed carrying ca-pacity thereafter (Trinkel et al 2010) Several other African lionpopulations that experienced various degrees of recovery

subsequently became relatively stable or exhibited decliningtrends (Bauer et al 2015)Our estimates of quasi‐extinction probabilities were lower than

those reported by Hostetler et al (2013) using a 2‐sex age‐structured matrix population model Lower probabilities ofquasi‐extinction than those reported by the earlier study forcomparable scenarios were likely a consequence of the fact thatthe estimates of abundance (McBride et al 2008) and thus thedensity‐dependent effects on survival of kittens and old panthers(gt10 yr) have changed in recent years Additionally these earlyestimates of the probability of quasi‐extinction and of time toquasi‐extinction did not consider genetic erosion that will in-evitably occur without management interventionUnder the MVM scenario the equilibrium population size was

twice that attained under the MPC scenario The MVM popula-tion estimates are approximately twice the MPCs (Fig 2) as aresult the simulated population under the MVM scenario achieveda higher equilibrium size Probability of quasi‐extinction under theMVM scenario was generally greater than that under the MPCscenario This result is a consequence of the fact that the estimateddensity dependence on kitten survival based on the MPC scenario isstronger than that for theMVM scenario (regression coefficients forabundance for the MPC scenario= minus0068 vs minus0002 for theMVM scenario Appendix B Table B1) Consequently simulatedpopulations under the MPC scenario have a stronger tendency tomove toward the equilibrium population size which inherentlyreduces the quasi‐extinction probability (eg Royama 1992)As expected for long‐lived mammals (Heppell et al 2000 Oli

and Dobson 2003) the population growth rate was proportio-nately most sensitive to changes in survival of prime adultsfollowed by that in younger age classes Global sensitivity ana-lysis in contrast showed that under all scenarios extinctionparameters were particularly sensitive to changes in survival ofFlorida panther kittens This result is a consequence of the highsensitivity of the population growth rate to kitten survival(Fig 9) and a larger standard error of kitten survival (Table 3)Results of our sensitivity analyses indicate that future researchshould focus on acquiring additional information on early lifestages to improve the accuracy and precision of estimates ofkitten survival Our results suggest that demographic variables(or their environmental drivers) that strongly influence extinc-tion risks are not necessarily those with the largest elasticityvalues A study of island fox (Bakker et al 2009) found thatextinction risk was strongly influenced by survival modifierparameters that had low elasticity values

Genetic Management When and How to InterveneThe use of genetic introgression as a management tool has al-ways been controversial and the probable effectiveness it mighthave in preventing the imminent demise of the panther popu-lation was initially debated (Creel 2006 Maehr et al 2006Pimm et al 2006) These debates notwithstanding the pantherpopulation had suffered from genetic morphological and bio-medical correlates of inbreeding before this management in-tervention (Johnson et al 2010 Onorato et al 2010) whereasthe post‐introgression period has been characterized by a sub-stantial reduction in the biomedical correlates of inbreeding andimprovements in demographic vigor and abundance (McBride

28 Wildlife Monographs bull 203

et al 2008 Hostetler et al 2010 2013 Johnson et al 2010Benson et al 2011 McClintock et al 2015) Nevertheless thepopulation remains small and isolated habitat is still being lostand there is no current possibility of natural gene flow betweenthis and other North American puma populations thus therecurrence of inbreeding‐related problems and subsequent po-pulation declines are inevitable The question therefore is notwhether genetic management of the Florida panther populationis needed but when and how it should be implementedWithout genetic introgression probabilities of quasi‐extinc-

tion were substantially greater than those from models thatincorporated periodic genetic introgression events (Fig 15)highlighting the importance of genetic management in smallisolated populations When 5 female pumas are added to theFlorida panther population every 10 years genetic variationincreased substantially under the MPC and MVM scenariosThe IBM predicted that the equillibrium population size wouldbe substantially larger when 5 pumas were released into thepopulation every 10 years than populations without intervention(ie no introgression) Releasing 15 Texas females did not af-ford substantially greater benefit to the equilibrium populationsize than did releasing only 5 Texas females Instead addingmore individuals caused increasingly large fluctuations in po-pulation size due to the numerical increases and density de-pendence (Fig 14) possibly diminishing the benefits associatedwith increased genetic variationCompared with the no‐intervention scenario (ie no genetic

introgression implemented) the introduction of 5 female pumasevery 10 years reduced quasi‐extinction probabilities from 13to 4 and from 17 to 6 for the MPC and MVM scenariosrespectively The benefit of genetic‐introgression strategies de-creased as the interval between successive management inter-ventions increased and no benefit was evident once the intervalreached 80 years (Fig 15) Introgression reduced the averagetime until quasi‐extinction (Fig 18) This somewhat counter‐intuitive result can be explained by the fact that repeated geneticintrogression makes the population more robust over timewhich will reduce the probability of extinction Consequentlyfew simulated population trajectories that fall below the criticalthreshold do so early reducing the mean time to extinctionAlso density dependence has a stabilizing effect on populations(Royama 1992) populations tend toward equilibrium popula-tion sizes which reduces the probability over time of popula-tions falling below quasi‐extinction threshold However timeuntil quasi‐extinction must be interpreted in conjunction withthe probability of quasi‐extinction which is very low for allscenarios with genetic introgression (Fig 15)Cost is a significant impediment to the implementation of a

genetic introgression program because captures relocations andmonitoring efforts before and after introgression are expensive(Edmands 2007 Frankham 2010b 2015 2016) Costs may beacceptable to society if the management actions result in sub-stantial fitness benefits especially if the species of concern ispopular and charismatic (Frankham 2015) We estimated the100‐year cost of introgression strategies modeled here to rangefrom $50000 to $12 million More expensive strategies gen-erally led to larger decreases in the probability of quasi‐extinc-tion but cheaper strategies also substantially improved

population viability (ie reduced quasi‐extinction probabilityTable 6) One of the cheaper strategies (5 pumas released every20 years) which cost an estimated $200000 over 100 yearsreduced the probability of quasi‐extinction by 59 and 44 forthe MPC and MVM scenarios respectively But these pre-liminary cost estimates are based on expenses incurred duringthe 1995 introgression and include costs of capture quarantinetransportation release and post‐release monitoring of femalesonly Although the cost for future genetic interventions wouldlikely be higher than those reported here the relative differencesin cost between alternative intervention strategies should remainapproximately the sameOur ability to simulate genetics and link it to the viability of

the Florida panther population is based on the empirically es-timated relationship between individual heterozygosity andsurvival Such heterozygosity and fitness correlations have beenstudied for decades (David 1998) but have only recently beenused to predict population viability (Benson et al 2016) noneto our knowledge have incorporated cost into their modelingefforts The mechanisms underlying heterozygosity and fitnesscorrelations remain unclear (David 1998 Szulkin et al 2010)and their significance as a proxy for inbreeding depression hasbeen debated (Balloux et al 2004 Miller and Coltman 2014)Nevertheless evidence continues to mount demonstratingpositive relationships between heterozygosity and survival(Coulson et al 1998ab Hansson et al 2004 Silva et al 2005Acevedo‐Whitehouse et al 2006 Jensen et al 2007 Velandoet al 2015) and between heterozygosity and fecundity (Seddonet al 2004 Charpentier et al 2005 Agudo et al 2012 Velandoet al 2015) Although the reported correlations are generallyweak their consequences can be substantial if population growthand persistence parameters are highly sensitive to the affectedvital rates (Velando et al 2015) Compared with scenarios thatignore genetics incorporating the effects of inbreeding on sur-vival increased quasi‐extinction probabilities within a 100‐yeartimeframe from 15 to 13 and from 14 to 17 for theMPC and MVM scenarios respectively These results high-light the importance of incorporating genetics when analyzingthe viability of small isolated populationsAlthough conservation biologists have recognized the im-

portance of genetics in population viability many still fail toincorporate genetic factors into PVA models (Reed et al 2002)Explicit consideration of genetics in PVAs requires long‐termgenetic and demographic data This permits the assessment ofthe functional relationship between genetic variability and de-mographic parameters Population viability analysis softwarepackages such as VORTEX (Lacy 1993 2000) offer a powerfulframework for individual‐based simulations but they often donot adequately capture a speciesrsquo life history or genetics Basedon a thorough review of the importance of genetic factors inwildlife conservation Frankham et al (2014) concluded thatgenetic factors can strongly influence the dynamics and persis-tence of small andor fragmented populations They furthernoted that ldquoMost PVA‐based risk assessments ignore or in-adequately model genetic factorsrdquo and recommended that ldquoPVAshould routinely include realistic inbreeding depressionhelliprdquo(Frankham et al 201457) Our custom‐built IBM permitted usto develop a model that adequately captured the Florida panther

van de Kerk et al bull Florida Panther Population and Genetic Viability 29

life history and we could parameterize the model with our long‐term genetic and demographic dataThe explicit consideration of the effects of inbreeding on

Florida panther population dynamics and persistence representsan important step forward Nonetheless our model remainsimperfect First we neglected the possible effects of habitat lossdetrimental anthropogenic activities climate change the prob-ability of immigration of pumas from western populationsdisease or catastrophes on demographic rates and populationviability Although immigration from puma populations outsideFlorida is possible its probability is very low given the extensivechallenges the anthropogenically altered landscape would posefor such a migrant to make it successfully to South Florida Weomitted mutations from our model because we have no in-formation on mutation rates though we expect that they are lowbased on values reported for other mammals (Kumar and Sub-ramanian 2002) Finally we only considered possible effects ofinbreeding on age‐specific survival rates because there was noevidence that heterozygosity affected female reproduction Lossof heterozygosity can negatively affect reproduction becauseinbred male panthers can suffer from cryptorchidism which canadversely affect their fecundity However given the polygynousmating system we expect the Florida panther population growthis primarily female‐limitedAlthough it is generally accepted that genetic introgression

only temporarily relieves a population from inbreeding depres-sion we know of no cases in which it has been implementedmore than once to conserve a threatened wildlife populationOur study provides an example of how demographic and mo-lecular data can be used within an IBM framework to evaluatethe efficacy of alternative genetic management strategies via theFlorida panther as a case study Our model can be adjusted forand applied to other species and offers great potential as a toolfor genetic management of small isolated wildlife populations

MANAGEMENT IMPLICATIONS

Our results and those of Hostetler et al (2013) and Johnsonet al (2010) provide persuasive evidence that the genetic in-tervention implemented in 1995 prevented the demise of theFlorida panther and restored demographic vigor to the popu-lation The Florida panther population remains small and iso-lated from other puma populations the closest puma populationis located gt1000 km away in Texas and the intervening land-scape is highly developed and fragmented with many interstate(and other high‐traffic) highways and major urban centersTheory suggests that 1 migrant per generation is sufficient tominimize the loss of genetic diversity (eg Mills and Allendorf1996) However the possibility of successful migration of apuma to South Florida followed by successful mating every ~5years is very low considering the distance between Texas andFlorida (Hedrick 1995) and the many risks and barriers thatdispersing pumas face (Beier 1995 Maehr et al 2002 Thatcheret al 2006) Consequently the pantherrsquos long‐term persistencewill likely depend on subsequent timely genetic introgressionsgiven the near‐impossibility of natural gene flow from otherpuma populations To that end our study offers considerableinsights into benefits of different genetic management strategiesand associated costs

Without genetic management the Florida panther populationfaces a substantial risk of quasi‐extinction within 100 years (10ndash46 depending on quasi‐extinction threshold and scenarios)Twenty‐three years have passed since genetic introgression wasimplemented and the population appears healthy as indicatedby measures of genetic variation increases in the populationsize and improvements in age‐specific survival rates Ourfindings suggest that releasing more pumas more frequently doesnot necessarily improve persistence probability because such astrategy would cause larger fluctuations in population size(Fig 14) which negatively affects persistence probability Thusextinction risks faced by inbred populations of large carnivorescan be substantially reduced by releasing approximately 5 im-migrants every 2 decades We suggest that such a managementstrategy be followed only in conjunction with continued long‐term monitoring of the population Our analyses demonstratethat population growth and persistence are highly sensitive tokitten and female (adult and subadult) survival Continuing tocollect data on these demographic variables along with bio-metric and genetic sampling will remain important to pantherrecovery and vigilance in identifying issues associated with in-breeding depression The collection of these monitoring datashould allow managers to detect any demographic or geneticchanges in a timely fashion and respond with genetic in-trogression or other management actions if warrantedRecovery objectives as defined by the USFWS in its current

recovery plan (USFWS 2008) delineate the need for additionalpopulations in the historical range to downlist or delist theFlorida panther If the only extant population of Floridapanthers continued to grow it could serve as a source ofpanthers to be translocated to suitable habitat to seed addi-tional populations Maintaining current information on thegenetic health of the population and precise estimates of itssize will be important in assessing whether it can withstand theremoval of a subset of animals for translocation Although the2017 documentation of a female panther with kittens north ofthe current breeding range is encouraging in terms of thepotential for population expansionmdashgiven that no female hadbeen recorded north of the Caloosahatchee River since 1972mdashthe re‐establishment of separate new populations will mostlikely require translocations by wildlife managers and a lengthysociopolitical processConservation of threatened or endangered species such as the

Florida panther is inherently challenging For panthers and otherlarge carnivores that require vast extents of natural habitat topersist habitat is typically the most limiting factor that will de-termine whether recovery efforts succeed Although geneticmanagement invariably plays a key role in the long‐term persis-tence of the panther population even a healthy population caneventually succumb to the impacts of habitat loss The humanpopulation of Florida continues to be one of the fastest‐growingin the nation the United States Census Bureau currently ranksFlorida as the third most populous Therefore habitat loss isgoing to continue to be a key issue for panthers Diligence towardpreserving panther habitat in its historical range via conservationeasements or other means and trying to keep such areas inter-connected via corridors is a goal stakeholders such as governmentagencies sportsmen non‐governmental organizations ranchers

30 Wildlife Monographs bull 203

and other private landowners must work on collaboratively toachieve

ACKNOWLEDGMENTS

We thank D Shindle D Jennings T OrsquoMeara D Land andC Belden for valuable advice throughout the project We thankS Bass M Criffield M Cunningham D Giardina D JansenA Johnson D Land M Lotz C McBride Rocky McBrideRowdy McBride Roy McBride L Oberhofer S Schulze staffsat Everglades National Park and Big Cypress National Preserveand others for assistance with fieldwork We also thank JAustin M Conroy J Hines J Laake S Mills J Nichols JHines M Sunquist J Fryxell and T Coulson for advice andassistance regarding data analysis and panther biology TheMuseum of Texas Tech University Natural Science ResearchLaboratory provided samples from pumas from west Texas forcomparative genetic analyses M Ben‐David D Land WMcCown E Hellgren and 4 anonymous reviewers providedmany helpful comments on the manuscript We thank NereaUbierna Lopez for completing the Spanish translation of theabstract We are grateful to the Department of ResearchComputing University of Florida for excellent high‐perfor-mance computing services We would like to thank US Fishand Wildlife Service Florida Fish and Wildlife ConservationCommission (FWC) US National Park Service QuantitativeSpatial Ecology Evolution and Environment (QSE3) In-tegrative Graduate Education and Research Traineeship(IGERT) program funded by the National Science Foundationand the University of Florida for financially supporting thisstudy Funding for panther research conducted by the FWC issupported predominantly via donations accrued with vehicleregistrations for ldquoProtect the Pantherrdquo license plates

LITERATURE CITEDAcevedo‐Whitehouse K T R Spraker E Lyons S R Melin F Gulland RL Delong and W Amos 2006 Contrasting effects of heterozygosity onsurvival and hookworm resistance in California sea lion pups MolecularEcology 151973ndash1982

Adams J R L M Vucetich P W Hedrick R O Peterson and J AVucetich 2011 Genomic sweep and potential genetic rescue during limitingenvironmental conditions in an isolated wolf population Proceedings of theRoyal Society B Biological Sciences 2783336ndash3344

Agresti A 2013 Categorical data analysis Third edition John Wiley amp SonsHoboken New Jersey USA

Agudo R M Carrete M Alcaide C Rico F Hiraldo and J A Donaacutezar2012 Genetic diversity at neutral and adaptive loci determines individualfitness in a long‐lived territorial bird Proceedings of the Royal Society BBiological Sciences 2793241ndash3249

Albeke S E N P Nibbelink and M Ben‐David 2015 Modeling behavior bycoastal river otter (Lontra canadensis) in response to prey availability in PrinceWilliam Sound Alaska a spatially‐explicit individual‐based approach PLOSOne 10e0126208

Alho J S K Vaumllimaumlki and J Merilauml 2010 Rhh an R extension for estimatingmultilocus heterozygosity and heterozygosity‐heterozygosity correlationMolecular Ecology Resources 10720ndash722

Alvarez K 1993 Twilight of the panther Myakka River Publishing SarasotaFlorida USA

Aparicio J M J Ortego and P J Cordero 2006 What should we weigh toestimate heterozygosity alleles or loci Molecular Ecology 154659ndash4665

Bakker V J D F Doak G W Roemer D K Garcelon T J Coonan S AMorrison C Lynch K Ralls and R Shaw 2009 Incorporating ecologicaldrivers and uncertainty into a demographic population viability analysis for theisland fox Ecological Monographs 7977ndash108

Balloux F W Amos and T Coulson 2004 Does heterozygosity estimateinbreeding in real populations Molecular Ecology 133021ndash3031

Barone M A M E Roelke J Howard J L Brown A E Anderson and DE Wildt 1994 Reproductive characteristics of male Florida panthersmdashcomparative studies from Florida Texas Colorado Latin‐America andNorth‐American Zoos Journal of Mammalogy 75150ndash162

Bateson Z W P O Dunn S D Hull A E Henschen J A Johnson and LA Whittingham 2014 Genetic restoration of a threatened population ofgreater prairie‐chickens Biological Conservation 17412ndash19

Bauer H G Chapron K Nowell P Henschel P Funston L T B HunterD W Macdonald and C Packer 2015 Lion (Panthera leo) populations aredeclining rapidly across Africa except in intensively managed areas Pro-ceedings of National Academy of Sciences of the United States of America11214894ndash14899

Beier P 1995 Dispersal of juvenile cougars in fragmented habitat Journal ofWildlife Management 59228ndash237

Benson J F J A Hostetler D P Onorato W E Johnson M E Roelke SJ OrsquoBrien D Jansen and M K Oli 2011 Intentional genetic introgressioninfluences survival of adults and subadults in a small inbred felid populationJournal of Animal Ecology 80958ndash967

Benson J F P J Mahoney J A Sikich L E K Serieys J P Pollinger H BErnest and S P D Riley 2016 Interactions between demography geneticsand landscape connectivity increase extinction probability for a small popu-lation of large carnivores in a major metropolitan area Proceedings of theRoyal Society B Biological Sciences 28320160957

Beverly F 1874 The Florida panther Forest and Stream 3290Boyce M S C V Haridas C T Lee and the NCEAS Stochastic Demo-graphy Working Group 2006 Demography in an increasingly variable worldTrends in Ecology amp Evolution 21141ndash148

Brook B W J J OrsquoGrady A P Chapman M A Burgman H R Akcakayaand R Frankham 2000 Predictive accuracy of population viability analysis inconservation biology Nature 404385ndash387

Brys R and H Jacquemyn 2016 Severe outbreeding and inbreeding de-pression maintain mating system differentiation in Epipactis (Orchidaceae)Journal of Evolutionary Biology 29352ndash359

Burke J M and M L Arnold 2001 Genetics and the fitness of hybridsAnnual Review of Genetics 3531ndash52

Burnham K P 1993 A theory for combined analysis of ring recovery andrecapture data Pages 199ndash213 in J‐D Lebreton and P M North editorsMarked individuals in the study of bird population Springer‐Verlag NewYork New York USA

Burnham K P and D R Anderson 2002 Model selection and multimodelinference a practical information‐theoretic approach Second edition SpringerVerlag New York New York USA

Caswell H 2001 Matrix population models construction analysis and in-terpretation Sinauer Associates Sunderland Massachusetts USA

Charpentier M J M Setchell F Prugnolle L A Knapp E J Wickings PPeignot and M Hossaert‐McKey 2005 Genetic diversity and reproductivesuccess in mandrills (Mandrillus sphinx) Proceedings of the NationalAcademy of Sciences of the United States of America 10216723ndash16728

Cooch E and G C White editors 2018 Program MARK a gentle in-troduction lthttpwwwphidotorgsoftwaremarkdocsbookgt Accessed 7Apr 2016

Cooley H S R B Wielgus G M Koehler H S Robinson and B TMaletzke 2009 Does hunting regulate cougar populations A test of thecompensatory mortality hypothesis Ecology 902913ndash2921

Coulson T E A Catchpole S D Albon B J Morgan J M Pemberton TH Clutton‐Brock M J Crawley and B T Grenfell 2001 Age sex densitywinter weather and population crashes in Soay sheep Science 2921528ndash1531

Coulson T J Pemberton S Albon M Beaumont T Marshall J Slate FGuinness and T Clutton‐Brock 1998a Microsatellites reveal heterosis in reddeer Proceedings of the Royal Society B Biological Sciences 265489ndash495

Coulson T N S D Albon J M Pemberton J Slate F E Guinness and TH Clutton‐Brock 1998b Genotype by environment interactions in wintersurvival in red deer Journal of Animal Ecology 67434ndash445

Cox D 1972 Regression models and life‐tables Journal of the Royal StatisticalSociety Series B Statistical Methodology 34187ndash220

Creel S 2006 Recovery of the Florida panthermdashgenetic rescue demographic rescueor both Response to Pimm et al (2006) Animal Conservation 9125ndash126

Crnokrak P and D A Roff 1999 Inbreeding depression in the wild Heredity83260ndash270

Crow J 1948 Alternative hypotheses of hybrid vigor Genetics 33477ndash487

van de Kerk et al bull Florida Panther Population and Genetic Viability 31

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 5: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

anthropogenic alteration of the landscape and unregulatedhunting As early as the 1870s panthers were rarely encounteredand considered mythical in some portions of their range (Beverly1874) The decline in the population of panthers eventuallycaptured the attention of lawmakers subsequently they receivedpartial legal protection as a game animal in 1950 and completelegal protection in Florida by 1958 (Onorato et al 2010) By the1960s the core of remaining wilderness within the Big Cypressregion of South Florida where the last group of breedingpanthers was thought to persist was further affected by theconstruction of Alligator Alley This roadway along with theTamiami Trail allowed access to these once inaccessible areasAll these factors ultimately led to the listing of the Floridapanther as endangered on the inaugural list of endangeredspecies on 11 March 1967 (US Federal Register 1967) andsubsequent protection under the Endangered Species Act of1973 (Public Law 93‐205) This designation invariably raisedawareness of the plight of the panther and served as a catalyst toinitiate research to avoid what appeared to be their imminentextinction (Onorato et al 2010)In 1973 the World Wildlife Fund initiated a survey that re-

sulted in the capture of a single female panther and doc-umentation of their sign in just a few locations in South Florida(Nowak and McBride 1974) The information accrued duringthis survey would ultimately lead to the commencement of along‐term research project on the Florida panther in 1981 by

what is now known as the Florida Fish and Wildlife Con-servation Commission (FWC) Over the next decade a multi-tude of studies were published that helped improve knowledgeof panther reproduction (Maehr et al 1989 Barone et al 1994)diet (Maehr 1990) and genetics (OrsquoBrien et al 1990 Roelkeet al 1993) among other topics Early research was essential foridentifying challenges faced by the remaining panthers mostimportantly the impacts of the continued loss of habitat isola-tion of the population and inbreeding depression (Onoratoet al 2010) In combination these factors would force wildlifemanagers to contemplate the implementation of genetic in-trogression to avert the extinction of the pantherBy the early 1990s the Florida panther population had declined

to a minimum of 20ndash30 individuals (McBride et al 2008) Severalstudies documented low levels of genetic diversity and expressedtraits thought to be correlated with inbreeding depression (Roelkeet al 1993 Johnson et al 2010 Onorato et al 2010) includingcowlicks (mid‐dorsal pelage whorls) and kinked tails (Wilkinset al 1997) These morphological traits probably had little directimpact on fitness of Florida panthers More concerning were theproportion of individuals in the population in the early 1990s thatwere unilaterally or bilaterally cryptorchid (Roelke et al 1993)possessed poor sperm quality (Barone et al 1994) sufferedcompromised immune systems (OrsquoBrien 1994) exhibited atrialseptal defects (Roelke et al 1993) and possessed some of thelowest levels of genetic variation observed in wild felids (Driscoll

Figure 1 Historical and current (2015) breeding range of the Florida panther in the southeastern United States Delineation of current breeding range is describedin Frakes et al (2015)

van de Kerk et al bull Florida Panther Population and Genetic Viability 7

et al 2002) Taken together these factors suggested that thelong‐term prospects for the persistence of the Florida pantherwere poorSubsequently the United States Fish and Wildlife Service

(USFWS) conducted an assessment that led to the delineation of 4possible management options 1) initiate a captive breeding pro-gram 2) introduce genetic material from captive stockpurported to contain both pure Florida panther and SouthAmerican puma lineages 3) translocate wild female pumas (Pumaconcolor stanleyana) from Texas USA into South Florida or 4)take no action at all (USFWS 1994) Collaborative discussionsinvolving academics non‐governmental organizations and stateand federal wildlife agencies eventually led to the selection of op-tion 3 and a plan for genetic introgression that involved the releaseof 8 female pumas from Texas into South Florida was im-plemented in 1995 (Seal 1994 Onorato et al 2010)Continued research conducted after the introgression interven-

tion provided clear evidence of the benefits accrued to the pantherpopulation including increased levels of genetic diversity declinesin the frequency of morphological and biomedical correlates ofinbreeding depression and the substantial increase of the popu-lation size (McBride et al 2008 Johnson et al 2010 McClintocket al 2015) Furthermore Hostetler et al (2013) showed that thepost‐introgression population growth was driven primarily byhigher survival rates associated with admixed panthers (offspringof pairings between the female Texas pumas and male Floridapanthers and subsequent generations of those progeny) followinggenetic introgression The apparent success of genetic introgres-sion in preventing the imminent extinction of the Florida pantherpopulation is encouraging but the population remains small andisolated and continues to face substantial challenges especially thelarge‐scale loss of habitatAlmost 5 panther generations (generation time= 45 years

Hostetler et al 2013) have passed since the implementation of thegenetic introgression program The panther population has beenmonitored continuously during this time period leading to asubstantial amount of additional data and biological insights sincethe demographic assessment made immediately following the in-trogression (Hostetler et al 2010 2012 2013 Benson et al 2011)Given that the effects of genetic introgression are not permanent itis important to discern whether the positive impacts of this man-agement initiative are continuing or waning Decline in the benefitsof the introgression could reduce the population growth rate andprobability of persistence At the same time panther habitat israpidly lost to a variety of human activities (eg land development)and environmental changes associated with invasive species andclimate change (Land et al 2008 Dorcas et al 2012 Frakes et al2015) When long‐term monitoring data are available it is prudentto periodically evaluate demographic rates population growth andpersistence parameters so that changes are detected and timelymanagement action can be implementedTwo population modeling approaches have become popular

among ecologists and wildlife managers for studying the dynamicsand persistence of age‐ or stage‐structured populations matrixpopulation models and individual‐based models (IBMs Brooket al 2000 Morris and Doak 2002 Morrison et al 2016) Matrixpopulation models are the generalized discrete time‐version of theLotka‐Euler equation which describes the dynamics of age‐ or

stage‐structured populations of plants animals and humans (Lotka1924 Caswell 2001) They have become standard tools for mod-eling human and wildlife populations (Tuljapurkar 1990 Caswell2001 Keyfitz and Caswell 2005) because 1) they are powerful andflexible permitting adequate representation of life history of a widevariety of organisms including those with complex life histories 2)parameters for these models can be empirically estimated usingstandard analytical methods such as multi‐state capture‐recapturemethods (eg Williams et al 2002) 3) it is relatively straight‐forward to include the effects of factors such as environmental anddemographic stochasticity and density dependence 4) importantdemographic quantities such as asymptotic population growth ratestable age‐ or stage distribution and generation time can be easilycalculated from these models 5) they permit prospective and ret-rospective perturbation analyses which have been proven useful inwildlife conservation and management 6) they are computer‐friendly do not require extensive programming experience andtheir implementation using software packages such as R (R CoreTeam 2016) and MATLAB (MathWorks 2014) is straight‐for-ward and 7) they have sound theoretical foundation offeringanalytical solutions to many aspects of population modeling Weused matrix population models for 1) calculating the aforemen-tioned demographic quantities 2) performing sensitivity analysesinvolving asymptotic population growth rate and 3) comparativepurposes to check the results obtained from equivalent simulation‐based modelsDespite many advantages offered by matrix population models it

is difficult to incorporate attributes of individuals such as geneticsor behavior within that framework Agent‐based or IBMs (Grimm1999 Grimm and Railsback 2005 McLane et al 2011 Railsbackand Grimm 2011 Albeke et al 2015) offer an alternative modelingframework with tremendous flexibility Individual‐based modelsare computer simulation models that rely on a bottom‐up approachthat begins by explicitly considering the components of a system(ie individuals) population‐level properties emerge from the be-havior of and interactions among discrete individuals UsingIBMs it is possible to explicitly represent attributes of individualssuch as movement mating or genetics (McLane et al 2011Railsback and Grimm 2011 Albeke et al 2015) We used in-dividual‐based population models as the primary populationmodeling approach because they offered a framework for an ex-plicit consideration of genetics both at individual and populationlevelsOur overall goal was to provide a comprehensive demographic

assessment of the sole extant population of the Florida pantherusing long‐term (1981ndash2015) field data and novel modelingtechniques Our objectives were 1) to estimate demographicparameters for the Florida panther and investigate whetherpositive impacts of the 1995 genetic introgression remained inthe population and if so to what extent 2) to estimate popu-lation growth and persistence parameters using an IBM tailoredto the Florida pantherrsquos life history and 3) to evaluate thesensitivity of population growth and persistence parameters tovital demographic ratesThe Florida panther population remains small (le230 adults and

subadults FWC 2017) with no realistic possibility of natural geneflow so the loss of genetic variation and concomitant increase ininbreeding‐related problems are inevitable Ensuring the long‐term

8 Wildlife Monographs bull 203

persistence of the population will likely necessitate periodic geneticmanagement interventions Therefore our final objective was to 4)evaluate genetic and population‐dynamic consequences of variedgenetic introgression strategies and from these identify manage-ment strategies that are affordable and effective at improvingprospects of Florida panther persistence To achieve this objectivewe extended the individual‐based population model to track in-dividual genetics based on empirical allele frequencies taking ad-vantage of the long‐term demographic and genetic data that havebeen collected on Florida panthers since 1981 We used individualheterozygosity to inform vital rates based on empirically estimatedrelationships between individual heterozygosity and demographicparameters We then simulated population trajectories trackedindividual heterozygosity over time and estimated quasi‐extinctiontimes and probabilities without introgression and with introgres-sion for a range of intervals and number of immigrants perintrogression event We compared the genetic and demographicbenefits of implementing alternative genetic introgression scenariosand ranked them by the level of improvement in population per-sistence parameters and estimated costs because funding is a per-sistent challenge to conservation planningManagement of imperiled species is a complex process invol-

ving many stakeholders and plagued by layers of uncertainties(Runge 2011) It would therefore seem that management ofthreatened and endangered species would be a perfect case foradaptive resource management approach (ARM) because ARMfocuses on structured decision making for recurrent decisionsmade under uncertainty (Williams et al 2002 Runge 2011)Because it combines structured decision making with learningprudent application of ARM reduces uncertainty over time andcan lead to optimal management actions (Williams and Brown2012 2016) Key ingredients of a successful adaptive manage-ment program include stakeholder engagement clearly definedand mutually agreed‐upon objectives alternative managementactions and objective decision rules (Nichols et al 2007Williams and Brown 2012 2016) Unfortunately Florida pan-ther stakeholders have divergent views about the state of thesystem (eg panther abundance) and it is a challenge to havemutually agreed‐upon objectives management actions or deci-sion rules acceptable to all or most stakeholders AlthoughFlorida panther conservation efforts would be best served withinthe ARM framework in the long run impediments to its im-plementation (Runge 2011 Williams and Brown 2012) areunlikely to be overcome in the near future

STUDY AREA

The breeding population of Florida panthers mainly persists as asingle population South of the Caloosahatchee River (approxi-mately 267133degN latitude 815566degW longitude see Fig 1)One exception is a female that was documented just north of theCaloosahatchee River in 2016 (and subsequently documentedwith kittens in 2017) by FWC the first validation of a femalepanther north of the River since 1973 (FWC unpublisheddata) The breeding range in South Florida is topographicallyflat has a subtropical climate and is characterized by permanentand ephemeral wetlands influenced by rains from May throughOctober (Duever et al 1986) The study area contains a variety

of wildland land cover types including hardwood hammockscypress forests pine flatwoods freshwater marshes prairies andgrasslands (Davis 1943) and land characterized by human ac-tivity including citrus groves croplands pastureland rockmining and residential developments (Onorato et al 2011)The area is intersected by a multitude of roads that fragmentpanther habitat Most notable among these is Interstate 75(Alligator Alley) which was expanded from 2 to 4 lanes in 1990when fencing and wildlife crossings were constructed with theintention of minimizing harmful effects of road expansion onlocal wildlife (Foster and Humphrey 1995) A large portion ofthe area that comprises South Florida is under public ownershipmainly as federal or state lands Everglades National Park BigCypress National Preserve Florida Panther National WildlifeRefuge Picayune Strand State Forest and Fakahatchee StrandPreserve State Park alone encompass 9745 km2 though not allof the area is considered suitable for panthers (eg large ex-panses of sawgrass marshes in the western Everglades) Frakeset al (2015) delineated 5579 km2 of adult panther habitat inSouth Florida 1399 km2 of which are in private ownership Amajority of these private lands are located in the northern extentof the breeding range Although some private lands may beprotected (eg conservation easements) other areas are sus-ceptible to incompatible land uses such as rock mining or re-sidential developments The breeding range in South Florida isbounded to the east and west by large metropolitan areas in-clusive of Miami‐Fort Lauderdale‐Pompano Beach and CapeCoral‐Fort Myers‐Naples‐Marco Island‐Immokalee respec-tively that are populated by gt6000000 people (httpswwwcensusgov accessed 24 Jan 2019) These characteristics of thestudy area highlight the recovery challenges faced by panthers

METHODS

Field MethodsSince 1981 Florida panthers have been captured radiocollared andtracked by the FWC and National Park Service staff (Table 1)Capture and handling protocols followed guidelines of theAmerican Society of Mammalogists (Sikes 2016) Livestock Pro-tection Company (Alpine TX USA) provided trained hounds andhoundsmen for captures Teams either treed or bayed panthers onthe ground and then darted them with a 3‐ml compressed‐air dartfired from a CO2‐powered rifle Immobilization drugs have variedduring the tenure of the project but most recently teamsimmobilized panthers with a combination of ketamine HCl (10mgkg Doc Lanersquos Veterinary Pharmacy Lexington KY USA) andxylazine HCl (1mgkg Doc Lanersquos Veterinary Pharmacy) Fol-lowing immobilization teams lowered treed panthers to the groundby a rope or caught panthers with a net in some cases they used aportable cushion (McCown et al 1990) to further mitigate theimpact of a fall Capture teams administered propofol (PropoFlotradeAbbott Laboratories Abbott Park IL USA) intravenously either asa bolus or continuous drip to maintain anesthesia They adminis-tered midazolam HCl (003mgkg) intramuscularly or intravenouslyto supplement anesthesia in some panthers Panthers recovered in ashaded area away from water In some cases capture teams reversedxylazine HCl with yohimbine HCl (Yobinereg Lloyd IncShenandoah IA USA) at 25 its recommended dose

van de Kerk et al bull Florida Panther Population and Genetic Viability 9

Capture teams determined sex and age of panthers marked themwith an ear tattoo and a subcutaneous passive integrated trans-ponder (PIT‐tag) and fitted them with a very high frequency(VHF) or global positioning system (GPS) radio‐collar equippedwith a mortality sensor The GPS collars were programmed toattempt to collect a position at a variety of time intervals(range= 1ndash12 hr) depending on objectives of concurrent studiesobservers routinely tracked panthers with VHF radiocollars from afixed‐wing aircraft 3 times per week When a radio‐collar trans-mitted a mortality signal or when a location did not change forseveral days observers investigated the site to establish the pan-therrsquos fate When observers found a dead panther they assessed thecause of death based on an examination of the carcass and sur-rounding area if possible They then transported carcasses to anexperienced veterinarian for necropsy and a histopathological ex-amination to attempt to assign the cause of death Starting in 1995observers continually assessed successive locations of radio‐collaredfemales to determine if they initiated denning behavior Three to 4fixes at the same location (VHF collars) or successive returns to thesame location over a weeklong period (GPS collars) served as anindicator of a possible den Observers then located dens moreprecisely via triangulation with ground telemetry They visited denswhen the dam left the site typically to make a kill Teams capturedkittens by hand captured kittens ranged in age from 4 to 35 dayspost‐partum Teams counted sexed weighed dewormed andpermanently marked kittens with a subcutaneous PIT‐tag

Genetic AnalysisTeams collected blood tissue or hair from the majority ofcaptured kittens subadult and adult Florida panthers for DNAsamples The United States Department of Agriculture ForestService National Genomics Center for Wildlife and FishConservation (Missoula MT USA) processed samplesTechnicians used Qiagen DNeasy Blood and Tissue Kits(Qiagen Valencia CA USA) to extract whole genomic DNAfrom blood and tissue samples and used a slight modification(overnight incubation of sample in lysis buffer and proteinase Kon a rocker at 60deg elution of DNA in 50 μl of buffer) of theQiagen DNA extraction protocol (Mills et al 2000) to extractDNA from hair samples Technicians used a panel of 16 mi-crosatellite loci (Fca090 Fca133 Fca243 F124 F37 Fca075Fca559 Fca057 Fca081 Fca566 F42 Fca043 Fca161

Fca293 Fca369 and Fca668) identified by Menotti‐Raymondet al (1999) which had been previously applied to Floridapanther samples (Johnson et al 2010) Technicians amplifiedDNA samples in 10‐microl reaction volumes that included 10 microl ofDNA 1times reaction buffer (Applied Biosystems Life Technolo-gies Corporation Grand Island NY USA) 20 mM of MgCl2200 microM of each dNTP 1 microM of reverse primer 1 microM of dye‐labeled forward primer 15 mgml of bovine serum albumin(BSA) and 1U Taq polymerase (Applied Biosystems) Thepolymerase chain reactions (PCRs) included a thermal profile of94degC for 5 minutes followed by 36 cycles of 94degC for 60 seconds55degC for 60 seconds and 72degC for 30 seconds Techniciansvisualized the resulting PCR products on a LI‐COR DNAanalyzer (LI‐COR Biotechnology Lincoln NE USA) Tech-nicians collected genotypes from hair samples (n= 16) using amultitube approach (Taberlet et al 1996) error‐checked geno-types using Program DROPOUT (McKelvey and Schwartz2005) and combined them with genotypes from tissue andblood samples (n= 487) We evaluated all 16 loci for variousdescriptive statistics for the radio‐collared sample of adult andsubadult panthers We excluded genotypes of kittens from thatanalysis because related individuals are known to result in theoverrepresentation of certain alleles (Marshall et al 1998) Wedetermined the number of alleles observed heterozygosity andexpected heterozygosity using GenAlex (Peakall and Smouse2006 2012) We assessed allelic richness at each locus in Fstat293 (Goudet 1995) We assessed conformance of genotypedata from these 16 loci to HardyndashWeinberg assumptionslinkage disequilibrium and null alleles (Appendix A availableonline in Supporting Information) The Genomics Center alsoprovided us with genotype data from the same 16 loci for 41pumas from western Texas the source population for thegenetic rescue programEvidence of inbreeding depression in wild populations is

commonly determined on the basis of heterozygosityndashfitnesscorrelations (Silva et al 2005 Ortego et al 2007 Mainguyet al 2009 Johnson et al 2011) so we also estimated theheterozygosity of each individual panther We calculatedhomozygosity by loci (HL) which varies between 0 (all lociheterozygous) and 1 (all loci homozygous Aparicio et al 2006)using the Rhh package (Alho et al 2010) in Program R (R CoreTeam 2016) A microsatellite locus has more weight in HL

Table 1 Number of Florida panthers by age sex and genetic ancestry categories used in subsequent demographic analyses during the study period (1981ndash2013) inSouth Florida USA

Kittensa Subadults and adultsb

Males Females Total Males Females Total

PIT‐taggedc 215 180 395 Radiocollaredc 108 101 209Recovered 11 4 15 Subadult 72 50 122Litter failed 47 38 85 Prime adult 65 95 160Recaptured 25 30 55 Old adult 13 20 33Canonical 27 20 47 Canonical 43 29 72F1 admixed 6 12 18 F1 admixed 2 6 8Other admixed 130 105 235 Other admixed 45 53 98

a Kitten samples included those that were PIT‐tagged recovered (ie panthers that were found dead) part of a failed litter or recaptured as subadults or adultsb Radiocollared subadult and adult panthers are presented as sample size within each age class and ancestry classc Combined number of panthers of different ancestries is less than the number radiocollared or tagged with a subcutaneous passive integrated transponder (PIT‐tagged) because we did not conduct genetic sampling on all panthers Combined number of panthers in different age classes is greater than the number ofradiocollared panthers because some individuals were counted in ge2 age classes as animals aged

10 Wildlife Monographs bull 203

when it is more informative (ie has more alleles) and has aneven distribution of allele frequencies For our demographicanalyses we calculated individual heterozygosity for each in-dividual panther as 1 minusHL As do most indices of individualheterozygosity HL takes into account the average allele fre-quency in the population (Aparicio et al 2006) Thus duringsimulations the allele frequency in the population changesthroughout an individualrsquos lifetime which causes its hetero-zygosity also to change This is not biologically plausible becausegenetic properties are retained throughout an individualrsquos life-time Therefore for use in model simulations we simply cal-culated heterozygosity for each individual as the proportion ofheterozygous lociWe also used genotype data to implement a Bayesian clus-

tering analysis in Program STRUCTURE (Pritchard et al2000) to infer ancestral clusters of panthers This method uses aMarkov chain Monte Carlo (MCMC) method to determine thenumber of genetic clusters (K) in the sample while also pro-viding an individualrsquos percentage of ancestry (q values) allocatedto each cluster Runs for this analysis included a 100000MCMC burn‐in period followed by 500000 MCMC iterationsfor 1ndash10 ancestral genetic clusters (K= 1 to 10) with 10 re-petitions for each K To determine the most likely number of Kgenetic clusters we used the logarithm of the probability of thedata (log Pr(D)∣K Pritchard et al 2000) and estimates of ΔK(Evanno et al 2005) in Program STRUCTURE HAR-VESTER (Earl and VonHoldt 2011) We then used q valuesprovided in runs for the best K to assign individual panthers aseither canonical (in most cases ge90 pre‐introgression ancestrysee Johnson et al 2010) or admixed (lt90 pre‐introgressionancestry) We recognized admixed panthers that were the off-spring of matings between a Texas female puma and a canonicalmale panther as F1 admixed panthers for analyses

Demographic ParametersSurvival of kittensmdashMost kittens PIT‐tagged at den sites were

never encountered again A small proportion of the PIT‐taggedkittens were recovered dead or were recaptured as subadults oradults and radiocollared so that their fate could be monitoredWe also identified instances of complete litter failure when afemale died within 9 months after giving birth (the minimumage of independence for kittens) or when a female wasdocumented to have another litter less than 12 months aftergiving birth (the minimum age of independence plus a 3‐monthgestation period Hostetler et al 2010) Thus our data setconsisted of 1) live recaptures and subsequent radio‐tracking ofpanthers of various ages that had been PIT‐tagged as kittens 2)recovery of dead panthers of various ages that had been PIT‐tagged as kitten 3) live captures and subsequent radio‐trackingof other panthers and 4) instances of complete litter failure(Table 1) We analyzed these data using Burnhamrsquos live‐recapture dead‐recovery modeling framework (Burnham 1993Williams et al 2002) to estimate survival of kittens (0ndash1‐yearold) and to test covariate effects on this parameter We includeddata on individuals encountered dead or alive at an older age inour analyses because their encounter histories also providedinformation on kitten survival We used a live‐dead data inputformat coded with a 1‐year time step (Cooch and White 2018)

All known litter failures occurred within the first year of thekittensrsquo lives so we treated all litter failures as recoveries withinthe first year We treated panthers that would have died if wehad not removed them to captivity as having died on the date ofremoval Data preparation and analytical approach are describedin more detail by Hostetler et al (2010)In addition to survival probability (S) the Burnham model

incorporates parameters for recapture probability (p) recoveryprobability (r) and site fidelity (f) We set r and p to 10 forradio‐tracked individuals because their status was known eachyear We also fixed f at 10 for all individuals because the re-capture and recovery areas were the same and encompassed theentire range of the Florida panther Based on findings of Hos-tetler et al (2010) we used a base model that 1) allowed survivalto differ between kittens and older panthers 2) allowed survivalto differ between sexes and among age classes (females ages 1and 2 ge3 males ages 1 2 and 3 ge4) 3) constrained recaptureprobability to be the same for all uncollared panthers and 4)allowed recovery rates to differ between kittens and uncollaredolder panthers Even with our additional data this model re-mained the best fit for a range of models for recapture recoveryand survival of kittens subadults and adults of various age‐classcategories (Hostetler et al 2010)Subadult and adult survivalmdashTo estimate survival for panthers

ge1 year of age we analyzed radio‐tracking data using a Coxproportional hazard‐modeling framework (Cox 1972 Therneauand Grambsch 2000) We followed procedures outlined byBenson et al (2011) for data preparation and analysis Brieflywe right‐censored panthers on the last day of observation beforetheir collar failed When an individual aged into a new age classwhile being tracked we created 2 data entries 1 ending the dayit entered the new age class the other starting on that day Weused the FlemingndashHarrington method to generate survivalestimates from the Cox analysis and the Huber sandwichmethod to estimate robust standard errors (Therneau andGrambsch 2000 Benson et al 2011)We considered 3 age classes subadults (1ndash25 yr for females and

1ndash35 yr for males) prime adults (25ndash10 yr for females and35ndash10 yr for males) and old adults (gt10 yr old for both sexes)after Benson et al (2011) We estimated overall survival using abase model that allowed survival to differ between subadultfemales subadult males prime adult females prime adult malesand old adults of either sex (old age was additive with sex other-wise age and sex were interactive) This model provided the best fitto the data relative to a range of models considering differences insurvival between prime‐adult and older‐adult age classes both ad-ditively and interactively with sex (Benson et al 2011)To examine whether survival rates observed in recent years

differed from those observed immediately following introgres-sion we also estimated age‐specific survival rates for 2 nearlyequal periods of time immediately post‐introgression (1995ndash2004) and the more recent period (2005ndash2013) For theseanalyses we separated the radio‐tracking data into the 2 periodsand estimated survival rates for each period separately using thebase model Similar analyses for kitten survival were not possiblebecause of data limitationsProbability of reproductionmdashWe used telemetry data obtained

from radio‐tracked females that reproduced at least once to

van de Kerk et al bull Florida Panther Population and Genetic Viability 11

estimate annual probability of reproduction of subadult prime‐adult and old‐adult females using complementary logndashlogregression (Agresti 2013) following methods described indetail by Hostetler et al (2012) Briefly we modeled theprobability of a female panther giving birth (b) as a function ofcovariates as

γ= minus [minus ( )]b z1 exp exp i i

where zi is a row vector of covariates (see below for covariateinformation) for female panther i and γ is a column vector ofmodel coefficients (Appendix B Table B2) To account fordifferences in observation time we included log(m) as an offsetin the model where m is the number of months a femalepanther was radio‐tracked (Hostetler et al 2012)

Reproductive outputmdashTo estimate the number of kittensproduced by reproductive females we used cumulative logitregression (Min and Agresti 2005 Agresti 2013) Weconsidered a model that includes J categories for the numberof offspring with j denoting the number of offspring ( j= 1 2hellip J and J= 6) provided that a female reproduces Weconsidered J= 6 which is greater than the largest litter sizeobserved in our study (4 kittens) because females can produce 2or more litters per year if litters fail We modeled the probabilitythat a female i produces a litter of size at most j (Yi) as

le⎧

⎨⎩

δ( ) = == hellip minus

=

( )+ θ βminus +Y jj J

j JPr

1 2 1

1i ij e

1

1 xj i

where θj is the intercept for the annual number of kittens pro-duced by a female i being at most j xi is a row vector of cov-ariates for female panther i (see below) and β is a column vectorof model coefficients The probability that Yi will be exactly j isgiven by

⎧⎨⎩

πδ

δ δ( = ) = =

=

minus = hellip( minus )Y j

jj J

Pr 1

2 i ij

i

ij i j

1

1

Finally the expected annual number of kittens produced byfemale i (νi) is given by

sumν π==

j i

j

J

ij

1

Additional details regarding the estimation and modelingof the annual number of kittens produced can be found inHostetler et al (2012)

CovariatesmdashIn addition to sex and age we tested for the effect ofabundance genetic ancestry and individual heterozygosity on theaforementioned demographic parameters (model coefficients forkitten and adult survival are given by η and ι respectively AppendixB Table B2) We used a minimum population count (MPC) basedon radio‐tracking data and field evidence of subadult and adult (ieindependent age) panthers as an index of abundance (McBride et al2008) For these analyses we did not use the population sizeestimates reported by McClintock et al (2015) because unlike theminimum counts they did not cover the entire study period

(McBride et al 2008 R T McBride and C McBride RancherrsquosSupply Incorporated unpublished report Fig 2)To test if demographic parameters differed depending on an-

cestry we considered 3 ancestry models dividing the panthers into1) F1 admixed and all other panthers 2) canonical and admixed(including F1 admixed) panthers and 3) F1 admixed canonicaland other admixed panthers We quantified genetic diversity usingindividual heterozygosity as described previously We tested for theeffect of genetic covariates using a subset of kittens that were ge-netically sampled We calculated model‐averaged estimates ofmodel parameters on the scale in which they were estimated(Appendix B available online in Supporting Information)

Cause‐specific mortalitymdashWe performed cause‐specificmortality analysis only on dead radio‐collared panthers becausedetection rates of uncollared panther mortalities are highlycorrelated with the cause of death (eg vehicle collision)Following Benson et al (2011) we attributed each mortality to1 of 4 causes 1) vehicle collision 2) intraspecific aggression 3)other causes (including known causes such as diseases injuriesand infections unrelated to the first 2 causes) and 4) unknown(ie mortalities for which we could not assign a cause) Wethen used the non‐parametric cumulative incidence functionestimator a generalization of the staggered‐entry KaplanndashMeiermethod of survival estimation to estimate cause‐specificmortality rates for male and female panthers in different ageclasses (Pollock et al 1989 Heisey and Patterson 2006 Bensonet al 2011) We compared cause‐specific mortality rates betweensexes age classes and ancestries

Statistical inferencemdashWe used an information‐theoreticapproach (Akaikersquos information criterion [AIC]) for modelselection and statistical inference (Burnham and Anderson 2002)For the analysis of kitten survival we adjusted the AIC foroverdispersion and small sample size (QAICc details in Hostetleret al 2010) We performed all statistical analyses in Program R (RCore Team 2016) We used Burnhamrsquos live recapturendashdead

0

100

200

300

1985 1990 1995 2000 2005 2010Year

Indi

vidu

als Method

MPC

MVM

Figure 2 The Florida panther minimum population count (MPC) and populationsize estimated using motor vehicle collision mortalities (MVM) during our studyperiod Minimum population counts are from McBride et al (2008) and R TMcBride and C McBride (Rancherrsquos Supply Incorporated unpublished report)Estimates based on MVM are from McClintock et al (2015) The solid vertical lineindicates the year of genetic introgression

12 Wildlife Monographs bull 203

recovery model (Burnham 1993) to estimate and model kittensurvival using the RMark package version 2113 (Laake 2013) asan interface for Program MARK (White and Burnham 1999)We implemented the Cox proportional hazard model for esti-

mation and modeling of survival of subadults and adults using the Rpackage SURVIVAL (Therneau and Grambsch 2000) We per-formed cause‐specific mortality analyses using an R version of S‐PLUS code provided by Heisey and Patterson (2006)To account for model selection uncertainty we calculated

model‐averaged parameter estimates across all models using theR package AICcmodavg (Mazerolle 2017) using AIC weights asmodel weights (Burnham and Anderson 2002) We used modelswithout the categorical ancestry variables for model averagingand estimated parameters based on the average value for thecontinuous covariates (individual heterozygosity and abundanceindex) Because only a subset of the kittens was geneticallysampled we estimated 2 kitten survival rates First we estimatedkitten survival based on the data set that included all kittens(overall kitten survival) Second we estimated kitten survivalusing a subset of the data that only included kittens that weregenetically sampled (survival of genetically sampled kittens)

Matrix Population ModelsWe investigated Florida panther population dynamics andpersistence using 2 complementary modeling frameworks ma-trix population models (Caswell 2001) and IBMs (Grimm andRailsback 2005 McLane et al 2011 Railsback and Grimm2011 Albeke et al 2015) Matrix population models offer aflexible and powerful framework for investigating the dynamicsand persistence of age‐ or stage‐structured populations (egCaswell 2001 Robinson et al 2008 Kohira et al 2009 Hunteret al 2010 Hostetler et al 2012) With a fully developed theorybehind them these models also serve as a check for resultsobtained from simulation‐based models such as IBMs We usedan age‐structured‐matrix population model for deterministic andstochastic demographic analyses and for preliminary investiga-tions of Florida panther population viability (Caswell 2001Morris and Doak 2002)For deterministic and stochastic demographic analyses we

parameterized a female‐only 19 times 19 age‐specific populationprojection matrix of the form

⎢⎢⎢⎢⎢

⋯ ⋯

⋯ ⋯

⋱ ⋱ ⋮

⋮ ⋱ ⋱ ⋱ ⋮

⎥⎥⎥⎥⎥

( )=t

F F F

P

P

P

A0 0

0

0 0 0

t t t

t

t

t

1 2 19

1

2

18

where Fi and Pi are the age‐specific fertility and survival ratesrespectively and ( )tA is a time‐varying (or stochastic) popula-tion projection matrix We assumed that the longevity and ageof last reproduction of female Florida panthers was 185 yearswe based this assumption on the observation that the oldestfemale in our study was 186 years old Florida panthers re-produce year‐round thus we used birth‐flow methods to esti-mate Fi and Pi values following Caswell (2001) and Hostetleret al (2013) We estimated Pi as

= +P S S i i i 1

where Si is the age‐specific survival probability We estimated Fi as

⎛⎝

⎞⎠

=+ +F S

m Pm

2i k

i i i 1

where Sk is the kitten survival probability and mi is the age‐specific fecundity rate which was estimated as

ν=m b f i i i k

where bi is the breeding probability for a female in age class iduring time step t νi is the annual number of kittens producedby a breeding female in age class i and fk is the proportionfemale kittens at birth (assumed to be 05)We estimated the finite deterministic (λ) and stochastic annual

population growth rate (λs) and stochastic sensitivities and elasti-cities following methods described in detail by Caswell (2001) Toestimate λs we assumed identically and independently distributedenvironments (Caswell 2001 Morris and Doak 2002 Haridas andTuljapurkar 2005) and used a parametric bootstrapping approachto obtain a sequence of demographic parameters for 50000 ma-trices We then estimated log(λs) from this sequence of matrices(Tuljapurkar 1990 Caswell 2001) and exponentiated it to obtain λsWe calculated the sensitivity and elasticity of λs to changes in lower‐level vital demographic rates as per Caswell (2001) and Haridas andTuljapurkar (2005)For population projection and population viability analysis (PVA)

simulations we expanded the population projection matrix into a34 times 34 2‐sex age‐structured matrix to include female and maleFlorida panthers (Caswell 2001 Hostetler et al 2013) Malescontribute to the population only through survival in this modelingframework We assumed maximum longevity for males to be 145years of age because the oldest male in our study was 144 years oldThe population projection matrix was of the following form

⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢

⋯ ⋯ ⋯ ⋯

⋯ ⋯ ⋯ ⋯

⋱ ⋱ ⋮ ⋮ ⋱ ⋱ ⋮

⋮ ⋱ ⋱ ⋱ ⋮ ⋮ ⋱ ⋱ ⋮

⋯ ⋯ ⋯

⋯ ⋯ ⋯ ⋯

⋯ ⋯ ⋱ ⋱ ⋮

⋮ ⋮ ⋱ ⋱ ⋮ ⋱ ⋱ ⋮

⋯ ⋯ ⋯

⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥

( )=

( ) ( ) ( )

( )

( )

( )

( ) ( ) ( )

( )

( )

t

F N F N F N

P N

P N

P N

M N M N M N

Q N

Q N

A n

0 0

0 0 0 0

0

0 0 0 0 0

0 0

0 0 0

00 0 0 0 0

t t t

t

t

t

t t t

t

t

1 2 19

1

2

18

1 2 19

1

14

where Pi and Qi are age‐specific survival rates of female and maleFlorida panthers respectively and Fi and Mi are the rates atwhich female and male kittens are produced by females of ageclass i respectively Here the population projection matrix isboth time‐varying and density‐dependent (Caswell 2001)We incorporated the influence of parameter uncertainty en-

vironmental stochasticity and density dependence on Floridapanther population dynamics and persistence following the ap-proach outlined by Hostetler et al (2013) we used the sameapproach in the IBM analysis which is described below UnlikeIBMs matrix models do not intrinsically include demographicstochasticity so we explicitly included the influence of

van de Kerk et al bull Florida Panther Population and Genetic Viability 13

demographic stochasticity by simulating the fates of individualsas described in Caswell (2001)Population projections using matrix models necessitate a

vector of initial sex‐ and age‐specific abundance Because suchdata are currently unavailable for the Florida panther we esti-mated it based on the stable sex and age distribution and MPCin 2013 Using point estimates of the demographic parameterswe estimated the stable sex and age distribution for the 2‐sexdeterministic and density‐independent population projectionmatrix We then multiplied the stable sex and age distributionby the MPC in 2013 to get the starting population vector n(0)We projected the population over time as n(t+ 1)=A(tn)n(t)where A(tn) indicates the time‐specific density‐dependentpopulation projection matrix (Caswell 2001)We had 2 indices of abundance available the MPC (McBride

et al 2008) and population size estimated using motor vehiclecollision mortalities (MVM McClintock et al 2015) These 2indices are substantially different because MPC does not ac-count for imperfect detection whereas MVM accounts forimperfect detection and provides an estimate of population size(Fig 2) Therefore we examined 2 scenarios one using densitydependence based on MPC and the other using density de-pendence based on MVM We ran 1000 bootstraps of para-meter values and 1000 simulations per bootstrap for each sce-nario We projected population trajectories over 200 years andthen estimated population growth rates probabilities ofquasi‐extinction and time until quasi‐extinction and comparedthese with the results obtained from the IBM (see below) Weran scenarios with a quasi‐extinction threshold of 10 and 30panthers We used the mean of probabilities of quasi‐extinction(PQE) across simulations as the point estimate for PQE and the5th and 95th percentiles across simulations as the 90 con-fidence interval Because the confidence intervals werequantile‐based but the point estimates were not it is possiblewith skewed distributions for the point estimates to be outsidethe confidence interval We implemented the matrix

population models in the R computing environment (R CoreTeam 2016)

IBMs and PVABecause of well‐developed theory ease of implementation andthe modeling flexibility that they offer matrix populationmodels have become the model of choice for investigating thedynamics and persistence of age‐ and stage‐structured popula-tions (Caswell 2001) However it is difficult to incorporateattributes of individuals such as genetics and behavior usingmatrix models Quantifying the effects of genetic erosion on theFlorida panther population dynamics and persistence andevaluating benefits and costs of alternative genetic managementscenarios were important goals of our study Because IBMs relyon a bottom‐up approach that begins by explicitly consideringindividuals (McLane et al 2011 Railsback and Grimm 2011Albeke et al 2015) it is possible to represent genetic attributesof each individual and to track genetic variation over time Weused IBMs as the primary modeling framework in this studybecause they allowed for explicit consideration of genetics bothat individual and population levels and population viabilityassessment under alternative genetic management scenariosWe developed an IBM (Grimm 1999 Grimm and Railsback

2005 McLane et al 2011 Railsback and Grimm 2011 Albekeet al 2015) tailored to the life history of the Florida panther(Fig 3) We simulated every individual from birth until deathbased on a set of rules describing fates (eg birth and death) andattributes (eg genetics) of individuals at each annual time stepAs in other IBMs population‐level processes (eg populationsize and genetic variation) emerged from fates of and interac-tions among individuals (Grimm 1999 Railsback and Grimm2011) We developed an overview design concepts and details(ODD) protocol (Grimm et al 2006 2010) to describe ourIBM (Appendix C available online in Supporting Information)and provide pseudocodes for individual‐based simulations(Appendix D available online in Supporting Information)

For each panther at time = t

Kitten (lt1 yr)

Reproduce

Survive

Kitten(s) born

Kitten

Sex and age class

Subadult male

Prime adult male

Old adult male

Subadult female

Prime adult female

Old adult female

Age one year

No Male

Female

Yes

No

Yes

t=t+1

Yes

lt35 yrs

35-10 yrs

gt10 yrs

lt25 yrs

25-10 yrs

gt10 yrs

Figure 3 Schematic representation of Florida panther life history We tailored the individual‐based population model to represent the life‐history patterndepicted here

14 Wildlife Monographs bull 203

Dynamics and persistence of populations are influenced bymany factors such as demographic and environmental sto-chasticity and density dependence these effects and parametricuncertainty must be considered in population models whenpossible (Caswell 2001 Boyce et al 2006 Bakker et al 2009Hostetler et al 2013) Because by definition IBMs simulate thelife history of individuals in a population they inherently in-corporate the effect of demographic stochasticity Our analysesrevealed that survival of kittens subadults and adults and theprobability of reproduction varied over time so we incorporatedenvironmental stochasticity in the model for these variablesfollowing Hostetler et al (2013) Likewise we found evidenceof density‐dependent effects on kitten survival Because theMPC might be an underestimate of population size (McBrideet al 2008) we performed simulations using both the MPC andMVM abundance indicesTo account for model selection and parametric uncertainty we

used a Monte Carlo simulation approach For each bootstrapsample we selected a model based on the AIC (or QAICc)weights We then sampled the intercept and slope for kittensurvival (on the logit scale) subadult and adult survival (on thelogit scale with log‐hazard effect sizes) and probability of re-production (on the complementary logndashlog scale Appendix B)from multivariate normal distributions with mean vectors equalto the estimated parameters and variance‐covariance matricesequal to the sampling variance‐covariance matrices We thentransformed the results to nominal scale and converted the es-timates to age‐specific survival and reproduction parameters asdescribed in Hostetler et al (2013)We ran 1000 simulations for 1000 parametric bootstraps for

200 years for the MPC and the MVM scenario for both thestochastic 2‐sex matrix‐population model and the IBM Wetracked the population size at each time step and estimatedquasi‐extinction probabilities and times based on the populationtrajectory We considered the population to be quasi‐extinct ifindividuals of only 1 sex remained or if the population size fellbelow critical thresholds set at 10 and 30 individuals We alsoestimated expected time to quasi‐extinction for both thresholdsdefined as the mean and median time until a population be-comes quasi‐extinct conditioned on quasi‐extinction We im-plemented the IBM using the RNetLogo package version 10ndash2(Thiele et al 2012 Thiele 2014) as an interface for the IBMplatform NetLogo (Wilensky 1999)

Sensitivity AnalysisAn important component of demographic analysis is sensitivityanalysis which is the quantification of the sensitivity of amodelrsquos outcome to changes in the input parameters (Caswell2001 Bakker et al 2009 Thiele et al 2014) Using an IBM weperformed global sensitivity analyses for 2 (ie MPC andMVM) density‐dependent scenarios to assess how sensitivequasi‐extinction times and probabilities were to changes (anduncertainties) in the estimated demographic rates We usedLatin hypercube sampling to sample parameters from the entireparameter space (Marino et al 2008 Thiele et al 2014) Wesampled intercept and slope for kitten survival (on logit scale)subadult and adult survival (on logit scale with log‐hazard effectsizes) and probability of reproduction (on complementary

logndashlog scale) from normal distributions We sampled theprobability distribution for the number of kittens producedannually (on the real scale) from a Dirichlet distribution themultivariate generalization of the beta distribution (Kotz et al2000) We sampled 1000 different parameter sets and ran 1000simulations for each scenario We calculated the probability ofquasi‐extinction within 200 years for each density‐dependencescenario (MPC or MVM) using a quasi‐extinction threshold of10 individuals and then calculated the partial rank correlationcoefficient between each of those and each parameter trans-formed to the real scale (Bakker et al 2009 Hostetler et al2013) Partial rank correlation coefficients quantify the sensi-tivity of the probability of quasi‐extinction to input variables(ie survival and reproductive parameters) with higher absolutevalues indicating stronger influence of that variable on quasi‐extinction probabilities

Genetic ManagementOne of our goals was to estimate the benefits (improvements ingenetic variation demographic parameters and populationgrowth and persistence parameters) and costs (financial costs ofcapture transportation quarantine release and monitoring ofintroduced panthers) of genetic management To achieve thisobjective we extended the IBM to include a genetic componentTo simulate individual‐ and population‐level heterozygosity overtime we assigned each panther alleles for 16 microsatellite locibased on the empirical allele frequency in the population (esti-mated from genetic samples collected during 2008ndash2015) whenwe initialized the model (Pierson et al 2015) At each time stepwe simulated Mendelian inheritance for kittens produced suchthat each kitten randomly inherited alleles from its mother andfrom a male panther randomly chosen to have putatively siredthe litter We did not explicitly force inbreeding to occur butthe probability of inbreeding naturally increased as the popula-tion size decreasedIntrogression strategiesmdashWe modeled genetic introgression as a

result of the release of 2‐year‐old female pumas from Texas intothe simulated Florida panther population For the geneticintrogression implemented in 1995 Texas females between 15and 25 years of age were released because females at that agehave lower mortality rates (Benson et al 2011) higherreproductive potential and also because it is much easier todetect with certainty the successful reproduction by females thanby males The ability to more closely monitor reproduction andthe birth of kittens is an important consideration in reducing theprobability of a genomic sweep of Texas alleles into the Floridapanther population Wildlife managers removed the last 2surviving Texas females from the wild population in Florida in2003 to reduce the possibility of genomic sweep To simulatethis strategy in the model we removed any Texas females thatwere still alive 5 years after each introgression event Weassigned Texas females alleles based on the allele frequency ofthe Texas population (based on genotypes from 48 samplescollected in west Texas that was inclusive of the pumas releasedin South Florida in 1995) when they entered the Florida pantherpopulation We could not estimate demographic rates separatelyfor the released Texas females because of small sample size (only8 females were released in 1995) therefore we assumed that

van de Kerk et al bull Florida Panther Population and Genetic Viability 15

survival and reproductive rates and the effects of covariates onthese rates for the released Texas females were identical to thoseof female Florida panthersThe impact of genetic introgression depends on the frequency

of introgression events and the number of individuals introducedat each attempt Thus we defined introgression strategies basedon the number of Texas females to be released (0 5 10 or 15)and the interval between genetic introgressions (10 20 40 and80 yr) for a total of 13 strategies We simulated each manage-ment strategy with density dependence estimated using theMPC and MVM abundance indices

We sampled 1000 parameter sets and for each of these setswe ran all introgression strategies 1000 times for 200 years Weapplied the same parametric uncertainty and environmentalstochasticity across all management scenarios to allow for bettercomparison of introgression strategies At each time step werecorded the projected population size and average individualheterozygosity We calculated the heterozygosity for each in-dividual at birth based on the alleles it inherited This individualheterozygosity then informed the survival rate of that individualbased on the empirically estimated relationship between in-dividual heterozygosity and age‐specific survival rates Hetero-zygosity did not change throughout an individualrsquos lifetime butits effect on survival changed as individuals aged because theeffect of individual heterozygosity on survival rate differedamong age classes Survival rates of genetically sampled kittenswere biased high because some kittens were sampled later in lifewhen they had already survived for some time To correct forthis bias we adjusted kitten survival rates predicted by theheterozygosity model proportionally From the population tra-jectories we estimated the probability of quasi‐extinction (de-fined as the probability that the population will fall below 10 or30 individuals or that individuals of only 1 sex will remain)

Cost‐benefit analysismdashExperts estimated financial costs ofintrogression based on their experience with the geneticintrogression executed in 1995 and we adjusted estimates tocurrent costs to account for inflation (R T McBride RancherrsquosSupply Incorporated M Cunningham and D P OnoratoFlorida Fish and Wildlife Conservation Commission personalcommunication) Costs included capturing and transportingfemale pumas from the Texas source population caring for thepumas while they were in quarantine and post‐introgressionmonitoring To compare the financial costs of the differentscenarios we also calculated the average costs incurred over

Table 2 Sample size (n) number of alleles (Na) allelic richness (AR) observedheterozygosity (Ho) and expected heterozygosity (He) at 16 microsatellite loci inradio‐collared Florida panthers sampled from 1981 to 2013 in South FloridaUSA We calculated these values from a subset (n= 137) of the samples used inthe analyses and they include only adult and subadult panthers We excludedkittens because they can result in an overrepresentation of certain alleles

Locus n Na AR Ho He

FCA090 129 5 50 0333 0401FCA133 136 3 30 0618 0608FCA243 136 5 50 0419 0487F124 137 7 69 0708 0729F37 129 4 40 0395 0397FCA075 131 5 50 0534 0598FCA559 133 5 50 0496 0503FCA057 134 6 59 0679 0624FCA081 134 7 69 0209 0206FCA566 130 6 60 0462 0475F42 133 10 100 0737 0766FCA043 136 5 49 0596 0588FCA161 132 6 60 0500 0515FCA293 132 4 40 0614 0624FCA369 131 6 60 0573 0567FCA668 133 7 70 0406 0462Mean 1329 57 57 0517 0535

Figure 4 Proportional membership (q) of radio‐collared Florida panthers (n= 218) and pumas from Texas USA (n= 7) in 2 clusters identified by STRUCTUREEach individual animal is represented by a separate vertical bar Yellow indicates canonical panther ancestry and blue represents admixed ancestry The admixedancestry of the Texas females and F1 generation panthers that were radiocollared are highlighted red and green respectively to demarcate those groups The pre‐introgression period is inclusive of panthers radiocollared from 10 February 1981 to 6 March 1996 The post‐introgression era inclusive of the F1 panthers includesanimals radiocollared from 4 March 1997 to 18 February 2015 in South Florida USA

16 Wildlife Monographs bull 203

100 years assuming that the population would be managedaccording to a particular strategy Our goal with theseexploratory analyses was to evaluate the relative costs andbenefits of alternative management scenarios

RESULTS

Genetic VariationWe assessed genetic variation at 16 microsatellite loci for 137DNA samples collected from adult and subadult panthers(1981ndash2013) that were captured and subsequently radiocollaredThe number of alleles at these loci ranged from 3ndash10 and allelicrichness averaged 57 (Table 2) Average expected hetero-zygosity for this sample was 0535 and average observed het-erozygosity was 0517 (Table 2)We determined individual heterozygosity (1 minusHL) and an-

cestry for the complete sample of panther genotypes (n= 503)collected from 1981 to 2015 for use as covariates in subsequentanalyses Average individual heterozygosity was 0559 andranged from 0ndash0980 Our STRUCTURE analysis providedstrong support for K= 2 clusters based on the probability of thedata (log Pr(D)|K) and ΔK in STRUCTURE HARVESTERBased on this result we categorized the ancestry of each pantheras either canonical (n= 123) or admixed (n= 380) using valuesof proportional membership (q Fig 4) The average individualheterozygosity of canonical panthers was 0386plusmn 0012 (SE) foradmixed panthers it was 0615plusmn 0007

Survival and Cause‐Specific Mortality RatesOf the 395 kittens that were PIT‐tagged and released 55 werelater encountered alive 15 were recovered dead and 85 werepart of failed litters (Table 1) We radio‐tracked 209 subadult oradult panthers for a total of 244195 panther‐days and docu-mented 126 mortalities (Table 1)Survival depended strongly on age and kittens had the lowest

overall rate (032plusmn 009 Fig 5A Table 3) The model‐averagedkitten survival was 047plusmn 009 when we included only geneti-cally sampled individuals in the analysis (Table 3) as statedabove this estimate is biased high because many kittenswere genetically sampled when they were recaptured alive orrecovered dead at older ages We found no evidence for sex‐specific differences in kitten survival but females survived betterthan males in all older age classes For females subadults hadthe highest survival rates followed by prime adults and oldadults For males prime adults had the highest survival ratefollowed by subadults and old adults (Fig 5A and Table 3)For panthers of all ages there was substantial evidence that

ancestry affected survival with F1 admixed panthers of all ageshaving the highest rates and canonical individuals having thelowest (Fig 5B) The combined AIC weights of models thatincluded an ancestry covariate were 0880 for kittens and 0992for older panthers The survival rates of subadult prime‐adultand old‐adult panthers in the period immediately after the in-trogression (1995ndash2004) were similar to those in more recentyears (2005ndash2013 Fig 5C)After genetic introgression individual panther heterozygosity

increased (Fig 6) and varied among ancestry categories individualheterozygosity was the highest for F1 panthers (078plusmn 001)

A

B

C

Figure 5 Overall age‐ and sex‐specific survival rates (plusmnSE A) age‐ and sex‐specific survival rates (plusmnSE) for Florida panthers of canonical F1 admixed andother admixed ancestry (B) and age‐ and sex‐specific survival estimates (plusmnSE)for subadult and adult Florida panthers in the period immediately post‐introgression (1995ndash2004) and more recently (2005ndash2013 C) in SouthFlorida USA

Table 3 Model‐averaged estimates (plusmnSE) of demographic parameters for theFlorida panther obtained using data collected during 1981ndash2013 in SouthFlorida USA

Parameter Sex Age class Estimate

Survival Both Kitten 032plusmn 009a

Female Subadult 097plusmn 002Prime adult 086+ 003Old adult 078plusmn 009

Male Subadult 066plusmn 006Prime adult 077plusmn 005Old adult 065plusmn 010

Probability of reproduction Female Subadult 035plusmn 008Prime adult 050plusmn 005Old adult 025plusmn 006

Kittens produced annually Female Subadult 280plusmn 005Prime adult 267plusmn 001Old adult 228plusmn 013

a Overall kitten survival (based on all kittens in our data set including thosethat were not genetically sampled) Estimate of survival of geneticallysampled kittens was 047plusmn 009 this estimate is biased high because manykittens were genetically sampled when they were recaptured alive or re-covered dead at older ages

van de Kerk et al bull Florida Panther Population and Genetic Viability 17

followed by those for non‐F1 admixed (062plusmn 001) and canonical(039plusmn 001) panthers Individual heterozygosity positively affectedsurvival rates of kittens (slope parameter η= 1455 95CI= 0505ndash2405) Individual heterozygosity also positively af-fected survival of subadult and adult panthers but the evidence forthis effect was weaker because the confidence interval for the ha-zard ratio overlapped 10 (hazard ratio exp(ɩ)= 0311 95CI= 0092ndash1054 Table 4 and Fig 7)The MPC increased after genetic introgression (Fig 2) This

abundance index was also included in top‐ranking modelsindicating that population density affected survival (Table 4)Abundance reduced the survival of kittens (slope parameterη= minus0035 95 CI= minus0049 to minus0021) evidence was weak forthe effect of abundance on survival of subadult and adult panthers(hazard ratio exp(ɩ)= 1002 95 CI= 0995ndash1010 Fig 7) Re-gression coefficients associated with the effect of abundance andindividual heterozygosity on demographic parameters are presentedin Table B1 (available online in Supporting Information)The most frequently observed causes of death for radio‐

collared panthers were vehicle collision and intraspecific ag-gression Cause‐specific mortality analyses revealed thatmortality rates among possible causes were generally similar forthe 2 sexes but that males were more likely to die from in-traspecific aggression than were females (Fig 8A) Male mor-tality from intraspecific aggression was higher for subadult andold adult panthers than for prime adults (Fig 8B) For bothmales and females canonical panthers were more likely to diefrom intraspecific aggression than were admixed panthers(Fig 8C)

Reproductive ParametersOf 101 radio‐collared females 67 reproduced at least once duringour monitoring we used these individuals to assess the annualprobability of reproduction The top‐ranked model for the annual

probability of reproduction included age effect only suggesting thatadding ancestry or abundance did not improve model parsimony(Table 4) Model‐averaged annual probabilities of reproductiondiffered between female subadults (035plusmn 008) prime adults(050plusmn 005) and older adults (025plusmn 006)The most parsimonious model for the number of kittens

produced annually by females that produced at least 1 litter (ieconditional on reproduction) included effects of age ancestryand abundance a model that included only the effect of age wasalmost equally supported (Table 4) The model‐averagednumber of kittens produced annually conditional on re-production was 280plusmn 075 for subadults 267plusmn 043 for primeadults and 228plusmn 083 for old adults Confidence intervals forthe slope parameter (β) overlapped 0 for both abundance(β= 0010 95 CI= minus0001 to 0021) and ancestry (β= minus073795 CI= minus1637 to 0162)

Population Dynamics and PersistenceDeterministic and stochastic demographymdashThe deterministic (λ)

and stochastic (λs) annual population growth rate calculated usingthe matrix population model were 106 (5th and 95th percentiles099ndash114) and 104 (072ndash141) respectively The generation timewas 47 years A female starting in age class one (kitten) wasexpected to live another 58 years and produce 10 kittens onaverage during her lifetime Overall λs was proportionately(elasticity) most sensitive to changes in female prime adultsurvival on the absolute scale (sensitivity) however it was mostsensitive to changes in kitten survival (Fig 9) Populationtrajectories probabilities of quasi‐extinction and times untilquasi‐extinction estimated using the matrix population modeland IBM for comparable scenarios were nearly identical for acritical quasi‐extinction threshold of 10 panthers (Figs 10 and 11)and for a critical quasi‐extinction threshold of 30 panthers(Appendix E available online in Supporting Information)Minimum population count scenariomdashUnder the MPC scenario

(ie density dependence estimated using the minimumpopulation count data) with a critical threshold of 10panthers the population size reached 99 (72ndash104) and 100(77ndash105) at 200 years (Fig 10) as predicted by the IBM and thematrix model respectively The cumulative quasi‐extinctionprobabilities within 100 years based on the MPC scenario were15 (IBM 0ndash48) and 30 (matrix model 0ndash76) Theseprobabilities increased to 25 (IBM 0ndash114) and 38 (matrixmodel 0ndash154) by year 200 (Fig 10) The mean times untilquasi‐extinction were 15 years (IBM 0ndash67) and 15 years (matrixmodel 0ndash72) conditioned on going quasi‐extinct within 100years Expected times until quasi‐extinction conditioned onquasi‐extinction within 200 years were higher 34 years (IBM0ndash137) and 32 years (matrix model 0ndash134 Fig 10)Motor vehicle mortality scenariomdashUnder the MVM scenario

(ie density dependence estimated using estimates of pantherpopulation size from McClintock et al 2015) with a criticalthreshold of 10 panthers the projected population reached 188(142ndash217) and 190 (143ndash219) at 200 years as predicted by theIBM and the matrix model respectively (Fig 11) Thecumulative quasi‐extinction probabilities within 100 years were14 (IBM 0ndash08) and 13 (matrix model 0ndash06) Theseprobabilities increased to 20 (IBM 0ndash17) and 19 (matrix

000

025

050

075

100

1995 2000 2005 2010Year of birth

Indi

vidu

al h

eter

ozyg

osity

Figure 6 Temporal changes in the individual heterozygosity of Floridapanthers born during the period 1995ndash2013 in South Florida USA Pointsare measurements solid line is linear regression shaded area is 95 confidenceinterval of slope Slope= 0006plusmn 0001 R2= 009

18 Wildlife Monographs bull 203

model 0ndash16) within year 200 (Fig 11) The mean times until quasi‐extinction based on the MVM scenario were 5 years (IBM 0ndash61)and 6 years (matrix model 0ndash64) conditioned on going quasi‐extinctwithin 100 years Expected times until quasi‐extinction conditionedon quasi‐extinction within 200 years were 11 years (IBM 0ndash113)and 11 years (matrix model 0ndash112 Fig 11)

Sensitivity of extinction probability to demographic parametersmdashThe partial rank correlation coefficients between quasi‐extinctionprobabilities and demographic parameters were negative anincrease in any of the demographic parameters decreased thequasi‐extinction probability Probabilities of quasi‐extinction within200 years were most sensitive to changes in kitten survival for bothscenarios (Fig 12) followed by survival of subadult females in theMVM scenarios and survival of prime adult females in the MPCscenario The probability of reproduction of prime adult femaleshad the third‐largest partial rank correlation coefficient with quasi‐extinction probability for both scenarios

Benefits and Costs of Genetic ManagementMinimum population count scenariomdashThe MPC scenario with

a critical threshold of 10 panthers predicted that withoutmanagement intervention average individual heterozygositywould decrease reaching 057 (049ndash064) at 100 years and053 (042ndash062) at 200 years (Fig 13) The population woulddecrease slightly reaching an average of 79 (41ndash99) at 100 yearsand 76 (43ndash98) at 200 years (Fig 14) The probabilities of quasi‐extinction under the no‐intervention MPC scenario when weconsidered the effects of genetic erosion were 13 (0ndash99) at 100years and 23 (0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with

the translocation of 5 Texas females average individualheterozygosity under the MPC scenario increased and thenstabilized at 068 (064ndash072) at 100 years and remained atthat level at 200 years (Fig 13) The population reached anaverage of 88 (62ndash104) at 100 years and stayed the same at200 years (Fig 14) The probabilities of quasi‐extinctionunder this introgression strategy were 6 (0ndash53) within 100years and 7 (0ndash82) within 200 years (Fig 15) Results ofanalyses using a critical threshold of 30 panthers revealed asimilar pattern of relative benefits However the prob-abilities of quasi‐extinction were substantially higher (ge45)across all scenarios (Appendix E)

Motor vehicle mortality scenariomdashWithout managementintervention the average individual heterozygosity predictedby the MVM scenario and a critical threshold of 10 panthersdecreased to 060 (046ndash069) at 100 years and to 059 (039ndash070) at 200 years (Fig 13) The population increased slightlyand reached an equilibrium at 154 individuals (46ndash217) at 100years and 157 (57ndash218) at 200 years (Fig 14) The probabilitiesof quasi‐extinction were 17 (0ndash100) at 100 years and 22(0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with 5

female pumas from Texas average individual heterozygosityincreased slightly reaching 067 (060ndash073) at 100 years and068 (059ndash075) at 200 years (Fig 13) Under this strategy thepopulation reached an average of 164 individuals (44ndash226) at

Table 4 Model comparison results testing for the effects of several covariateson survival of all kittens survival of genetically sampled kittens survival ofsubadult and adult Florida panthers probability of reproduction and kittensproduced annually for Florida panthers sampled from 1981ndash2013 in SouthFlorida USA

Model Parameters ΔAICa Weight

Kitten survival (all data)Base+ F1b+ abundancec 10 000 0988Base+ abundance 9 1018 0006Base+ F1 9 1035 0005Base 8 2711 lt0001Base+ sexd 9 2912 lt0001Base+ litter sizee 9 2914 lt0001

Kitten survival (genetically sampled kittens only)Base+ F1+ abundance 10 000 0541Base+ F1CanAdmf+ abundance 11 099 0329Base+Hetg+ abundance 10 310 0115Base+CanAdmh+ abundance 10 791 0010Base+ abundance 9 944 0005Base+ F1 9 1638 lt0001Base+ F1CanAdm 10 1837 lt0001Base+Het 9 2872 lt0001Base 8 3296 lt0001Base+CanAdm 9 3348 lt0001

Subadult and adult survivalBase+ F1CanAdm+ abundance 7 000 0360Base+ F1 5 053 0276Base+ F1CanAdm 6 105 0213Base+ F1+ abundance 6 222 0119Base+CanAdm+ abundance 6 575 0020Base+CanAdm 5 908 0004Base+Het 5 964 0003Base+Het+ abundance 6 984 0003Base 4 1118 0001Base+ abundance 5 1278 0001

Probability of reproductionAge 3 000 0242Age+CanAdm 4 012 0228Age+ abundance 4 173 0102Age+ F1 4 190 0094Age+CanAdm+ abundance 5 216 0082Age+Het 4 219 0081Age+ F1CanAdm 5 232 0076Age+ F1+ abundance 5 372 0038Age+Het+ abundance 5 399 0033Age+ F1CanAdm+ abundance 6 444 0026

Kittens produced annuallyAge+ F1+ abundance 5 000 0194Age 3 060 0144Age+ abundance 4 061 0143Age+ F1 4 097 0120Age+CanAdm 4 139 0097Age+ F1CanAdm+ abundance 6 196 0073Age+ F1CanAdm 5 231 0061Age+CanAdm+ abundance 5 238 0059Age+Het+ abundance 5 250 0056Age+Het 4 259 0053

a Akaikersquos information criterion (AIC) for kitten survival models was adjustedfor small sample size and overdispersion (QAICc)

b F1 divides Florida panthers into 2 groups F1 admixed and all other panthersc Index of Florida panther abundanced Sex of an individual Florida panther (male or female)e Size of the litter in which a Florida panther kitten was bornf F1CanAdm divides panthers into 3 groups F1 admixed canonical andother admixed panthers

g Het is the individual heterozygosityh CanAdm divides panthers into 2 groups canonical and admixed (includingF1 admixed) panthers

van de Kerk et al bull Florida Panther Population and Genetic Viability 19

100 years and 170 (67ndash231) at 200 years based on the MVMscenario (Fig 14) The probabilities of quasi‐extinction underthis introgression strategy were 10 (0ndash99) at 100 years and12 (0ndash99) at 200 years (Fig 15)The projected population size and average individual hetero-

zygosity reached equilibrium levels with periodic fluctuations inthese values when introgression was implemented (Figs 16ndash17)Genetic introgression decreased the time until quasi‐extinctionacross all scenarios (Fig 18) Results of analyses using a criticalthreshold of 30 panthers revealed a similar pattern of relative ben-efits However the probabilities of quasi‐extinction were sub-stantially higher (ge45) across all scenarios (Appendix E)

Addition of pumas from Texas increased both the populationsize and average individual heterozygosity consequentlyintrogression caused periodic increases in average individualheterozygosity and population size at the interval at which theintrogression occurred Whereas average individual hetero-zygosity gradually decreased following introgression densitydependence caused the population size to fluctuate especiallywhen a larger number of pumas from Texas were released Thepositive effect of genetic introgression initiatives on quasi‐extinction probabilities decreased as the time interval betweenthese events increased More frequent genetic introgressions ledto greater increases in average individual heterozygosity and

Old adult

Prime adult

Subadult

Kitten

025 050 075

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Individual heterozygosity

Ann

ual s

urvi

val r

ate

Old adult

Prime adult

Subadult

Kitten

40 60 80 100 120

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Abundance index

Ann

ual s

urvi

val r

ate

Kittens Males Females

Figure 7 The effect of individual heterozygosity (left) and the abundance index (right) on Florida panther age‐ and sex‐specific survival estimates (shaded area isSE) in South Florida USA 1981ndash2013 Abundance index data are from McBride et al (2008) and R T McBride and C McBride (Rancherrsquos Supply Incorporatedunpublished report)

20 Wildlife Monographs bull 203

equilibrium population size while reducing the probability ofquasi‐extinction Comparatively performing genetic introgres-sion less often but with more individuals per release was lesseffective at improving the long‐term prospects for the pantherpopulation (Figs 13ndash15)

Financial cost and persistence benefitsmdashThe total estimated costof 1 genetic introgression was $40000 $80000 or $120000for releasing 5 10 or 15 pumas from Texas respectively(Table 5) The total cost incurred over 100 years for eachstrategy ranged from $50000 to $12 million (Table 6)The most expensive strategy involved the introduction of15 pumas every 10 years this strategy reduced the probability

of quasi‐extinction by 58 and 40 under the MPC and MVMscenarios respectively (Table 6) Introducing 5 pumas every80 years cost the least but improvements in populationpersistence under this strategy were insubstantial (Table 6)Releasing more panthers more frequently caused increasedpopulation fluctuations but did not necessarily reduce the quasi‐extinction probability (Figs 14 and 15 Table 6)

DISCUSSION

The duration (34 yr of data) and sample sizes associated with theFlorida Panther Project allowed us to obtain a comprehensiveunderstanding of genetic variation demographic parameters

A

B

C

Figure 8 Cause‐specific mortality rates (plusmnSE) for male and female Florida panthers (A) subadult prime‐adult and old‐adult Florida panthers of both sexes (B)and canonical and admixed Florida panthers of both sexes (C) in South Florida USA 1981ndash2013 Causes of mortality are vehicle collision intraspecific aggression(ISA) other causes (known causes such as diseases injuries and infections unrelated to the first 2 causes) and unknown cause (mortalities for which we could notassign a cause)

van de Kerk et al bull Florida Panther Population and Genetic Viability 21

probability of population persistence and the duration of thepositive impacts of genetic restoration on this endangered po-pulation The Florida panther population was in dire straits inthe early 1990s and our results clearly demonstrated that the

implementation of the introgression project in 1995 subse-quently accrued a series of genetic and demographic improve-ments that have led to the change from a declining population toone that is growing and expanding its current breeding range

Sensitivity Elasticity

0 1 2 3 00 02 04

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

Para

met

er

Figure 9 Sensitivity and elasticity (proportional sensitivity) of stochastic population growth rate (λs) of the Florida panther to vital demographic parameters in SouthFlorida USA 1981ndash2013 Horizontal lines represent 5th and 95th percentiles

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

70

80

90

100

110

0

5

10

15

0

50

100

Years

StatisticMeanMedian

Figure 10 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the minimum population count(MPC) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

125

150

175

200

00

05

10

15

20

0

30

60

90

Years

StatisticMeanMedian

Figure 11 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the motor vehicle mortality(MVM) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

22 Wildlife Monographs bull 203

MPC MVM

minus08 minus06 minus04 minus02 00 minus08 minus06 minus04 minus02 00

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Male subadult survival

Male prime adult survival

Male old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

PRCC

Para

met

er

Figure 12 Partial rank correlation coefficient (PRCC) between quasi‐extinction probabilities and the demographic parameters of the Florida panther for theminimum population count (MPC) and motor vehicle mortality (MVM) scenarios Error bars indicate standard deviation Based on data collected in South FloridaUSA 1981ndash2013

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 13 Average projected individual heterozygosities in the Florida panther population over 200 years without genetic management intervention and with eachof the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include theintroduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years Average individual heterozygosity peaks when pumas are added to the Floridapanther population Based on data collected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 23

Our research has further validated the success of genetic in-trogression by quantifying the reduction in the probability ofextinction of the panther population because of this manage-ment initiative These findings all bode well for continuedprogress toward recovery That said the Florida panther po-pulation remains small and isolated from other populations ofpuma from western North America thereby denoting that theeventual impact of genetic drift and inbreeding may negativelyaffect the population in the future Realizing this will continueto be an issue that managers will have to contemplate wemodeled the impact of varied genetic management scenarios todetermine the frequency of introgression along with the numberof panthers that should be released during each initiative tominimize cost and maximize benefits to the population Theseresults will prove invaluable to managers attempting to conservethe Florida panther

Genetic Variation Demographic Parametersand Persistence of Heterotic Benefits from GeneticIntrogressionOur measure of individual heterozygosity (1 minusHL) was higheron average for the admixed (0615) panthers than for those ofcanonical ancestry (0386) Levels of individual heterozygosityfor admixed panthers were comparable to levels observed in asample of pumas from west Texas (x plusmn SE= 0695plusmn 0019n= 41) which further demonstrates the improvement in levelsof genetic variation in the Florida population since 1995 The

correlation between HL and several demographic parametershas been noted in wild populations including cheetahs (Terrellet al 2016) and several species of birds such as European shags(Phalacrocorax aristotelis male and female survival probabilityVelando et al 2015) and Egyptian vulture (Neophron percnop-terus age at recruitment Agudo et al 2012) If we focus only onpanthers captured and radiocollared from 2012ndash2015 (FP211ndashFP240) all of which had estimated birth years ge2005 (ge10 yrpost‐introgression) those panthers had an average individualheterozygosity level of 0618plusmn 0029 highlighting the elevatedlevels of genetic variation that continue to persist in cohorts ofpanthers 5 generations after the release of female pumas fromTexasThe proportional membership values (q) from our

STRUCTURE analysis allowed us to delineate 2 clusters ofancestry for Florida panthers (Fig 4) During the pre‐in-trogression era the population was dominated by panthers thatretained a high percentage of the canonical ancestry which wasaffected by inbreeding depression The post‐introgression eraancestry values are comprised of many admixed individuals inessence demonstrating the success of the introgression in termsof increasing genetic variation in the population and mi-micking gene flow that historically occurred with panthers andother populations of pumas (Onorato et al 2010) A somewhatanalogous situation although abetted by natural immigrationwas evident in the ancestral variation in a small population ofpuma intersected by a major freeway in California USA In

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

75

80

85

90

95

100

130

150

170

190

Years

Popu

latio

n si

ze

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 14 Average projected population sizes in the Florida panther population over 200 years without intervention and with each of the genetic introgressionstrategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include the introduction of 5 10 or 15pumas from Texas USA every 10 20 40 or 80 years Peaks occur when pumas are added to the Florida panther population Based on data collected in SouthFlorida USA 1981ndash2013

24 Wildlife Monographs bull 203

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

25

50

75

100

0

25

50

75

100

Introgression interval (yrs)

PQ

E (

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 15 Average probability of quasi‐extinction (PQE) of the Florida panther population within 100 years and within 200 years without genetic managementintervention (N) and with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenariosIntrogression strategies include the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Errorbars indicate 5th and 95th percentiles Based on data collected in South Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

50

60

70

80

90

100

110

50

100

150

200

Years

Popu

latio

n si

ze

Figure 16 Average projected Florida panther population trajectories under a geneticintrogression strategy involving the release of 5 female pumas from Texas USA every20 years for the minimum population count (MPC) and motor vehicle mortality(MVM) scenarios Shaded area indicates 5th and 95th percentiles and isrepresentative of confidence intervals for other scenarios Based on data collected inSouth Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Figure 17 Average projected Florida panther population‐level heterozygositiesunder a genetic introgression strategy involving the release of 5 female pumas fromTexas USA every 20 years for the minimum population count (MPC) and motorvehicle mortality (MVM) scenarios Shaded area bounds the 5th and 95th percentilesand is representative of confidence intervals for other scenarios Based on datacollected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 25

this case the migration of just a single male across the freewayto the south resulted in an increase in the number of in-dividuals with admixed ancestry and substantially improvedgenetic diversity of the subpopulation south of the freeway(Riley et al 2014)Our estimate of overall kitten survival (0321plusmn 0056) was

similar to that reported by Hostetler et al (2010) who used13 years of post‐introgression data (0323plusmn 0071) These valuesare lower than kitten survival estimates derived from severalpuma populations in western North America (044ndash091 Loganand Sweanor 2001 Lambert et al 2006 Laundre et al 2007Robinson et al 2008 Ruth et al 2011) When we included onlydata for individuals that had been genetically sampled our es-timate was 15 higher The proportion of admixed kittens thatwere genetically sampled increased as the population expandedwhich could have resulted in higher estimates of kitten survivalbecause admixed kittens survive better than canonical kittensKitten survival was strongly influenced by genetic ancestry withsurvival rates of F1 kittens being almost twice that of canonicalkittens (Fig 5B) Consistent with the findings of Hostetler et al(2010) increases in heterozygosity coincided with increasedkitten survival whereas population density negatively affectedkitten survival Population density often has a strong negativeimpact on survival of young age classes due to infanticide andcannibalism which has been reported from several carnivore

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

50

100

150

0

50

100

150

Introgression interval (yrs)

TQE

(yrs

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 18 Average times until quasi‐extinction (TQE) for the Florida panther population within 100 years and within 200 years without genetic management interventionand with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategiesinclude the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Error bars indicate 5th and 95thpercentiles Based on data collected in South Florida USA 1981ndash2013

Table 5 Average cost (2017 US dollars) of introgression strategies employedover 100 years that involve the introduction of 5 10 or 15 pumas from TexasUSA into the Florida panther population in South Florida USA at an in-creasingly higher frequency Costs of capture (including transportation) quar-antine and post‐introgression monitoring were estimated by Roy McBrideMark Cunningham and Dave Onorato respectively

Frequency Component 5 pumas 10 pumas 15 pumas

Once Capture 25000 50000 75000

Quarantine 5000 10000 15000

Monitoring 10000 20000 30000

Total 40000 80000 120000

Every 80 years Capture 31250 62500 93750

Quarantine 6250 12500 18750

Monitoring 12500 25000 37500

Total 50000 100000 150000

Every 40 years Capture 62500 125000 187500

Quarantine 12500 25000 37500

Monitoring 25000 50000 75000

Total 100000 200000 300000

Every 20 years Capture 125000 250000 375000

Quarantine 25000 50000 75000

Monitoring 50000 100000 150000

Total 200000 400000 600000

Every 10 years Capture 250000 500000 750000

Quarantine 50000 100000 150000

Monitoring 100000 200000 300000

Total 400000 800000 1200000

26 Wildlife Monographs bull 203

species including polar bears (Ursus maritimus Derocher andWiig 1999) black bears (Ursus americanus Garrison et al 2007)and pumas (Logan and Sweanor 2001)Our estimates of sex‐ and age‐specific survival rates of subadult

and adult panthers (Table 3) and the effects of covariates on theserates were similar to those reported by Benson et al (2011) derivedfrom data collected through 2006 Our survival rates for males(065ndash077) were lower than females (078ndash097) for all 3 ageclasses (subadult prime adult and old adult) which is the typicaltrend observed in most puma populations (eg Ruth et al 19982011) However an unhunted puma population occupying afragmented urban landscape in southern California exhibited lowersurvival (0586 females 0525 males Vickers et al 2015) perhapsas a result of severe habitat fragmentation and the lack of extensiveswaths of public land protected in perpetuity Most populations ofpuma in western North America are typically subject to variedlevels of management via regulated hunting which often is biasedtoward the take of males (Cooley et al 2009 Ruth et al 2011)Consequently annual survival estimates from hunted populationsin northeast Washington (0679ndash0733 females 0341 malesRobinson et al 2008) and southeastern Arizona (0685 females0584 males Cunningham et al 2001) were generally lower thanthose we estimated for Florida panthersCause‐specific mortality analysis showed that male panthers

experienced a substantially greater mortality risk from in-traspecific aggression This differs from the pattern observedduring a long‐term study in Arizona where females were at ahigher risk of intraspecific mortality than males (46 of maleand 53 of female mortality Logan and Sweanor 2001)Mortality associated with intraspecific strife has been docu-mented in hunted and unhunted populations (this study Loganand Sweanor 2001 Stoner et al 2010) and its root cause hasbeen difficult to determine (eg population density and preypopulation trends) That being the case finding ways to reducewhat is often a major source of mortality for puma populationscontinues to pose a challenge to managersCanonical panthers were substantially more likely to die from

intraspecific aggression than were admixed panthers This mayat least partly explain the difference in survival between admixed

and canonical panthers Canonical panthers are known to bemore likely to suffer from varied correlates of inbreeding de-pression than admixed animals including atrial septal defects(Johnson et al 2010) and compromised immune systems(Roelke et al 1993) These factors have the potential to affectfitness and perhaps dominance of individuals when engaged inan intraspecific encounter A somewhat similar scenario wasobserved by Hogg et al (2006) in a population of bighorn sheepthat were part of an introgression project Admixed males in thispopulation had a greater probability of paternity than males thatwere less outbred This included coursing males with higherlevels of admixture being markedly more successful at fightingmales that were defending females in estrous (Hogg et al 2006)Heterosis resulting from genetic introgression is expected to be

the strongest in F1 individuals (Shull 1908 Crow 1948 Burkeand Arnold 2001 Waller 2015) and to progressively decline insubsequent generations (Pickup et al 2013 Frankham 2015)Given that admixed (especially F1) panthers survived better thancanonical panthers and that individual heterozygosity improvedsurvival of both sexes and all age classes our data provide evi-dence for heterotic advantage whereby individuals of mixedancestry had higher fitness (Shull 1908 Crow 1948) Our resultsare consistent with findings of other studies that involved in-tentional genetic introgression for conservation purposes Forexample Hogg et al (2006) detected substantial improvementsin survival and reproduction of introgressed bighorn sheep andMadsen et al (1999 2004) reported a dramatic increase in thepopulation of adders after genetic introgressionKnowledge of the persistence of benefits to a population ac-

crued via genetic introgression is essential for determiningwhen additional introgression may be warranted There is adepauperate amount of information regarding this topic and theidentification of factors that may influence persistence of thebenefits of genetic introgression (Tallmon et al 2004 Hedricket al 2014 Whiteley et al 2015) For example in minks(Neovison vison) Thirstrup et al (2014) found that outcrossingled to larger litters in F2 and F3 but this benefit disappeared insubsequent generations In a population of gray wolves Hedricket al (2014) suggested that benefits of genetic introgression may

Table 6 Costs and benefits accumulated over 100 years for genetic introgression at different intervals and with different numbers of pumas from Texas USAintroduced into the Florida panther population in South Florida USA Costs (2017 US dollars) include capture and transportation quarantine and post‐introgression monitoring Benefits are represented as average probability of quasi‐extinction (PQE and 5th and 95th percentiles) and changes (Δ) therein under thescenario using the minimum population count (McBride et al 2008 MPC) and the scenario using the motor vehicle mortality population estimates (McClintocket al 2015 MVM) The table is sorted by percent change in probability of quasi‐extinction under the MPC scenario

Interval (yr) Pumas Cost MPC PQE () ΔMPC PQE () MVM PQE () ΔMVM PQE ()

ndash 0 0 13 (0ndash99) 0 17 (0ndash100) 080 10 100000 12 (0ndash99) 10 (0ndash58) 16 (0ndash100) 5 (0ndash38)80 15 150000 12 (0ndash99) 13 (0ndash65) 15 (0ndash100) 8 (0ndash56)80 5 50000 12 (0ndash99) 14 (0ndash73) 15 (0ndash99) 9 (0ndash41)40 15 300000 10 (0ndash98) 26 (0ndash104) 14 (0ndash99) 13 (0ndash60)40 10 200000 9 (0ndash94) 35 (0ndash183) 13 (0ndash99) 21 (0ndash95)40 5 100000 8 (0ndash89) 39 (0ndash211) 12 (0ndash99) 26 (0ndash124)20 15 600000 8 (0ndash88) 42 (0ndash257) 13 (0ndash99) 24 (0ndash123)20 10 400000 6 (0ndash67) 54 (0ndash318) 10 (0ndash99) 37 (0ndash217)10 15 1200000 6 (0ndash53) 58 (0ndash372) 10 (0ndash99) 40 (0ndash208)20 5 200000 5 (0ndash50) 59 (0ndash338) 9 (0ndash99) 44 (0ndash352)10 10 800000 4 (0ndash16) 69 (0ndash489) 8 (0ndash92) 55 (0ndash401)10 5 400000 4 (0ndash7) 73 (0ndash502) 6 (0ndash71) 63 (0ndash503)

van de Kerk et al bull Florida Panther Population and Genetic Viability 27

wane after 2 or 3 generations The WrightndashFisher model ofgenetic drift predicts a geometric decline in expected hetero-zygosity at the rate of (1 minus 12Ne) per generation where Ne is theeffective population size (Hartl and Clark 1997 Frankham et al2014) Thus it seems logical to assume that the duration of thebenefits of introgression is limited especially in a species per-sisting in a small population such as Florida panthersWe expected Florida panther survival in the most recent years

(2005ndash2013) post‐introgression to differ from that observed in thedecade following the introgression in 1995 because pantherabundance and heterozygosity factors known to influence panthersurvival (Hostetler et al 2010 Benson et al 2011) have changed(Figs 2 and 6) We found that subadult and adult survival rates inrecent years (2005ndash2013) were similar to those in the period im-mediately following introgression (1995ndash2004 Fig 5C) overallkitten survival was similar to estimates of Hostetler et al (2010) Infact the pattern of covariate effects on Florida panther survivalreported by Benson et al (2011) and Hostetler et al (2010) hasbarely changed The negative effects of population density andpositive effects of heterozygosity may counterbalance survival ratesreducing population fluctuations Our result that benefits of geneticrestoration have remained in the population nearly for 5 genera-tions post‐intervention was surprising but also encouraging Pos-sible explanations for this observation may include 1) genetic ero-sion occurs at slower rate than previously thought 2) introducing 5migrants in 1 genetic intervention may have beneficial effects si-milar to those from 1 migrant per generation over multiple gen-erations or 3) other and as yet unknown factors may reduce therate of genetic erosion and subsequent demographic effects Adensity‐dependent effect on kitten survival could also explain whynon‐F1 admixed kittens did not survive substantially better thandid canonical kittens most kittens sampled from 2005ndash2013 wereadmixed and their survival was lower possibly because of a density‐dependent effect as the Florida panther population continuouslygrew during that period Trinkel et al (2010) also found thatpopulation density and inbreeding coefficient interacted to affectdemographic parameters and growth rate of an African lion po-pulation Likewise population density interacted with winterweather to affect the survival and population growth rates in Soaysheep (Ovis aries Milner et al 1999 Coulson et al 2001)

Population Dynamics and PersistenceThe finite deterministic and stochastic growth rates were gt10reflecting that much of the data used in this study were collectedfollowing genetic introgression in 1995 during which time thepopulation increased substantially Because our demographicparameter estimates did not change markedly in comparison tothose of Hostetler et al (2013) the similarity between thepopulation growth estimates from these studies was expectedBut these estimates reflect population growth during thedata‐collection period and do not predict future growth In factestimates of population size reported by McClintock et al(2015) suggest that population growth may already be slowing(Fig 2) A similar pattern was observed in an introduced po-pulation of African lion which increased steadily from 1995until 2001 then fluctuated around the presumed carrying ca-pacity thereafter (Trinkel et al 2010) Several other African lionpopulations that experienced various degrees of recovery

subsequently became relatively stable or exhibited decliningtrends (Bauer et al 2015)Our estimates of quasi‐extinction probabilities were lower than

those reported by Hostetler et al (2013) using a 2‐sex age‐structured matrix population model Lower probabilities ofquasi‐extinction than those reported by the earlier study forcomparable scenarios were likely a consequence of the fact thatthe estimates of abundance (McBride et al 2008) and thus thedensity‐dependent effects on survival of kittens and old panthers(gt10 yr) have changed in recent years Additionally these earlyestimates of the probability of quasi‐extinction and of time toquasi‐extinction did not consider genetic erosion that will in-evitably occur without management interventionUnder the MVM scenario the equilibrium population size was

twice that attained under the MPC scenario The MVM popula-tion estimates are approximately twice the MPCs (Fig 2) as aresult the simulated population under the MVM scenario achieveda higher equilibrium size Probability of quasi‐extinction under theMVM scenario was generally greater than that under the MPCscenario This result is a consequence of the fact that the estimateddensity dependence on kitten survival based on the MPC scenario isstronger than that for theMVM scenario (regression coefficients forabundance for the MPC scenario= minus0068 vs minus0002 for theMVM scenario Appendix B Table B1) Consequently simulatedpopulations under the MPC scenario have a stronger tendency tomove toward the equilibrium population size which inherentlyreduces the quasi‐extinction probability (eg Royama 1992)As expected for long‐lived mammals (Heppell et al 2000 Oli

and Dobson 2003) the population growth rate was proportio-nately most sensitive to changes in survival of prime adultsfollowed by that in younger age classes Global sensitivity ana-lysis in contrast showed that under all scenarios extinctionparameters were particularly sensitive to changes in survival ofFlorida panther kittens This result is a consequence of the highsensitivity of the population growth rate to kitten survival(Fig 9) and a larger standard error of kitten survival (Table 3)Results of our sensitivity analyses indicate that future researchshould focus on acquiring additional information on early lifestages to improve the accuracy and precision of estimates ofkitten survival Our results suggest that demographic variables(or their environmental drivers) that strongly influence extinc-tion risks are not necessarily those with the largest elasticityvalues A study of island fox (Bakker et al 2009) found thatextinction risk was strongly influenced by survival modifierparameters that had low elasticity values

Genetic Management When and How to InterveneThe use of genetic introgression as a management tool has al-ways been controversial and the probable effectiveness it mighthave in preventing the imminent demise of the panther popu-lation was initially debated (Creel 2006 Maehr et al 2006Pimm et al 2006) These debates notwithstanding the pantherpopulation had suffered from genetic morphological and bio-medical correlates of inbreeding before this management in-tervention (Johnson et al 2010 Onorato et al 2010) whereasthe post‐introgression period has been characterized by a sub-stantial reduction in the biomedical correlates of inbreeding andimprovements in demographic vigor and abundance (McBride

28 Wildlife Monographs bull 203

et al 2008 Hostetler et al 2010 2013 Johnson et al 2010Benson et al 2011 McClintock et al 2015) Nevertheless thepopulation remains small and isolated habitat is still being lostand there is no current possibility of natural gene flow betweenthis and other North American puma populations thus therecurrence of inbreeding‐related problems and subsequent po-pulation declines are inevitable The question therefore is notwhether genetic management of the Florida panther populationis needed but when and how it should be implementedWithout genetic introgression probabilities of quasi‐extinc-

tion were substantially greater than those from models thatincorporated periodic genetic introgression events (Fig 15)highlighting the importance of genetic management in smallisolated populations When 5 female pumas are added to theFlorida panther population every 10 years genetic variationincreased substantially under the MPC and MVM scenariosThe IBM predicted that the equillibrium population size wouldbe substantially larger when 5 pumas were released into thepopulation every 10 years than populations without intervention(ie no introgression) Releasing 15 Texas females did not af-ford substantially greater benefit to the equilibrium populationsize than did releasing only 5 Texas females Instead addingmore individuals caused increasingly large fluctuations in po-pulation size due to the numerical increases and density de-pendence (Fig 14) possibly diminishing the benefits associatedwith increased genetic variationCompared with the no‐intervention scenario (ie no genetic

introgression implemented) the introduction of 5 female pumasevery 10 years reduced quasi‐extinction probabilities from 13to 4 and from 17 to 6 for the MPC and MVM scenariosrespectively The benefit of genetic‐introgression strategies de-creased as the interval between successive management inter-ventions increased and no benefit was evident once the intervalreached 80 years (Fig 15) Introgression reduced the averagetime until quasi‐extinction (Fig 18) This somewhat counter‐intuitive result can be explained by the fact that repeated geneticintrogression makes the population more robust over timewhich will reduce the probability of extinction Consequentlyfew simulated population trajectories that fall below the criticalthreshold do so early reducing the mean time to extinctionAlso density dependence has a stabilizing effect on populations(Royama 1992) populations tend toward equilibrium popula-tion sizes which reduces the probability over time of popula-tions falling below quasi‐extinction threshold However timeuntil quasi‐extinction must be interpreted in conjunction withthe probability of quasi‐extinction which is very low for allscenarios with genetic introgression (Fig 15)Cost is a significant impediment to the implementation of a

genetic introgression program because captures relocations andmonitoring efforts before and after introgression are expensive(Edmands 2007 Frankham 2010b 2015 2016) Costs may beacceptable to society if the management actions result in sub-stantial fitness benefits especially if the species of concern ispopular and charismatic (Frankham 2015) We estimated the100‐year cost of introgression strategies modeled here to rangefrom $50000 to $12 million More expensive strategies gen-erally led to larger decreases in the probability of quasi‐extinc-tion but cheaper strategies also substantially improved

population viability (ie reduced quasi‐extinction probabilityTable 6) One of the cheaper strategies (5 pumas released every20 years) which cost an estimated $200000 over 100 yearsreduced the probability of quasi‐extinction by 59 and 44 forthe MPC and MVM scenarios respectively But these pre-liminary cost estimates are based on expenses incurred duringthe 1995 introgression and include costs of capture quarantinetransportation release and post‐release monitoring of femalesonly Although the cost for future genetic interventions wouldlikely be higher than those reported here the relative differencesin cost between alternative intervention strategies should remainapproximately the sameOur ability to simulate genetics and link it to the viability of

the Florida panther population is based on the empirically es-timated relationship between individual heterozygosity andsurvival Such heterozygosity and fitness correlations have beenstudied for decades (David 1998) but have only recently beenused to predict population viability (Benson et al 2016) noneto our knowledge have incorporated cost into their modelingefforts The mechanisms underlying heterozygosity and fitnesscorrelations remain unclear (David 1998 Szulkin et al 2010)and their significance as a proxy for inbreeding depression hasbeen debated (Balloux et al 2004 Miller and Coltman 2014)Nevertheless evidence continues to mount demonstratingpositive relationships between heterozygosity and survival(Coulson et al 1998ab Hansson et al 2004 Silva et al 2005Acevedo‐Whitehouse et al 2006 Jensen et al 2007 Velandoet al 2015) and between heterozygosity and fecundity (Seddonet al 2004 Charpentier et al 2005 Agudo et al 2012 Velandoet al 2015) Although the reported correlations are generallyweak their consequences can be substantial if population growthand persistence parameters are highly sensitive to the affectedvital rates (Velando et al 2015) Compared with scenarios thatignore genetics incorporating the effects of inbreeding on sur-vival increased quasi‐extinction probabilities within a 100‐yeartimeframe from 15 to 13 and from 14 to 17 for theMPC and MVM scenarios respectively These results high-light the importance of incorporating genetics when analyzingthe viability of small isolated populationsAlthough conservation biologists have recognized the im-

portance of genetics in population viability many still fail toincorporate genetic factors into PVA models (Reed et al 2002)Explicit consideration of genetics in PVAs requires long‐termgenetic and demographic data This permits the assessment ofthe functional relationship between genetic variability and de-mographic parameters Population viability analysis softwarepackages such as VORTEX (Lacy 1993 2000) offer a powerfulframework for individual‐based simulations but they often donot adequately capture a speciesrsquo life history or genetics Basedon a thorough review of the importance of genetic factors inwildlife conservation Frankham et al (2014) concluded thatgenetic factors can strongly influence the dynamics and persis-tence of small andor fragmented populations They furthernoted that ldquoMost PVA‐based risk assessments ignore or in-adequately model genetic factorsrdquo and recommended that ldquoPVAshould routinely include realistic inbreeding depressionhelliprdquo(Frankham et al 201457) Our custom‐built IBM permitted usto develop a model that adequately captured the Florida panther

van de Kerk et al bull Florida Panther Population and Genetic Viability 29

life history and we could parameterize the model with our long‐term genetic and demographic dataThe explicit consideration of the effects of inbreeding on

Florida panther population dynamics and persistence representsan important step forward Nonetheless our model remainsimperfect First we neglected the possible effects of habitat lossdetrimental anthropogenic activities climate change the prob-ability of immigration of pumas from western populationsdisease or catastrophes on demographic rates and populationviability Although immigration from puma populations outsideFlorida is possible its probability is very low given the extensivechallenges the anthropogenically altered landscape would posefor such a migrant to make it successfully to South Florida Weomitted mutations from our model because we have no in-formation on mutation rates though we expect that they are lowbased on values reported for other mammals (Kumar and Sub-ramanian 2002) Finally we only considered possible effects ofinbreeding on age‐specific survival rates because there was noevidence that heterozygosity affected female reproduction Lossof heterozygosity can negatively affect reproduction becauseinbred male panthers can suffer from cryptorchidism which canadversely affect their fecundity However given the polygynousmating system we expect the Florida panther population growthis primarily female‐limitedAlthough it is generally accepted that genetic introgression

only temporarily relieves a population from inbreeding depres-sion we know of no cases in which it has been implementedmore than once to conserve a threatened wildlife populationOur study provides an example of how demographic and mo-lecular data can be used within an IBM framework to evaluatethe efficacy of alternative genetic management strategies via theFlorida panther as a case study Our model can be adjusted forand applied to other species and offers great potential as a toolfor genetic management of small isolated wildlife populations

MANAGEMENT IMPLICATIONS

Our results and those of Hostetler et al (2013) and Johnsonet al (2010) provide persuasive evidence that the genetic in-tervention implemented in 1995 prevented the demise of theFlorida panther and restored demographic vigor to the popu-lation The Florida panther population remains small and iso-lated from other puma populations the closest puma populationis located gt1000 km away in Texas and the intervening land-scape is highly developed and fragmented with many interstate(and other high‐traffic) highways and major urban centersTheory suggests that 1 migrant per generation is sufficient tominimize the loss of genetic diversity (eg Mills and Allendorf1996) However the possibility of successful migration of apuma to South Florida followed by successful mating every ~5years is very low considering the distance between Texas andFlorida (Hedrick 1995) and the many risks and barriers thatdispersing pumas face (Beier 1995 Maehr et al 2002 Thatcheret al 2006) Consequently the pantherrsquos long‐term persistencewill likely depend on subsequent timely genetic introgressionsgiven the near‐impossibility of natural gene flow from otherpuma populations To that end our study offers considerableinsights into benefits of different genetic management strategiesand associated costs

Without genetic management the Florida panther populationfaces a substantial risk of quasi‐extinction within 100 years (10ndash46 depending on quasi‐extinction threshold and scenarios)Twenty‐three years have passed since genetic introgression wasimplemented and the population appears healthy as indicatedby measures of genetic variation increases in the populationsize and improvements in age‐specific survival rates Ourfindings suggest that releasing more pumas more frequently doesnot necessarily improve persistence probability because such astrategy would cause larger fluctuations in population size(Fig 14) which negatively affects persistence probability Thusextinction risks faced by inbred populations of large carnivorescan be substantially reduced by releasing approximately 5 im-migrants every 2 decades We suggest that such a managementstrategy be followed only in conjunction with continued long‐term monitoring of the population Our analyses demonstratethat population growth and persistence are highly sensitive tokitten and female (adult and subadult) survival Continuing tocollect data on these demographic variables along with bio-metric and genetic sampling will remain important to pantherrecovery and vigilance in identifying issues associated with in-breeding depression The collection of these monitoring datashould allow managers to detect any demographic or geneticchanges in a timely fashion and respond with genetic in-trogression or other management actions if warrantedRecovery objectives as defined by the USFWS in its current

recovery plan (USFWS 2008) delineate the need for additionalpopulations in the historical range to downlist or delist theFlorida panther If the only extant population of Floridapanthers continued to grow it could serve as a source ofpanthers to be translocated to suitable habitat to seed addi-tional populations Maintaining current information on thegenetic health of the population and precise estimates of itssize will be important in assessing whether it can withstand theremoval of a subset of animals for translocation Although the2017 documentation of a female panther with kittens north ofthe current breeding range is encouraging in terms of thepotential for population expansionmdashgiven that no female hadbeen recorded north of the Caloosahatchee River since 1972mdashthe re‐establishment of separate new populations will mostlikely require translocations by wildlife managers and a lengthysociopolitical processConservation of threatened or endangered species such as the

Florida panther is inherently challenging For panthers and otherlarge carnivores that require vast extents of natural habitat topersist habitat is typically the most limiting factor that will de-termine whether recovery efforts succeed Although geneticmanagement invariably plays a key role in the long‐term persis-tence of the panther population even a healthy population caneventually succumb to the impacts of habitat loss The humanpopulation of Florida continues to be one of the fastest‐growingin the nation the United States Census Bureau currently ranksFlorida as the third most populous Therefore habitat loss isgoing to continue to be a key issue for panthers Diligence towardpreserving panther habitat in its historical range via conservationeasements or other means and trying to keep such areas inter-connected via corridors is a goal stakeholders such as governmentagencies sportsmen non‐governmental organizations ranchers

30 Wildlife Monographs bull 203

and other private landowners must work on collaboratively toachieve

ACKNOWLEDGMENTS

We thank D Shindle D Jennings T OrsquoMeara D Land andC Belden for valuable advice throughout the project We thankS Bass M Criffield M Cunningham D Giardina D JansenA Johnson D Land M Lotz C McBride Rocky McBrideRowdy McBride Roy McBride L Oberhofer S Schulze staffsat Everglades National Park and Big Cypress National Preserveand others for assistance with fieldwork We also thank JAustin M Conroy J Hines J Laake S Mills J Nichols JHines M Sunquist J Fryxell and T Coulson for advice andassistance regarding data analysis and panther biology TheMuseum of Texas Tech University Natural Science ResearchLaboratory provided samples from pumas from west Texas forcomparative genetic analyses M Ben‐David D Land WMcCown E Hellgren and 4 anonymous reviewers providedmany helpful comments on the manuscript We thank NereaUbierna Lopez for completing the Spanish translation of theabstract We are grateful to the Department of ResearchComputing University of Florida for excellent high‐perfor-mance computing services We would like to thank US Fishand Wildlife Service Florida Fish and Wildlife ConservationCommission (FWC) US National Park Service QuantitativeSpatial Ecology Evolution and Environment (QSE3) In-tegrative Graduate Education and Research Traineeship(IGERT) program funded by the National Science Foundationand the University of Florida for financially supporting thisstudy Funding for panther research conducted by the FWC issupported predominantly via donations accrued with vehicleregistrations for ldquoProtect the Pantherrdquo license plates

LITERATURE CITEDAcevedo‐Whitehouse K T R Spraker E Lyons S R Melin F Gulland RL Delong and W Amos 2006 Contrasting effects of heterozygosity onsurvival and hookworm resistance in California sea lion pups MolecularEcology 151973ndash1982

Adams J R L M Vucetich P W Hedrick R O Peterson and J AVucetich 2011 Genomic sweep and potential genetic rescue during limitingenvironmental conditions in an isolated wolf population Proceedings of theRoyal Society B Biological Sciences 2783336ndash3344

Agresti A 2013 Categorical data analysis Third edition John Wiley amp SonsHoboken New Jersey USA

Agudo R M Carrete M Alcaide C Rico F Hiraldo and J A Donaacutezar2012 Genetic diversity at neutral and adaptive loci determines individualfitness in a long‐lived territorial bird Proceedings of the Royal Society BBiological Sciences 2793241ndash3249

Albeke S E N P Nibbelink and M Ben‐David 2015 Modeling behavior bycoastal river otter (Lontra canadensis) in response to prey availability in PrinceWilliam Sound Alaska a spatially‐explicit individual‐based approach PLOSOne 10e0126208

Alho J S K Vaumllimaumlki and J Merilauml 2010 Rhh an R extension for estimatingmultilocus heterozygosity and heterozygosity‐heterozygosity correlationMolecular Ecology Resources 10720ndash722

Alvarez K 1993 Twilight of the panther Myakka River Publishing SarasotaFlorida USA

Aparicio J M J Ortego and P J Cordero 2006 What should we weigh toestimate heterozygosity alleles or loci Molecular Ecology 154659ndash4665

Bakker V J D F Doak G W Roemer D K Garcelon T J Coonan S AMorrison C Lynch K Ralls and R Shaw 2009 Incorporating ecologicaldrivers and uncertainty into a demographic population viability analysis for theisland fox Ecological Monographs 7977ndash108

Balloux F W Amos and T Coulson 2004 Does heterozygosity estimateinbreeding in real populations Molecular Ecology 133021ndash3031

Barone M A M E Roelke J Howard J L Brown A E Anderson and DE Wildt 1994 Reproductive characteristics of male Florida panthersmdashcomparative studies from Florida Texas Colorado Latin‐America andNorth‐American Zoos Journal of Mammalogy 75150ndash162

Bateson Z W P O Dunn S D Hull A E Henschen J A Johnson and LA Whittingham 2014 Genetic restoration of a threatened population ofgreater prairie‐chickens Biological Conservation 17412ndash19

Bauer H G Chapron K Nowell P Henschel P Funston L T B HunterD W Macdonald and C Packer 2015 Lion (Panthera leo) populations aredeclining rapidly across Africa except in intensively managed areas Pro-ceedings of National Academy of Sciences of the United States of America11214894ndash14899

Beier P 1995 Dispersal of juvenile cougars in fragmented habitat Journal ofWildlife Management 59228ndash237

Benson J F J A Hostetler D P Onorato W E Johnson M E Roelke SJ OrsquoBrien D Jansen and M K Oli 2011 Intentional genetic introgressioninfluences survival of adults and subadults in a small inbred felid populationJournal of Animal Ecology 80958ndash967

Benson J F P J Mahoney J A Sikich L E K Serieys J P Pollinger H BErnest and S P D Riley 2016 Interactions between demography geneticsand landscape connectivity increase extinction probability for a small popu-lation of large carnivores in a major metropolitan area Proceedings of theRoyal Society B Biological Sciences 28320160957

Beverly F 1874 The Florida panther Forest and Stream 3290Boyce M S C V Haridas C T Lee and the NCEAS Stochastic Demo-graphy Working Group 2006 Demography in an increasingly variable worldTrends in Ecology amp Evolution 21141ndash148

Brook B W J J OrsquoGrady A P Chapman M A Burgman H R Akcakayaand R Frankham 2000 Predictive accuracy of population viability analysis inconservation biology Nature 404385ndash387

Brys R and H Jacquemyn 2016 Severe outbreeding and inbreeding de-pression maintain mating system differentiation in Epipactis (Orchidaceae)Journal of Evolutionary Biology 29352ndash359

Burke J M and M L Arnold 2001 Genetics and the fitness of hybridsAnnual Review of Genetics 3531ndash52

Burnham K P 1993 A theory for combined analysis of ring recovery andrecapture data Pages 199ndash213 in J‐D Lebreton and P M North editorsMarked individuals in the study of bird population Springer‐Verlag NewYork New York USA

Burnham K P and D R Anderson 2002 Model selection and multimodelinference a practical information‐theoretic approach Second edition SpringerVerlag New York New York USA

Caswell H 2001 Matrix population models construction analysis and in-terpretation Sinauer Associates Sunderland Massachusetts USA

Charpentier M J M Setchell F Prugnolle L A Knapp E J Wickings PPeignot and M Hossaert‐McKey 2005 Genetic diversity and reproductivesuccess in mandrills (Mandrillus sphinx) Proceedings of the NationalAcademy of Sciences of the United States of America 10216723ndash16728

Cooch E and G C White editors 2018 Program MARK a gentle in-troduction lthttpwwwphidotorgsoftwaremarkdocsbookgt Accessed 7Apr 2016

Cooley H S R B Wielgus G M Koehler H S Robinson and B TMaletzke 2009 Does hunting regulate cougar populations A test of thecompensatory mortality hypothesis Ecology 902913ndash2921

Coulson T E A Catchpole S D Albon B J Morgan J M Pemberton TH Clutton‐Brock M J Crawley and B T Grenfell 2001 Age sex densitywinter weather and population crashes in Soay sheep Science 2921528ndash1531

Coulson T J Pemberton S Albon M Beaumont T Marshall J Slate FGuinness and T Clutton‐Brock 1998a Microsatellites reveal heterosis in reddeer Proceedings of the Royal Society B Biological Sciences 265489ndash495

Coulson T N S D Albon J M Pemberton J Slate F E Guinness and TH Clutton‐Brock 1998b Genotype by environment interactions in wintersurvival in red deer Journal of Animal Ecology 67434ndash445

Cox D 1972 Regression models and life‐tables Journal of the Royal StatisticalSociety Series B Statistical Methodology 34187ndash220

Creel S 2006 Recovery of the Florida panthermdashgenetic rescue demographic rescueor both Response to Pimm et al (2006) Animal Conservation 9125ndash126

Crnokrak P and D A Roff 1999 Inbreeding depression in the wild Heredity83260ndash270

Crow J 1948 Alternative hypotheses of hybrid vigor Genetics 33477ndash487

van de Kerk et al bull Florida Panther Population and Genetic Viability 31

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 6: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

et al 2002) Taken together these factors suggested that thelong‐term prospects for the persistence of the Florida pantherwere poorSubsequently the United States Fish and Wildlife Service

(USFWS) conducted an assessment that led to the delineation of 4possible management options 1) initiate a captive breeding pro-gram 2) introduce genetic material from captive stockpurported to contain both pure Florida panther and SouthAmerican puma lineages 3) translocate wild female pumas (Pumaconcolor stanleyana) from Texas USA into South Florida or 4)take no action at all (USFWS 1994) Collaborative discussionsinvolving academics non‐governmental organizations and stateand federal wildlife agencies eventually led to the selection of op-tion 3 and a plan for genetic introgression that involved the releaseof 8 female pumas from Texas into South Florida was im-plemented in 1995 (Seal 1994 Onorato et al 2010)Continued research conducted after the introgression interven-

tion provided clear evidence of the benefits accrued to the pantherpopulation including increased levels of genetic diversity declinesin the frequency of morphological and biomedical correlates ofinbreeding depression and the substantial increase of the popu-lation size (McBride et al 2008 Johnson et al 2010 McClintocket al 2015) Furthermore Hostetler et al (2013) showed that thepost‐introgression population growth was driven primarily byhigher survival rates associated with admixed panthers (offspringof pairings between the female Texas pumas and male Floridapanthers and subsequent generations of those progeny) followinggenetic introgression The apparent success of genetic introgres-sion in preventing the imminent extinction of the Florida pantherpopulation is encouraging but the population remains small andisolated and continues to face substantial challenges especially thelarge‐scale loss of habitatAlmost 5 panther generations (generation time= 45 years

Hostetler et al 2013) have passed since the implementation of thegenetic introgression program The panther population has beenmonitored continuously during this time period leading to asubstantial amount of additional data and biological insights sincethe demographic assessment made immediately following the in-trogression (Hostetler et al 2010 2012 2013 Benson et al 2011)Given that the effects of genetic introgression are not permanent itis important to discern whether the positive impacts of this man-agement initiative are continuing or waning Decline in the benefitsof the introgression could reduce the population growth rate andprobability of persistence At the same time panther habitat israpidly lost to a variety of human activities (eg land development)and environmental changes associated with invasive species andclimate change (Land et al 2008 Dorcas et al 2012 Frakes et al2015) When long‐term monitoring data are available it is prudentto periodically evaluate demographic rates population growth andpersistence parameters so that changes are detected and timelymanagement action can be implementedTwo population modeling approaches have become popular

among ecologists and wildlife managers for studying the dynamicsand persistence of age‐ or stage‐structured populations matrixpopulation models and individual‐based models (IBMs Brooket al 2000 Morris and Doak 2002 Morrison et al 2016) Matrixpopulation models are the generalized discrete time‐version of theLotka‐Euler equation which describes the dynamics of age‐ or

stage‐structured populations of plants animals and humans (Lotka1924 Caswell 2001) They have become standard tools for mod-eling human and wildlife populations (Tuljapurkar 1990 Caswell2001 Keyfitz and Caswell 2005) because 1) they are powerful andflexible permitting adequate representation of life history of a widevariety of organisms including those with complex life histories 2)parameters for these models can be empirically estimated usingstandard analytical methods such as multi‐state capture‐recapturemethods (eg Williams et al 2002) 3) it is relatively straight‐forward to include the effects of factors such as environmental anddemographic stochasticity and density dependence 4) importantdemographic quantities such as asymptotic population growth ratestable age‐ or stage distribution and generation time can be easilycalculated from these models 5) they permit prospective and ret-rospective perturbation analyses which have been proven useful inwildlife conservation and management 6) they are computer‐friendly do not require extensive programming experience andtheir implementation using software packages such as R (R CoreTeam 2016) and MATLAB (MathWorks 2014) is straight‐for-ward and 7) they have sound theoretical foundation offeringanalytical solutions to many aspects of population modeling Weused matrix population models for 1) calculating the aforemen-tioned demographic quantities 2) performing sensitivity analysesinvolving asymptotic population growth rate and 3) comparativepurposes to check the results obtained from equivalent simulation‐based modelsDespite many advantages offered by matrix population models it

is difficult to incorporate attributes of individuals such as geneticsor behavior within that framework Agent‐based or IBMs (Grimm1999 Grimm and Railsback 2005 McLane et al 2011 Railsbackand Grimm 2011 Albeke et al 2015) offer an alternative modelingframework with tremendous flexibility Individual‐based modelsare computer simulation models that rely on a bottom‐up approachthat begins by explicitly considering the components of a system(ie individuals) population‐level properties emerge from the be-havior of and interactions among discrete individuals UsingIBMs it is possible to explicitly represent attributes of individualssuch as movement mating or genetics (McLane et al 2011Railsback and Grimm 2011 Albeke et al 2015) We used in-dividual‐based population models as the primary populationmodeling approach because they offered a framework for an ex-plicit consideration of genetics both at individual and populationlevelsOur overall goal was to provide a comprehensive demographic

assessment of the sole extant population of the Florida pantherusing long‐term (1981ndash2015) field data and novel modelingtechniques Our objectives were 1) to estimate demographicparameters for the Florida panther and investigate whetherpositive impacts of the 1995 genetic introgression remained inthe population and if so to what extent 2) to estimate popu-lation growth and persistence parameters using an IBM tailoredto the Florida pantherrsquos life history and 3) to evaluate thesensitivity of population growth and persistence parameters tovital demographic ratesThe Florida panther population remains small (le230 adults and

subadults FWC 2017) with no realistic possibility of natural geneflow so the loss of genetic variation and concomitant increase ininbreeding‐related problems are inevitable Ensuring the long‐term

8 Wildlife Monographs bull 203

persistence of the population will likely necessitate periodic geneticmanagement interventions Therefore our final objective was to 4)evaluate genetic and population‐dynamic consequences of variedgenetic introgression strategies and from these identify manage-ment strategies that are affordable and effective at improvingprospects of Florida panther persistence To achieve this objectivewe extended the individual‐based population model to track in-dividual genetics based on empirical allele frequencies taking ad-vantage of the long‐term demographic and genetic data that havebeen collected on Florida panthers since 1981 We used individualheterozygosity to inform vital rates based on empirically estimatedrelationships between individual heterozygosity and demographicparameters We then simulated population trajectories trackedindividual heterozygosity over time and estimated quasi‐extinctiontimes and probabilities without introgression and with introgres-sion for a range of intervals and number of immigrants perintrogression event We compared the genetic and demographicbenefits of implementing alternative genetic introgression scenariosand ranked them by the level of improvement in population per-sistence parameters and estimated costs because funding is a per-sistent challenge to conservation planningManagement of imperiled species is a complex process invol-

ving many stakeholders and plagued by layers of uncertainties(Runge 2011) It would therefore seem that management ofthreatened and endangered species would be a perfect case foradaptive resource management approach (ARM) because ARMfocuses on structured decision making for recurrent decisionsmade under uncertainty (Williams et al 2002 Runge 2011)Because it combines structured decision making with learningprudent application of ARM reduces uncertainty over time andcan lead to optimal management actions (Williams and Brown2012 2016) Key ingredients of a successful adaptive manage-ment program include stakeholder engagement clearly definedand mutually agreed‐upon objectives alternative managementactions and objective decision rules (Nichols et al 2007Williams and Brown 2012 2016) Unfortunately Florida pan-ther stakeholders have divergent views about the state of thesystem (eg panther abundance) and it is a challenge to havemutually agreed‐upon objectives management actions or deci-sion rules acceptable to all or most stakeholders AlthoughFlorida panther conservation efforts would be best served withinthe ARM framework in the long run impediments to its im-plementation (Runge 2011 Williams and Brown 2012) areunlikely to be overcome in the near future

STUDY AREA

The breeding population of Florida panthers mainly persists as asingle population South of the Caloosahatchee River (approxi-mately 267133degN latitude 815566degW longitude see Fig 1)One exception is a female that was documented just north of theCaloosahatchee River in 2016 (and subsequently documentedwith kittens in 2017) by FWC the first validation of a femalepanther north of the River since 1973 (FWC unpublisheddata) The breeding range in South Florida is topographicallyflat has a subtropical climate and is characterized by permanentand ephemeral wetlands influenced by rains from May throughOctober (Duever et al 1986) The study area contains a variety

of wildland land cover types including hardwood hammockscypress forests pine flatwoods freshwater marshes prairies andgrasslands (Davis 1943) and land characterized by human ac-tivity including citrus groves croplands pastureland rockmining and residential developments (Onorato et al 2011)The area is intersected by a multitude of roads that fragmentpanther habitat Most notable among these is Interstate 75(Alligator Alley) which was expanded from 2 to 4 lanes in 1990when fencing and wildlife crossings were constructed with theintention of minimizing harmful effects of road expansion onlocal wildlife (Foster and Humphrey 1995) A large portion ofthe area that comprises South Florida is under public ownershipmainly as federal or state lands Everglades National Park BigCypress National Preserve Florida Panther National WildlifeRefuge Picayune Strand State Forest and Fakahatchee StrandPreserve State Park alone encompass 9745 km2 though not allof the area is considered suitable for panthers (eg large ex-panses of sawgrass marshes in the western Everglades) Frakeset al (2015) delineated 5579 km2 of adult panther habitat inSouth Florida 1399 km2 of which are in private ownership Amajority of these private lands are located in the northern extentof the breeding range Although some private lands may beprotected (eg conservation easements) other areas are sus-ceptible to incompatible land uses such as rock mining or re-sidential developments The breeding range in South Florida isbounded to the east and west by large metropolitan areas in-clusive of Miami‐Fort Lauderdale‐Pompano Beach and CapeCoral‐Fort Myers‐Naples‐Marco Island‐Immokalee respec-tively that are populated by gt6000000 people (httpswwwcensusgov accessed 24 Jan 2019) These characteristics of thestudy area highlight the recovery challenges faced by panthers

METHODS

Field MethodsSince 1981 Florida panthers have been captured radiocollared andtracked by the FWC and National Park Service staff (Table 1)Capture and handling protocols followed guidelines of theAmerican Society of Mammalogists (Sikes 2016) Livestock Pro-tection Company (Alpine TX USA) provided trained hounds andhoundsmen for captures Teams either treed or bayed panthers onthe ground and then darted them with a 3‐ml compressed‐air dartfired from a CO2‐powered rifle Immobilization drugs have variedduring the tenure of the project but most recently teamsimmobilized panthers with a combination of ketamine HCl (10mgkg Doc Lanersquos Veterinary Pharmacy Lexington KY USA) andxylazine HCl (1mgkg Doc Lanersquos Veterinary Pharmacy) Fol-lowing immobilization teams lowered treed panthers to the groundby a rope or caught panthers with a net in some cases they used aportable cushion (McCown et al 1990) to further mitigate theimpact of a fall Capture teams administered propofol (PropoFlotradeAbbott Laboratories Abbott Park IL USA) intravenously either asa bolus or continuous drip to maintain anesthesia They adminis-tered midazolam HCl (003mgkg) intramuscularly or intravenouslyto supplement anesthesia in some panthers Panthers recovered in ashaded area away from water In some cases capture teams reversedxylazine HCl with yohimbine HCl (Yobinereg Lloyd IncShenandoah IA USA) at 25 its recommended dose

van de Kerk et al bull Florida Panther Population and Genetic Viability 9

Capture teams determined sex and age of panthers marked themwith an ear tattoo and a subcutaneous passive integrated trans-ponder (PIT‐tag) and fitted them with a very high frequency(VHF) or global positioning system (GPS) radio‐collar equippedwith a mortality sensor The GPS collars were programmed toattempt to collect a position at a variety of time intervals(range= 1ndash12 hr) depending on objectives of concurrent studiesobservers routinely tracked panthers with VHF radiocollars from afixed‐wing aircraft 3 times per week When a radio‐collar trans-mitted a mortality signal or when a location did not change forseveral days observers investigated the site to establish the pan-therrsquos fate When observers found a dead panther they assessed thecause of death based on an examination of the carcass and sur-rounding area if possible They then transported carcasses to anexperienced veterinarian for necropsy and a histopathological ex-amination to attempt to assign the cause of death Starting in 1995observers continually assessed successive locations of radio‐collaredfemales to determine if they initiated denning behavior Three to 4fixes at the same location (VHF collars) or successive returns to thesame location over a weeklong period (GPS collars) served as anindicator of a possible den Observers then located dens moreprecisely via triangulation with ground telemetry They visited denswhen the dam left the site typically to make a kill Teams capturedkittens by hand captured kittens ranged in age from 4 to 35 dayspost‐partum Teams counted sexed weighed dewormed andpermanently marked kittens with a subcutaneous PIT‐tag

Genetic AnalysisTeams collected blood tissue or hair from the majority ofcaptured kittens subadult and adult Florida panthers for DNAsamples The United States Department of Agriculture ForestService National Genomics Center for Wildlife and FishConservation (Missoula MT USA) processed samplesTechnicians used Qiagen DNeasy Blood and Tissue Kits(Qiagen Valencia CA USA) to extract whole genomic DNAfrom blood and tissue samples and used a slight modification(overnight incubation of sample in lysis buffer and proteinase Kon a rocker at 60deg elution of DNA in 50 μl of buffer) of theQiagen DNA extraction protocol (Mills et al 2000) to extractDNA from hair samples Technicians used a panel of 16 mi-crosatellite loci (Fca090 Fca133 Fca243 F124 F37 Fca075Fca559 Fca057 Fca081 Fca566 F42 Fca043 Fca161

Fca293 Fca369 and Fca668) identified by Menotti‐Raymondet al (1999) which had been previously applied to Floridapanther samples (Johnson et al 2010) Technicians amplifiedDNA samples in 10‐microl reaction volumes that included 10 microl ofDNA 1times reaction buffer (Applied Biosystems Life Technolo-gies Corporation Grand Island NY USA) 20 mM of MgCl2200 microM of each dNTP 1 microM of reverse primer 1 microM of dye‐labeled forward primer 15 mgml of bovine serum albumin(BSA) and 1U Taq polymerase (Applied Biosystems) Thepolymerase chain reactions (PCRs) included a thermal profile of94degC for 5 minutes followed by 36 cycles of 94degC for 60 seconds55degC for 60 seconds and 72degC for 30 seconds Techniciansvisualized the resulting PCR products on a LI‐COR DNAanalyzer (LI‐COR Biotechnology Lincoln NE USA) Tech-nicians collected genotypes from hair samples (n= 16) using amultitube approach (Taberlet et al 1996) error‐checked geno-types using Program DROPOUT (McKelvey and Schwartz2005) and combined them with genotypes from tissue andblood samples (n= 487) We evaluated all 16 loci for variousdescriptive statistics for the radio‐collared sample of adult andsubadult panthers We excluded genotypes of kittens from thatanalysis because related individuals are known to result in theoverrepresentation of certain alleles (Marshall et al 1998) Wedetermined the number of alleles observed heterozygosity andexpected heterozygosity using GenAlex (Peakall and Smouse2006 2012) We assessed allelic richness at each locus in Fstat293 (Goudet 1995) We assessed conformance of genotypedata from these 16 loci to HardyndashWeinberg assumptionslinkage disequilibrium and null alleles (Appendix A availableonline in Supporting Information) The Genomics Center alsoprovided us with genotype data from the same 16 loci for 41pumas from western Texas the source population for thegenetic rescue programEvidence of inbreeding depression in wild populations is

commonly determined on the basis of heterozygosityndashfitnesscorrelations (Silva et al 2005 Ortego et al 2007 Mainguyet al 2009 Johnson et al 2011) so we also estimated theheterozygosity of each individual panther We calculatedhomozygosity by loci (HL) which varies between 0 (all lociheterozygous) and 1 (all loci homozygous Aparicio et al 2006)using the Rhh package (Alho et al 2010) in Program R (R CoreTeam 2016) A microsatellite locus has more weight in HL

Table 1 Number of Florida panthers by age sex and genetic ancestry categories used in subsequent demographic analyses during the study period (1981ndash2013) inSouth Florida USA

Kittensa Subadults and adultsb

Males Females Total Males Females Total

PIT‐taggedc 215 180 395 Radiocollaredc 108 101 209Recovered 11 4 15 Subadult 72 50 122Litter failed 47 38 85 Prime adult 65 95 160Recaptured 25 30 55 Old adult 13 20 33Canonical 27 20 47 Canonical 43 29 72F1 admixed 6 12 18 F1 admixed 2 6 8Other admixed 130 105 235 Other admixed 45 53 98

a Kitten samples included those that were PIT‐tagged recovered (ie panthers that were found dead) part of a failed litter or recaptured as subadults or adultsb Radiocollared subadult and adult panthers are presented as sample size within each age class and ancestry classc Combined number of panthers of different ancestries is less than the number radiocollared or tagged with a subcutaneous passive integrated transponder (PIT‐tagged) because we did not conduct genetic sampling on all panthers Combined number of panthers in different age classes is greater than the number ofradiocollared panthers because some individuals were counted in ge2 age classes as animals aged

10 Wildlife Monographs bull 203

when it is more informative (ie has more alleles) and has aneven distribution of allele frequencies For our demographicanalyses we calculated individual heterozygosity for each in-dividual panther as 1 minusHL As do most indices of individualheterozygosity HL takes into account the average allele fre-quency in the population (Aparicio et al 2006) Thus duringsimulations the allele frequency in the population changesthroughout an individualrsquos lifetime which causes its hetero-zygosity also to change This is not biologically plausible becausegenetic properties are retained throughout an individualrsquos life-time Therefore for use in model simulations we simply cal-culated heterozygosity for each individual as the proportion ofheterozygous lociWe also used genotype data to implement a Bayesian clus-

tering analysis in Program STRUCTURE (Pritchard et al2000) to infer ancestral clusters of panthers This method uses aMarkov chain Monte Carlo (MCMC) method to determine thenumber of genetic clusters (K) in the sample while also pro-viding an individualrsquos percentage of ancestry (q values) allocatedto each cluster Runs for this analysis included a 100000MCMC burn‐in period followed by 500000 MCMC iterationsfor 1ndash10 ancestral genetic clusters (K= 1 to 10) with 10 re-petitions for each K To determine the most likely number of Kgenetic clusters we used the logarithm of the probability of thedata (log Pr(D)∣K Pritchard et al 2000) and estimates of ΔK(Evanno et al 2005) in Program STRUCTURE HAR-VESTER (Earl and VonHoldt 2011) We then used q valuesprovided in runs for the best K to assign individual panthers aseither canonical (in most cases ge90 pre‐introgression ancestrysee Johnson et al 2010) or admixed (lt90 pre‐introgressionancestry) We recognized admixed panthers that were the off-spring of matings between a Texas female puma and a canonicalmale panther as F1 admixed panthers for analyses

Demographic ParametersSurvival of kittensmdashMost kittens PIT‐tagged at den sites were

never encountered again A small proportion of the PIT‐taggedkittens were recovered dead or were recaptured as subadults oradults and radiocollared so that their fate could be monitoredWe also identified instances of complete litter failure when afemale died within 9 months after giving birth (the minimumage of independence for kittens) or when a female wasdocumented to have another litter less than 12 months aftergiving birth (the minimum age of independence plus a 3‐monthgestation period Hostetler et al 2010) Thus our data setconsisted of 1) live recaptures and subsequent radio‐tracking ofpanthers of various ages that had been PIT‐tagged as kittens 2)recovery of dead panthers of various ages that had been PIT‐tagged as kitten 3) live captures and subsequent radio‐trackingof other panthers and 4) instances of complete litter failure(Table 1) We analyzed these data using Burnhamrsquos live‐recapture dead‐recovery modeling framework (Burnham 1993Williams et al 2002) to estimate survival of kittens (0ndash1‐yearold) and to test covariate effects on this parameter We includeddata on individuals encountered dead or alive at an older age inour analyses because their encounter histories also providedinformation on kitten survival We used a live‐dead data inputformat coded with a 1‐year time step (Cooch and White 2018)

All known litter failures occurred within the first year of thekittensrsquo lives so we treated all litter failures as recoveries withinthe first year We treated panthers that would have died if wehad not removed them to captivity as having died on the date ofremoval Data preparation and analytical approach are describedin more detail by Hostetler et al (2010)In addition to survival probability (S) the Burnham model

incorporates parameters for recapture probability (p) recoveryprobability (r) and site fidelity (f) We set r and p to 10 forradio‐tracked individuals because their status was known eachyear We also fixed f at 10 for all individuals because the re-capture and recovery areas were the same and encompassed theentire range of the Florida panther Based on findings of Hos-tetler et al (2010) we used a base model that 1) allowed survivalto differ between kittens and older panthers 2) allowed survivalto differ between sexes and among age classes (females ages 1and 2 ge3 males ages 1 2 and 3 ge4) 3) constrained recaptureprobability to be the same for all uncollared panthers and 4)allowed recovery rates to differ between kittens and uncollaredolder panthers Even with our additional data this model re-mained the best fit for a range of models for recapture recoveryand survival of kittens subadults and adults of various age‐classcategories (Hostetler et al 2010)Subadult and adult survivalmdashTo estimate survival for panthers

ge1 year of age we analyzed radio‐tracking data using a Coxproportional hazard‐modeling framework (Cox 1972 Therneauand Grambsch 2000) We followed procedures outlined byBenson et al (2011) for data preparation and analysis Brieflywe right‐censored panthers on the last day of observation beforetheir collar failed When an individual aged into a new age classwhile being tracked we created 2 data entries 1 ending the dayit entered the new age class the other starting on that day Weused the FlemingndashHarrington method to generate survivalestimates from the Cox analysis and the Huber sandwichmethod to estimate robust standard errors (Therneau andGrambsch 2000 Benson et al 2011)We considered 3 age classes subadults (1ndash25 yr for females and

1ndash35 yr for males) prime adults (25ndash10 yr for females and35ndash10 yr for males) and old adults (gt10 yr old for both sexes)after Benson et al (2011) We estimated overall survival using abase model that allowed survival to differ between subadultfemales subadult males prime adult females prime adult malesand old adults of either sex (old age was additive with sex other-wise age and sex were interactive) This model provided the best fitto the data relative to a range of models considering differences insurvival between prime‐adult and older‐adult age classes both ad-ditively and interactively with sex (Benson et al 2011)To examine whether survival rates observed in recent years

differed from those observed immediately following introgres-sion we also estimated age‐specific survival rates for 2 nearlyequal periods of time immediately post‐introgression (1995ndash2004) and the more recent period (2005ndash2013) For theseanalyses we separated the radio‐tracking data into the 2 periodsand estimated survival rates for each period separately using thebase model Similar analyses for kitten survival were not possiblebecause of data limitationsProbability of reproductionmdashWe used telemetry data obtained

from radio‐tracked females that reproduced at least once to

van de Kerk et al bull Florida Panther Population and Genetic Viability 11

estimate annual probability of reproduction of subadult prime‐adult and old‐adult females using complementary logndashlogregression (Agresti 2013) following methods described indetail by Hostetler et al (2012) Briefly we modeled theprobability of a female panther giving birth (b) as a function ofcovariates as

γ= minus [minus ( )]b z1 exp exp i i

where zi is a row vector of covariates (see below for covariateinformation) for female panther i and γ is a column vector ofmodel coefficients (Appendix B Table B2) To account fordifferences in observation time we included log(m) as an offsetin the model where m is the number of months a femalepanther was radio‐tracked (Hostetler et al 2012)

Reproductive outputmdashTo estimate the number of kittensproduced by reproductive females we used cumulative logitregression (Min and Agresti 2005 Agresti 2013) Weconsidered a model that includes J categories for the numberof offspring with j denoting the number of offspring ( j= 1 2hellip J and J= 6) provided that a female reproduces Weconsidered J= 6 which is greater than the largest litter sizeobserved in our study (4 kittens) because females can produce 2or more litters per year if litters fail We modeled the probabilitythat a female i produces a litter of size at most j (Yi) as

le⎧

⎨⎩

δ( ) = == hellip minus

=

( )+ θ βminus +Y jj J

j JPr

1 2 1

1i ij e

1

1 xj i

where θj is the intercept for the annual number of kittens pro-duced by a female i being at most j xi is a row vector of cov-ariates for female panther i (see below) and β is a column vectorof model coefficients The probability that Yi will be exactly j isgiven by

⎧⎨⎩

πδ

δ δ( = ) = =

=

minus = hellip( minus )Y j

jj J

Pr 1

2 i ij

i

ij i j

1

1

Finally the expected annual number of kittens produced byfemale i (νi) is given by

sumν π==

j i

j

J

ij

1

Additional details regarding the estimation and modelingof the annual number of kittens produced can be found inHostetler et al (2012)

CovariatesmdashIn addition to sex and age we tested for the effect ofabundance genetic ancestry and individual heterozygosity on theaforementioned demographic parameters (model coefficients forkitten and adult survival are given by η and ι respectively AppendixB Table B2) We used a minimum population count (MPC) basedon radio‐tracking data and field evidence of subadult and adult (ieindependent age) panthers as an index of abundance (McBride et al2008) For these analyses we did not use the population sizeestimates reported by McClintock et al (2015) because unlike theminimum counts they did not cover the entire study period

(McBride et al 2008 R T McBride and C McBride RancherrsquosSupply Incorporated unpublished report Fig 2)To test if demographic parameters differed depending on an-

cestry we considered 3 ancestry models dividing the panthers into1) F1 admixed and all other panthers 2) canonical and admixed(including F1 admixed) panthers and 3) F1 admixed canonicaland other admixed panthers We quantified genetic diversity usingindividual heterozygosity as described previously We tested for theeffect of genetic covariates using a subset of kittens that were ge-netically sampled We calculated model‐averaged estimates ofmodel parameters on the scale in which they were estimated(Appendix B available online in Supporting Information)

Cause‐specific mortalitymdashWe performed cause‐specificmortality analysis only on dead radio‐collared panthers becausedetection rates of uncollared panther mortalities are highlycorrelated with the cause of death (eg vehicle collision)Following Benson et al (2011) we attributed each mortality to1 of 4 causes 1) vehicle collision 2) intraspecific aggression 3)other causes (including known causes such as diseases injuriesand infections unrelated to the first 2 causes) and 4) unknown(ie mortalities for which we could not assign a cause) Wethen used the non‐parametric cumulative incidence functionestimator a generalization of the staggered‐entry KaplanndashMeiermethod of survival estimation to estimate cause‐specificmortality rates for male and female panthers in different ageclasses (Pollock et al 1989 Heisey and Patterson 2006 Bensonet al 2011) We compared cause‐specific mortality rates betweensexes age classes and ancestries

Statistical inferencemdashWe used an information‐theoreticapproach (Akaikersquos information criterion [AIC]) for modelselection and statistical inference (Burnham and Anderson 2002)For the analysis of kitten survival we adjusted the AIC foroverdispersion and small sample size (QAICc details in Hostetleret al 2010) We performed all statistical analyses in Program R (RCore Team 2016) We used Burnhamrsquos live recapturendashdead

0

100

200

300

1985 1990 1995 2000 2005 2010Year

Indi

vidu

als Method

MPC

MVM

Figure 2 The Florida panther minimum population count (MPC) and populationsize estimated using motor vehicle collision mortalities (MVM) during our studyperiod Minimum population counts are from McBride et al (2008) and R TMcBride and C McBride (Rancherrsquos Supply Incorporated unpublished report)Estimates based on MVM are from McClintock et al (2015) The solid vertical lineindicates the year of genetic introgression

12 Wildlife Monographs bull 203

recovery model (Burnham 1993) to estimate and model kittensurvival using the RMark package version 2113 (Laake 2013) asan interface for Program MARK (White and Burnham 1999)We implemented the Cox proportional hazard model for esti-

mation and modeling of survival of subadults and adults using the Rpackage SURVIVAL (Therneau and Grambsch 2000) We per-formed cause‐specific mortality analyses using an R version of S‐PLUS code provided by Heisey and Patterson (2006)To account for model selection uncertainty we calculated

model‐averaged parameter estimates across all models using theR package AICcmodavg (Mazerolle 2017) using AIC weights asmodel weights (Burnham and Anderson 2002) We used modelswithout the categorical ancestry variables for model averagingand estimated parameters based on the average value for thecontinuous covariates (individual heterozygosity and abundanceindex) Because only a subset of the kittens was geneticallysampled we estimated 2 kitten survival rates First we estimatedkitten survival based on the data set that included all kittens(overall kitten survival) Second we estimated kitten survivalusing a subset of the data that only included kittens that weregenetically sampled (survival of genetically sampled kittens)

Matrix Population ModelsWe investigated Florida panther population dynamics andpersistence using 2 complementary modeling frameworks ma-trix population models (Caswell 2001) and IBMs (Grimm andRailsback 2005 McLane et al 2011 Railsback and Grimm2011 Albeke et al 2015) Matrix population models offer aflexible and powerful framework for investigating the dynamicsand persistence of age‐ or stage‐structured populations (egCaswell 2001 Robinson et al 2008 Kohira et al 2009 Hunteret al 2010 Hostetler et al 2012) With a fully developed theorybehind them these models also serve as a check for resultsobtained from simulation‐based models such as IBMs We usedan age‐structured‐matrix population model for deterministic andstochastic demographic analyses and for preliminary investiga-tions of Florida panther population viability (Caswell 2001Morris and Doak 2002)For deterministic and stochastic demographic analyses we

parameterized a female‐only 19 times 19 age‐specific populationprojection matrix of the form

⎢⎢⎢⎢⎢

⋯ ⋯

⋯ ⋯

⋱ ⋱ ⋮

⋮ ⋱ ⋱ ⋱ ⋮

⎥⎥⎥⎥⎥

( )=t

F F F

P

P

P

A0 0

0

0 0 0

t t t

t

t

t

1 2 19

1

2

18

where Fi and Pi are the age‐specific fertility and survival ratesrespectively and ( )tA is a time‐varying (or stochastic) popula-tion projection matrix We assumed that the longevity and ageof last reproduction of female Florida panthers was 185 yearswe based this assumption on the observation that the oldestfemale in our study was 186 years old Florida panthers re-produce year‐round thus we used birth‐flow methods to esti-mate Fi and Pi values following Caswell (2001) and Hostetleret al (2013) We estimated Pi as

= +P S S i i i 1

where Si is the age‐specific survival probability We estimated Fi as

⎛⎝

⎞⎠

=+ +F S

m Pm

2i k

i i i 1

where Sk is the kitten survival probability and mi is the age‐specific fecundity rate which was estimated as

ν=m b f i i i k

where bi is the breeding probability for a female in age class iduring time step t νi is the annual number of kittens producedby a breeding female in age class i and fk is the proportionfemale kittens at birth (assumed to be 05)We estimated the finite deterministic (λ) and stochastic annual

population growth rate (λs) and stochastic sensitivities and elasti-cities following methods described in detail by Caswell (2001) Toestimate λs we assumed identically and independently distributedenvironments (Caswell 2001 Morris and Doak 2002 Haridas andTuljapurkar 2005) and used a parametric bootstrapping approachto obtain a sequence of demographic parameters for 50000 ma-trices We then estimated log(λs) from this sequence of matrices(Tuljapurkar 1990 Caswell 2001) and exponentiated it to obtain λsWe calculated the sensitivity and elasticity of λs to changes in lower‐level vital demographic rates as per Caswell (2001) and Haridas andTuljapurkar (2005)For population projection and population viability analysis (PVA)

simulations we expanded the population projection matrix into a34 times 34 2‐sex age‐structured matrix to include female and maleFlorida panthers (Caswell 2001 Hostetler et al 2013) Malescontribute to the population only through survival in this modelingframework We assumed maximum longevity for males to be 145years of age because the oldest male in our study was 144 years oldThe population projection matrix was of the following form

⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢

⋯ ⋯ ⋯ ⋯

⋯ ⋯ ⋯ ⋯

⋱ ⋱ ⋮ ⋮ ⋱ ⋱ ⋮

⋮ ⋱ ⋱ ⋱ ⋮ ⋮ ⋱ ⋱ ⋮

⋯ ⋯ ⋯

⋯ ⋯ ⋯ ⋯

⋯ ⋯ ⋱ ⋱ ⋮

⋮ ⋮ ⋱ ⋱ ⋮ ⋱ ⋱ ⋮

⋯ ⋯ ⋯

⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥

( )=

( ) ( ) ( )

( )

( )

( )

( ) ( ) ( )

( )

( )

t

F N F N F N

P N

P N

P N

M N M N M N

Q N

Q N

A n

0 0

0 0 0 0

0

0 0 0 0 0

0 0

0 0 0

00 0 0 0 0

t t t

t

t

t

t t t

t

t

1 2 19

1

2

18

1 2 19

1

14

where Pi and Qi are age‐specific survival rates of female and maleFlorida panthers respectively and Fi and Mi are the rates atwhich female and male kittens are produced by females of ageclass i respectively Here the population projection matrix isboth time‐varying and density‐dependent (Caswell 2001)We incorporated the influence of parameter uncertainty en-

vironmental stochasticity and density dependence on Floridapanther population dynamics and persistence following the ap-proach outlined by Hostetler et al (2013) we used the sameapproach in the IBM analysis which is described below UnlikeIBMs matrix models do not intrinsically include demographicstochasticity so we explicitly included the influence of

van de Kerk et al bull Florida Panther Population and Genetic Viability 13

demographic stochasticity by simulating the fates of individualsas described in Caswell (2001)Population projections using matrix models necessitate a

vector of initial sex‐ and age‐specific abundance Because suchdata are currently unavailable for the Florida panther we esti-mated it based on the stable sex and age distribution and MPCin 2013 Using point estimates of the demographic parameterswe estimated the stable sex and age distribution for the 2‐sexdeterministic and density‐independent population projectionmatrix We then multiplied the stable sex and age distributionby the MPC in 2013 to get the starting population vector n(0)We projected the population over time as n(t+ 1)=A(tn)n(t)where A(tn) indicates the time‐specific density‐dependentpopulation projection matrix (Caswell 2001)We had 2 indices of abundance available the MPC (McBride

et al 2008) and population size estimated using motor vehiclecollision mortalities (MVM McClintock et al 2015) These 2indices are substantially different because MPC does not ac-count for imperfect detection whereas MVM accounts forimperfect detection and provides an estimate of population size(Fig 2) Therefore we examined 2 scenarios one using densitydependence based on MPC and the other using density de-pendence based on MVM We ran 1000 bootstraps of para-meter values and 1000 simulations per bootstrap for each sce-nario We projected population trajectories over 200 years andthen estimated population growth rates probabilities ofquasi‐extinction and time until quasi‐extinction and comparedthese with the results obtained from the IBM (see below) Weran scenarios with a quasi‐extinction threshold of 10 and 30panthers We used the mean of probabilities of quasi‐extinction(PQE) across simulations as the point estimate for PQE and the5th and 95th percentiles across simulations as the 90 con-fidence interval Because the confidence intervals werequantile‐based but the point estimates were not it is possiblewith skewed distributions for the point estimates to be outsidethe confidence interval We implemented the matrix

population models in the R computing environment (R CoreTeam 2016)

IBMs and PVABecause of well‐developed theory ease of implementation andthe modeling flexibility that they offer matrix populationmodels have become the model of choice for investigating thedynamics and persistence of age‐ and stage‐structured popula-tions (Caswell 2001) However it is difficult to incorporateattributes of individuals such as genetics and behavior usingmatrix models Quantifying the effects of genetic erosion on theFlorida panther population dynamics and persistence andevaluating benefits and costs of alternative genetic managementscenarios were important goals of our study Because IBMs relyon a bottom‐up approach that begins by explicitly consideringindividuals (McLane et al 2011 Railsback and Grimm 2011Albeke et al 2015) it is possible to represent genetic attributesof each individual and to track genetic variation over time Weused IBMs as the primary modeling framework in this studybecause they allowed for explicit consideration of genetics bothat individual and population levels and population viabilityassessment under alternative genetic management scenariosWe developed an IBM (Grimm 1999 Grimm and Railsback

2005 McLane et al 2011 Railsback and Grimm 2011 Albekeet al 2015) tailored to the life history of the Florida panther(Fig 3) We simulated every individual from birth until deathbased on a set of rules describing fates (eg birth and death) andattributes (eg genetics) of individuals at each annual time stepAs in other IBMs population‐level processes (eg populationsize and genetic variation) emerged from fates of and interac-tions among individuals (Grimm 1999 Railsback and Grimm2011) We developed an overview design concepts and details(ODD) protocol (Grimm et al 2006 2010) to describe ourIBM (Appendix C available online in Supporting Information)and provide pseudocodes for individual‐based simulations(Appendix D available online in Supporting Information)

For each panther at time = t

Kitten (lt1 yr)

Reproduce

Survive

Kitten(s) born

Kitten

Sex and age class

Subadult male

Prime adult male

Old adult male

Subadult female

Prime adult female

Old adult female

Age one year

No Male

Female

Yes

No

Yes

t=t+1

Yes

lt35 yrs

35-10 yrs

gt10 yrs

lt25 yrs

25-10 yrs

gt10 yrs

Figure 3 Schematic representation of Florida panther life history We tailored the individual‐based population model to represent the life‐history patterndepicted here

14 Wildlife Monographs bull 203

Dynamics and persistence of populations are influenced bymany factors such as demographic and environmental sto-chasticity and density dependence these effects and parametricuncertainty must be considered in population models whenpossible (Caswell 2001 Boyce et al 2006 Bakker et al 2009Hostetler et al 2013) Because by definition IBMs simulate thelife history of individuals in a population they inherently in-corporate the effect of demographic stochasticity Our analysesrevealed that survival of kittens subadults and adults and theprobability of reproduction varied over time so we incorporatedenvironmental stochasticity in the model for these variablesfollowing Hostetler et al (2013) Likewise we found evidenceof density‐dependent effects on kitten survival Because theMPC might be an underestimate of population size (McBrideet al 2008) we performed simulations using both the MPC andMVM abundance indicesTo account for model selection and parametric uncertainty we

used a Monte Carlo simulation approach For each bootstrapsample we selected a model based on the AIC (or QAICc)weights We then sampled the intercept and slope for kittensurvival (on the logit scale) subadult and adult survival (on thelogit scale with log‐hazard effect sizes) and probability of re-production (on the complementary logndashlog scale Appendix B)from multivariate normal distributions with mean vectors equalto the estimated parameters and variance‐covariance matricesequal to the sampling variance‐covariance matrices We thentransformed the results to nominal scale and converted the es-timates to age‐specific survival and reproduction parameters asdescribed in Hostetler et al (2013)We ran 1000 simulations for 1000 parametric bootstraps for

200 years for the MPC and the MVM scenario for both thestochastic 2‐sex matrix‐population model and the IBM Wetracked the population size at each time step and estimatedquasi‐extinction probabilities and times based on the populationtrajectory We considered the population to be quasi‐extinct ifindividuals of only 1 sex remained or if the population size fellbelow critical thresholds set at 10 and 30 individuals We alsoestimated expected time to quasi‐extinction for both thresholdsdefined as the mean and median time until a population be-comes quasi‐extinct conditioned on quasi‐extinction We im-plemented the IBM using the RNetLogo package version 10ndash2(Thiele et al 2012 Thiele 2014) as an interface for the IBMplatform NetLogo (Wilensky 1999)

Sensitivity AnalysisAn important component of demographic analysis is sensitivityanalysis which is the quantification of the sensitivity of amodelrsquos outcome to changes in the input parameters (Caswell2001 Bakker et al 2009 Thiele et al 2014) Using an IBM weperformed global sensitivity analyses for 2 (ie MPC andMVM) density‐dependent scenarios to assess how sensitivequasi‐extinction times and probabilities were to changes (anduncertainties) in the estimated demographic rates We usedLatin hypercube sampling to sample parameters from the entireparameter space (Marino et al 2008 Thiele et al 2014) Wesampled intercept and slope for kitten survival (on logit scale)subadult and adult survival (on logit scale with log‐hazard effectsizes) and probability of reproduction (on complementary

logndashlog scale) from normal distributions We sampled theprobability distribution for the number of kittens producedannually (on the real scale) from a Dirichlet distribution themultivariate generalization of the beta distribution (Kotz et al2000) We sampled 1000 different parameter sets and ran 1000simulations for each scenario We calculated the probability ofquasi‐extinction within 200 years for each density‐dependencescenario (MPC or MVM) using a quasi‐extinction threshold of10 individuals and then calculated the partial rank correlationcoefficient between each of those and each parameter trans-formed to the real scale (Bakker et al 2009 Hostetler et al2013) Partial rank correlation coefficients quantify the sensi-tivity of the probability of quasi‐extinction to input variables(ie survival and reproductive parameters) with higher absolutevalues indicating stronger influence of that variable on quasi‐extinction probabilities

Genetic ManagementOne of our goals was to estimate the benefits (improvements ingenetic variation demographic parameters and populationgrowth and persistence parameters) and costs (financial costs ofcapture transportation quarantine release and monitoring ofintroduced panthers) of genetic management To achieve thisobjective we extended the IBM to include a genetic componentTo simulate individual‐ and population‐level heterozygosity overtime we assigned each panther alleles for 16 microsatellite locibased on the empirical allele frequency in the population (esti-mated from genetic samples collected during 2008ndash2015) whenwe initialized the model (Pierson et al 2015) At each time stepwe simulated Mendelian inheritance for kittens produced suchthat each kitten randomly inherited alleles from its mother andfrom a male panther randomly chosen to have putatively siredthe litter We did not explicitly force inbreeding to occur butthe probability of inbreeding naturally increased as the popula-tion size decreasedIntrogression strategiesmdashWe modeled genetic introgression as a

result of the release of 2‐year‐old female pumas from Texas intothe simulated Florida panther population For the geneticintrogression implemented in 1995 Texas females between 15and 25 years of age were released because females at that agehave lower mortality rates (Benson et al 2011) higherreproductive potential and also because it is much easier todetect with certainty the successful reproduction by females thanby males The ability to more closely monitor reproduction andthe birth of kittens is an important consideration in reducing theprobability of a genomic sweep of Texas alleles into the Floridapanther population Wildlife managers removed the last 2surviving Texas females from the wild population in Florida in2003 to reduce the possibility of genomic sweep To simulatethis strategy in the model we removed any Texas females thatwere still alive 5 years after each introgression event Weassigned Texas females alleles based on the allele frequency ofthe Texas population (based on genotypes from 48 samplescollected in west Texas that was inclusive of the pumas releasedin South Florida in 1995) when they entered the Florida pantherpopulation We could not estimate demographic rates separatelyfor the released Texas females because of small sample size (only8 females were released in 1995) therefore we assumed that

van de Kerk et al bull Florida Panther Population and Genetic Viability 15

survival and reproductive rates and the effects of covariates onthese rates for the released Texas females were identical to thoseof female Florida panthersThe impact of genetic introgression depends on the frequency

of introgression events and the number of individuals introducedat each attempt Thus we defined introgression strategies basedon the number of Texas females to be released (0 5 10 or 15)and the interval between genetic introgressions (10 20 40 and80 yr) for a total of 13 strategies We simulated each manage-ment strategy with density dependence estimated using theMPC and MVM abundance indices

We sampled 1000 parameter sets and for each of these setswe ran all introgression strategies 1000 times for 200 years Weapplied the same parametric uncertainty and environmentalstochasticity across all management scenarios to allow for bettercomparison of introgression strategies At each time step werecorded the projected population size and average individualheterozygosity We calculated the heterozygosity for each in-dividual at birth based on the alleles it inherited This individualheterozygosity then informed the survival rate of that individualbased on the empirically estimated relationship between in-dividual heterozygosity and age‐specific survival rates Hetero-zygosity did not change throughout an individualrsquos lifetime butits effect on survival changed as individuals aged because theeffect of individual heterozygosity on survival rate differedamong age classes Survival rates of genetically sampled kittenswere biased high because some kittens were sampled later in lifewhen they had already survived for some time To correct forthis bias we adjusted kitten survival rates predicted by theheterozygosity model proportionally From the population tra-jectories we estimated the probability of quasi‐extinction (de-fined as the probability that the population will fall below 10 or30 individuals or that individuals of only 1 sex will remain)

Cost‐benefit analysismdashExperts estimated financial costs ofintrogression based on their experience with the geneticintrogression executed in 1995 and we adjusted estimates tocurrent costs to account for inflation (R T McBride RancherrsquosSupply Incorporated M Cunningham and D P OnoratoFlorida Fish and Wildlife Conservation Commission personalcommunication) Costs included capturing and transportingfemale pumas from the Texas source population caring for thepumas while they were in quarantine and post‐introgressionmonitoring To compare the financial costs of the differentscenarios we also calculated the average costs incurred over

Table 2 Sample size (n) number of alleles (Na) allelic richness (AR) observedheterozygosity (Ho) and expected heterozygosity (He) at 16 microsatellite loci inradio‐collared Florida panthers sampled from 1981 to 2013 in South FloridaUSA We calculated these values from a subset (n= 137) of the samples used inthe analyses and they include only adult and subadult panthers We excludedkittens because they can result in an overrepresentation of certain alleles

Locus n Na AR Ho He

FCA090 129 5 50 0333 0401FCA133 136 3 30 0618 0608FCA243 136 5 50 0419 0487F124 137 7 69 0708 0729F37 129 4 40 0395 0397FCA075 131 5 50 0534 0598FCA559 133 5 50 0496 0503FCA057 134 6 59 0679 0624FCA081 134 7 69 0209 0206FCA566 130 6 60 0462 0475F42 133 10 100 0737 0766FCA043 136 5 49 0596 0588FCA161 132 6 60 0500 0515FCA293 132 4 40 0614 0624FCA369 131 6 60 0573 0567FCA668 133 7 70 0406 0462Mean 1329 57 57 0517 0535

Figure 4 Proportional membership (q) of radio‐collared Florida panthers (n= 218) and pumas from Texas USA (n= 7) in 2 clusters identified by STRUCTUREEach individual animal is represented by a separate vertical bar Yellow indicates canonical panther ancestry and blue represents admixed ancestry The admixedancestry of the Texas females and F1 generation panthers that were radiocollared are highlighted red and green respectively to demarcate those groups The pre‐introgression period is inclusive of panthers radiocollared from 10 February 1981 to 6 March 1996 The post‐introgression era inclusive of the F1 panthers includesanimals radiocollared from 4 March 1997 to 18 February 2015 in South Florida USA

16 Wildlife Monographs bull 203

100 years assuming that the population would be managedaccording to a particular strategy Our goal with theseexploratory analyses was to evaluate the relative costs andbenefits of alternative management scenarios

RESULTS

Genetic VariationWe assessed genetic variation at 16 microsatellite loci for 137DNA samples collected from adult and subadult panthers(1981ndash2013) that were captured and subsequently radiocollaredThe number of alleles at these loci ranged from 3ndash10 and allelicrichness averaged 57 (Table 2) Average expected hetero-zygosity for this sample was 0535 and average observed het-erozygosity was 0517 (Table 2)We determined individual heterozygosity (1 minusHL) and an-

cestry for the complete sample of panther genotypes (n= 503)collected from 1981 to 2015 for use as covariates in subsequentanalyses Average individual heterozygosity was 0559 andranged from 0ndash0980 Our STRUCTURE analysis providedstrong support for K= 2 clusters based on the probability of thedata (log Pr(D)|K) and ΔK in STRUCTURE HARVESTERBased on this result we categorized the ancestry of each pantheras either canonical (n= 123) or admixed (n= 380) using valuesof proportional membership (q Fig 4) The average individualheterozygosity of canonical panthers was 0386plusmn 0012 (SE) foradmixed panthers it was 0615plusmn 0007

Survival and Cause‐Specific Mortality RatesOf the 395 kittens that were PIT‐tagged and released 55 werelater encountered alive 15 were recovered dead and 85 werepart of failed litters (Table 1) We radio‐tracked 209 subadult oradult panthers for a total of 244195 panther‐days and docu-mented 126 mortalities (Table 1)Survival depended strongly on age and kittens had the lowest

overall rate (032plusmn 009 Fig 5A Table 3) The model‐averagedkitten survival was 047plusmn 009 when we included only geneti-cally sampled individuals in the analysis (Table 3) as statedabove this estimate is biased high because many kittenswere genetically sampled when they were recaptured alive orrecovered dead at older ages We found no evidence for sex‐specific differences in kitten survival but females survived betterthan males in all older age classes For females subadults hadthe highest survival rates followed by prime adults and oldadults For males prime adults had the highest survival ratefollowed by subadults and old adults (Fig 5A and Table 3)For panthers of all ages there was substantial evidence that

ancestry affected survival with F1 admixed panthers of all ageshaving the highest rates and canonical individuals having thelowest (Fig 5B) The combined AIC weights of models thatincluded an ancestry covariate were 0880 for kittens and 0992for older panthers The survival rates of subadult prime‐adultand old‐adult panthers in the period immediately after the in-trogression (1995ndash2004) were similar to those in more recentyears (2005ndash2013 Fig 5C)After genetic introgression individual panther heterozygosity

increased (Fig 6) and varied among ancestry categories individualheterozygosity was the highest for F1 panthers (078plusmn 001)

A

B

C

Figure 5 Overall age‐ and sex‐specific survival rates (plusmnSE A) age‐ and sex‐specific survival rates (plusmnSE) for Florida panthers of canonical F1 admixed andother admixed ancestry (B) and age‐ and sex‐specific survival estimates (plusmnSE)for subadult and adult Florida panthers in the period immediately post‐introgression (1995ndash2004) and more recently (2005ndash2013 C) in SouthFlorida USA

Table 3 Model‐averaged estimates (plusmnSE) of demographic parameters for theFlorida panther obtained using data collected during 1981ndash2013 in SouthFlorida USA

Parameter Sex Age class Estimate

Survival Both Kitten 032plusmn 009a

Female Subadult 097plusmn 002Prime adult 086+ 003Old adult 078plusmn 009

Male Subadult 066plusmn 006Prime adult 077plusmn 005Old adult 065plusmn 010

Probability of reproduction Female Subadult 035plusmn 008Prime adult 050plusmn 005Old adult 025plusmn 006

Kittens produced annually Female Subadult 280plusmn 005Prime adult 267plusmn 001Old adult 228plusmn 013

a Overall kitten survival (based on all kittens in our data set including thosethat were not genetically sampled) Estimate of survival of geneticallysampled kittens was 047plusmn 009 this estimate is biased high because manykittens were genetically sampled when they were recaptured alive or re-covered dead at older ages

van de Kerk et al bull Florida Panther Population and Genetic Viability 17

followed by those for non‐F1 admixed (062plusmn 001) and canonical(039plusmn 001) panthers Individual heterozygosity positively affectedsurvival rates of kittens (slope parameter η= 1455 95CI= 0505ndash2405) Individual heterozygosity also positively af-fected survival of subadult and adult panthers but the evidence forthis effect was weaker because the confidence interval for the ha-zard ratio overlapped 10 (hazard ratio exp(ɩ)= 0311 95CI= 0092ndash1054 Table 4 and Fig 7)The MPC increased after genetic introgression (Fig 2) This

abundance index was also included in top‐ranking modelsindicating that population density affected survival (Table 4)Abundance reduced the survival of kittens (slope parameterη= minus0035 95 CI= minus0049 to minus0021) evidence was weak forthe effect of abundance on survival of subadult and adult panthers(hazard ratio exp(ɩ)= 1002 95 CI= 0995ndash1010 Fig 7) Re-gression coefficients associated with the effect of abundance andindividual heterozygosity on demographic parameters are presentedin Table B1 (available online in Supporting Information)The most frequently observed causes of death for radio‐

collared panthers were vehicle collision and intraspecific ag-gression Cause‐specific mortality analyses revealed thatmortality rates among possible causes were generally similar forthe 2 sexes but that males were more likely to die from in-traspecific aggression than were females (Fig 8A) Male mor-tality from intraspecific aggression was higher for subadult andold adult panthers than for prime adults (Fig 8B) For bothmales and females canonical panthers were more likely to diefrom intraspecific aggression than were admixed panthers(Fig 8C)

Reproductive ParametersOf 101 radio‐collared females 67 reproduced at least once duringour monitoring we used these individuals to assess the annualprobability of reproduction The top‐ranked model for the annual

probability of reproduction included age effect only suggesting thatadding ancestry or abundance did not improve model parsimony(Table 4) Model‐averaged annual probabilities of reproductiondiffered between female subadults (035plusmn 008) prime adults(050plusmn 005) and older adults (025plusmn 006)The most parsimonious model for the number of kittens

produced annually by females that produced at least 1 litter (ieconditional on reproduction) included effects of age ancestryand abundance a model that included only the effect of age wasalmost equally supported (Table 4) The model‐averagednumber of kittens produced annually conditional on re-production was 280plusmn 075 for subadults 267plusmn 043 for primeadults and 228plusmn 083 for old adults Confidence intervals forthe slope parameter (β) overlapped 0 for both abundance(β= 0010 95 CI= minus0001 to 0021) and ancestry (β= minus073795 CI= minus1637 to 0162)

Population Dynamics and PersistenceDeterministic and stochastic demographymdashThe deterministic (λ)

and stochastic (λs) annual population growth rate calculated usingthe matrix population model were 106 (5th and 95th percentiles099ndash114) and 104 (072ndash141) respectively The generation timewas 47 years A female starting in age class one (kitten) wasexpected to live another 58 years and produce 10 kittens onaverage during her lifetime Overall λs was proportionately(elasticity) most sensitive to changes in female prime adultsurvival on the absolute scale (sensitivity) however it was mostsensitive to changes in kitten survival (Fig 9) Populationtrajectories probabilities of quasi‐extinction and times untilquasi‐extinction estimated using the matrix population modeland IBM for comparable scenarios were nearly identical for acritical quasi‐extinction threshold of 10 panthers (Figs 10 and 11)and for a critical quasi‐extinction threshold of 30 panthers(Appendix E available online in Supporting Information)Minimum population count scenariomdashUnder the MPC scenario

(ie density dependence estimated using the minimumpopulation count data) with a critical threshold of 10panthers the population size reached 99 (72ndash104) and 100(77ndash105) at 200 years (Fig 10) as predicted by the IBM and thematrix model respectively The cumulative quasi‐extinctionprobabilities within 100 years based on the MPC scenario were15 (IBM 0ndash48) and 30 (matrix model 0ndash76) Theseprobabilities increased to 25 (IBM 0ndash114) and 38 (matrixmodel 0ndash154) by year 200 (Fig 10) The mean times untilquasi‐extinction were 15 years (IBM 0ndash67) and 15 years (matrixmodel 0ndash72) conditioned on going quasi‐extinct within 100years Expected times until quasi‐extinction conditioned onquasi‐extinction within 200 years were higher 34 years (IBM0ndash137) and 32 years (matrix model 0ndash134 Fig 10)Motor vehicle mortality scenariomdashUnder the MVM scenario

(ie density dependence estimated using estimates of pantherpopulation size from McClintock et al 2015) with a criticalthreshold of 10 panthers the projected population reached 188(142ndash217) and 190 (143ndash219) at 200 years as predicted by theIBM and the matrix model respectively (Fig 11) Thecumulative quasi‐extinction probabilities within 100 years were14 (IBM 0ndash08) and 13 (matrix model 0ndash06) Theseprobabilities increased to 20 (IBM 0ndash17) and 19 (matrix

000

025

050

075

100

1995 2000 2005 2010Year of birth

Indi

vidu

al h

eter

ozyg

osity

Figure 6 Temporal changes in the individual heterozygosity of Floridapanthers born during the period 1995ndash2013 in South Florida USA Pointsare measurements solid line is linear regression shaded area is 95 confidenceinterval of slope Slope= 0006plusmn 0001 R2= 009

18 Wildlife Monographs bull 203

model 0ndash16) within year 200 (Fig 11) The mean times until quasi‐extinction based on the MVM scenario were 5 years (IBM 0ndash61)and 6 years (matrix model 0ndash64) conditioned on going quasi‐extinctwithin 100 years Expected times until quasi‐extinction conditionedon quasi‐extinction within 200 years were 11 years (IBM 0ndash113)and 11 years (matrix model 0ndash112 Fig 11)

Sensitivity of extinction probability to demographic parametersmdashThe partial rank correlation coefficients between quasi‐extinctionprobabilities and demographic parameters were negative anincrease in any of the demographic parameters decreased thequasi‐extinction probability Probabilities of quasi‐extinction within200 years were most sensitive to changes in kitten survival for bothscenarios (Fig 12) followed by survival of subadult females in theMVM scenarios and survival of prime adult females in the MPCscenario The probability of reproduction of prime adult femaleshad the third‐largest partial rank correlation coefficient with quasi‐extinction probability for both scenarios

Benefits and Costs of Genetic ManagementMinimum population count scenariomdashThe MPC scenario with

a critical threshold of 10 panthers predicted that withoutmanagement intervention average individual heterozygositywould decrease reaching 057 (049ndash064) at 100 years and053 (042ndash062) at 200 years (Fig 13) The population woulddecrease slightly reaching an average of 79 (41ndash99) at 100 yearsand 76 (43ndash98) at 200 years (Fig 14) The probabilities of quasi‐extinction under the no‐intervention MPC scenario when weconsidered the effects of genetic erosion were 13 (0ndash99) at 100years and 23 (0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with

the translocation of 5 Texas females average individualheterozygosity under the MPC scenario increased and thenstabilized at 068 (064ndash072) at 100 years and remained atthat level at 200 years (Fig 13) The population reached anaverage of 88 (62ndash104) at 100 years and stayed the same at200 years (Fig 14) The probabilities of quasi‐extinctionunder this introgression strategy were 6 (0ndash53) within 100years and 7 (0ndash82) within 200 years (Fig 15) Results ofanalyses using a critical threshold of 30 panthers revealed asimilar pattern of relative benefits However the prob-abilities of quasi‐extinction were substantially higher (ge45)across all scenarios (Appendix E)

Motor vehicle mortality scenariomdashWithout managementintervention the average individual heterozygosity predictedby the MVM scenario and a critical threshold of 10 panthersdecreased to 060 (046ndash069) at 100 years and to 059 (039ndash070) at 200 years (Fig 13) The population increased slightlyand reached an equilibrium at 154 individuals (46ndash217) at 100years and 157 (57ndash218) at 200 years (Fig 14) The probabilitiesof quasi‐extinction were 17 (0ndash100) at 100 years and 22(0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with 5

female pumas from Texas average individual heterozygosityincreased slightly reaching 067 (060ndash073) at 100 years and068 (059ndash075) at 200 years (Fig 13) Under this strategy thepopulation reached an average of 164 individuals (44ndash226) at

Table 4 Model comparison results testing for the effects of several covariateson survival of all kittens survival of genetically sampled kittens survival ofsubadult and adult Florida panthers probability of reproduction and kittensproduced annually for Florida panthers sampled from 1981ndash2013 in SouthFlorida USA

Model Parameters ΔAICa Weight

Kitten survival (all data)Base+ F1b+ abundancec 10 000 0988Base+ abundance 9 1018 0006Base+ F1 9 1035 0005Base 8 2711 lt0001Base+ sexd 9 2912 lt0001Base+ litter sizee 9 2914 lt0001

Kitten survival (genetically sampled kittens only)Base+ F1+ abundance 10 000 0541Base+ F1CanAdmf+ abundance 11 099 0329Base+Hetg+ abundance 10 310 0115Base+CanAdmh+ abundance 10 791 0010Base+ abundance 9 944 0005Base+ F1 9 1638 lt0001Base+ F1CanAdm 10 1837 lt0001Base+Het 9 2872 lt0001Base 8 3296 lt0001Base+CanAdm 9 3348 lt0001

Subadult and adult survivalBase+ F1CanAdm+ abundance 7 000 0360Base+ F1 5 053 0276Base+ F1CanAdm 6 105 0213Base+ F1+ abundance 6 222 0119Base+CanAdm+ abundance 6 575 0020Base+CanAdm 5 908 0004Base+Het 5 964 0003Base+Het+ abundance 6 984 0003Base 4 1118 0001Base+ abundance 5 1278 0001

Probability of reproductionAge 3 000 0242Age+CanAdm 4 012 0228Age+ abundance 4 173 0102Age+ F1 4 190 0094Age+CanAdm+ abundance 5 216 0082Age+Het 4 219 0081Age+ F1CanAdm 5 232 0076Age+ F1+ abundance 5 372 0038Age+Het+ abundance 5 399 0033Age+ F1CanAdm+ abundance 6 444 0026

Kittens produced annuallyAge+ F1+ abundance 5 000 0194Age 3 060 0144Age+ abundance 4 061 0143Age+ F1 4 097 0120Age+CanAdm 4 139 0097Age+ F1CanAdm+ abundance 6 196 0073Age+ F1CanAdm 5 231 0061Age+CanAdm+ abundance 5 238 0059Age+Het+ abundance 5 250 0056Age+Het 4 259 0053

a Akaikersquos information criterion (AIC) for kitten survival models was adjustedfor small sample size and overdispersion (QAICc)

b F1 divides Florida panthers into 2 groups F1 admixed and all other panthersc Index of Florida panther abundanced Sex of an individual Florida panther (male or female)e Size of the litter in which a Florida panther kitten was bornf F1CanAdm divides panthers into 3 groups F1 admixed canonical andother admixed panthers

g Het is the individual heterozygosityh CanAdm divides panthers into 2 groups canonical and admixed (includingF1 admixed) panthers

van de Kerk et al bull Florida Panther Population and Genetic Viability 19

100 years and 170 (67ndash231) at 200 years based on the MVMscenario (Fig 14) The probabilities of quasi‐extinction underthis introgression strategy were 10 (0ndash99) at 100 years and12 (0ndash99) at 200 years (Fig 15)The projected population size and average individual hetero-

zygosity reached equilibrium levels with periodic fluctuations inthese values when introgression was implemented (Figs 16ndash17)Genetic introgression decreased the time until quasi‐extinctionacross all scenarios (Fig 18) Results of analyses using a criticalthreshold of 30 panthers revealed a similar pattern of relative ben-efits However the probabilities of quasi‐extinction were sub-stantially higher (ge45) across all scenarios (Appendix E)

Addition of pumas from Texas increased both the populationsize and average individual heterozygosity consequentlyintrogression caused periodic increases in average individualheterozygosity and population size at the interval at which theintrogression occurred Whereas average individual hetero-zygosity gradually decreased following introgression densitydependence caused the population size to fluctuate especiallywhen a larger number of pumas from Texas were released Thepositive effect of genetic introgression initiatives on quasi‐extinction probabilities decreased as the time interval betweenthese events increased More frequent genetic introgressions ledto greater increases in average individual heterozygosity and

Old adult

Prime adult

Subadult

Kitten

025 050 075

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Individual heterozygosity

Ann

ual s

urvi

val r

ate

Old adult

Prime adult

Subadult

Kitten

40 60 80 100 120

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Abundance index

Ann

ual s

urvi

val r

ate

Kittens Males Females

Figure 7 The effect of individual heterozygosity (left) and the abundance index (right) on Florida panther age‐ and sex‐specific survival estimates (shaded area isSE) in South Florida USA 1981ndash2013 Abundance index data are from McBride et al (2008) and R T McBride and C McBride (Rancherrsquos Supply Incorporatedunpublished report)

20 Wildlife Monographs bull 203

equilibrium population size while reducing the probability ofquasi‐extinction Comparatively performing genetic introgres-sion less often but with more individuals per release was lesseffective at improving the long‐term prospects for the pantherpopulation (Figs 13ndash15)

Financial cost and persistence benefitsmdashThe total estimated costof 1 genetic introgression was $40000 $80000 or $120000for releasing 5 10 or 15 pumas from Texas respectively(Table 5) The total cost incurred over 100 years for eachstrategy ranged from $50000 to $12 million (Table 6)The most expensive strategy involved the introduction of15 pumas every 10 years this strategy reduced the probability

of quasi‐extinction by 58 and 40 under the MPC and MVMscenarios respectively (Table 6) Introducing 5 pumas every80 years cost the least but improvements in populationpersistence under this strategy were insubstantial (Table 6)Releasing more panthers more frequently caused increasedpopulation fluctuations but did not necessarily reduce the quasi‐extinction probability (Figs 14 and 15 Table 6)

DISCUSSION

The duration (34 yr of data) and sample sizes associated with theFlorida Panther Project allowed us to obtain a comprehensiveunderstanding of genetic variation demographic parameters

A

B

C

Figure 8 Cause‐specific mortality rates (plusmnSE) for male and female Florida panthers (A) subadult prime‐adult and old‐adult Florida panthers of both sexes (B)and canonical and admixed Florida panthers of both sexes (C) in South Florida USA 1981ndash2013 Causes of mortality are vehicle collision intraspecific aggression(ISA) other causes (known causes such as diseases injuries and infections unrelated to the first 2 causes) and unknown cause (mortalities for which we could notassign a cause)

van de Kerk et al bull Florida Panther Population and Genetic Viability 21

probability of population persistence and the duration of thepositive impacts of genetic restoration on this endangered po-pulation The Florida panther population was in dire straits inthe early 1990s and our results clearly demonstrated that the

implementation of the introgression project in 1995 subse-quently accrued a series of genetic and demographic improve-ments that have led to the change from a declining population toone that is growing and expanding its current breeding range

Sensitivity Elasticity

0 1 2 3 00 02 04

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

Para

met

er

Figure 9 Sensitivity and elasticity (proportional sensitivity) of stochastic population growth rate (λs) of the Florida panther to vital demographic parameters in SouthFlorida USA 1981ndash2013 Horizontal lines represent 5th and 95th percentiles

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

70

80

90

100

110

0

5

10

15

0

50

100

Years

StatisticMeanMedian

Figure 10 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the minimum population count(MPC) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

125

150

175

200

00

05

10

15

20

0

30

60

90

Years

StatisticMeanMedian

Figure 11 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the motor vehicle mortality(MVM) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

22 Wildlife Monographs bull 203

MPC MVM

minus08 minus06 minus04 minus02 00 minus08 minus06 minus04 minus02 00

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Male subadult survival

Male prime adult survival

Male old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

PRCC

Para

met

er

Figure 12 Partial rank correlation coefficient (PRCC) between quasi‐extinction probabilities and the demographic parameters of the Florida panther for theminimum population count (MPC) and motor vehicle mortality (MVM) scenarios Error bars indicate standard deviation Based on data collected in South FloridaUSA 1981ndash2013

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 13 Average projected individual heterozygosities in the Florida panther population over 200 years without genetic management intervention and with eachof the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include theintroduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years Average individual heterozygosity peaks when pumas are added to the Floridapanther population Based on data collected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 23

Our research has further validated the success of genetic in-trogression by quantifying the reduction in the probability ofextinction of the panther population because of this manage-ment initiative These findings all bode well for continuedprogress toward recovery That said the Florida panther po-pulation remains small and isolated from other populations ofpuma from western North America thereby denoting that theeventual impact of genetic drift and inbreeding may negativelyaffect the population in the future Realizing this will continueto be an issue that managers will have to contemplate wemodeled the impact of varied genetic management scenarios todetermine the frequency of introgression along with the numberof panthers that should be released during each initiative tominimize cost and maximize benefits to the population Theseresults will prove invaluable to managers attempting to conservethe Florida panther

Genetic Variation Demographic Parametersand Persistence of Heterotic Benefits from GeneticIntrogressionOur measure of individual heterozygosity (1 minusHL) was higheron average for the admixed (0615) panthers than for those ofcanonical ancestry (0386) Levels of individual heterozygosityfor admixed panthers were comparable to levels observed in asample of pumas from west Texas (x plusmn SE= 0695plusmn 0019n= 41) which further demonstrates the improvement in levelsof genetic variation in the Florida population since 1995 The

correlation between HL and several demographic parametershas been noted in wild populations including cheetahs (Terrellet al 2016) and several species of birds such as European shags(Phalacrocorax aristotelis male and female survival probabilityVelando et al 2015) and Egyptian vulture (Neophron percnop-terus age at recruitment Agudo et al 2012) If we focus only onpanthers captured and radiocollared from 2012ndash2015 (FP211ndashFP240) all of which had estimated birth years ge2005 (ge10 yrpost‐introgression) those panthers had an average individualheterozygosity level of 0618plusmn 0029 highlighting the elevatedlevels of genetic variation that continue to persist in cohorts ofpanthers 5 generations after the release of female pumas fromTexasThe proportional membership values (q) from our

STRUCTURE analysis allowed us to delineate 2 clusters ofancestry for Florida panthers (Fig 4) During the pre‐in-trogression era the population was dominated by panthers thatretained a high percentage of the canonical ancestry which wasaffected by inbreeding depression The post‐introgression eraancestry values are comprised of many admixed individuals inessence demonstrating the success of the introgression in termsof increasing genetic variation in the population and mi-micking gene flow that historically occurred with panthers andother populations of pumas (Onorato et al 2010) A somewhatanalogous situation although abetted by natural immigrationwas evident in the ancestral variation in a small population ofpuma intersected by a major freeway in California USA In

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

75

80

85

90

95

100

130

150

170

190

Years

Popu

latio

n si

ze

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 14 Average projected population sizes in the Florida panther population over 200 years without intervention and with each of the genetic introgressionstrategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include the introduction of 5 10 or 15pumas from Texas USA every 10 20 40 or 80 years Peaks occur when pumas are added to the Florida panther population Based on data collected in SouthFlorida USA 1981ndash2013

24 Wildlife Monographs bull 203

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

25

50

75

100

0

25

50

75

100

Introgression interval (yrs)

PQ

E (

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 15 Average probability of quasi‐extinction (PQE) of the Florida panther population within 100 years and within 200 years without genetic managementintervention (N) and with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenariosIntrogression strategies include the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Errorbars indicate 5th and 95th percentiles Based on data collected in South Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

50

60

70

80

90

100

110

50

100

150

200

Years

Popu

latio

n si

ze

Figure 16 Average projected Florida panther population trajectories under a geneticintrogression strategy involving the release of 5 female pumas from Texas USA every20 years for the minimum population count (MPC) and motor vehicle mortality(MVM) scenarios Shaded area indicates 5th and 95th percentiles and isrepresentative of confidence intervals for other scenarios Based on data collected inSouth Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Figure 17 Average projected Florida panther population‐level heterozygositiesunder a genetic introgression strategy involving the release of 5 female pumas fromTexas USA every 20 years for the minimum population count (MPC) and motorvehicle mortality (MVM) scenarios Shaded area bounds the 5th and 95th percentilesand is representative of confidence intervals for other scenarios Based on datacollected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 25

this case the migration of just a single male across the freewayto the south resulted in an increase in the number of in-dividuals with admixed ancestry and substantially improvedgenetic diversity of the subpopulation south of the freeway(Riley et al 2014)Our estimate of overall kitten survival (0321plusmn 0056) was

similar to that reported by Hostetler et al (2010) who used13 years of post‐introgression data (0323plusmn 0071) These valuesare lower than kitten survival estimates derived from severalpuma populations in western North America (044ndash091 Loganand Sweanor 2001 Lambert et al 2006 Laundre et al 2007Robinson et al 2008 Ruth et al 2011) When we included onlydata for individuals that had been genetically sampled our es-timate was 15 higher The proportion of admixed kittens thatwere genetically sampled increased as the population expandedwhich could have resulted in higher estimates of kitten survivalbecause admixed kittens survive better than canonical kittensKitten survival was strongly influenced by genetic ancestry withsurvival rates of F1 kittens being almost twice that of canonicalkittens (Fig 5B) Consistent with the findings of Hostetler et al(2010) increases in heterozygosity coincided with increasedkitten survival whereas population density negatively affectedkitten survival Population density often has a strong negativeimpact on survival of young age classes due to infanticide andcannibalism which has been reported from several carnivore

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

50

100

150

0

50

100

150

Introgression interval (yrs)

TQE

(yrs

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 18 Average times until quasi‐extinction (TQE) for the Florida panther population within 100 years and within 200 years without genetic management interventionand with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategiesinclude the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Error bars indicate 5th and 95thpercentiles Based on data collected in South Florida USA 1981ndash2013

Table 5 Average cost (2017 US dollars) of introgression strategies employedover 100 years that involve the introduction of 5 10 or 15 pumas from TexasUSA into the Florida panther population in South Florida USA at an in-creasingly higher frequency Costs of capture (including transportation) quar-antine and post‐introgression monitoring were estimated by Roy McBrideMark Cunningham and Dave Onorato respectively

Frequency Component 5 pumas 10 pumas 15 pumas

Once Capture 25000 50000 75000

Quarantine 5000 10000 15000

Monitoring 10000 20000 30000

Total 40000 80000 120000

Every 80 years Capture 31250 62500 93750

Quarantine 6250 12500 18750

Monitoring 12500 25000 37500

Total 50000 100000 150000

Every 40 years Capture 62500 125000 187500

Quarantine 12500 25000 37500

Monitoring 25000 50000 75000

Total 100000 200000 300000

Every 20 years Capture 125000 250000 375000

Quarantine 25000 50000 75000

Monitoring 50000 100000 150000

Total 200000 400000 600000

Every 10 years Capture 250000 500000 750000

Quarantine 50000 100000 150000

Monitoring 100000 200000 300000

Total 400000 800000 1200000

26 Wildlife Monographs bull 203

species including polar bears (Ursus maritimus Derocher andWiig 1999) black bears (Ursus americanus Garrison et al 2007)and pumas (Logan and Sweanor 2001)Our estimates of sex‐ and age‐specific survival rates of subadult

and adult panthers (Table 3) and the effects of covariates on theserates were similar to those reported by Benson et al (2011) derivedfrom data collected through 2006 Our survival rates for males(065ndash077) were lower than females (078ndash097) for all 3 ageclasses (subadult prime adult and old adult) which is the typicaltrend observed in most puma populations (eg Ruth et al 19982011) However an unhunted puma population occupying afragmented urban landscape in southern California exhibited lowersurvival (0586 females 0525 males Vickers et al 2015) perhapsas a result of severe habitat fragmentation and the lack of extensiveswaths of public land protected in perpetuity Most populations ofpuma in western North America are typically subject to variedlevels of management via regulated hunting which often is biasedtoward the take of males (Cooley et al 2009 Ruth et al 2011)Consequently annual survival estimates from hunted populationsin northeast Washington (0679ndash0733 females 0341 malesRobinson et al 2008) and southeastern Arizona (0685 females0584 males Cunningham et al 2001) were generally lower thanthose we estimated for Florida panthersCause‐specific mortality analysis showed that male panthers

experienced a substantially greater mortality risk from in-traspecific aggression This differs from the pattern observedduring a long‐term study in Arizona where females were at ahigher risk of intraspecific mortality than males (46 of maleand 53 of female mortality Logan and Sweanor 2001)Mortality associated with intraspecific strife has been docu-mented in hunted and unhunted populations (this study Loganand Sweanor 2001 Stoner et al 2010) and its root cause hasbeen difficult to determine (eg population density and preypopulation trends) That being the case finding ways to reducewhat is often a major source of mortality for puma populationscontinues to pose a challenge to managersCanonical panthers were substantially more likely to die from

intraspecific aggression than were admixed panthers This mayat least partly explain the difference in survival between admixed

and canonical panthers Canonical panthers are known to bemore likely to suffer from varied correlates of inbreeding de-pression than admixed animals including atrial septal defects(Johnson et al 2010) and compromised immune systems(Roelke et al 1993) These factors have the potential to affectfitness and perhaps dominance of individuals when engaged inan intraspecific encounter A somewhat similar scenario wasobserved by Hogg et al (2006) in a population of bighorn sheepthat were part of an introgression project Admixed males in thispopulation had a greater probability of paternity than males thatwere less outbred This included coursing males with higherlevels of admixture being markedly more successful at fightingmales that were defending females in estrous (Hogg et al 2006)Heterosis resulting from genetic introgression is expected to be

the strongest in F1 individuals (Shull 1908 Crow 1948 Burkeand Arnold 2001 Waller 2015) and to progressively decline insubsequent generations (Pickup et al 2013 Frankham 2015)Given that admixed (especially F1) panthers survived better thancanonical panthers and that individual heterozygosity improvedsurvival of both sexes and all age classes our data provide evi-dence for heterotic advantage whereby individuals of mixedancestry had higher fitness (Shull 1908 Crow 1948) Our resultsare consistent with findings of other studies that involved in-tentional genetic introgression for conservation purposes Forexample Hogg et al (2006) detected substantial improvementsin survival and reproduction of introgressed bighorn sheep andMadsen et al (1999 2004) reported a dramatic increase in thepopulation of adders after genetic introgressionKnowledge of the persistence of benefits to a population ac-

crued via genetic introgression is essential for determiningwhen additional introgression may be warranted There is adepauperate amount of information regarding this topic and theidentification of factors that may influence persistence of thebenefits of genetic introgression (Tallmon et al 2004 Hedricket al 2014 Whiteley et al 2015) For example in minks(Neovison vison) Thirstrup et al (2014) found that outcrossingled to larger litters in F2 and F3 but this benefit disappeared insubsequent generations In a population of gray wolves Hedricket al (2014) suggested that benefits of genetic introgression may

Table 6 Costs and benefits accumulated over 100 years for genetic introgression at different intervals and with different numbers of pumas from Texas USAintroduced into the Florida panther population in South Florida USA Costs (2017 US dollars) include capture and transportation quarantine and post‐introgression monitoring Benefits are represented as average probability of quasi‐extinction (PQE and 5th and 95th percentiles) and changes (Δ) therein under thescenario using the minimum population count (McBride et al 2008 MPC) and the scenario using the motor vehicle mortality population estimates (McClintocket al 2015 MVM) The table is sorted by percent change in probability of quasi‐extinction under the MPC scenario

Interval (yr) Pumas Cost MPC PQE () ΔMPC PQE () MVM PQE () ΔMVM PQE ()

ndash 0 0 13 (0ndash99) 0 17 (0ndash100) 080 10 100000 12 (0ndash99) 10 (0ndash58) 16 (0ndash100) 5 (0ndash38)80 15 150000 12 (0ndash99) 13 (0ndash65) 15 (0ndash100) 8 (0ndash56)80 5 50000 12 (0ndash99) 14 (0ndash73) 15 (0ndash99) 9 (0ndash41)40 15 300000 10 (0ndash98) 26 (0ndash104) 14 (0ndash99) 13 (0ndash60)40 10 200000 9 (0ndash94) 35 (0ndash183) 13 (0ndash99) 21 (0ndash95)40 5 100000 8 (0ndash89) 39 (0ndash211) 12 (0ndash99) 26 (0ndash124)20 15 600000 8 (0ndash88) 42 (0ndash257) 13 (0ndash99) 24 (0ndash123)20 10 400000 6 (0ndash67) 54 (0ndash318) 10 (0ndash99) 37 (0ndash217)10 15 1200000 6 (0ndash53) 58 (0ndash372) 10 (0ndash99) 40 (0ndash208)20 5 200000 5 (0ndash50) 59 (0ndash338) 9 (0ndash99) 44 (0ndash352)10 10 800000 4 (0ndash16) 69 (0ndash489) 8 (0ndash92) 55 (0ndash401)10 5 400000 4 (0ndash7) 73 (0ndash502) 6 (0ndash71) 63 (0ndash503)

van de Kerk et al bull Florida Panther Population and Genetic Viability 27

wane after 2 or 3 generations The WrightndashFisher model ofgenetic drift predicts a geometric decline in expected hetero-zygosity at the rate of (1 minus 12Ne) per generation where Ne is theeffective population size (Hartl and Clark 1997 Frankham et al2014) Thus it seems logical to assume that the duration of thebenefits of introgression is limited especially in a species per-sisting in a small population such as Florida panthersWe expected Florida panther survival in the most recent years

(2005ndash2013) post‐introgression to differ from that observed in thedecade following the introgression in 1995 because pantherabundance and heterozygosity factors known to influence panthersurvival (Hostetler et al 2010 Benson et al 2011) have changed(Figs 2 and 6) We found that subadult and adult survival rates inrecent years (2005ndash2013) were similar to those in the period im-mediately following introgression (1995ndash2004 Fig 5C) overallkitten survival was similar to estimates of Hostetler et al (2010) Infact the pattern of covariate effects on Florida panther survivalreported by Benson et al (2011) and Hostetler et al (2010) hasbarely changed The negative effects of population density andpositive effects of heterozygosity may counterbalance survival ratesreducing population fluctuations Our result that benefits of geneticrestoration have remained in the population nearly for 5 genera-tions post‐intervention was surprising but also encouraging Pos-sible explanations for this observation may include 1) genetic ero-sion occurs at slower rate than previously thought 2) introducing 5migrants in 1 genetic intervention may have beneficial effects si-milar to those from 1 migrant per generation over multiple gen-erations or 3) other and as yet unknown factors may reduce therate of genetic erosion and subsequent demographic effects Adensity‐dependent effect on kitten survival could also explain whynon‐F1 admixed kittens did not survive substantially better thandid canonical kittens most kittens sampled from 2005ndash2013 wereadmixed and their survival was lower possibly because of a density‐dependent effect as the Florida panther population continuouslygrew during that period Trinkel et al (2010) also found thatpopulation density and inbreeding coefficient interacted to affectdemographic parameters and growth rate of an African lion po-pulation Likewise population density interacted with winterweather to affect the survival and population growth rates in Soaysheep (Ovis aries Milner et al 1999 Coulson et al 2001)

Population Dynamics and PersistenceThe finite deterministic and stochastic growth rates were gt10reflecting that much of the data used in this study were collectedfollowing genetic introgression in 1995 during which time thepopulation increased substantially Because our demographicparameter estimates did not change markedly in comparison tothose of Hostetler et al (2013) the similarity between thepopulation growth estimates from these studies was expectedBut these estimates reflect population growth during thedata‐collection period and do not predict future growth In factestimates of population size reported by McClintock et al(2015) suggest that population growth may already be slowing(Fig 2) A similar pattern was observed in an introduced po-pulation of African lion which increased steadily from 1995until 2001 then fluctuated around the presumed carrying ca-pacity thereafter (Trinkel et al 2010) Several other African lionpopulations that experienced various degrees of recovery

subsequently became relatively stable or exhibited decliningtrends (Bauer et al 2015)Our estimates of quasi‐extinction probabilities were lower than

those reported by Hostetler et al (2013) using a 2‐sex age‐structured matrix population model Lower probabilities ofquasi‐extinction than those reported by the earlier study forcomparable scenarios were likely a consequence of the fact thatthe estimates of abundance (McBride et al 2008) and thus thedensity‐dependent effects on survival of kittens and old panthers(gt10 yr) have changed in recent years Additionally these earlyestimates of the probability of quasi‐extinction and of time toquasi‐extinction did not consider genetic erosion that will in-evitably occur without management interventionUnder the MVM scenario the equilibrium population size was

twice that attained under the MPC scenario The MVM popula-tion estimates are approximately twice the MPCs (Fig 2) as aresult the simulated population under the MVM scenario achieveda higher equilibrium size Probability of quasi‐extinction under theMVM scenario was generally greater than that under the MPCscenario This result is a consequence of the fact that the estimateddensity dependence on kitten survival based on the MPC scenario isstronger than that for theMVM scenario (regression coefficients forabundance for the MPC scenario= minus0068 vs minus0002 for theMVM scenario Appendix B Table B1) Consequently simulatedpopulations under the MPC scenario have a stronger tendency tomove toward the equilibrium population size which inherentlyreduces the quasi‐extinction probability (eg Royama 1992)As expected for long‐lived mammals (Heppell et al 2000 Oli

and Dobson 2003) the population growth rate was proportio-nately most sensitive to changes in survival of prime adultsfollowed by that in younger age classes Global sensitivity ana-lysis in contrast showed that under all scenarios extinctionparameters were particularly sensitive to changes in survival ofFlorida panther kittens This result is a consequence of the highsensitivity of the population growth rate to kitten survival(Fig 9) and a larger standard error of kitten survival (Table 3)Results of our sensitivity analyses indicate that future researchshould focus on acquiring additional information on early lifestages to improve the accuracy and precision of estimates ofkitten survival Our results suggest that demographic variables(or their environmental drivers) that strongly influence extinc-tion risks are not necessarily those with the largest elasticityvalues A study of island fox (Bakker et al 2009) found thatextinction risk was strongly influenced by survival modifierparameters that had low elasticity values

Genetic Management When and How to InterveneThe use of genetic introgression as a management tool has al-ways been controversial and the probable effectiveness it mighthave in preventing the imminent demise of the panther popu-lation was initially debated (Creel 2006 Maehr et al 2006Pimm et al 2006) These debates notwithstanding the pantherpopulation had suffered from genetic morphological and bio-medical correlates of inbreeding before this management in-tervention (Johnson et al 2010 Onorato et al 2010) whereasthe post‐introgression period has been characterized by a sub-stantial reduction in the biomedical correlates of inbreeding andimprovements in demographic vigor and abundance (McBride

28 Wildlife Monographs bull 203

et al 2008 Hostetler et al 2010 2013 Johnson et al 2010Benson et al 2011 McClintock et al 2015) Nevertheless thepopulation remains small and isolated habitat is still being lostand there is no current possibility of natural gene flow betweenthis and other North American puma populations thus therecurrence of inbreeding‐related problems and subsequent po-pulation declines are inevitable The question therefore is notwhether genetic management of the Florida panther populationis needed but when and how it should be implementedWithout genetic introgression probabilities of quasi‐extinc-

tion were substantially greater than those from models thatincorporated periodic genetic introgression events (Fig 15)highlighting the importance of genetic management in smallisolated populations When 5 female pumas are added to theFlorida panther population every 10 years genetic variationincreased substantially under the MPC and MVM scenariosThe IBM predicted that the equillibrium population size wouldbe substantially larger when 5 pumas were released into thepopulation every 10 years than populations without intervention(ie no introgression) Releasing 15 Texas females did not af-ford substantially greater benefit to the equilibrium populationsize than did releasing only 5 Texas females Instead addingmore individuals caused increasingly large fluctuations in po-pulation size due to the numerical increases and density de-pendence (Fig 14) possibly diminishing the benefits associatedwith increased genetic variationCompared with the no‐intervention scenario (ie no genetic

introgression implemented) the introduction of 5 female pumasevery 10 years reduced quasi‐extinction probabilities from 13to 4 and from 17 to 6 for the MPC and MVM scenariosrespectively The benefit of genetic‐introgression strategies de-creased as the interval between successive management inter-ventions increased and no benefit was evident once the intervalreached 80 years (Fig 15) Introgression reduced the averagetime until quasi‐extinction (Fig 18) This somewhat counter‐intuitive result can be explained by the fact that repeated geneticintrogression makes the population more robust over timewhich will reduce the probability of extinction Consequentlyfew simulated population trajectories that fall below the criticalthreshold do so early reducing the mean time to extinctionAlso density dependence has a stabilizing effect on populations(Royama 1992) populations tend toward equilibrium popula-tion sizes which reduces the probability over time of popula-tions falling below quasi‐extinction threshold However timeuntil quasi‐extinction must be interpreted in conjunction withthe probability of quasi‐extinction which is very low for allscenarios with genetic introgression (Fig 15)Cost is a significant impediment to the implementation of a

genetic introgression program because captures relocations andmonitoring efforts before and after introgression are expensive(Edmands 2007 Frankham 2010b 2015 2016) Costs may beacceptable to society if the management actions result in sub-stantial fitness benefits especially if the species of concern ispopular and charismatic (Frankham 2015) We estimated the100‐year cost of introgression strategies modeled here to rangefrom $50000 to $12 million More expensive strategies gen-erally led to larger decreases in the probability of quasi‐extinc-tion but cheaper strategies also substantially improved

population viability (ie reduced quasi‐extinction probabilityTable 6) One of the cheaper strategies (5 pumas released every20 years) which cost an estimated $200000 over 100 yearsreduced the probability of quasi‐extinction by 59 and 44 forthe MPC and MVM scenarios respectively But these pre-liminary cost estimates are based on expenses incurred duringthe 1995 introgression and include costs of capture quarantinetransportation release and post‐release monitoring of femalesonly Although the cost for future genetic interventions wouldlikely be higher than those reported here the relative differencesin cost between alternative intervention strategies should remainapproximately the sameOur ability to simulate genetics and link it to the viability of

the Florida panther population is based on the empirically es-timated relationship between individual heterozygosity andsurvival Such heterozygosity and fitness correlations have beenstudied for decades (David 1998) but have only recently beenused to predict population viability (Benson et al 2016) noneto our knowledge have incorporated cost into their modelingefforts The mechanisms underlying heterozygosity and fitnesscorrelations remain unclear (David 1998 Szulkin et al 2010)and their significance as a proxy for inbreeding depression hasbeen debated (Balloux et al 2004 Miller and Coltman 2014)Nevertheless evidence continues to mount demonstratingpositive relationships between heterozygosity and survival(Coulson et al 1998ab Hansson et al 2004 Silva et al 2005Acevedo‐Whitehouse et al 2006 Jensen et al 2007 Velandoet al 2015) and between heterozygosity and fecundity (Seddonet al 2004 Charpentier et al 2005 Agudo et al 2012 Velandoet al 2015) Although the reported correlations are generallyweak their consequences can be substantial if population growthand persistence parameters are highly sensitive to the affectedvital rates (Velando et al 2015) Compared with scenarios thatignore genetics incorporating the effects of inbreeding on sur-vival increased quasi‐extinction probabilities within a 100‐yeartimeframe from 15 to 13 and from 14 to 17 for theMPC and MVM scenarios respectively These results high-light the importance of incorporating genetics when analyzingthe viability of small isolated populationsAlthough conservation biologists have recognized the im-

portance of genetics in population viability many still fail toincorporate genetic factors into PVA models (Reed et al 2002)Explicit consideration of genetics in PVAs requires long‐termgenetic and demographic data This permits the assessment ofthe functional relationship between genetic variability and de-mographic parameters Population viability analysis softwarepackages such as VORTEX (Lacy 1993 2000) offer a powerfulframework for individual‐based simulations but they often donot adequately capture a speciesrsquo life history or genetics Basedon a thorough review of the importance of genetic factors inwildlife conservation Frankham et al (2014) concluded thatgenetic factors can strongly influence the dynamics and persis-tence of small andor fragmented populations They furthernoted that ldquoMost PVA‐based risk assessments ignore or in-adequately model genetic factorsrdquo and recommended that ldquoPVAshould routinely include realistic inbreeding depressionhelliprdquo(Frankham et al 201457) Our custom‐built IBM permitted usto develop a model that adequately captured the Florida panther

van de Kerk et al bull Florida Panther Population and Genetic Viability 29

life history and we could parameterize the model with our long‐term genetic and demographic dataThe explicit consideration of the effects of inbreeding on

Florida panther population dynamics and persistence representsan important step forward Nonetheless our model remainsimperfect First we neglected the possible effects of habitat lossdetrimental anthropogenic activities climate change the prob-ability of immigration of pumas from western populationsdisease or catastrophes on demographic rates and populationviability Although immigration from puma populations outsideFlorida is possible its probability is very low given the extensivechallenges the anthropogenically altered landscape would posefor such a migrant to make it successfully to South Florida Weomitted mutations from our model because we have no in-formation on mutation rates though we expect that they are lowbased on values reported for other mammals (Kumar and Sub-ramanian 2002) Finally we only considered possible effects ofinbreeding on age‐specific survival rates because there was noevidence that heterozygosity affected female reproduction Lossof heterozygosity can negatively affect reproduction becauseinbred male panthers can suffer from cryptorchidism which canadversely affect their fecundity However given the polygynousmating system we expect the Florida panther population growthis primarily female‐limitedAlthough it is generally accepted that genetic introgression

only temporarily relieves a population from inbreeding depres-sion we know of no cases in which it has been implementedmore than once to conserve a threatened wildlife populationOur study provides an example of how demographic and mo-lecular data can be used within an IBM framework to evaluatethe efficacy of alternative genetic management strategies via theFlorida panther as a case study Our model can be adjusted forand applied to other species and offers great potential as a toolfor genetic management of small isolated wildlife populations

MANAGEMENT IMPLICATIONS

Our results and those of Hostetler et al (2013) and Johnsonet al (2010) provide persuasive evidence that the genetic in-tervention implemented in 1995 prevented the demise of theFlorida panther and restored demographic vigor to the popu-lation The Florida panther population remains small and iso-lated from other puma populations the closest puma populationis located gt1000 km away in Texas and the intervening land-scape is highly developed and fragmented with many interstate(and other high‐traffic) highways and major urban centersTheory suggests that 1 migrant per generation is sufficient tominimize the loss of genetic diversity (eg Mills and Allendorf1996) However the possibility of successful migration of apuma to South Florida followed by successful mating every ~5years is very low considering the distance between Texas andFlorida (Hedrick 1995) and the many risks and barriers thatdispersing pumas face (Beier 1995 Maehr et al 2002 Thatcheret al 2006) Consequently the pantherrsquos long‐term persistencewill likely depend on subsequent timely genetic introgressionsgiven the near‐impossibility of natural gene flow from otherpuma populations To that end our study offers considerableinsights into benefits of different genetic management strategiesand associated costs

Without genetic management the Florida panther populationfaces a substantial risk of quasi‐extinction within 100 years (10ndash46 depending on quasi‐extinction threshold and scenarios)Twenty‐three years have passed since genetic introgression wasimplemented and the population appears healthy as indicatedby measures of genetic variation increases in the populationsize and improvements in age‐specific survival rates Ourfindings suggest that releasing more pumas more frequently doesnot necessarily improve persistence probability because such astrategy would cause larger fluctuations in population size(Fig 14) which negatively affects persistence probability Thusextinction risks faced by inbred populations of large carnivorescan be substantially reduced by releasing approximately 5 im-migrants every 2 decades We suggest that such a managementstrategy be followed only in conjunction with continued long‐term monitoring of the population Our analyses demonstratethat population growth and persistence are highly sensitive tokitten and female (adult and subadult) survival Continuing tocollect data on these demographic variables along with bio-metric and genetic sampling will remain important to pantherrecovery and vigilance in identifying issues associated with in-breeding depression The collection of these monitoring datashould allow managers to detect any demographic or geneticchanges in a timely fashion and respond with genetic in-trogression or other management actions if warrantedRecovery objectives as defined by the USFWS in its current

recovery plan (USFWS 2008) delineate the need for additionalpopulations in the historical range to downlist or delist theFlorida panther If the only extant population of Floridapanthers continued to grow it could serve as a source ofpanthers to be translocated to suitable habitat to seed addi-tional populations Maintaining current information on thegenetic health of the population and precise estimates of itssize will be important in assessing whether it can withstand theremoval of a subset of animals for translocation Although the2017 documentation of a female panther with kittens north ofthe current breeding range is encouraging in terms of thepotential for population expansionmdashgiven that no female hadbeen recorded north of the Caloosahatchee River since 1972mdashthe re‐establishment of separate new populations will mostlikely require translocations by wildlife managers and a lengthysociopolitical processConservation of threatened or endangered species such as the

Florida panther is inherently challenging For panthers and otherlarge carnivores that require vast extents of natural habitat topersist habitat is typically the most limiting factor that will de-termine whether recovery efforts succeed Although geneticmanagement invariably plays a key role in the long‐term persis-tence of the panther population even a healthy population caneventually succumb to the impacts of habitat loss The humanpopulation of Florida continues to be one of the fastest‐growingin the nation the United States Census Bureau currently ranksFlorida as the third most populous Therefore habitat loss isgoing to continue to be a key issue for panthers Diligence towardpreserving panther habitat in its historical range via conservationeasements or other means and trying to keep such areas inter-connected via corridors is a goal stakeholders such as governmentagencies sportsmen non‐governmental organizations ranchers

30 Wildlife Monographs bull 203

and other private landowners must work on collaboratively toachieve

ACKNOWLEDGMENTS

We thank D Shindle D Jennings T OrsquoMeara D Land andC Belden for valuable advice throughout the project We thankS Bass M Criffield M Cunningham D Giardina D JansenA Johnson D Land M Lotz C McBride Rocky McBrideRowdy McBride Roy McBride L Oberhofer S Schulze staffsat Everglades National Park and Big Cypress National Preserveand others for assistance with fieldwork We also thank JAustin M Conroy J Hines J Laake S Mills J Nichols JHines M Sunquist J Fryxell and T Coulson for advice andassistance regarding data analysis and panther biology TheMuseum of Texas Tech University Natural Science ResearchLaboratory provided samples from pumas from west Texas forcomparative genetic analyses M Ben‐David D Land WMcCown E Hellgren and 4 anonymous reviewers providedmany helpful comments on the manuscript We thank NereaUbierna Lopez for completing the Spanish translation of theabstract We are grateful to the Department of ResearchComputing University of Florida for excellent high‐perfor-mance computing services We would like to thank US Fishand Wildlife Service Florida Fish and Wildlife ConservationCommission (FWC) US National Park Service QuantitativeSpatial Ecology Evolution and Environment (QSE3) In-tegrative Graduate Education and Research Traineeship(IGERT) program funded by the National Science Foundationand the University of Florida for financially supporting thisstudy Funding for panther research conducted by the FWC issupported predominantly via donations accrued with vehicleregistrations for ldquoProtect the Pantherrdquo license plates

LITERATURE CITEDAcevedo‐Whitehouse K T R Spraker E Lyons S R Melin F Gulland RL Delong and W Amos 2006 Contrasting effects of heterozygosity onsurvival and hookworm resistance in California sea lion pups MolecularEcology 151973ndash1982

Adams J R L M Vucetich P W Hedrick R O Peterson and J AVucetich 2011 Genomic sweep and potential genetic rescue during limitingenvironmental conditions in an isolated wolf population Proceedings of theRoyal Society B Biological Sciences 2783336ndash3344

Agresti A 2013 Categorical data analysis Third edition John Wiley amp SonsHoboken New Jersey USA

Agudo R M Carrete M Alcaide C Rico F Hiraldo and J A Donaacutezar2012 Genetic diversity at neutral and adaptive loci determines individualfitness in a long‐lived territorial bird Proceedings of the Royal Society BBiological Sciences 2793241ndash3249

Albeke S E N P Nibbelink and M Ben‐David 2015 Modeling behavior bycoastal river otter (Lontra canadensis) in response to prey availability in PrinceWilliam Sound Alaska a spatially‐explicit individual‐based approach PLOSOne 10e0126208

Alho J S K Vaumllimaumlki and J Merilauml 2010 Rhh an R extension for estimatingmultilocus heterozygosity and heterozygosity‐heterozygosity correlationMolecular Ecology Resources 10720ndash722

Alvarez K 1993 Twilight of the panther Myakka River Publishing SarasotaFlorida USA

Aparicio J M J Ortego and P J Cordero 2006 What should we weigh toestimate heterozygosity alleles or loci Molecular Ecology 154659ndash4665

Bakker V J D F Doak G W Roemer D K Garcelon T J Coonan S AMorrison C Lynch K Ralls and R Shaw 2009 Incorporating ecologicaldrivers and uncertainty into a demographic population viability analysis for theisland fox Ecological Monographs 7977ndash108

Balloux F W Amos and T Coulson 2004 Does heterozygosity estimateinbreeding in real populations Molecular Ecology 133021ndash3031

Barone M A M E Roelke J Howard J L Brown A E Anderson and DE Wildt 1994 Reproductive characteristics of male Florida panthersmdashcomparative studies from Florida Texas Colorado Latin‐America andNorth‐American Zoos Journal of Mammalogy 75150ndash162

Bateson Z W P O Dunn S D Hull A E Henschen J A Johnson and LA Whittingham 2014 Genetic restoration of a threatened population ofgreater prairie‐chickens Biological Conservation 17412ndash19

Bauer H G Chapron K Nowell P Henschel P Funston L T B HunterD W Macdonald and C Packer 2015 Lion (Panthera leo) populations aredeclining rapidly across Africa except in intensively managed areas Pro-ceedings of National Academy of Sciences of the United States of America11214894ndash14899

Beier P 1995 Dispersal of juvenile cougars in fragmented habitat Journal ofWildlife Management 59228ndash237

Benson J F J A Hostetler D P Onorato W E Johnson M E Roelke SJ OrsquoBrien D Jansen and M K Oli 2011 Intentional genetic introgressioninfluences survival of adults and subadults in a small inbred felid populationJournal of Animal Ecology 80958ndash967

Benson J F P J Mahoney J A Sikich L E K Serieys J P Pollinger H BErnest and S P D Riley 2016 Interactions between demography geneticsand landscape connectivity increase extinction probability for a small popu-lation of large carnivores in a major metropolitan area Proceedings of theRoyal Society B Biological Sciences 28320160957

Beverly F 1874 The Florida panther Forest and Stream 3290Boyce M S C V Haridas C T Lee and the NCEAS Stochastic Demo-graphy Working Group 2006 Demography in an increasingly variable worldTrends in Ecology amp Evolution 21141ndash148

Brook B W J J OrsquoGrady A P Chapman M A Burgman H R Akcakayaand R Frankham 2000 Predictive accuracy of population viability analysis inconservation biology Nature 404385ndash387

Brys R and H Jacquemyn 2016 Severe outbreeding and inbreeding de-pression maintain mating system differentiation in Epipactis (Orchidaceae)Journal of Evolutionary Biology 29352ndash359

Burke J M and M L Arnold 2001 Genetics and the fitness of hybridsAnnual Review of Genetics 3531ndash52

Burnham K P 1993 A theory for combined analysis of ring recovery andrecapture data Pages 199ndash213 in J‐D Lebreton and P M North editorsMarked individuals in the study of bird population Springer‐Verlag NewYork New York USA

Burnham K P and D R Anderson 2002 Model selection and multimodelinference a practical information‐theoretic approach Second edition SpringerVerlag New York New York USA

Caswell H 2001 Matrix population models construction analysis and in-terpretation Sinauer Associates Sunderland Massachusetts USA

Charpentier M J M Setchell F Prugnolle L A Knapp E J Wickings PPeignot and M Hossaert‐McKey 2005 Genetic diversity and reproductivesuccess in mandrills (Mandrillus sphinx) Proceedings of the NationalAcademy of Sciences of the United States of America 10216723ndash16728

Cooch E and G C White editors 2018 Program MARK a gentle in-troduction lthttpwwwphidotorgsoftwaremarkdocsbookgt Accessed 7Apr 2016

Cooley H S R B Wielgus G M Koehler H S Robinson and B TMaletzke 2009 Does hunting regulate cougar populations A test of thecompensatory mortality hypothesis Ecology 902913ndash2921

Coulson T E A Catchpole S D Albon B J Morgan J M Pemberton TH Clutton‐Brock M J Crawley and B T Grenfell 2001 Age sex densitywinter weather and population crashes in Soay sheep Science 2921528ndash1531

Coulson T J Pemberton S Albon M Beaumont T Marshall J Slate FGuinness and T Clutton‐Brock 1998a Microsatellites reveal heterosis in reddeer Proceedings of the Royal Society B Biological Sciences 265489ndash495

Coulson T N S D Albon J M Pemberton J Slate F E Guinness and TH Clutton‐Brock 1998b Genotype by environment interactions in wintersurvival in red deer Journal of Animal Ecology 67434ndash445

Cox D 1972 Regression models and life‐tables Journal of the Royal StatisticalSociety Series B Statistical Methodology 34187ndash220

Creel S 2006 Recovery of the Florida panthermdashgenetic rescue demographic rescueor both Response to Pimm et al (2006) Animal Conservation 9125ndash126

Crnokrak P and D A Roff 1999 Inbreeding depression in the wild Heredity83260ndash270

Crow J 1948 Alternative hypotheses of hybrid vigor Genetics 33477ndash487

van de Kerk et al bull Florida Panther Population and Genetic Viability 31

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 7: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

persistence of the population will likely necessitate periodic geneticmanagement interventions Therefore our final objective was to 4)evaluate genetic and population‐dynamic consequences of variedgenetic introgression strategies and from these identify manage-ment strategies that are affordable and effective at improvingprospects of Florida panther persistence To achieve this objectivewe extended the individual‐based population model to track in-dividual genetics based on empirical allele frequencies taking ad-vantage of the long‐term demographic and genetic data that havebeen collected on Florida panthers since 1981 We used individualheterozygosity to inform vital rates based on empirically estimatedrelationships between individual heterozygosity and demographicparameters We then simulated population trajectories trackedindividual heterozygosity over time and estimated quasi‐extinctiontimes and probabilities without introgression and with introgres-sion for a range of intervals and number of immigrants perintrogression event We compared the genetic and demographicbenefits of implementing alternative genetic introgression scenariosand ranked them by the level of improvement in population per-sistence parameters and estimated costs because funding is a per-sistent challenge to conservation planningManagement of imperiled species is a complex process invol-

ving many stakeholders and plagued by layers of uncertainties(Runge 2011) It would therefore seem that management ofthreatened and endangered species would be a perfect case foradaptive resource management approach (ARM) because ARMfocuses on structured decision making for recurrent decisionsmade under uncertainty (Williams et al 2002 Runge 2011)Because it combines structured decision making with learningprudent application of ARM reduces uncertainty over time andcan lead to optimal management actions (Williams and Brown2012 2016) Key ingredients of a successful adaptive manage-ment program include stakeholder engagement clearly definedand mutually agreed‐upon objectives alternative managementactions and objective decision rules (Nichols et al 2007Williams and Brown 2012 2016) Unfortunately Florida pan-ther stakeholders have divergent views about the state of thesystem (eg panther abundance) and it is a challenge to havemutually agreed‐upon objectives management actions or deci-sion rules acceptable to all or most stakeholders AlthoughFlorida panther conservation efforts would be best served withinthe ARM framework in the long run impediments to its im-plementation (Runge 2011 Williams and Brown 2012) areunlikely to be overcome in the near future

STUDY AREA

The breeding population of Florida panthers mainly persists as asingle population South of the Caloosahatchee River (approxi-mately 267133degN latitude 815566degW longitude see Fig 1)One exception is a female that was documented just north of theCaloosahatchee River in 2016 (and subsequently documentedwith kittens in 2017) by FWC the first validation of a femalepanther north of the River since 1973 (FWC unpublisheddata) The breeding range in South Florida is topographicallyflat has a subtropical climate and is characterized by permanentand ephemeral wetlands influenced by rains from May throughOctober (Duever et al 1986) The study area contains a variety

of wildland land cover types including hardwood hammockscypress forests pine flatwoods freshwater marshes prairies andgrasslands (Davis 1943) and land characterized by human ac-tivity including citrus groves croplands pastureland rockmining and residential developments (Onorato et al 2011)The area is intersected by a multitude of roads that fragmentpanther habitat Most notable among these is Interstate 75(Alligator Alley) which was expanded from 2 to 4 lanes in 1990when fencing and wildlife crossings were constructed with theintention of minimizing harmful effects of road expansion onlocal wildlife (Foster and Humphrey 1995) A large portion ofthe area that comprises South Florida is under public ownershipmainly as federal or state lands Everglades National Park BigCypress National Preserve Florida Panther National WildlifeRefuge Picayune Strand State Forest and Fakahatchee StrandPreserve State Park alone encompass 9745 km2 though not allof the area is considered suitable for panthers (eg large ex-panses of sawgrass marshes in the western Everglades) Frakeset al (2015) delineated 5579 km2 of adult panther habitat inSouth Florida 1399 km2 of which are in private ownership Amajority of these private lands are located in the northern extentof the breeding range Although some private lands may beprotected (eg conservation easements) other areas are sus-ceptible to incompatible land uses such as rock mining or re-sidential developments The breeding range in South Florida isbounded to the east and west by large metropolitan areas in-clusive of Miami‐Fort Lauderdale‐Pompano Beach and CapeCoral‐Fort Myers‐Naples‐Marco Island‐Immokalee respec-tively that are populated by gt6000000 people (httpswwwcensusgov accessed 24 Jan 2019) These characteristics of thestudy area highlight the recovery challenges faced by panthers

METHODS

Field MethodsSince 1981 Florida panthers have been captured radiocollared andtracked by the FWC and National Park Service staff (Table 1)Capture and handling protocols followed guidelines of theAmerican Society of Mammalogists (Sikes 2016) Livestock Pro-tection Company (Alpine TX USA) provided trained hounds andhoundsmen for captures Teams either treed or bayed panthers onthe ground and then darted them with a 3‐ml compressed‐air dartfired from a CO2‐powered rifle Immobilization drugs have variedduring the tenure of the project but most recently teamsimmobilized panthers with a combination of ketamine HCl (10mgkg Doc Lanersquos Veterinary Pharmacy Lexington KY USA) andxylazine HCl (1mgkg Doc Lanersquos Veterinary Pharmacy) Fol-lowing immobilization teams lowered treed panthers to the groundby a rope or caught panthers with a net in some cases they used aportable cushion (McCown et al 1990) to further mitigate theimpact of a fall Capture teams administered propofol (PropoFlotradeAbbott Laboratories Abbott Park IL USA) intravenously either asa bolus or continuous drip to maintain anesthesia They adminis-tered midazolam HCl (003mgkg) intramuscularly or intravenouslyto supplement anesthesia in some panthers Panthers recovered in ashaded area away from water In some cases capture teams reversedxylazine HCl with yohimbine HCl (Yobinereg Lloyd IncShenandoah IA USA) at 25 its recommended dose

van de Kerk et al bull Florida Panther Population and Genetic Viability 9

Capture teams determined sex and age of panthers marked themwith an ear tattoo and a subcutaneous passive integrated trans-ponder (PIT‐tag) and fitted them with a very high frequency(VHF) or global positioning system (GPS) radio‐collar equippedwith a mortality sensor The GPS collars were programmed toattempt to collect a position at a variety of time intervals(range= 1ndash12 hr) depending on objectives of concurrent studiesobservers routinely tracked panthers with VHF radiocollars from afixed‐wing aircraft 3 times per week When a radio‐collar trans-mitted a mortality signal or when a location did not change forseveral days observers investigated the site to establish the pan-therrsquos fate When observers found a dead panther they assessed thecause of death based on an examination of the carcass and sur-rounding area if possible They then transported carcasses to anexperienced veterinarian for necropsy and a histopathological ex-amination to attempt to assign the cause of death Starting in 1995observers continually assessed successive locations of radio‐collaredfemales to determine if they initiated denning behavior Three to 4fixes at the same location (VHF collars) or successive returns to thesame location over a weeklong period (GPS collars) served as anindicator of a possible den Observers then located dens moreprecisely via triangulation with ground telemetry They visited denswhen the dam left the site typically to make a kill Teams capturedkittens by hand captured kittens ranged in age from 4 to 35 dayspost‐partum Teams counted sexed weighed dewormed andpermanently marked kittens with a subcutaneous PIT‐tag

Genetic AnalysisTeams collected blood tissue or hair from the majority ofcaptured kittens subadult and adult Florida panthers for DNAsamples The United States Department of Agriculture ForestService National Genomics Center for Wildlife and FishConservation (Missoula MT USA) processed samplesTechnicians used Qiagen DNeasy Blood and Tissue Kits(Qiagen Valencia CA USA) to extract whole genomic DNAfrom blood and tissue samples and used a slight modification(overnight incubation of sample in lysis buffer and proteinase Kon a rocker at 60deg elution of DNA in 50 μl of buffer) of theQiagen DNA extraction protocol (Mills et al 2000) to extractDNA from hair samples Technicians used a panel of 16 mi-crosatellite loci (Fca090 Fca133 Fca243 F124 F37 Fca075Fca559 Fca057 Fca081 Fca566 F42 Fca043 Fca161

Fca293 Fca369 and Fca668) identified by Menotti‐Raymondet al (1999) which had been previously applied to Floridapanther samples (Johnson et al 2010) Technicians amplifiedDNA samples in 10‐microl reaction volumes that included 10 microl ofDNA 1times reaction buffer (Applied Biosystems Life Technolo-gies Corporation Grand Island NY USA) 20 mM of MgCl2200 microM of each dNTP 1 microM of reverse primer 1 microM of dye‐labeled forward primer 15 mgml of bovine serum albumin(BSA) and 1U Taq polymerase (Applied Biosystems) Thepolymerase chain reactions (PCRs) included a thermal profile of94degC for 5 minutes followed by 36 cycles of 94degC for 60 seconds55degC for 60 seconds and 72degC for 30 seconds Techniciansvisualized the resulting PCR products on a LI‐COR DNAanalyzer (LI‐COR Biotechnology Lincoln NE USA) Tech-nicians collected genotypes from hair samples (n= 16) using amultitube approach (Taberlet et al 1996) error‐checked geno-types using Program DROPOUT (McKelvey and Schwartz2005) and combined them with genotypes from tissue andblood samples (n= 487) We evaluated all 16 loci for variousdescriptive statistics for the radio‐collared sample of adult andsubadult panthers We excluded genotypes of kittens from thatanalysis because related individuals are known to result in theoverrepresentation of certain alleles (Marshall et al 1998) Wedetermined the number of alleles observed heterozygosity andexpected heterozygosity using GenAlex (Peakall and Smouse2006 2012) We assessed allelic richness at each locus in Fstat293 (Goudet 1995) We assessed conformance of genotypedata from these 16 loci to HardyndashWeinberg assumptionslinkage disequilibrium and null alleles (Appendix A availableonline in Supporting Information) The Genomics Center alsoprovided us with genotype data from the same 16 loci for 41pumas from western Texas the source population for thegenetic rescue programEvidence of inbreeding depression in wild populations is

commonly determined on the basis of heterozygosityndashfitnesscorrelations (Silva et al 2005 Ortego et al 2007 Mainguyet al 2009 Johnson et al 2011) so we also estimated theheterozygosity of each individual panther We calculatedhomozygosity by loci (HL) which varies between 0 (all lociheterozygous) and 1 (all loci homozygous Aparicio et al 2006)using the Rhh package (Alho et al 2010) in Program R (R CoreTeam 2016) A microsatellite locus has more weight in HL

Table 1 Number of Florida panthers by age sex and genetic ancestry categories used in subsequent demographic analyses during the study period (1981ndash2013) inSouth Florida USA

Kittensa Subadults and adultsb

Males Females Total Males Females Total

PIT‐taggedc 215 180 395 Radiocollaredc 108 101 209Recovered 11 4 15 Subadult 72 50 122Litter failed 47 38 85 Prime adult 65 95 160Recaptured 25 30 55 Old adult 13 20 33Canonical 27 20 47 Canonical 43 29 72F1 admixed 6 12 18 F1 admixed 2 6 8Other admixed 130 105 235 Other admixed 45 53 98

a Kitten samples included those that were PIT‐tagged recovered (ie panthers that were found dead) part of a failed litter or recaptured as subadults or adultsb Radiocollared subadult and adult panthers are presented as sample size within each age class and ancestry classc Combined number of panthers of different ancestries is less than the number radiocollared or tagged with a subcutaneous passive integrated transponder (PIT‐tagged) because we did not conduct genetic sampling on all panthers Combined number of panthers in different age classes is greater than the number ofradiocollared panthers because some individuals were counted in ge2 age classes as animals aged

10 Wildlife Monographs bull 203

when it is more informative (ie has more alleles) and has aneven distribution of allele frequencies For our demographicanalyses we calculated individual heterozygosity for each in-dividual panther as 1 minusHL As do most indices of individualheterozygosity HL takes into account the average allele fre-quency in the population (Aparicio et al 2006) Thus duringsimulations the allele frequency in the population changesthroughout an individualrsquos lifetime which causes its hetero-zygosity also to change This is not biologically plausible becausegenetic properties are retained throughout an individualrsquos life-time Therefore for use in model simulations we simply cal-culated heterozygosity for each individual as the proportion ofheterozygous lociWe also used genotype data to implement a Bayesian clus-

tering analysis in Program STRUCTURE (Pritchard et al2000) to infer ancestral clusters of panthers This method uses aMarkov chain Monte Carlo (MCMC) method to determine thenumber of genetic clusters (K) in the sample while also pro-viding an individualrsquos percentage of ancestry (q values) allocatedto each cluster Runs for this analysis included a 100000MCMC burn‐in period followed by 500000 MCMC iterationsfor 1ndash10 ancestral genetic clusters (K= 1 to 10) with 10 re-petitions for each K To determine the most likely number of Kgenetic clusters we used the logarithm of the probability of thedata (log Pr(D)∣K Pritchard et al 2000) and estimates of ΔK(Evanno et al 2005) in Program STRUCTURE HAR-VESTER (Earl and VonHoldt 2011) We then used q valuesprovided in runs for the best K to assign individual panthers aseither canonical (in most cases ge90 pre‐introgression ancestrysee Johnson et al 2010) or admixed (lt90 pre‐introgressionancestry) We recognized admixed panthers that were the off-spring of matings between a Texas female puma and a canonicalmale panther as F1 admixed panthers for analyses

Demographic ParametersSurvival of kittensmdashMost kittens PIT‐tagged at den sites were

never encountered again A small proportion of the PIT‐taggedkittens were recovered dead or were recaptured as subadults oradults and radiocollared so that their fate could be monitoredWe also identified instances of complete litter failure when afemale died within 9 months after giving birth (the minimumage of independence for kittens) or when a female wasdocumented to have another litter less than 12 months aftergiving birth (the minimum age of independence plus a 3‐monthgestation period Hostetler et al 2010) Thus our data setconsisted of 1) live recaptures and subsequent radio‐tracking ofpanthers of various ages that had been PIT‐tagged as kittens 2)recovery of dead panthers of various ages that had been PIT‐tagged as kitten 3) live captures and subsequent radio‐trackingof other panthers and 4) instances of complete litter failure(Table 1) We analyzed these data using Burnhamrsquos live‐recapture dead‐recovery modeling framework (Burnham 1993Williams et al 2002) to estimate survival of kittens (0ndash1‐yearold) and to test covariate effects on this parameter We includeddata on individuals encountered dead or alive at an older age inour analyses because their encounter histories also providedinformation on kitten survival We used a live‐dead data inputformat coded with a 1‐year time step (Cooch and White 2018)

All known litter failures occurred within the first year of thekittensrsquo lives so we treated all litter failures as recoveries withinthe first year We treated panthers that would have died if wehad not removed them to captivity as having died on the date ofremoval Data preparation and analytical approach are describedin more detail by Hostetler et al (2010)In addition to survival probability (S) the Burnham model

incorporates parameters for recapture probability (p) recoveryprobability (r) and site fidelity (f) We set r and p to 10 forradio‐tracked individuals because their status was known eachyear We also fixed f at 10 for all individuals because the re-capture and recovery areas were the same and encompassed theentire range of the Florida panther Based on findings of Hos-tetler et al (2010) we used a base model that 1) allowed survivalto differ between kittens and older panthers 2) allowed survivalto differ between sexes and among age classes (females ages 1and 2 ge3 males ages 1 2 and 3 ge4) 3) constrained recaptureprobability to be the same for all uncollared panthers and 4)allowed recovery rates to differ between kittens and uncollaredolder panthers Even with our additional data this model re-mained the best fit for a range of models for recapture recoveryand survival of kittens subadults and adults of various age‐classcategories (Hostetler et al 2010)Subadult and adult survivalmdashTo estimate survival for panthers

ge1 year of age we analyzed radio‐tracking data using a Coxproportional hazard‐modeling framework (Cox 1972 Therneauand Grambsch 2000) We followed procedures outlined byBenson et al (2011) for data preparation and analysis Brieflywe right‐censored panthers on the last day of observation beforetheir collar failed When an individual aged into a new age classwhile being tracked we created 2 data entries 1 ending the dayit entered the new age class the other starting on that day Weused the FlemingndashHarrington method to generate survivalestimates from the Cox analysis and the Huber sandwichmethod to estimate robust standard errors (Therneau andGrambsch 2000 Benson et al 2011)We considered 3 age classes subadults (1ndash25 yr for females and

1ndash35 yr for males) prime adults (25ndash10 yr for females and35ndash10 yr for males) and old adults (gt10 yr old for both sexes)after Benson et al (2011) We estimated overall survival using abase model that allowed survival to differ between subadultfemales subadult males prime adult females prime adult malesand old adults of either sex (old age was additive with sex other-wise age and sex were interactive) This model provided the best fitto the data relative to a range of models considering differences insurvival between prime‐adult and older‐adult age classes both ad-ditively and interactively with sex (Benson et al 2011)To examine whether survival rates observed in recent years

differed from those observed immediately following introgres-sion we also estimated age‐specific survival rates for 2 nearlyequal periods of time immediately post‐introgression (1995ndash2004) and the more recent period (2005ndash2013) For theseanalyses we separated the radio‐tracking data into the 2 periodsand estimated survival rates for each period separately using thebase model Similar analyses for kitten survival were not possiblebecause of data limitationsProbability of reproductionmdashWe used telemetry data obtained

from radio‐tracked females that reproduced at least once to

van de Kerk et al bull Florida Panther Population and Genetic Viability 11

estimate annual probability of reproduction of subadult prime‐adult and old‐adult females using complementary logndashlogregression (Agresti 2013) following methods described indetail by Hostetler et al (2012) Briefly we modeled theprobability of a female panther giving birth (b) as a function ofcovariates as

γ= minus [minus ( )]b z1 exp exp i i

where zi is a row vector of covariates (see below for covariateinformation) for female panther i and γ is a column vector ofmodel coefficients (Appendix B Table B2) To account fordifferences in observation time we included log(m) as an offsetin the model where m is the number of months a femalepanther was radio‐tracked (Hostetler et al 2012)

Reproductive outputmdashTo estimate the number of kittensproduced by reproductive females we used cumulative logitregression (Min and Agresti 2005 Agresti 2013) Weconsidered a model that includes J categories for the numberof offspring with j denoting the number of offspring ( j= 1 2hellip J and J= 6) provided that a female reproduces Weconsidered J= 6 which is greater than the largest litter sizeobserved in our study (4 kittens) because females can produce 2or more litters per year if litters fail We modeled the probabilitythat a female i produces a litter of size at most j (Yi) as

le⎧

⎨⎩

δ( ) = == hellip minus

=

( )+ θ βminus +Y jj J

j JPr

1 2 1

1i ij e

1

1 xj i

where θj is the intercept for the annual number of kittens pro-duced by a female i being at most j xi is a row vector of cov-ariates for female panther i (see below) and β is a column vectorof model coefficients The probability that Yi will be exactly j isgiven by

⎧⎨⎩

πδ

δ δ( = ) = =

=

minus = hellip( minus )Y j

jj J

Pr 1

2 i ij

i

ij i j

1

1

Finally the expected annual number of kittens produced byfemale i (νi) is given by

sumν π==

j i

j

J

ij

1

Additional details regarding the estimation and modelingof the annual number of kittens produced can be found inHostetler et al (2012)

CovariatesmdashIn addition to sex and age we tested for the effect ofabundance genetic ancestry and individual heterozygosity on theaforementioned demographic parameters (model coefficients forkitten and adult survival are given by η and ι respectively AppendixB Table B2) We used a minimum population count (MPC) basedon radio‐tracking data and field evidence of subadult and adult (ieindependent age) panthers as an index of abundance (McBride et al2008) For these analyses we did not use the population sizeestimates reported by McClintock et al (2015) because unlike theminimum counts they did not cover the entire study period

(McBride et al 2008 R T McBride and C McBride RancherrsquosSupply Incorporated unpublished report Fig 2)To test if demographic parameters differed depending on an-

cestry we considered 3 ancestry models dividing the panthers into1) F1 admixed and all other panthers 2) canonical and admixed(including F1 admixed) panthers and 3) F1 admixed canonicaland other admixed panthers We quantified genetic diversity usingindividual heterozygosity as described previously We tested for theeffect of genetic covariates using a subset of kittens that were ge-netically sampled We calculated model‐averaged estimates ofmodel parameters on the scale in which they were estimated(Appendix B available online in Supporting Information)

Cause‐specific mortalitymdashWe performed cause‐specificmortality analysis only on dead radio‐collared panthers becausedetection rates of uncollared panther mortalities are highlycorrelated with the cause of death (eg vehicle collision)Following Benson et al (2011) we attributed each mortality to1 of 4 causes 1) vehicle collision 2) intraspecific aggression 3)other causes (including known causes such as diseases injuriesand infections unrelated to the first 2 causes) and 4) unknown(ie mortalities for which we could not assign a cause) Wethen used the non‐parametric cumulative incidence functionestimator a generalization of the staggered‐entry KaplanndashMeiermethod of survival estimation to estimate cause‐specificmortality rates for male and female panthers in different ageclasses (Pollock et al 1989 Heisey and Patterson 2006 Bensonet al 2011) We compared cause‐specific mortality rates betweensexes age classes and ancestries

Statistical inferencemdashWe used an information‐theoreticapproach (Akaikersquos information criterion [AIC]) for modelselection and statistical inference (Burnham and Anderson 2002)For the analysis of kitten survival we adjusted the AIC foroverdispersion and small sample size (QAICc details in Hostetleret al 2010) We performed all statistical analyses in Program R (RCore Team 2016) We used Burnhamrsquos live recapturendashdead

0

100

200

300

1985 1990 1995 2000 2005 2010Year

Indi

vidu

als Method

MPC

MVM

Figure 2 The Florida panther minimum population count (MPC) and populationsize estimated using motor vehicle collision mortalities (MVM) during our studyperiod Minimum population counts are from McBride et al (2008) and R TMcBride and C McBride (Rancherrsquos Supply Incorporated unpublished report)Estimates based on MVM are from McClintock et al (2015) The solid vertical lineindicates the year of genetic introgression

12 Wildlife Monographs bull 203

recovery model (Burnham 1993) to estimate and model kittensurvival using the RMark package version 2113 (Laake 2013) asan interface for Program MARK (White and Burnham 1999)We implemented the Cox proportional hazard model for esti-

mation and modeling of survival of subadults and adults using the Rpackage SURVIVAL (Therneau and Grambsch 2000) We per-formed cause‐specific mortality analyses using an R version of S‐PLUS code provided by Heisey and Patterson (2006)To account for model selection uncertainty we calculated

model‐averaged parameter estimates across all models using theR package AICcmodavg (Mazerolle 2017) using AIC weights asmodel weights (Burnham and Anderson 2002) We used modelswithout the categorical ancestry variables for model averagingand estimated parameters based on the average value for thecontinuous covariates (individual heterozygosity and abundanceindex) Because only a subset of the kittens was geneticallysampled we estimated 2 kitten survival rates First we estimatedkitten survival based on the data set that included all kittens(overall kitten survival) Second we estimated kitten survivalusing a subset of the data that only included kittens that weregenetically sampled (survival of genetically sampled kittens)

Matrix Population ModelsWe investigated Florida panther population dynamics andpersistence using 2 complementary modeling frameworks ma-trix population models (Caswell 2001) and IBMs (Grimm andRailsback 2005 McLane et al 2011 Railsback and Grimm2011 Albeke et al 2015) Matrix population models offer aflexible and powerful framework for investigating the dynamicsand persistence of age‐ or stage‐structured populations (egCaswell 2001 Robinson et al 2008 Kohira et al 2009 Hunteret al 2010 Hostetler et al 2012) With a fully developed theorybehind them these models also serve as a check for resultsobtained from simulation‐based models such as IBMs We usedan age‐structured‐matrix population model for deterministic andstochastic demographic analyses and for preliminary investiga-tions of Florida panther population viability (Caswell 2001Morris and Doak 2002)For deterministic and stochastic demographic analyses we

parameterized a female‐only 19 times 19 age‐specific populationprojection matrix of the form

⎢⎢⎢⎢⎢

⋯ ⋯

⋯ ⋯

⋱ ⋱ ⋮

⋮ ⋱ ⋱ ⋱ ⋮

⎥⎥⎥⎥⎥

( )=t

F F F

P

P

P

A0 0

0

0 0 0

t t t

t

t

t

1 2 19

1

2

18

where Fi and Pi are the age‐specific fertility and survival ratesrespectively and ( )tA is a time‐varying (or stochastic) popula-tion projection matrix We assumed that the longevity and ageof last reproduction of female Florida panthers was 185 yearswe based this assumption on the observation that the oldestfemale in our study was 186 years old Florida panthers re-produce year‐round thus we used birth‐flow methods to esti-mate Fi and Pi values following Caswell (2001) and Hostetleret al (2013) We estimated Pi as

= +P S S i i i 1

where Si is the age‐specific survival probability We estimated Fi as

⎛⎝

⎞⎠

=+ +F S

m Pm

2i k

i i i 1

where Sk is the kitten survival probability and mi is the age‐specific fecundity rate which was estimated as

ν=m b f i i i k

where bi is the breeding probability for a female in age class iduring time step t νi is the annual number of kittens producedby a breeding female in age class i and fk is the proportionfemale kittens at birth (assumed to be 05)We estimated the finite deterministic (λ) and stochastic annual

population growth rate (λs) and stochastic sensitivities and elasti-cities following methods described in detail by Caswell (2001) Toestimate λs we assumed identically and independently distributedenvironments (Caswell 2001 Morris and Doak 2002 Haridas andTuljapurkar 2005) and used a parametric bootstrapping approachto obtain a sequence of demographic parameters for 50000 ma-trices We then estimated log(λs) from this sequence of matrices(Tuljapurkar 1990 Caswell 2001) and exponentiated it to obtain λsWe calculated the sensitivity and elasticity of λs to changes in lower‐level vital demographic rates as per Caswell (2001) and Haridas andTuljapurkar (2005)For population projection and population viability analysis (PVA)

simulations we expanded the population projection matrix into a34 times 34 2‐sex age‐structured matrix to include female and maleFlorida panthers (Caswell 2001 Hostetler et al 2013) Malescontribute to the population only through survival in this modelingframework We assumed maximum longevity for males to be 145years of age because the oldest male in our study was 144 years oldThe population projection matrix was of the following form

⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢

⋯ ⋯ ⋯ ⋯

⋯ ⋯ ⋯ ⋯

⋱ ⋱ ⋮ ⋮ ⋱ ⋱ ⋮

⋮ ⋱ ⋱ ⋱ ⋮ ⋮ ⋱ ⋱ ⋮

⋯ ⋯ ⋯

⋯ ⋯ ⋯ ⋯

⋯ ⋯ ⋱ ⋱ ⋮

⋮ ⋮ ⋱ ⋱ ⋮ ⋱ ⋱ ⋮

⋯ ⋯ ⋯

⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥

( )=

( ) ( ) ( )

( )

( )

( )

( ) ( ) ( )

( )

( )

t

F N F N F N

P N

P N

P N

M N M N M N

Q N

Q N

A n

0 0

0 0 0 0

0

0 0 0 0 0

0 0

0 0 0

00 0 0 0 0

t t t

t

t

t

t t t

t

t

1 2 19

1

2

18

1 2 19

1

14

where Pi and Qi are age‐specific survival rates of female and maleFlorida panthers respectively and Fi and Mi are the rates atwhich female and male kittens are produced by females of ageclass i respectively Here the population projection matrix isboth time‐varying and density‐dependent (Caswell 2001)We incorporated the influence of parameter uncertainty en-

vironmental stochasticity and density dependence on Floridapanther population dynamics and persistence following the ap-proach outlined by Hostetler et al (2013) we used the sameapproach in the IBM analysis which is described below UnlikeIBMs matrix models do not intrinsically include demographicstochasticity so we explicitly included the influence of

van de Kerk et al bull Florida Panther Population and Genetic Viability 13

demographic stochasticity by simulating the fates of individualsas described in Caswell (2001)Population projections using matrix models necessitate a

vector of initial sex‐ and age‐specific abundance Because suchdata are currently unavailable for the Florida panther we esti-mated it based on the stable sex and age distribution and MPCin 2013 Using point estimates of the demographic parameterswe estimated the stable sex and age distribution for the 2‐sexdeterministic and density‐independent population projectionmatrix We then multiplied the stable sex and age distributionby the MPC in 2013 to get the starting population vector n(0)We projected the population over time as n(t+ 1)=A(tn)n(t)where A(tn) indicates the time‐specific density‐dependentpopulation projection matrix (Caswell 2001)We had 2 indices of abundance available the MPC (McBride

et al 2008) and population size estimated using motor vehiclecollision mortalities (MVM McClintock et al 2015) These 2indices are substantially different because MPC does not ac-count for imperfect detection whereas MVM accounts forimperfect detection and provides an estimate of population size(Fig 2) Therefore we examined 2 scenarios one using densitydependence based on MPC and the other using density de-pendence based on MVM We ran 1000 bootstraps of para-meter values and 1000 simulations per bootstrap for each sce-nario We projected population trajectories over 200 years andthen estimated population growth rates probabilities ofquasi‐extinction and time until quasi‐extinction and comparedthese with the results obtained from the IBM (see below) Weran scenarios with a quasi‐extinction threshold of 10 and 30panthers We used the mean of probabilities of quasi‐extinction(PQE) across simulations as the point estimate for PQE and the5th and 95th percentiles across simulations as the 90 con-fidence interval Because the confidence intervals werequantile‐based but the point estimates were not it is possiblewith skewed distributions for the point estimates to be outsidethe confidence interval We implemented the matrix

population models in the R computing environment (R CoreTeam 2016)

IBMs and PVABecause of well‐developed theory ease of implementation andthe modeling flexibility that they offer matrix populationmodels have become the model of choice for investigating thedynamics and persistence of age‐ and stage‐structured popula-tions (Caswell 2001) However it is difficult to incorporateattributes of individuals such as genetics and behavior usingmatrix models Quantifying the effects of genetic erosion on theFlorida panther population dynamics and persistence andevaluating benefits and costs of alternative genetic managementscenarios were important goals of our study Because IBMs relyon a bottom‐up approach that begins by explicitly consideringindividuals (McLane et al 2011 Railsback and Grimm 2011Albeke et al 2015) it is possible to represent genetic attributesof each individual and to track genetic variation over time Weused IBMs as the primary modeling framework in this studybecause they allowed for explicit consideration of genetics bothat individual and population levels and population viabilityassessment under alternative genetic management scenariosWe developed an IBM (Grimm 1999 Grimm and Railsback

2005 McLane et al 2011 Railsback and Grimm 2011 Albekeet al 2015) tailored to the life history of the Florida panther(Fig 3) We simulated every individual from birth until deathbased on a set of rules describing fates (eg birth and death) andattributes (eg genetics) of individuals at each annual time stepAs in other IBMs population‐level processes (eg populationsize and genetic variation) emerged from fates of and interac-tions among individuals (Grimm 1999 Railsback and Grimm2011) We developed an overview design concepts and details(ODD) protocol (Grimm et al 2006 2010) to describe ourIBM (Appendix C available online in Supporting Information)and provide pseudocodes for individual‐based simulations(Appendix D available online in Supporting Information)

For each panther at time = t

Kitten (lt1 yr)

Reproduce

Survive

Kitten(s) born

Kitten

Sex and age class

Subadult male

Prime adult male

Old adult male

Subadult female

Prime adult female

Old adult female

Age one year

No Male

Female

Yes

No

Yes

t=t+1

Yes

lt35 yrs

35-10 yrs

gt10 yrs

lt25 yrs

25-10 yrs

gt10 yrs

Figure 3 Schematic representation of Florida panther life history We tailored the individual‐based population model to represent the life‐history patterndepicted here

14 Wildlife Monographs bull 203

Dynamics and persistence of populations are influenced bymany factors such as demographic and environmental sto-chasticity and density dependence these effects and parametricuncertainty must be considered in population models whenpossible (Caswell 2001 Boyce et al 2006 Bakker et al 2009Hostetler et al 2013) Because by definition IBMs simulate thelife history of individuals in a population they inherently in-corporate the effect of demographic stochasticity Our analysesrevealed that survival of kittens subadults and adults and theprobability of reproduction varied over time so we incorporatedenvironmental stochasticity in the model for these variablesfollowing Hostetler et al (2013) Likewise we found evidenceof density‐dependent effects on kitten survival Because theMPC might be an underestimate of population size (McBrideet al 2008) we performed simulations using both the MPC andMVM abundance indicesTo account for model selection and parametric uncertainty we

used a Monte Carlo simulation approach For each bootstrapsample we selected a model based on the AIC (or QAICc)weights We then sampled the intercept and slope for kittensurvival (on the logit scale) subadult and adult survival (on thelogit scale with log‐hazard effect sizes) and probability of re-production (on the complementary logndashlog scale Appendix B)from multivariate normal distributions with mean vectors equalto the estimated parameters and variance‐covariance matricesequal to the sampling variance‐covariance matrices We thentransformed the results to nominal scale and converted the es-timates to age‐specific survival and reproduction parameters asdescribed in Hostetler et al (2013)We ran 1000 simulations for 1000 parametric bootstraps for

200 years for the MPC and the MVM scenario for both thestochastic 2‐sex matrix‐population model and the IBM Wetracked the population size at each time step and estimatedquasi‐extinction probabilities and times based on the populationtrajectory We considered the population to be quasi‐extinct ifindividuals of only 1 sex remained or if the population size fellbelow critical thresholds set at 10 and 30 individuals We alsoestimated expected time to quasi‐extinction for both thresholdsdefined as the mean and median time until a population be-comes quasi‐extinct conditioned on quasi‐extinction We im-plemented the IBM using the RNetLogo package version 10ndash2(Thiele et al 2012 Thiele 2014) as an interface for the IBMplatform NetLogo (Wilensky 1999)

Sensitivity AnalysisAn important component of demographic analysis is sensitivityanalysis which is the quantification of the sensitivity of amodelrsquos outcome to changes in the input parameters (Caswell2001 Bakker et al 2009 Thiele et al 2014) Using an IBM weperformed global sensitivity analyses for 2 (ie MPC andMVM) density‐dependent scenarios to assess how sensitivequasi‐extinction times and probabilities were to changes (anduncertainties) in the estimated demographic rates We usedLatin hypercube sampling to sample parameters from the entireparameter space (Marino et al 2008 Thiele et al 2014) Wesampled intercept and slope for kitten survival (on logit scale)subadult and adult survival (on logit scale with log‐hazard effectsizes) and probability of reproduction (on complementary

logndashlog scale) from normal distributions We sampled theprobability distribution for the number of kittens producedannually (on the real scale) from a Dirichlet distribution themultivariate generalization of the beta distribution (Kotz et al2000) We sampled 1000 different parameter sets and ran 1000simulations for each scenario We calculated the probability ofquasi‐extinction within 200 years for each density‐dependencescenario (MPC or MVM) using a quasi‐extinction threshold of10 individuals and then calculated the partial rank correlationcoefficient between each of those and each parameter trans-formed to the real scale (Bakker et al 2009 Hostetler et al2013) Partial rank correlation coefficients quantify the sensi-tivity of the probability of quasi‐extinction to input variables(ie survival and reproductive parameters) with higher absolutevalues indicating stronger influence of that variable on quasi‐extinction probabilities

Genetic ManagementOne of our goals was to estimate the benefits (improvements ingenetic variation demographic parameters and populationgrowth and persistence parameters) and costs (financial costs ofcapture transportation quarantine release and monitoring ofintroduced panthers) of genetic management To achieve thisobjective we extended the IBM to include a genetic componentTo simulate individual‐ and population‐level heterozygosity overtime we assigned each panther alleles for 16 microsatellite locibased on the empirical allele frequency in the population (esti-mated from genetic samples collected during 2008ndash2015) whenwe initialized the model (Pierson et al 2015) At each time stepwe simulated Mendelian inheritance for kittens produced suchthat each kitten randomly inherited alleles from its mother andfrom a male panther randomly chosen to have putatively siredthe litter We did not explicitly force inbreeding to occur butthe probability of inbreeding naturally increased as the popula-tion size decreasedIntrogression strategiesmdashWe modeled genetic introgression as a

result of the release of 2‐year‐old female pumas from Texas intothe simulated Florida panther population For the geneticintrogression implemented in 1995 Texas females between 15and 25 years of age were released because females at that agehave lower mortality rates (Benson et al 2011) higherreproductive potential and also because it is much easier todetect with certainty the successful reproduction by females thanby males The ability to more closely monitor reproduction andthe birth of kittens is an important consideration in reducing theprobability of a genomic sweep of Texas alleles into the Floridapanther population Wildlife managers removed the last 2surviving Texas females from the wild population in Florida in2003 to reduce the possibility of genomic sweep To simulatethis strategy in the model we removed any Texas females thatwere still alive 5 years after each introgression event Weassigned Texas females alleles based on the allele frequency ofthe Texas population (based on genotypes from 48 samplescollected in west Texas that was inclusive of the pumas releasedin South Florida in 1995) when they entered the Florida pantherpopulation We could not estimate demographic rates separatelyfor the released Texas females because of small sample size (only8 females were released in 1995) therefore we assumed that

van de Kerk et al bull Florida Panther Population and Genetic Viability 15

survival and reproductive rates and the effects of covariates onthese rates for the released Texas females were identical to thoseof female Florida panthersThe impact of genetic introgression depends on the frequency

of introgression events and the number of individuals introducedat each attempt Thus we defined introgression strategies basedon the number of Texas females to be released (0 5 10 or 15)and the interval between genetic introgressions (10 20 40 and80 yr) for a total of 13 strategies We simulated each manage-ment strategy with density dependence estimated using theMPC and MVM abundance indices

We sampled 1000 parameter sets and for each of these setswe ran all introgression strategies 1000 times for 200 years Weapplied the same parametric uncertainty and environmentalstochasticity across all management scenarios to allow for bettercomparison of introgression strategies At each time step werecorded the projected population size and average individualheterozygosity We calculated the heterozygosity for each in-dividual at birth based on the alleles it inherited This individualheterozygosity then informed the survival rate of that individualbased on the empirically estimated relationship between in-dividual heterozygosity and age‐specific survival rates Hetero-zygosity did not change throughout an individualrsquos lifetime butits effect on survival changed as individuals aged because theeffect of individual heterozygosity on survival rate differedamong age classes Survival rates of genetically sampled kittenswere biased high because some kittens were sampled later in lifewhen they had already survived for some time To correct forthis bias we adjusted kitten survival rates predicted by theheterozygosity model proportionally From the population tra-jectories we estimated the probability of quasi‐extinction (de-fined as the probability that the population will fall below 10 or30 individuals or that individuals of only 1 sex will remain)

Cost‐benefit analysismdashExperts estimated financial costs ofintrogression based on their experience with the geneticintrogression executed in 1995 and we adjusted estimates tocurrent costs to account for inflation (R T McBride RancherrsquosSupply Incorporated M Cunningham and D P OnoratoFlorida Fish and Wildlife Conservation Commission personalcommunication) Costs included capturing and transportingfemale pumas from the Texas source population caring for thepumas while they were in quarantine and post‐introgressionmonitoring To compare the financial costs of the differentscenarios we also calculated the average costs incurred over

Table 2 Sample size (n) number of alleles (Na) allelic richness (AR) observedheterozygosity (Ho) and expected heterozygosity (He) at 16 microsatellite loci inradio‐collared Florida panthers sampled from 1981 to 2013 in South FloridaUSA We calculated these values from a subset (n= 137) of the samples used inthe analyses and they include only adult and subadult panthers We excludedkittens because they can result in an overrepresentation of certain alleles

Locus n Na AR Ho He

FCA090 129 5 50 0333 0401FCA133 136 3 30 0618 0608FCA243 136 5 50 0419 0487F124 137 7 69 0708 0729F37 129 4 40 0395 0397FCA075 131 5 50 0534 0598FCA559 133 5 50 0496 0503FCA057 134 6 59 0679 0624FCA081 134 7 69 0209 0206FCA566 130 6 60 0462 0475F42 133 10 100 0737 0766FCA043 136 5 49 0596 0588FCA161 132 6 60 0500 0515FCA293 132 4 40 0614 0624FCA369 131 6 60 0573 0567FCA668 133 7 70 0406 0462Mean 1329 57 57 0517 0535

Figure 4 Proportional membership (q) of radio‐collared Florida panthers (n= 218) and pumas from Texas USA (n= 7) in 2 clusters identified by STRUCTUREEach individual animal is represented by a separate vertical bar Yellow indicates canonical panther ancestry and blue represents admixed ancestry The admixedancestry of the Texas females and F1 generation panthers that were radiocollared are highlighted red and green respectively to demarcate those groups The pre‐introgression period is inclusive of panthers radiocollared from 10 February 1981 to 6 March 1996 The post‐introgression era inclusive of the F1 panthers includesanimals radiocollared from 4 March 1997 to 18 February 2015 in South Florida USA

16 Wildlife Monographs bull 203

100 years assuming that the population would be managedaccording to a particular strategy Our goal with theseexploratory analyses was to evaluate the relative costs andbenefits of alternative management scenarios

RESULTS

Genetic VariationWe assessed genetic variation at 16 microsatellite loci for 137DNA samples collected from adult and subadult panthers(1981ndash2013) that were captured and subsequently radiocollaredThe number of alleles at these loci ranged from 3ndash10 and allelicrichness averaged 57 (Table 2) Average expected hetero-zygosity for this sample was 0535 and average observed het-erozygosity was 0517 (Table 2)We determined individual heterozygosity (1 minusHL) and an-

cestry for the complete sample of panther genotypes (n= 503)collected from 1981 to 2015 for use as covariates in subsequentanalyses Average individual heterozygosity was 0559 andranged from 0ndash0980 Our STRUCTURE analysis providedstrong support for K= 2 clusters based on the probability of thedata (log Pr(D)|K) and ΔK in STRUCTURE HARVESTERBased on this result we categorized the ancestry of each pantheras either canonical (n= 123) or admixed (n= 380) using valuesof proportional membership (q Fig 4) The average individualheterozygosity of canonical panthers was 0386plusmn 0012 (SE) foradmixed panthers it was 0615plusmn 0007

Survival and Cause‐Specific Mortality RatesOf the 395 kittens that were PIT‐tagged and released 55 werelater encountered alive 15 were recovered dead and 85 werepart of failed litters (Table 1) We radio‐tracked 209 subadult oradult panthers for a total of 244195 panther‐days and docu-mented 126 mortalities (Table 1)Survival depended strongly on age and kittens had the lowest

overall rate (032plusmn 009 Fig 5A Table 3) The model‐averagedkitten survival was 047plusmn 009 when we included only geneti-cally sampled individuals in the analysis (Table 3) as statedabove this estimate is biased high because many kittenswere genetically sampled when they were recaptured alive orrecovered dead at older ages We found no evidence for sex‐specific differences in kitten survival but females survived betterthan males in all older age classes For females subadults hadthe highest survival rates followed by prime adults and oldadults For males prime adults had the highest survival ratefollowed by subadults and old adults (Fig 5A and Table 3)For panthers of all ages there was substantial evidence that

ancestry affected survival with F1 admixed panthers of all ageshaving the highest rates and canonical individuals having thelowest (Fig 5B) The combined AIC weights of models thatincluded an ancestry covariate were 0880 for kittens and 0992for older panthers The survival rates of subadult prime‐adultand old‐adult panthers in the period immediately after the in-trogression (1995ndash2004) were similar to those in more recentyears (2005ndash2013 Fig 5C)After genetic introgression individual panther heterozygosity

increased (Fig 6) and varied among ancestry categories individualheterozygosity was the highest for F1 panthers (078plusmn 001)

A

B

C

Figure 5 Overall age‐ and sex‐specific survival rates (plusmnSE A) age‐ and sex‐specific survival rates (plusmnSE) for Florida panthers of canonical F1 admixed andother admixed ancestry (B) and age‐ and sex‐specific survival estimates (plusmnSE)for subadult and adult Florida panthers in the period immediately post‐introgression (1995ndash2004) and more recently (2005ndash2013 C) in SouthFlorida USA

Table 3 Model‐averaged estimates (plusmnSE) of demographic parameters for theFlorida panther obtained using data collected during 1981ndash2013 in SouthFlorida USA

Parameter Sex Age class Estimate

Survival Both Kitten 032plusmn 009a

Female Subadult 097plusmn 002Prime adult 086+ 003Old adult 078plusmn 009

Male Subadult 066plusmn 006Prime adult 077plusmn 005Old adult 065plusmn 010

Probability of reproduction Female Subadult 035plusmn 008Prime adult 050plusmn 005Old adult 025plusmn 006

Kittens produced annually Female Subadult 280plusmn 005Prime adult 267plusmn 001Old adult 228plusmn 013

a Overall kitten survival (based on all kittens in our data set including thosethat were not genetically sampled) Estimate of survival of geneticallysampled kittens was 047plusmn 009 this estimate is biased high because manykittens were genetically sampled when they were recaptured alive or re-covered dead at older ages

van de Kerk et al bull Florida Panther Population and Genetic Viability 17

followed by those for non‐F1 admixed (062plusmn 001) and canonical(039plusmn 001) panthers Individual heterozygosity positively affectedsurvival rates of kittens (slope parameter η= 1455 95CI= 0505ndash2405) Individual heterozygosity also positively af-fected survival of subadult and adult panthers but the evidence forthis effect was weaker because the confidence interval for the ha-zard ratio overlapped 10 (hazard ratio exp(ɩ)= 0311 95CI= 0092ndash1054 Table 4 and Fig 7)The MPC increased after genetic introgression (Fig 2) This

abundance index was also included in top‐ranking modelsindicating that population density affected survival (Table 4)Abundance reduced the survival of kittens (slope parameterη= minus0035 95 CI= minus0049 to minus0021) evidence was weak forthe effect of abundance on survival of subadult and adult panthers(hazard ratio exp(ɩ)= 1002 95 CI= 0995ndash1010 Fig 7) Re-gression coefficients associated with the effect of abundance andindividual heterozygosity on demographic parameters are presentedin Table B1 (available online in Supporting Information)The most frequently observed causes of death for radio‐

collared panthers were vehicle collision and intraspecific ag-gression Cause‐specific mortality analyses revealed thatmortality rates among possible causes were generally similar forthe 2 sexes but that males were more likely to die from in-traspecific aggression than were females (Fig 8A) Male mor-tality from intraspecific aggression was higher for subadult andold adult panthers than for prime adults (Fig 8B) For bothmales and females canonical panthers were more likely to diefrom intraspecific aggression than were admixed panthers(Fig 8C)

Reproductive ParametersOf 101 radio‐collared females 67 reproduced at least once duringour monitoring we used these individuals to assess the annualprobability of reproduction The top‐ranked model for the annual

probability of reproduction included age effect only suggesting thatadding ancestry or abundance did not improve model parsimony(Table 4) Model‐averaged annual probabilities of reproductiondiffered between female subadults (035plusmn 008) prime adults(050plusmn 005) and older adults (025plusmn 006)The most parsimonious model for the number of kittens

produced annually by females that produced at least 1 litter (ieconditional on reproduction) included effects of age ancestryand abundance a model that included only the effect of age wasalmost equally supported (Table 4) The model‐averagednumber of kittens produced annually conditional on re-production was 280plusmn 075 for subadults 267plusmn 043 for primeadults and 228plusmn 083 for old adults Confidence intervals forthe slope parameter (β) overlapped 0 for both abundance(β= 0010 95 CI= minus0001 to 0021) and ancestry (β= minus073795 CI= minus1637 to 0162)

Population Dynamics and PersistenceDeterministic and stochastic demographymdashThe deterministic (λ)

and stochastic (λs) annual population growth rate calculated usingthe matrix population model were 106 (5th and 95th percentiles099ndash114) and 104 (072ndash141) respectively The generation timewas 47 years A female starting in age class one (kitten) wasexpected to live another 58 years and produce 10 kittens onaverage during her lifetime Overall λs was proportionately(elasticity) most sensitive to changes in female prime adultsurvival on the absolute scale (sensitivity) however it was mostsensitive to changes in kitten survival (Fig 9) Populationtrajectories probabilities of quasi‐extinction and times untilquasi‐extinction estimated using the matrix population modeland IBM for comparable scenarios were nearly identical for acritical quasi‐extinction threshold of 10 panthers (Figs 10 and 11)and for a critical quasi‐extinction threshold of 30 panthers(Appendix E available online in Supporting Information)Minimum population count scenariomdashUnder the MPC scenario

(ie density dependence estimated using the minimumpopulation count data) with a critical threshold of 10panthers the population size reached 99 (72ndash104) and 100(77ndash105) at 200 years (Fig 10) as predicted by the IBM and thematrix model respectively The cumulative quasi‐extinctionprobabilities within 100 years based on the MPC scenario were15 (IBM 0ndash48) and 30 (matrix model 0ndash76) Theseprobabilities increased to 25 (IBM 0ndash114) and 38 (matrixmodel 0ndash154) by year 200 (Fig 10) The mean times untilquasi‐extinction were 15 years (IBM 0ndash67) and 15 years (matrixmodel 0ndash72) conditioned on going quasi‐extinct within 100years Expected times until quasi‐extinction conditioned onquasi‐extinction within 200 years were higher 34 years (IBM0ndash137) and 32 years (matrix model 0ndash134 Fig 10)Motor vehicle mortality scenariomdashUnder the MVM scenario

(ie density dependence estimated using estimates of pantherpopulation size from McClintock et al 2015) with a criticalthreshold of 10 panthers the projected population reached 188(142ndash217) and 190 (143ndash219) at 200 years as predicted by theIBM and the matrix model respectively (Fig 11) Thecumulative quasi‐extinction probabilities within 100 years were14 (IBM 0ndash08) and 13 (matrix model 0ndash06) Theseprobabilities increased to 20 (IBM 0ndash17) and 19 (matrix

000

025

050

075

100

1995 2000 2005 2010Year of birth

Indi

vidu

al h

eter

ozyg

osity

Figure 6 Temporal changes in the individual heterozygosity of Floridapanthers born during the period 1995ndash2013 in South Florida USA Pointsare measurements solid line is linear regression shaded area is 95 confidenceinterval of slope Slope= 0006plusmn 0001 R2= 009

18 Wildlife Monographs bull 203

model 0ndash16) within year 200 (Fig 11) The mean times until quasi‐extinction based on the MVM scenario were 5 years (IBM 0ndash61)and 6 years (matrix model 0ndash64) conditioned on going quasi‐extinctwithin 100 years Expected times until quasi‐extinction conditionedon quasi‐extinction within 200 years were 11 years (IBM 0ndash113)and 11 years (matrix model 0ndash112 Fig 11)

Sensitivity of extinction probability to demographic parametersmdashThe partial rank correlation coefficients between quasi‐extinctionprobabilities and demographic parameters were negative anincrease in any of the demographic parameters decreased thequasi‐extinction probability Probabilities of quasi‐extinction within200 years were most sensitive to changes in kitten survival for bothscenarios (Fig 12) followed by survival of subadult females in theMVM scenarios and survival of prime adult females in the MPCscenario The probability of reproduction of prime adult femaleshad the third‐largest partial rank correlation coefficient with quasi‐extinction probability for both scenarios

Benefits and Costs of Genetic ManagementMinimum population count scenariomdashThe MPC scenario with

a critical threshold of 10 panthers predicted that withoutmanagement intervention average individual heterozygositywould decrease reaching 057 (049ndash064) at 100 years and053 (042ndash062) at 200 years (Fig 13) The population woulddecrease slightly reaching an average of 79 (41ndash99) at 100 yearsand 76 (43ndash98) at 200 years (Fig 14) The probabilities of quasi‐extinction under the no‐intervention MPC scenario when weconsidered the effects of genetic erosion were 13 (0ndash99) at 100years and 23 (0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with

the translocation of 5 Texas females average individualheterozygosity under the MPC scenario increased and thenstabilized at 068 (064ndash072) at 100 years and remained atthat level at 200 years (Fig 13) The population reached anaverage of 88 (62ndash104) at 100 years and stayed the same at200 years (Fig 14) The probabilities of quasi‐extinctionunder this introgression strategy were 6 (0ndash53) within 100years and 7 (0ndash82) within 200 years (Fig 15) Results ofanalyses using a critical threshold of 30 panthers revealed asimilar pattern of relative benefits However the prob-abilities of quasi‐extinction were substantially higher (ge45)across all scenarios (Appendix E)

Motor vehicle mortality scenariomdashWithout managementintervention the average individual heterozygosity predictedby the MVM scenario and a critical threshold of 10 panthersdecreased to 060 (046ndash069) at 100 years and to 059 (039ndash070) at 200 years (Fig 13) The population increased slightlyand reached an equilibrium at 154 individuals (46ndash217) at 100years and 157 (57ndash218) at 200 years (Fig 14) The probabilitiesof quasi‐extinction were 17 (0ndash100) at 100 years and 22(0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with 5

female pumas from Texas average individual heterozygosityincreased slightly reaching 067 (060ndash073) at 100 years and068 (059ndash075) at 200 years (Fig 13) Under this strategy thepopulation reached an average of 164 individuals (44ndash226) at

Table 4 Model comparison results testing for the effects of several covariateson survival of all kittens survival of genetically sampled kittens survival ofsubadult and adult Florida panthers probability of reproduction and kittensproduced annually for Florida panthers sampled from 1981ndash2013 in SouthFlorida USA

Model Parameters ΔAICa Weight

Kitten survival (all data)Base+ F1b+ abundancec 10 000 0988Base+ abundance 9 1018 0006Base+ F1 9 1035 0005Base 8 2711 lt0001Base+ sexd 9 2912 lt0001Base+ litter sizee 9 2914 lt0001

Kitten survival (genetically sampled kittens only)Base+ F1+ abundance 10 000 0541Base+ F1CanAdmf+ abundance 11 099 0329Base+Hetg+ abundance 10 310 0115Base+CanAdmh+ abundance 10 791 0010Base+ abundance 9 944 0005Base+ F1 9 1638 lt0001Base+ F1CanAdm 10 1837 lt0001Base+Het 9 2872 lt0001Base 8 3296 lt0001Base+CanAdm 9 3348 lt0001

Subadult and adult survivalBase+ F1CanAdm+ abundance 7 000 0360Base+ F1 5 053 0276Base+ F1CanAdm 6 105 0213Base+ F1+ abundance 6 222 0119Base+CanAdm+ abundance 6 575 0020Base+CanAdm 5 908 0004Base+Het 5 964 0003Base+Het+ abundance 6 984 0003Base 4 1118 0001Base+ abundance 5 1278 0001

Probability of reproductionAge 3 000 0242Age+CanAdm 4 012 0228Age+ abundance 4 173 0102Age+ F1 4 190 0094Age+CanAdm+ abundance 5 216 0082Age+Het 4 219 0081Age+ F1CanAdm 5 232 0076Age+ F1+ abundance 5 372 0038Age+Het+ abundance 5 399 0033Age+ F1CanAdm+ abundance 6 444 0026

Kittens produced annuallyAge+ F1+ abundance 5 000 0194Age 3 060 0144Age+ abundance 4 061 0143Age+ F1 4 097 0120Age+CanAdm 4 139 0097Age+ F1CanAdm+ abundance 6 196 0073Age+ F1CanAdm 5 231 0061Age+CanAdm+ abundance 5 238 0059Age+Het+ abundance 5 250 0056Age+Het 4 259 0053

a Akaikersquos information criterion (AIC) for kitten survival models was adjustedfor small sample size and overdispersion (QAICc)

b F1 divides Florida panthers into 2 groups F1 admixed and all other panthersc Index of Florida panther abundanced Sex of an individual Florida panther (male or female)e Size of the litter in which a Florida panther kitten was bornf F1CanAdm divides panthers into 3 groups F1 admixed canonical andother admixed panthers

g Het is the individual heterozygosityh CanAdm divides panthers into 2 groups canonical and admixed (includingF1 admixed) panthers

van de Kerk et al bull Florida Panther Population and Genetic Viability 19

100 years and 170 (67ndash231) at 200 years based on the MVMscenario (Fig 14) The probabilities of quasi‐extinction underthis introgression strategy were 10 (0ndash99) at 100 years and12 (0ndash99) at 200 years (Fig 15)The projected population size and average individual hetero-

zygosity reached equilibrium levels with periodic fluctuations inthese values when introgression was implemented (Figs 16ndash17)Genetic introgression decreased the time until quasi‐extinctionacross all scenarios (Fig 18) Results of analyses using a criticalthreshold of 30 panthers revealed a similar pattern of relative ben-efits However the probabilities of quasi‐extinction were sub-stantially higher (ge45) across all scenarios (Appendix E)

Addition of pumas from Texas increased both the populationsize and average individual heterozygosity consequentlyintrogression caused periodic increases in average individualheterozygosity and population size at the interval at which theintrogression occurred Whereas average individual hetero-zygosity gradually decreased following introgression densitydependence caused the population size to fluctuate especiallywhen a larger number of pumas from Texas were released Thepositive effect of genetic introgression initiatives on quasi‐extinction probabilities decreased as the time interval betweenthese events increased More frequent genetic introgressions ledto greater increases in average individual heterozygosity and

Old adult

Prime adult

Subadult

Kitten

025 050 075

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Individual heterozygosity

Ann

ual s

urvi

val r

ate

Old adult

Prime adult

Subadult

Kitten

40 60 80 100 120

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Abundance index

Ann

ual s

urvi

val r

ate

Kittens Males Females

Figure 7 The effect of individual heterozygosity (left) and the abundance index (right) on Florida panther age‐ and sex‐specific survival estimates (shaded area isSE) in South Florida USA 1981ndash2013 Abundance index data are from McBride et al (2008) and R T McBride and C McBride (Rancherrsquos Supply Incorporatedunpublished report)

20 Wildlife Monographs bull 203

equilibrium population size while reducing the probability ofquasi‐extinction Comparatively performing genetic introgres-sion less often but with more individuals per release was lesseffective at improving the long‐term prospects for the pantherpopulation (Figs 13ndash15)

Financial cost and persistence benefitsmdashThe total estimated costof 1 genetic introgression was $40000 $80000 or $120000for releasing 5 10 or 15 pumas from Texas respectively(Table 5) The total cost incurred over 100 years for eachstrategy ranged from $50000 to $12 million (Table 6)The most expensive strategy involved the introduction of15 pumas every 10 years this strategy reduced the probability

of quasi‐extinction by 58 and 40 under the MPC and MVMscenarios respectively (Table 6) Introducing 5 pumas every80 years cost the least but improvements in populationpersistence under this strategy were insubstantial (Table 6)Releasing more panthers more frequently caused increasedpopulation fluctuations but did not necessarily reduce the quasi‐extinction probability (Figs 14 and 15 Table 6)

DISCUSSION

The duration (34 yr of data) and sample sizes associated with theFlorida Panther Project allowed us to obtain a comprehensiveunderstanding of genetic variation demographic parameters

A

B

C

Figure 8 Cause‐specific mortality rates (plusmnSE) for male and female Florida panthers (A) subadult prime‐adult and old‐adult Florida panthers of both sexes (B)and canonical and admixed Florida panthers of both sexes (C) in South Florida USA 1981ndash2013 Causes of mortality are vehicle collision intraspecific aggression(ISA) other causes (known causes such as diseases injuries and infections unrelated to the first 2 causes) and unknown cause (mortalities for which we could notassign a cause)

van de Kerk et al bull Florida Panther Population and Genetic Viability 21

probability of population persistence and the duration of thepositive impacts of genetic restoration on this endangered po-pulation The Florida panther population was in dire straits inthe early 1990s and our results clearly demonstrated that the

implementation of the introgression project in 1995 subse-quently accrued a series of genetic and demographic improve-ments that have led to the change from a declining population toone that is growing and expanding its current breeding range

Sensitivity Elasticity

0 1 2 3 00 02 04

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

Para

met

er

Figure 9 Sensitivity and elasticity (proportional sensitivity) of stochastic population growth rate (λs) of the Florida panther to vital demographic parameters in SouthFlorida USA 1981ndash2013 Horizontal lines represent 5th and 95th percentiles

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

70

80

90

100

110

0

5

10

15

0

50

100

Years

StatisticMeanMedian

Figure 10 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the minimum population count(MPC) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

125

150

175

200

00

05

10

15

20

0

30

60

90

Years

StatisticMeanMedian

Figure 11 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the motor vehicle mortality(MVM) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

22 Wildlife Monographs bull 203

MPC MVM

minus08 minus06 minus04 minus02 00 minus08 minus06 minus04 minus02 00

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Male subadult survival

Male prime adult survival

Male old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

PRCC

Para

met

er

Figure 12 Partial rank correlation coefficient (PRCC) between quasi‐extinction probabilities and the demographic parameters of the Florida panther for theminimum population count (MPC) and motor vehicle mortality (MVM) scenarios Error bars indicate standard deviation Based on data collected in South FloridaUSA 1981ndash2013

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 13 Average projected individual heterozygosities in the Florida panther population over 200 years without genetic management intervention and with eachof the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include theintroduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years Average individual heterozygosity peaks when pumas are added to the Floridapanther population Based on data collected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 23

Our research has further validated the success of genetic in-trogression by quantifying the reduction in the probability ofextinction of the panther population because of this manage-ment initiative These findings all bode well for continuedprogress toward recovery That said the Florida panther po-pulation remains small and isolated from other populations ofpuma from western North America thereby denoting that theeventual impact of genetic drift and inbreeding may negativelyaffect the population in the future Realizing this will continueto be an issue that managers will have to contemplate wemodeled the impact of varied genetic management scenarios todetermine the frequency of introgression along with the numberof panthers that should be released during each initiative tominimize cost and maximize benefits to the population Theseresults will prove invaluable to managers attempting to conservethe Florida panther

Genetic Variation Demographic Parametersand Persistence of Heterotic Benefits from GeneticIntrogressionOur measure of individual heterozygosity (1 minusHL) was higheron average for the admixed (0615) panthers than for those ofcanonical ancestry (0386) Levels of individual heterozygosityfor admixed panthers were comparable to levels observed in asample of pumas from west Texas (x plusmn SE= 0695plusmn 0019n= 41) which further demonstrates the improvement in levelsof genetic variation in the Florida population since 1995 The

correlation between HL and several demographic parametershas been noted in wild populations including cheetahs (Terrellet al 2016) and several species of birds such as European shags(Phalacrocorax aristotelis male and female survival probabilityVelando et al 2015) and Egyptian vulture (Neophron percnop-terus age at recruitment Agudo et al 2012) If we focus only onpanthers captured and radiocollared from 2012ndash2015 (FP211ndashFP240) all of which had estimated birth years ge2005 (ge10 yrpost‐introgression) those panthers had an average individualheterozygosity level of 0618plusmn 0029 highlighting the elevatedlevels of genetic variation that continue to persist in cohorts ofpanthers 5 generations after the release of female pumas fromTexasThe proportional membership values (q) from our

STRUCTURE analysis allowed us to delineate 2 clusters ofancestry for Florida panthers (Fig 4) During the pre‐in-trogression era the population was dominated by panthers thatretained a high percentage of the canonical ancestry which wasaffected by inbreeding depression The post‐introgression eraancestry values are comprised of many admixed individuals inessence demonstrating the success of the introgression in termsof increasing genetic variation in the population and mi-micking gene flow that historically occurred with panthers andother populations of pumas (Onorato et al 2010) A somewhatanalogous situation although abetted by natural immigrationwas evident in the ancestral variation in a small population ofpuma intersected by a major freeway in California USA In

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

75

80

85

90

95

100

130

150

170

190

Years

Popu

latio

n si

ze

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 14 Average projected population sizes in the Florida panther population over 200 years without intervention and with each of the genetic introgressionstrategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include the introduction of 5 10 or 15pumas from Texas USA every 10 20 40 or 80 years Peaks occur when pumas are added to the Florida panther population Based on data collected in SouthFlorida USA 1981ndash2013

24 Wildlife Monographs bull 203

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

25

50

75

100

0

25

50

75

100

Introgression interval (yrs)

PQ

E (

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 15 Average probability of quasi‐extinction (PQE) of the Florida panther population within 100 years and within 200 years without genetic managementintervention (N) and with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenariosIntrogression strategies include the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Errorbars indicate 5th and 95th percentiles Based on data collected in South Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

50

60

70

80

90

100

110

50

100

150

200

Years

Popu

latio

n si

ze

Figure 16 Average projected Florida panther population trajectories under a geneticintrogression strategy involving the release of 5 female pumas from Texas USA every20 years for the minimum population count (MPC) and motor vehicle mortality(MVM) scenarios Shaded area indicates 5th and 95th percentiles and isrepresentative of confidence intervals for other scenarios Based on data collected inSouth Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Figure 17 Average projected Florida panther population‐level heterozygositiesunder a genetic introgression strategy involving the release of 5 female pumas fromTexas USA every 20 years for the minimum population count (MPC) and motorvehicle mortality (MVM) scenarios Shaded area bounds the 5th and 95th percentilesand is representative of confidence intervals for other scenarios Based on datacollected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 25

this case the migration of just a single male across the freewayto the south resulted in an increase in the number of in-dividuals with admixed ancestry and substantially improvedgenetic diversity of the subpopulation south of the freeway(Riley et al 2014)Our estimate of overall kitten survival (0321plusmn 0056) was

similar to that reported by Hostetler et al (2010) who used13 years of post‐introgression data (0323plusmn 0071) These valuesare lower than kitten survival estimates derived from severalpuma populations in western North America (044ndash091 Loganand Sweanor 2001 Lambert et al 2006 Laundre et al 2007Robinson et al 2008 Ruth et al 2011) When we included onlydata for individuals that had been genetically sampled our es-timate was 15 higher The proportion of admixed kittens thatwere genetically sampled increased as the population expandedwhich could have resulted in higher estimates of kitten survivalbecause admixed kittens survive better than canonical kittensKitten survival was strongly influenced by genetic ancestry withsurvival rates of F1 kittens being almost twice that of canonicalkittens (Fig 5B) Consistent with the findings of Hostetler et al(2010) increases in heterozygosity coincided with increasedkitten survival whereas population density negatively affectedkitten survival Population density often has a strong negativeimpact on survival of young age classes due to infanticide andcannibalism which has been reported from several carnivore

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

50

100

150

0

50

100

150

Introgression interval (yrs)

TQE

(yrs

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 18 Average times until quasi‐extinction (TQE) for the Florida panther population within 100 years and within 200 years without genetic management interventionand with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategiesinclude the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Error bars indicate 5th and 95thpercentiles Based on data collected in South Florida USA 1981ndash2013

Table 5 Average cost (2017 US dollars) of introgression strategies employedover 100 years that involve the introduction of 5 10 or 15 pumas from TexasUSA into the Florida panther population in South Florida USA at an in-creasingly higher frequency Costs of capture (including transportation) quar-antine and post‐introgression monitoring were estimated by Roy McBrideMark Cunningham and Dave Onorato respectively

Frequency Component 5 pumas 10 pumas 15 pumas

Once Capture 25000 50000 75000

Quarantine 5000 10000 15000

Monitoring 10000 20000 30000

Total 40000 80000 120000

Every 80 years Capture 31250 62500 93750

Quarantine 6250 12500 18750

Monitoring 12500 25000 37500

Total 50000 100000 150000

Every 40 years Capture 62500 125000 187500

Quarantine 12500 25000 37500

Monitoring 25000 50000 75000

Total 100000 200000 300000

Every 20 years Capture 125000 250000 375000

Quarantine 25000 50000 75000

Monitoring 50000 100000 150000

Total 200000 400000 600000

Every 10 years Capture 250000 500000 750000

Quarantine 50000 100000 150000

Monitoring 100000 200000 300000

Total 400000 800000 1200000

26 Wildlife Monographs bull 203

species including polar bears (Ursus maritimus Derocher andWiig 1999) black bears (Ursus americanus Garrison et al 2007)and pumas (Logan and Sweanor 2001)Our estimates of sex‐ and age‐specific survival rates of subadult

and adult panthers (Table 3) and the effects of covariates on theserates were similar to those reported by Benson et al (2011) derivedfrom data collected through 2006 Our survival rates for males(065ndash077) were lower than females (078ndash097) for all 3 ageclasses (subadult prime adult and old adult) which is the typicaltrend observed in most puma populations (eg Ruth et al 19982011) However an unhunted puma population occupying afragmented urban landscape in southern California exhibited lowersurvival (0586 females 0525 males Vickers et al 2015) perhapsas a result of severe habitat fragmentation and the lack of extensiveswaths of public land protected in perpetuity Most populations ofpuma in western North America are typically subject to variedlevels of management via regulated hunting which often is biasedtoward the take of males (Cooley et al 2009 Ruth et al 2011)Consequently annual survival estimates from hunted populationsin northeast Washington (0679ndash0733 females 0341 malesRobinson et al 2008) and southeastern Arizona (0685 females0584 males Cunningham et al 2001) were generally lower thanthose we estimated for Florida panthersCause‐specific mortality analysis showed that male panthers

experienced a substantially greater mortality risk from in-traspecific aggression This differs from the pattern observedduring a long‐term study in Arizona where females were at ahigher risk of intraspecific mortality than males (46 of maleand 53 of female mortality Logan and Sweanor 2001)Mortality associated with intraspecific strife has been docu-mented in hunted and unhunted populations (this study Loganand Sweanor 2001 Stoner et al 2010) and its root cause hasbeen difficult to determine (eg population density and preypopulation trends) That being the case finding ways to reducewhat is often a major source of mortality for puma populationscontinues to pose a challenge to managersCanonical panthers were substantially more likely to die from

intraspecific aggression than were admixed panthers This mayat least partly explain the difference in survival between admixed

and canonical panthers Canonical panthers are known to bemore likely to suffer from varied correlates of inbreeding de-pression than admixed animals including atrial septal defects(Johnson et al 2010) and compromised immune systems(Roelke et al 1993) These factors have the potential to affectfitness and perhaps dominance of individuals when engaged inan intraspecific encounter A somewhat similar scenario wasobserved by Hogg et al (2006) in a population of bighorn sheepthat were part of an introgression project Admixed males in thispopulation had a greater probability of paternity than males thatwere less outbred This included coursing males with higherlevels of admixture being markedly more successful at fightingmales that were defending females in estrous (Hogg et al 2006)Heterosis resulting from genetic introgression is expected to be

the strongest in F1 individuals (Shull 1908 Crow 1948 Burkeand Arnold 2001 Waller 2015) and to progressively decline insubsequent generations (Pickup et al 2013 Frankham 2015)Given that admixed (especially F1) panthers survived better thancanonical panthers and that individual heterozygosity improvedsurvival of both sexes and all age classes our data provide evi-dence for heterotic advantage whereby individuals of mixedancestry had higher fitness (Shull 1908 Crow 1948) Our resultsare consistent with findings of other studies that involved in-tentional genetic introgression for conservation purposes Forexample Hogg et al (2006) detected substantial improvementsin survival and reproduction of introgressed bighorn sheep andMadsen et al (1999 2004) reported a dramatic increase in thepopulation of adders after genetic introgressionKnowledge of the persistence of benefits to a population ac-

crued via genetic introgression is essential for determiningwhen additional introgression may be warranted There is adepauperate amount of information regarding this topic and theidentification of factors that may influence persistence of thebenefits of genetic introgression (Tallmon et al 2004 Hedricket al 2014 Whiteley et al 2015) For example in minks(Neovison vison) Thirstrup et al (2014) found that outcrossingled to larger litters in F2 and F3 but this benefit disappeared insubsequent generations In a population of gray wolves Hedricket al (2014) suggested that benefits of genetic introgression may

Table 6 Costs and benefits accumulated over 100 years for genetic introgression at different intervals and with different numbers of pumas from Texas USAintroduced into the Florida panther population in South Florida USA Costs (2017 US dollars) include capture and transportation quarantine and post‐introgression monitoring Benefits are represented as average probability of quasi‐extinction (PQE and 5th and 95th percentiles) and changes (Δ) therein under thescenario using the minimum population count (McBride et al 2008 MPC) and the scenario using the motor vehicle mortality population estimates (McClintocket al 2015 MVM) The table is sorted by percent change in probability of quasi‐extinction under the MPC scenario

Interval (yr) Pumas Cost MPC PQE () ΔMPC PQE () MVM PQE () ΔMVM PQE ()

ndash 0 0 13 (0ndash99) 0 17 (0ndash100) 080 10 100000 12 (0ndash99) 10 (0ndash58) 16 (0ndash100) 5 (0ndash38)80 15 150000 12 (0ndash99) 13 (0ndash65) 15 (0ndash100) 8 (0ndash56)80 5 50000 12 (0ndash99) 14 (0ndash73) 15 (0ndash99) 9 (0ndash41)40 15 300000 10 (0ndash98) 26 (0ndash104) 14 (0ndash99) 13 (0ndash60)40 10 200000 9 (0ndash94) 35 (0ndash183) 13 (0ndash99) 21 (0ndash95)40 5 100000 8 (0ndash89) 39 (0ndash211) 12 (0ndash99) 26 (0ndash124)20 15 600000 8 (0ndash88) 42 (0ndash257) 13 (0ndash99) 24 (0ndash123)20 10 400000 6 (0ndash67) 54 (0ndash318) 10 (0ndash99) 37 (0ndash217)10 15 1200000 6 (0ndash53) 58 (0ndash372) 10 (0ndash99) 40 (0ndash208)20 5 200000 5 (0ndash50) 59 (0ndash338) 9 (0ndash99) 44 (0ndash352)10 10 800000 4 (0ndash16) 69 (0ndash489) 8 (0ndash92) 55 (0ndash401)10 5 400000 4 (0ndash7) 73 (0ndash502) 6 (0ndash71) 63 (0ndash503)

van de Kerk et al bull Florida Panther Population and Genetic Viability 27

wane after 2 or 3 generations The WrightndashFisher model ofgenetic drift predicts a geometric decline in expected hetero-zygosity at the rate of (1 minus 12Ne) per generation where Ne is theeffective population size (Hartl and Clark 1997 Frankham et al2014) Thus it seems logical to assume that the duration of thebenefits of introgression is limited especially in a species per-sisting in a small population such as Florida panthersWe expected Florida panther survival in the most recent years

(2005ndash2013) post‐introgression to differ from that observed in thedecade following the introgression in 1995 because pantherabundance and heterozygosity factors known to influence panthersurvival (Hostetler et al 2010 Benson et al 2011) have changed(Figs 2 and 6) We found that subadult and adult survival rates inrecent years (2005ndash2013) were similar to those in the period im-mediately following introgression (1995ndash2004 Fig 5C) overallkitten survival was similar to estimates of Hostetler et al (2010) Infact the pattern of covariate effects on Florida panther survivalreported by Benson et al (2011) and Hostetler et al (2010) hasbarely changed The negative effects of population density andpositive effects of heterozygosity may counterbalance survival ratesreducing population fluctuations Our result that benefits of geneticrestoration have remained in the population nearly for 5 genera-tions post‐intervention was surprising but also encouraging Pos-sible explanations for this observation may include 1) genetic ero-sion occurs at slower rate than previously thought 2) introducing 5migrants in 1 genetic intervention may have beneficial effects si-milar to those from 1 migrant per generation over multiple gen-erations or 3) other and as yet unknown factors may reduce therate of genetic erosion and subsequent demographic effects Adensity‐dependent effect on kitten survival could also explain whynon‐F1 admixed kittens did not survive substantially better thandid canonical kittens most kittens sampled from 2005ndash2013 wereadmixed and their survival was lower possibly because of a density‐dependent effect as the Florida panther population continuouslygrew during that period Trinkel et al (2010) also found thatpopulation density and inbreeding coefficient interacted to affectdemographic parameters and growth rate of an African lion po-pulation Likewise population density interacted with winterweather to affect the survival and population growth rates in Soaysheep (Ovis aries Milner et al 1999 Coulson et al 2001)

Population Dynamics and PersistenceThe finite deterministic and stochastic growth rates were gt10reflecting that much of the data used in this study were collectedfollowing genetic introgression in 1995 during which time thepopulation increased substantially Because our demographicparameter estimates did not change markedly in comparison tothose of Hostetler et al (2013) the similarity between thepopulation growth estimates from these studies was expectedBut these estimates reflect population growth during thedata‐collection period and do not predict future growth In factestimates of population size reported by McClintock et al(2015) suggest that population growth may already be slowing(Fig 2) A similar pattern was observed in an introduced po-pulation of African lion which increased steadily from 1995until 2001 then fluctuated around the presumed carrying ca-pacity thereafter (Trinkel et al 2010) Several other African lionpopulations that experienced various degrees of recovery

subsequently became relatively stable or exhibited decliningtrends (Bauer et al 2015)Our estimates of quasi‐extinction probabilities were lower than

those reported by Hostetler et al (2013) using a 2‐sex age‐structured matrix population model Lower probabilities ofquasi‐extinction than those reported by the earlier study forcomparable scenarios were likely a consequence of the fact thatthe estimates of abundance (McBride et al 2008) and thus thedensity‐dependent effects on survival of kittens and old panthers(gt10 yr) have changed in recent years Additionally these earlyestimates of the probability of quasi‐extinction and of time toquasi‐extinction did not consider genetic erosion that will in-evitably occur without management interventionUnder the MVM scenario the equilibrium population size was

twice that attained under the MPC scenario The MVM popula-tion estimates are approximately twice the MPCs (Fig 2) as aresult the simulated population under the MVM scenario achieveda higher equilibrium size Probability of quasi‐extinction under theMVM scenario was generally greater than that under the MPCscenario This result is a consequence of the fact that the estimateddensity dependence on kitten survival based on the MPC scenario isstronger than that for theMVM scenario (regression coefficients forabundance for the MPC scenario= minus0068 vs minus0002 for theMVM scenario Appendix B Table B1) Consequently simulatedpopulations under the MPC scenario have a stronger tendency tomove toward the equilibrium population size which inherentlyreduces the quasi‐extinction probability (eg Royama 1992)As expected for long‐lived mammals (Heppell et al 2000 Oli

and Dobson 2003) the population growth rate was proportio-nately most sensitive to changes in survival of prime adultsfollowed by that in younger age classes Global sensitivity ana-lysis in contrast showed that under all scenarios extinctionparameters were particularly sensitive to changes in survival ofFlorida panther kittens This result is a consequence of the highsensitivity of the population growth rate to kitten survival(Fig 9) and a larger standard error of kitten survival (Table 3)Results of our sensitivity analyses indicate that future researchshould focus on acquiring additional information on early lifestages to improve the accuracy and precision of estimates ofkitten survival Our results suggest that demographic variables(or their environmental drivers) that strongly influence extinc-tion risks are not necessarily those with the largest elasticityvalues A study of island fox (Bakker et al 2009) found thatextinction risk was strongly influenced by survival modifierparameters that had low elasticity values

Genetic Management When and How to InterveneThe use of genetic introgression as a management tool has al-ways been controversial and the probable effectiveness it mighthave in preventing the imminent demise of the panther popu-lation was initially debated (Creel 2006 Maehr et al 2006Pimm et al 2006) These debates notwithstanding the pantherpopulation had suffered from genetic morphological and bio-medical correlates of inbreeding before this management in-tervention (Johnson et al 2010 Onorato et al 2010) whereasthe post‐introgression period has been characterized by a sub-stantial reduction in the biomedical correlates of inbreeding andimprovements in demographic vigor and abundance (McBride

28 Wildlife Monographs bull 203

et al 2008 Hostetler et al 2010 2013 Johnson et al 2010Benson et al 2011 McClintock et al 2015) Nevertheless thepopulation remains small and isolated habitat is still being lostand there is no current possibility of natural gene flow betweenthis and other North American puma populations thus therecurrence of inbreeding‐related problems and subsequent po-pulation declines are inevitable The question therefore is notwhether genetic management of the Florida panther populationis needed but when and how it should be implementedWithout genetic introgression probabilities of quasi‐extinc-

tion were substantially greater than those from models thatincorporated periodic genetic introgression events (Fig 15)highlighting the importance of genetic management in smallisolated populations When 5 female pumas are added to theFlorida panther population every 10 years genetic variationincreased substantially under the MPC and MVM scenariosThe IBM predicted that the equillibrium population size wouldbe substantially larger when 5 pumas were released into thepopulation every 10 years than populations without intervention(ie no introgression) Releasing 15 Texas females did not af-ford substantially greater benefit to the equilibrium populationsize than did releasing only 5 Texas females Instead addingmore individuals caused increasingly large fluctuations in po-pulation size due to the numerical increases and density de-pendence (Fig 14) possibly diminishing the benefits associatedwith increased genetic variationCompared with the no‐intervention scenario (ie no genetic

introgression implemented) the introduction of 5 female pumasevery 10 years reduced quasi‐extinction probabilities from 13to 4 and from 17 to 6 for the MPC and MVM scenariosrespectively The benefit of genetic‐introgression strategies de-creased as the interval between successive management inter-ventions increased and no benefit was evident once the intervalreached 80 years (Fig 15) Introgression reduced the averagetime until quasi‐extinction (Fig 18) This somewhat counter‐intuitive result can be explained by the fact that repeated geneticintrogression makes the population more robust over timewhich will reduce the probability of extinction Consequentlyfew simulated population trajectories that fall below the criticalthreshold do so early reducing the mean time to extinctionAlso density dependence has a stabilizing effect on populations(Royama 1992) populations tend toward equilibrium popula-tion sizes which reduces the probability over time of popula-tions falling below quasi‐extinction threshold However timeuntil quasi‐extinction must be interpreted in conjunction withthe probability of quasi‐extinction which is very low for allscenarios with genetic introgression (Fig 15)Cost is a significant impediment to the implementation of a

genetic introgression program because captures relocations andmonitoring efforts before and after introgression are expensive(Edmands 2007 Frankham 2010b 2015 2016) Costs may beacceptable to society if the management actions result in sub-stantial fitness benefits especially if the species of concern ispopular and charismatic (Frankham 2015) We estimated the100‐year cost of introgression strategies modeled here to rangefrom $50000 to $12 million More expensive strategies gen-erally led to larger decreases in the probability of quasi‐extinc-tion but cheaper strategies also substantially improved

population viability (ie reduced quasi‐extinction probabilityTable 6) One of the cheaper strategies (5 pumas released every20 years) which cost an estimated $200000 over 100 yearsreduced the probability of quasi‐extinction by 59 and 44 forthe MPC and MVM scenarios respectively But these pre-liminary cost estimates are based on expenses incurred duringthe 1995 introgression and include costs of capture quarantinetransportation release and post‐release monitoring of femalesonly Although the cost for future genetic interventions wouldlikely be higher than those reported here the relative differencesin cost between alternative intervention strategies should remainapproximately the sameOur ability to simulate genetics and link it to the viability of

the Florida panther population is based on the empirically es-timated relationship between individual heterozygosity andsurvival Such heterozygosity and fitness correlations have beenstudied for decades (David 1998) but have only recently beenused to predict population viability (Benson et al 2016) noneto our knowledge have incorporated cost into their modelingefforts The mechanisms underlying heterozygosity and fitnesscorrelations remain unclear (David 1998 Szulkin et al 2010)and their significance as a proxy for inbreeding depression hasbeen debated (Balloux et al 2004 Miller and Coltman 2014)Nevertheless evidence continues to mount demonstratingpositive relationships between heterozygosity and survival(Coulson et al 1998ab Hansson et al 2004 Silva et al 2005Acevedo‐Whitehouse et al 2006 Jensen et al 2007 Velandoet al 2015) and between heterozygosity and fecundity (Seddonet al 2004 Charpentier et al 2005 Agudo et al 2012 Velandoet al 2015) Although the reported correlations are generallyweak their consequences can be substantial if population growthand persistence parameters are highly sensitive to the affectedvital rates (Velando et al 2015) Compared with scenarios thatignore genetics incorporating the effects of inbreeding on sur-vival increased quasi‐extinction probabilities within a 100‐yeartimeframe from 15 to 13 and from 14 to 17 for theMPC and MVM scenarios respectively These results high-light the importance of incorporating genetics when analyzingthe viability of small isolated populationsAlthough conservation biologists have recognized the im-

portance of genetics in population viability many still fail toincorporate genetic factors into PVA models (Reed et al 2002)Explicit consideration of genetics in PVAs requires long‐termgenetic and demographic data This permits the assessment ofthe functional relationship between genetic variability and de-mographic parameters Population viability analysis softwarepackages such as VORTEX (Lacy 1993 2000) offer a powerfulframework for individual‐based simulations but they often donot adequately capture a speciesrsquo life history or genetics Basedon a thorough review of the importance of genetic factors inwildlife conservation Frankham et al (2014) concluded thatgenetic factors can strongly influence the dynamics and persis-tence of small andor fragmented populations They furthernoted that ldquoMost PVA‐based risk assessments ignore or in-adequately model genetic factorsrdquo and recommended that ldquoPVAshould routinely include realistic inbreeding depressionhelliprdquo(Frankham et al 201457) Our custom‐built IBM permitted usto develop a model that adequately captured the Florida panther

van de Kerk et al bull Florida Panther Population and Genetic Viability 29

life history and we could parameterize the model with our long‐term genetic and demographic dataThe explicit consideration of the effects of inbreeding on

Florida panther population dynamics and persistence representsan important step forward Nonetheless our model remainsimperfect First we neglected the possible effects of habitat lossdetrimental anthropogenic activities climate change the prob-ability of immigration of pumas from western populationsdisease or catastrophes on demographic rates and populationviability Although immigration from puma populations outsideFlorida is possible its probability is very low given the extensivechallenges the anthropogenically altered landscape would posefor such a migrant to make it successfully to South Florida Weomitted mutations from our model because we have no in-formation on mutation rates though we expect that they are lowbased on values reported for other mammals (Kumar and Sub-ramanian 2002) Finally we only considered possible effects ofinbreeding on age‐specific survival rates because there was noevidence that heterozygosity affected female reproduction Lossof heterozygosity can negatively affect reproduction becauseinbred male panthers can suffer from cryptorchidism which canadversely affect their fecundity However given the polygynousmating system we expect the Florida panther population growthis primarily female‐limitedAlthough it is generally accepted that genetic introgression

only temporarily relieves a population from inbreeding depres-sion we know of no cases in which it has been implementedmore than once to conserve a threatened wildlife populationOur study provides an example of how demographic and mo-lecular data can be used within an IBM framework to evaluatethe efficacy of alternative genetic management strategies via theFlorida panther as a case study Our model can be adjusted forand applied to other species and offers great potential as a toolfor genetic management of small isolated wildlife populations

MANAGEMENT IMPLICATIONS

Our results and those of Hostetler et al (2013) and Johnsonet al (2010) provide persuasive evidence that the genetic in-tervention implemented in 1995 prevented the demise of theFlorida panther and restored demographic vigor to the popu-lation The Florida panther population remains small and iso-lated from other puma populations the closest puma populationis located gt1000 km away in Texas and the intervening land-scape is highly developed and fragmented with many interstate(and other high‐traffic) highways and major urban centersTheory suggests that 1 migrant per generation is sufficient tominimize the loss of genetic diversity (eg Mills and Allendorf1996) However the possibility of successful migration of apuma to South Florida followed by successful mating every ~5years is very low considering the distance between Texas andFlorida (Hedrick 1995) and the many risks and barriers thatdispersing pumas face (Beier 1995 Maehr et al 2002 Thatcheret al 2006) Consequently the pantherrsquos long‐term persistencewill likely depend on subsequent timely genetic introgressionsgiven the near‐impossibility of natural gene flow from otherpuma populations To that end our study offers considerableinsights into benefits of different genetic management strategiesand associated costs

Without genetic management the Florida panther populationfaces a substantial risk of quasi‐extinction within 100 years (10ndash46 depending on quasi‐extinction threshold and scenarios)Twenty‐three years have passed since genetic introgression wasimplemented and the population appears healthy as indicatedby measures of genetic variation increases in the populationsize and improvements in age‐specific survival rates Ourfindings suggest that releasing more pumas more frequently doesnot necessarily improve persistence probability because such astrategy would cause larger fluctuations in population size(Fig 14) which negatively affects persistence probability Thusextinction risks faced by inbred populations of large carnivorescan be substantially reduced by releasing approximately 5 im-migrants every 2 decades We suggest that such a managementstrategy be followed only in conjunction with continued long‐term monitoring of the population Our analyses demonstratethat population growth and persistence are highly sensitive tokitten and female (adult and subadult) survival Continuing tocollect data on these demographic variables along with bio-metric and genetic sampling will remain important to pantherrecovery and vigilance in identifying issues associated with in-breeding depression The collection of these monitoring datashould allow managers to detect any demographic or geneticchanges in a timely fashion and respond with genetic in-trogression or other management actions if warrantedRecovery objectives as defined by the USFWS in its current

recovery plan (USFWS 2008) delineate the need for additionalpopulations in the historical range to downlist or delist theFlorida panther If the only extant population of Floridapanthers continued to grow it could serve as a source ofpanthers to be translocated to suitable habitat to seed addi-tional populations Maintaining current information on thegenetic health of the population and precise estimates of itssize will be important in assessing whether it can withstand theremoval of a subset of animals for translocation Although the2017 documentation of a female panther with kittens north ofthe current breeding range is encouraging in terms of thepotential for population expansionmdashgiven that no female hadbeen recorded north of the Caloosahatchee River since 1972mdashthe re‐establishment of separate new populations will mostlikely require translocations by wildlife managers and a lengthysociopolitical processConservation of threatened or endangered species such as the

Florida panther is inherently challenging For panthers and otherlarge carnivores that require vast extents of natural habitat topersist habitat is typically the most limiting factor that will de-termine whether recovery efforts succeed Although geneticmanagement invariably plays a key role in the long‐term persis-tence of the panther population even a healthy population caneventually succumb to the impacts of habitat loss The humanpopulation of Florida continues to be one of the fastest‐growingin the nation the United States Census Bureau currently ranksFlorida as the third most populous Therefore habitat loss isgoing to continue to be a key issue for panthers Diligence towardpreserving panther habitat in its historical range via conservationeasements or other means and trying to keep such areas inter-connected via corridors is a goal stakeholders such as governmentagencies sportsmen non‐governmental organizations ranchers

30 Wildlife Monographs bull 203

and other private landowners must work on collaboratively toachieve

ACKNOWLEDGMENTS

We thank D Shindle D Jennings T OrsquoMeara D Land andC Belden for valuable advice throughout the project We thankS Bass M Criffield M Cunningham D Giardina D JansenA Johnson D Land M Lotz C McBride Rocky McBrideRowdy McBride Roy McBride L Oberhofer S Schulze staffsat Everglades National Park and Big Cypress National Preserveand others for assistance with fieldwork We also thank JAustin M Conroy J Hines J Laake S Mills J Nichols JHines M Sunquist J Fryxell and T Coulson for advice andassistance regarding data analysis and panther biology TheMuseum of Texas Tech University Natural Science ResearchLaboratory provided samples from pumas from west Texas forcomparative genetic analyses M Ben‐David D Land WMcCown E Hellgren and 4 anonymous reviewers providedmany helpful comments on the manuscript We thank NereaUbierna Lopez for completing the Spanish translation of theabstract We are grateful to the Department of ResearchComputing University of Florida for excellent high‐perfor-mance computing services We would like to thank US Fishand Wildlife Service Florida Fish and Wildlife ConservationCommission (FWC) US National Park Service QuantitativeSpatial Ecology Evolution and Environment (QSE3) In-tegrative Graduate Education and Research Traineeship(IGERT) program funded by the National Science Foundationand the University of Florida for financially supporting thisstudy Funding for panther research conducted by the FWC issupported predominantly via donations accrued with vehicleregistrations for ldquoProtect the Pantherrdquo license plates

LITERATURE CITEDAcevedo‐Whitehouse K T R Spraker E Lyons S R Melin F Gulland RL Delong and W Amos 2006 Contrasting effects of heterozygosity onsurvival and hookworm resistance in California sea lion pups MolecularEcology 151973ndash1982

Adams J R L M Vucetich P W Hedrick R O Peterson and J AVucetich 2011 Genomic sweep and potential genetic rescue during limitingenvironmental conditions in an isolated wolf population Proceedings of theRoyal Society B Biological Sciences 2783336ndash3344

Agresti A 2013 Categorical data analysis Third edition John Wiley amp SonsHoboken New Jersey USA

Agudo R M Carrete M Alcaide C Rico F Hiraldo and J A Donaacutezar2012 Genetic diversity at neutral and adaptive loci determines individualfitness in a long‐lived territorial bird Proceedings of the Royal Society BBiological Sciences 2793241ndash3249

Albeke S E N P Nibbelink and M Ben‐David 2015 Modeling behavior bycoastal river otter (Lontra canadensis) in response to prey availability in PrinceWilliam Sound Alaska a spatially‐explicit individual‐based approach PLOSOne 10e0126208

Alho J S K Vaumllimaumlki and J Merilauml 2010 Rhh an R extension for estimatingmultilocus heterozygosity and heterozygosity‐heterozygosity correlationMolecular Ecology Resources 10720ndash722

Alvarez K 1993 Twilight of the panther Myakka River Publishing SarasotaFlorida USA

Aparicio J M J Ortego and P J Cordero 2006 What should we weigh toestimate heterozygosity alleles or loci Molecular Ecology 154659ndash4665

Bakker V J D F Doak G W Roemer D K Garcelon T J Coonan S AMorrison C Lynch K Ralls and R Shaw 2009 Incorporating ecologicaldrivers and uncertainty into a demographic population viability analysis for theisland fox Ecological Monographs 7977ndash108

Balloux F W Amos and T Coulson 2004 Does heterozygosity estimateinbreeding in real populations Molecular Ecology 133021ndash3031

Barone M A M E Roelke J Howard J L Brown A E Anderson and DE Wildt 1994 Reproductive characteristics of male Florida panthersmdashcomparative studies from Florida Texas Colorado Latin‐America andNorth‐American Zoos Journal of Mammalogy 75150ndash162

Bateson Z W P O Dunn S D Hull A E Henschen J A Johnson and LA Whittingham 2014 Genetic restoration of a threatened population ofgreater prairie‐chickens Biological Conservation 17412ndash19

Bauer H G Chapron K Nowell P Henschel P Funston L T B HunterD W Macdonald and C Packer 2015 Lion (Panthera leo) populations aredeclining rapidly across Africa except in intensively managed areas Pro-ceedings of National Academy of Sciences of the United States of America11214894ndash14899

Beier P 1995 Dispersal of juvenile cougars in fragmented habitat Journal ofWildlife Management 59228ndash237

Benson J F J A Hostetler D P Onorato W E Johnson M E Roelke SJ OrsquoBrien D Jansen and M K Oli 2011 Intentional genetic introgressioninfluences survival of adults and subadults in a small inbred felid populationJournal of Animal Ecology 80958ndash967

Benson J F P J Mahoney J A Sikich L E K Serieys J P Pollinger H BErnest and S P D Riley 2016 Interactions between demography geneticsand landscape connectivity increase extinction probability for a small popu-lation of large carnivores in a major metropolitan area Proceedings of theRoyal Society B Biological Sciences 28320160957

Beverly F 1874 The Florida panther Forest and Stream 3290Boyce M S C V Haridas C T Lee and the NCEAS Stochastic Demo-graphy Working Group 2006 Demography in an increasingly variable worldTrends in Ecology amp Evolution 21141ndash148

Brook B W J J OrsquoGrady A P Chapman M A Burgman H R Akcakayaand R Frankham 2000 Predictive accuracy of population viability analysis inconservation biology Nature 404385ndash387

Brys R and H Jacquemyn 2016 Severe outbreeding and inbreeding de-pression maintain mating system differentiation in Epipactis (Orchidaceae)Journal of Evolutionary Biology 29352ndash359

Burke J M and M L Arnold 2001 Genetics and the fitness of hybridsAnnual Review of Genetics 3531ndash52

Burnham K P 1993 A theory for combined analysis of ring recovery andrecapture data Pages 199ndash213 in J‐D Lebreton and P M North editorsMarked individuals in the study of bird population Springer‐Verlag NewYork New York USA

Burnham K P and D R Anderson 2002 Model selection and multimodelinference a practical information‐theoretic approach Second edition SpringerVerlag New York New York USA

Caswell H 2001 Matrix population models construction analysis and in-terpretation Sinauer Associates Sunderland Massachusetts USA

Charpentier M J M Setchell F Prugnolle L A Knapp E J Wickings PPeignot and M Hossaert‐McKey 2005 Genetic diversity and reproductivesuccess in mandrills (Mandrillus sphinx) Proceedings of the NationalAcademy of Sciences of the United States of America 10216723ndash16728

Cooch E and G C White editors 2018 Program MARK a gentle in-troduction lthttpwwwphidotorgsoftwaremarkdocsbookgt Accessed 7Apr 2016

Cooley H S R B Wielgus G M Koehler H S Robinson and B TMaletzke 2009 Does hunting regulate cougar populations A test of thecompensatory mortality hypothesis Ecology 902913ndash2921

Coulson T E A Catchpole S D Albon B J Morgan J M Pemberton TH Clutton‐Brock M J Crawley and B T Grenfell 2001 Age sex densitywinter weather and population crashes in Soay sheep Science 2921528ndash1531

Coulson T J Pemberton S Albon M Beaumont T Marshall J Slate FGuinness and T Clutton‐Brock 1998a Microsatellites reveal heterosis in reddeer Proceedings of the Royal Society B Biological Sciences 265489ndash495

Coulson T N S D Albon J M Pemberton J Slate F E Guinness and TH Clutton‐Brock 1998b Genotype by environment interactions in wintersurvival in red deer Journal of Animal Ecology 67434ndash445

Cox D 1972 Regression models and life‐tables Journal of the Royal StatisticalSociety Series B Statistical Methodology 34187ndash220

Creel S 2006 Recovery of the Florida panthermdashgenetic rescue demographic rescueor both Response to Pimm et al (2006) Animal Conservation 9125ndash126

Crnokrak P and D A Roff 1999 Inbreeding depression in the wild Heredity83260ndash270

Crow J 1948 Alternative hypotheses of hybrid vigor Genetics 33477ndash487

van de Kerk et al bull Florida Panther Population and Genetic Viability 31

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 8: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

Capture teams determined sex and age of panthers marked themwith an ear tattoo and a subcutaneous passive integrated trans-ponder (PIT‐tag) and fitted them with a very high frequency(VHF) or global positioning system (GPS) radio‐collar equippedwith a mortality sensor The GPS collars were programmed toattempt to collect a position at a variety of time intervals(range= 1ndash12 hr) depending on objectives of concurrent studiesobservers routinely tracked panthers with VHF radiocollars from afixed‐wing aircraft 3 times per week When a radio‐collar trans-mitted a mortality signal or when a location did not change forseveral days observers investigated the site to establish the pan-therrsquos fate When observers found a dead panther they assessed thecause of death based on an examination of the carcass and sur-rounding area if possible They then transported carcasses to anexperienced veterinarian for necropsy and a histopathological ex-amination to attempt to assign the cause of death Starting in 1995observers continually assessed successive locations of radio‐collaredfemales to determine if they initiated denning behavior Three to 4fixes at the same location (VHF collars) or successive returns to thesame location over a weeklong period (GPS collars) served as anindicator of a possible den Observers then located dens moreprecisely via triangulation with ground telemetry They visited denswhen the dam left the site typically to make a kill Teams capturedkittens by hand captured kittens ranged in age from 4 to 35 dayspost‐partum Teams counted sexed weighed dewormed andpermanently marked kittens with a subcutaneous PIT‐tag

Genetic AnalysisTeams collected blood tissue or hair from the majority ofcaptured kittens subadult and adult Florida panthers for DNAsamples The United States Department of Agriculture ForestService National Genomics Center for Wildlife and FishConservation (Missoula MT USA) processed samplesTechnicians used Qiagen DNeasy Blood and Tissue Kits(Qiagen Valencia CA USA) to extract whole genomic DNAfrom blood and tissue samples and used a slight modification(overnight incubation of sample in lysis buffer and proteinase Kon a rocker at 60deg elution of DNA in 50 μl of buffer) of theQiagen DNA extraction protocol (Mills et al 2000) to extractDNA from hair samples Technicians used a panel of 16 mi-crosatellite loci (Fca090 Fca133 Fca243 F124 F37 Fca075Fca559 Fca057 Fca081 Fca566 F42 Fca043 Fca161

Fca293 Fca369 and Fca668) identified by Menotti‐Raymondet al (1999) which had been previously applied to Floridapanther samples (Johnson et al 2010) Technicians amplifiedDNA samples in 10‐microl reaction volumes that included 10 microl ofDNA 1times reaction buffer (Applied Biosystems Life Technolo-gies Corporation Grand Island NY USA) 20 mM of MgCl2200 microM of each dNTP 1 microM of reverse primer 1 microM of dye‐labeled forward primer 15 mgml of bovine serum albumin(BSA) and 1U Taq polymerase (Applied Biosystems) Thepolymerase chain reactions (PCRs) included a thermal profile of94degC for 5 minutes followed by 36 cycles of 94degC for 60 seconds55degC for 60 seconds and 72degC for 30 seconds Techniciansvisualized the resulting PCR products on a LI‐COR DNAanalyzer (LI‐COR Biotechnology Lincoln NE USA) Tech-nicians collected genotypes from hair samples (n= 16) using amultitube approach (Taberlet et al 1996) error‐checked geno-types using Program DROPOUT (McKelvey and Schwartz2005) and combined them with genotypes from tissue andblood samples (n= 487) We evaluated all 16 loci for variousdescriptive statistics for the radio‐collared sample of adult andsubadult panthers We excluded genotypes of kittens from thatanalysis because related individuals are known to result in theoverrepresentation of certain alleles (Marshall et al 1998) Wedetermined the number of alleles observed heterozygosity andexpected heterozygosity using GenAlex (Peakall and Smouse2006 2012) We assessed allelic richness at each locus in Fstat293 (Goudet 1995) We assessed conformance of genotypedata from these 16 loci to HardyndashWeinberg assumptionslinkage disequilibrium and null alleles (Appendix A availableonline in Supporting Information) The Genomics Center alsoprovided us with genotype data from the same 16 loci for 41pumas from western Texas the source population for thegenetic rescue programEvidence of inbreeding depression in wild populations is

commonly determined on the basis of heterozygosityndashfitnesscorrelations (Silva et al 2005 Ortego et al 2007 Mainguyet al 2009 Johnson et al 2011) so we also estimated theheterozygosity of each individual panther We calculatedhomozygosity by loci (HL) which varies between 0 (all lociheterozygous) and 1 (all loci homozygous Aparicio et al 2006)using the Rhh package (Alho et al 2010) in Program R (R CoreTeam 2016) A microsatellite locus has more weight in HL

Table 1 Number of Florida panthers by age sex and genetic ancestry categories used in subsequent demographic analyses during the study period (1981ndash2013) inSouth Florida USA

Kittensa Subadults and adultsb

Males Females Total Males Females Total

PIT‐taggedc 215 180 395 Radiocollaredc 108 101 209Recovered 11 4 15 Subadult 72 50 122Litter failed 47 38 85 Prime adult 65 95 160Recaptured 25 30 55 Old adult 13 20 33Canonical 27 20 47 Canonical 43 29 72F1 admixed 6 12 18 F1 admixed 2 6 8Other admixed 130 105 235 Other admixed 45 53 98

a Kitten samples included those that were PIT‐tagged recovered (ie panthers that were found dead) part of a failed litter or recaptured as subadults or adultsb Radiocollared subadult and adult panthers are presented as sample size within each age class and ancestry classc Combined number of panthers of different ancestries is less than the number radiocollared or tagged with a subcutaneous passive integrated transponder (PIT‐tagged) because we did not conduct genetic sampling on all panthers Combined number of panthers in different age classes is greater than the number ofradiocollared panthers because some individuals were counted in ge2 age classes as animals aged

10 Wildlife Monographs bull 203

when it is more informative (ie has more alleles) and has aneven distribution of allele frequencies For our demographicanalyses we calculated individual heterozygosity for each in-dividual panther as 1 minusHL As do most indices of individualheterozygosity HL takes into account the average allele fre-quency in the population (Aparicio et al 2006) Thus duringsimulations the allele frequency in the population changesthroughout an individualrsquos lifetime which causes its hetero-zygosity also to change This is not biologically plausible becausegenetic properties are retained throughout an individualrsquos life-time Therefore for use in model simulations we simply cal-culated heterozygosity for each individual as the proportion ofheterozygous lociWe also used genotype data to implement a Bayesian clus-

tering analysis in Program STRUCTURE (Pritchard et al2000) to infer ancestral clusters of panthers This method uses aMarkov chain Monte Carlo (MCMC) method to determine thenumber of genetic clusters (K) in the sample while also pro-viding an individualrsquos percentage of ancestry (q values) allocatedto each cluster Runs for this analysis included a 100000MCMC burn‐in period followed by 500000 MCMC iterationsfor 1ndash10 ancestral genetic clusters (K= 1 to 10) with 10 re-petitions for each K To determine the most likely number of Kgenetic clusters we used the logarithm of the probability of thedata (log Pr(D)∣K Pritchard et al 2000) and estimates of ΔK(Evanno et al 2005) in Program STRUCTURE HAR-VESTER (Earl and VonHoldt 2011) We then used q valuesprovided in runs for the best K to assign individual panthers aseither canonical (in most cases ge90 pre‐introgression ancestrysee Johnson et al 2010) or admixed (lt90 pre‐introgressionancestry) We recognized admixed panthers that were the off-spring of matings between a Texas female puma and a canonicalmale panther as F1 admixed panthers for analyses

Demographic ParametersSurvival of kittensmdashMost kittens PIT‐tagged at den sites were

never encountered again A small proportion of the PIT‐taggedkittens were recovered dead or were recaptured as subadults oradults and radiocollared so that their fate could be monitoredWe also identified instances of complete litter failure when afemale died within 9 months after giving birth (the minimumage of independence for kittens) or when a female wasdocumented to have another litter less than 12 months aftergiving birth (the minimum age of independence plus a 3‐monthgestation period Hostetler et al 2010) Thus our data setconsisted of 1) live recaptures and subsequent radio‐tracking ofpanthers of various ages that had been PIT‐tagged as kittens 2)recovery of dead panthers of various ages that had been PIT‐tagged as kitten 3) live captures and subsequent radio‐trackingof other panthers and 4) instances of complete litter failure(Table 1) We analyzed these data using Burnhamrsquos live‐recapture dead‐recovery modeling framework (Burnham 1993Williams et al 2002) to estimate survival of kittens (0ndash1‐yearold) and to test covariate effects on this parameter We includeddata on individuals encountered dead or alive at an older age inour analyses because their encounter histories also providedinformation on kitten survival We used a live‐dead data inputformat coded with a 1‐year time step (Cooch and White 2018)

All known litter failures occurred within the first year of thekittensrsquo lives so we treated all litter failures as recoveries withinthe first year We treated panthers that would have died if wehad not removed them to captivity as having died on the date ofremoval Data preparation and analytical approach are describedin more detail by Hostetler et al (2010)In addition to survival probability (S) the Burnham model

incorporates parameters for recapture probability (p) recoveryprobability (r) and site fidelity (f) We set r and p to 10 forradio‐tracked individuals because their status was known eachyear We also fixed f at 10 for all individuals because the re-capture and recovery areas were the same and encompassed theentire range of the Florida panther Based on findings of Hos-tetler et al (2010) we used a base model that 1) allowed survivalto differ between kittens and older panthers 2) allowed survivalto differ between sexes and among age classes (females ages 1and 2 ge3 males ages 1 2 and 3 ge4) 3) constrained recaptureprobability to be the same for all uncollared panthers and 4)allowed recovery rates to differ between kittens and uncollaredolder panthers Even with our additional data this model re-mained the best fit for a range of models for recapture recoveryand survival of kittens subadults and adults of various age‐classcategories (Hostetler et al 2010)Subadult and adult survivalmdashTo estimate survival for panthers

ge1 year of age we analyzed radio‐tracking data using a Coxproportional hazard‐modeling framework (Cox 1972 Therneauand Grambsch 2000) We followed procedures outlined byBenson et al (2011) for data preparation and analysis Brieflywe right‐censored panthers on the last day of observation beforetheir collar failed When an individual aged into a new age classwhile being tracked we created 2 data entries 1 ending the dayit entered the new age class the other starting on that day Weused the FlemingndashHarrington method to generate survivalestimates from the Cox analysis and the Huber sandwichmethod to estimate robust standard errors (Therneau andGrambsch 2000 Benson et al 2011)We considered 3 age classes subadults (1ndash25 yr for females and

1ndash35 yr for males) prime adults (25ndash10 yr for females and35ndash10 yr for males) and old adults (gt10 yr old for both sexes)after Benson et al (2011) We estimated overall survival using abase model that allowed survival to differ between subadultfemales subadult males prime adult females prime adult malesand old adults of either sex (old age was additive with sex other-wise age and sex were interactive) This model provided the best fitto the data relative to a range of models considering differences insurvival between prime‐adult and older‐adult age classes both ad-ditively and interactively with sex (Benson et al 2011)To examine whether survival rates observed in recent years

differed from those observed immediately following introgres-sion we also estimated age‐specific survival rates for 2 nearlyequal periods of time immediately post‐introgression (1995ndash2004) and the more recent period (2005ndash2013) For theseanalyses we separated the radio‐tracking data into the 2 periodsand estimated survival rates for each period separately using thebase model Similar analyses for kitten survival were not possiblebecause of data limitationsProbability of reproductionmdashWe used telemetry data obtained

from radio‐tracked females that reproduced at least once to

van de Kerk et al bull Florida Panther Population and Genetic Viability 11

estimate annual probability of reproduction of subadult prime‐adult and old‐adult females using complementary logndashlogregression (Agresti 2013) following methods described indetail by Hostetler et al (2012) Briefly we modeled theprobability of a female panther giving birth (b) as a function ofcovariates as

γ= minus [minus ( )]b z1 exp exp i i

where zi is a row vector of covariates (see below for covariateinformation) for female panther i and γ is a column vector ofmodel coefficients (Appendix B Table B2) To account fordifferences in observation time we included log(m) as an offsetin the model where m is the number of months a femalepanther was radio‐tracked (Hostetler et al 2012)

Reproductive outputmdashTo estimate the number of kittensproduced by reproductive females we used cumulative logitregression (Min and Agresti 2005 Agresti 2013) Weconsidered a model that includes J categories for the numberof offspring with j denoting the number of offspring ( j= 1 2hellip J and J= 6) provided that a female reproduces Weconsidered J= 6 which is greater than the largest litter sizeobserved in our study (4 kittens) because females can produce 2or more litters per year if litters fail We modeled the probabilitythat a female i produces a litter of size at most j (Yi) as

le⎧

⎨⎩

δ( ) = == hellip minus

=

( )+ θ βminus +Y jj J

j JPr

1 2 1

1i ij e

1

1 xj i

where θj is the intercept for the annual number of kittens pro-duced by a female i being at most j xi is a row vector of cov-ariates for female panther i (see below) and β is a column vectorof model coefficients The probability that Yi will be exactly j isgiven by

⎧⎨⎩

πδ

δ δ( = ) = =

=

minus = hellip( minus )Y j

jj J

Pr 1

2 i ij

i

ij i j

1

1

Finally the expected annual number of kittens produced byfemale i (νi) is given by

sumν π==

j i

j

J

ij

1

Additional details regarding the estimation and modelingof the annual number of kittens produced can be found inHostetler et al (2012)

CovariatesmdashIn addition to sex and age we tested for the effect ofabundance genetic ancestry and individual heterozygosity on theaforementioned demographic parameters (model coefficients forkitten and adult survival are given by η and ι respectively AppendixB Table B2) We used a minimum population count (MPC) basedon radio‐tracking data and field evidence of subadult and adult (ieindependent age) panthers as an index of abundance (McBride et al2008) For these analyses we did not use the population sizeestimates reported by McClintock et al (2015) because unlike theminimum counts they did not cover the entire study period

(McBride et al 2008 R T McBride and C McBride RancherrsquosSupply Incorporated unpublished report Fig 2)To test if demographic parameters differed depending on an-

cestry we considered 3 ancestry models dividing the panthers into1) F1 admixed and all other panthers 2) canonical and admixed(including F1 admixed) panthers and 3) F1 admixed canonicaland other admixed panthers We quantified genetic diversity usingindividual heterozygosity as described previously We tested for theeffect of genetic covariates using a subset of kittens that were ge-netically sampled We calculated model‐averaged estimates ofmodel parameters on the scale in which they were estimated(Appendix B available online in Supporting Information)

Cause‐specific mortalitymdashWe performed cause‐specificmortality analysis only on dead radio‐collared panthers becausedetection rates of uncollared panther mortalities are highlycorrelated with the cause of death (eg vehicle collision)Following Benson et al (2011) we attributed each mortality to1 of 4 causes 1) vehicle collision 2) intraspecific aggression 3)other causes (including known causes such as diseases injuriesand infections unrelated to the first 2 causes) and 4) unknown(ie mortalities for which we could not assign a cause) Wethen used the non‐parametric cumulative incidence functionestimator a generalization of the staggered‐entry KaplanndashMeiermethod of survival estimation to estimate cause‐specificmortality rates for male and female panthers in different ageclasses (Pollock et al 1989 Heisey and Patterson 2006 Bensonet al 2011) We compared cause‐specific mortality rates betweensexes age classes and ancestries

Statistical inferencemdashWe used an information‐theoreticapproach (Akaikersquos information criterion [AIC]) for modelselection and statistical inference (Burnham and Anderson 2002)For the analysis of kitten survival we adjusted the AIC foroverdispersion and small sample size (QAICc details in Hostetleret al 2010) We performed all statistical analyses in Program R (RCore Team 2016) We used Burnhamrsquos live recapturendashdead

0

100

200

300

1985 1990 1995 2000 2005 2010Year

Indi

vidu

als Method

MPC

MVM

Figure 2 The Florida panther minimum population count (MPC) and populationsize estimated using motor vehicle collision mortalities (MVM) during our studyperiod Minimum population counts are from McBride et al (2008) and R TMcBride and C McBride (Rancherrsquos Supply Incorporated unpublished report)Estimates based on MVM are from McClintock et al (2015) The solid vertical lineindicates the year of genetic introgression

12 Wildlife Monographs bull 203

recovery model (Burnham 1993) to estimate and model kittensurvival using the RMark package version 2113 (Laake 2013) asan interface for Program MARK (White and Burnham 1999)We implemented the Cox proportional hazard model for esti-

mation and modeling of survival of subadults and adults using the Rpackage SURVIVAL (Therneau and Grambsch 2000) We per-formed cause‐specific mortality analyses using an R version of S‐PLUS code provided by Heisey and Patterson (2006)To account for model selection uncertainty we calculated

model‐averaged parameter estimates across all models using theR package AICcmodavg (Mazerolle 2017) using AIC weights asmodel weights (Burnham and Anderson 2002) We used modelswithout the categorical ancestry variables for model averagingand estimated parameters based on the average value for thecontinuous covariates (individual heterozygosity and abundanceindex) Because only a subset of the kittens was geneticallysampled we estimated 2 kitten survival rates First we estimatedkitten survival based on the data set that included all kittens(overall kitten survival) Second we estimated kitten survivalusing a subset of the data that only included kittens that weregenetically sampled (survival of genetically sampled kittens)

Matrix Population ModelsWe investigated Florida panther population dynamics andpersistence using 2 complementary modeling frameworks ma-trix population models (Caswell 2001) and IBMs (Grimm andRailsback 2005 McLane et al 2011 Railsback and Grimm2011 Albeke et al 2015) Matrix population models offer aflexible and powerful framework for investigating the dynamicsand persistence of age‐ or stage‐structured populations (egCaswell 2001 Robinson et al 2008 Kohira et al 2009 Hunteret al 2010 Hostetler et al 2012) With a fully developed theorybehind them these models also serve as a check for resultsobtained from simulation‐based models such as IBMs We usedan age‐structured‐matrix population model for deterministic andstochastic demographic analyses and for preliminary investiga-tions of Florida panther population viability (Caswell 2001Morris and Doak 2002)For deterministic and stochastic demographic analyses we

parameterized a female‐only 19 times 19 age‐specific populationprojection matrix of the form

⎢⎢⎢⎢⎢

⋯ ⋯

⋯ ⋯

⋱ ⋱ ⋮

⋮ ⋱ ⋱ ⋱ ⋮

⎥⎥⎥⎥⎥

( )=t

F F F

P

P

P

A0 0

0

0 0 0

t t t

t

t

t

1 2 19

1

2

18

where Fi and Pi are the age‐specific fertility and survival ratesrespectively and ( )tA is a time‐varying (or stochastic) popula-tion projection matrix We assumed that the longevity and ageof last reproduction of female Florida panthers was 185 yearswe based this assumption on the observation that the oldestfemale in our study was 186 years old Florida panthers re-produce year‐round thus we used birth‐flow methods to esti-mate Fi and Pi values following Caswell (2001) and Hostetleret al (2013) We estimated Pi as

= +P S S i i i 1

where Si is the age‐specific survival probability We estimated Fi as

⎛⎝

⎞⎠

=+ +F S

m Pm

2i k

i i i 1

where Sk is the kitten survival probability and mi is the age‐specific fecundity rate which was estimated as

ν=m b f i i i k

where bi is the breeding probability for a female in age class iduring time step t νi is the annual number of kittens producedby a breeding female in age class i and fk is the proportionfemale kittens at birth (assumed to be 05)We estimated the finite deterministic (λ) and stochastic annual

population growth rate (λs) and stochastic sensitivities and elasti-cities following methods described in detail by Caswell (2001) Toestimate λs we assumed identically and independently distributedenvironments (Caswell 2001 Morris and Doak 2002 Haridas andTuljapurkar 2005) and used a parametric bootstrapping approachto obtain a sequence of demographic parameters for 50000 ma-trices We then estimated log(λs) from this sequence of matrices(Tuljapurkar 1990 Caswell 2001) and exponentiated it to obtain λsWe calculated the sensitivity and elasticity of λs to changes in lower‐level vital demographic rates as per Caswell (2001) and Haridas andTuljapurkar (2005)For population projection and population viability analysis (PVA)

simulations we expanded the population projection matrix into a34 times 34 2‐sex age‐structured matrix to include female and maleFlorida panthers (Caswell 2001 Hostetler et al 2013) Malescontribute to the population only through survival in this modelingframework We assumed maximum longevity for males to be 145years of age because the oldest male in our study was 144 years oldThe population projection matrix was of the following form

⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢

⋯ ⋯ ⋯ ⋯

⋯ ⋯ ⋯ ⋯

⋱ ⋱ ⋮ ⋮ ⋱ ⋱ ⋮

⋮ ⋱ ⋱ ⋱ ⋮ ⋮ ⋱ ⋱ ⋮

⋯ ⋯ ⋯

⋯ ⋯ ⋯ ⋯

⋯ ⋯ ⋱ ⋱ ⋮

⋮ ⋮ ⋱ ⋱ ⋮ ⋱ ⋱ ⋮

⋯ ⋯ ⋯

⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥

( )=

( ) ( ) ( )

( )

( )

( )

( ) ( ) ( )

( )

( )

t

F N F N F N

P N

P N

P N

M N M N M N

Q N

Q N

A n

0 0

0 0 0 0

0

0 0 0 0 0

0 0

0 0 0

00 0 0 0 0

t t t

t

t

t

t t t

t

t

1 2 19

1

2

18

1 2 19

1

14

where Pi and Qi are age‐specific survival rates of female and maleFlorida panthers respectively and Fi and Mi are the rates atwhich female and male kittens are produced by females of ageclass i respectively Here the population projection matrix isboth time‐varying and density‐dependent (Caswell 2001)We incorporated the influence of parameter uncertainty en-

vironmental stochasticity and density dependence on Floridapanther population dynamics and persistence following the ap-proach outlined by Hostetler et al (2013) we used the sameapproach in the IBM analysis which is described below UnlikeIBMs matrix models do not intrinsically include demographicstochasticity so we explicitly included the influence of

van de Kerk et al bull Florida Panther Population and Genetic Viability 13

demographic stochasticity by simulating the fates of individualsas described in Caswell (2001)Population projections using matrix models necessitate a

vector of initial sex‐ and age‐specific abundance Because suchdata are currently unavailable for the Florida panther we esti-mated it based on the stable sex and age distribution and MPCin 2013 Using point estimates of the demographic parameterswe estimated the stable sex and age distribution for the 2‐sexdeterministic and density‐independent population projectionmatrix We then multiplied the stable sex and age distributionby the MPC in 2013 to get the starting population vector n(0)We projected the population over time as n(t+ 1)=A(tn)n(t)where A(tn) indicates the time‐specific density‐dependentpopulation projection matrix (Caswell 2001)We had 2 indices of abundance available the MPC (McBride

et al 2008) and population size estimated using motor vehiclecollision mortalities (MVM McClintock et al 2015) These 2indices are substantially different because MPC does not ac-count for imperfect detection whereas MVM accounts forimperfect detection and provides an estimate of population size(Fig 2) Therefore we examined 2 scenarios one using densitydependence based on MPC and the other using density de-pendence based on MVM We ran 1000 bootstraps of para-meter values and 1000 simulations per bootstrap for each sce-nario We projected population trajectories over 200 years andthen estimated population growth rates probabilities ofquasi‐extinction and time until quasi‐extinction and comparedthese with the results obtained from the IBM (see below) Weran scenarios with a quasi‐extinction threshold of 10 and 30panthers We used the mean of probabilities of quasi‐extinction(PQE) across simulations as the point estimate for PQE and the5th and 95th percentiles across simulations as the 90 con-fidence interval Because the confidence intervals werequantile‐based but the point estimates were not it is possiblewith skewed distributions for the point estimates to be outsidethe confidence interval We implemented the matrix

population models in the R computing environment (R CoreTeam 2016)

IBMs and PVABecause of well‐developed theory ease of implementation andthe modeling flexibility that they offer matrix populationmodels have become the model of choice for investigating thedynamics and persistence of age‐ and stage‐structured popula-tions (Caswell 2001) However it is difficult to incorporateattributes of individuals such as genetics and behavior usingmatrix models Quantifying the effects of genetic erosion on theFlorida panther population dynamics and persistence andevaluating benefits and costs of alternative genetic managementscenarios were important goals of our study Because IBMs relyon a bottom‐up approach that begins by explicitly consideringindividuals (McLane et al 2011 Railsback and Grimm 2011Albeke et al 2015) it is possible to represent genetic attributesof each individual and to track genetic variation over time Weused IBMs as the primary modeling framework in this studybecause they allowed for explicit consideration of genetics bothat individual and population levels and population viabilityassessment under alternative genetic management scenariosWe developed an IBM (Grimm 1999 Grimm and Railsback

2005 McLane et al 2011 Railsback and Grimm 2011 Albekeet al 2015) tailored to the life history of the Florida panther(Fig 3) We simulated every individual from birth until deathbased on a set of rules describing fates (eg birth and death) andattributes (eg genetics) of individuals at each annual time stepAs in other IBMs population‐level processes (eg populationsize and genetic variation) emerged from fates of and interac-tions among individuals (Grimm 1999 Railsback and Grimm2011) We developed an overview design concepts and details(ODD) protocol (Grimm et al 2006 2010) to describe ourIBM (Appendix C available online in Supporting Information)and provide pseudocodes for individual‐based simulations(Appendix D available online in Supporting Information)

For each panther at time = t

Kitten (lt1 yr)

Reproduce

Survive

Kitten(s) born

Kitten

Sex and age class

Subadult male

Prime adult male

Old adult male

Subadult female

Prime adult female

Old adult female

Age one year

No Male

Female

Yes

No

Yes

t=t+1

Yes

lt35 yrs

35-10 yrs

gt10 yrs

lt25 yrs

25-10 yrs

gt10 yrs

Figure 3 Schematic representation of Florida panther life history We tailored the individual‐based population model to represent the life‐history patterndepicted here

14 Wildlife Monographs bull 203

Dynamics and persistence of populations are influenced bymany factors such as demographic and environmental sto-chasticity and density dependence these effects and parametricuncertainty must be considered in population models whenpossible (Caswell 2001 Boyce et al 2006 Bakker et al 2009Hostetler et al 2013) Because by definition IBMs simulate thelife history of individuals in a population they inherently in-corporate the effect of demographic stochasticity Our analysesrevealed that survival of kittens subadults and adults and theprobability of reproduction varied over time so we incorporatedenvironmental stochasticity in the model for these variablesfollowing Hostetler et al (2013) Likewise we found evidenceof density‐dependent effects on kitten survival Because theMPC might be an underestimate of population size (McBrideet al 2008) we performed simulations using both the MPC andMVM abundance indicesTo account for model selection and parametric uncertainty we

used a Monte Carlo simulation approach For each bootstrapsample we selected a model based on the AIC (or QAICc)weights We then sampled the intercept and slope for kittensurvival (on the logit scale) subadult and adult survival (on thelogit scale with log‐hazard effect sizes) and probability of re-production (on the complementary logndashlog scale Appendix B)from multivariate normal distributions with mean vectors equalto the estimated parameters and variance‐covariance matricesequal to the sampling variance‐covariance matrices We thentransformed the results to nominal scale and converted the es-timates to age‐specific survival and reproduction parameters asdescribed in Hostetler et al (2013)We ran 1000 simulations for 1000 parametric bootstraps for

200 years for the MPC and the MVM scenario for both thestochastic 2‐sex matrix‐population model and the IBM Wetracked the population size at each time step and estimatedquasi‐extinction probabilities and times based on the populationtrajectory We considered the population to be quasi‐extinct ifindividuals of only 1 sex remained or if the population size fellbelow critical thresholds set at 10 and 30 individuals We alsoestimated expected time to quasi‐extinction for both thresholdsdefined as the mean and median time until a population be-comes quasi‐extinct conditioned on quasi‐extinction We im-plemented the IBM using the RNetLogo package version 10ndash2(Thiele et al 2012 Thiele 2014) as an interface for the IBMplatform NetLogo (Wilensky 1999)

Sensitivity AnalysisAn important component of demographic analysis is sensitivityanalysis which is the quantification of the sensitivity of amodelrsquos outcome to changes in the input parameters (Caswell2001 Bakker et al 2009 Thiele et al 2014) Using an IBM weperformed global sensitivity analyses for 2 (ie MPC andMVM) density‐dependent scenarios to assess how sensitivequasi‐extinction times and probabilities were to changes (anduncertainties) in the estimated demographic rates We usedLatin hypercube sampling to sample parameters from the entireparameter space (Marino et al 2008 Thiele et al 2014) Wesampled intercept and slope for kitten survival (on logit scale)subadult and adult survival (on logit scale with log‐hazard effectsizes) and probability of reproduction (on complementary

logndashlog scale) from normal distributions We sampled theprobability distribution for the number of kittens producedannually (on the real scale) from a Dirichlet distribution themultivariate generalization of the beta distribution (Kotz et al2000) We sampled 1000 different parameter sets and ran 1000simulations for each scenario We calculated the probability ofquasi‐extinction within 200 years for each density‐dependencescenario (MPC or MVM) using a quasi‐extinction threshold of10 individuals and then calculated the partial rank correlationcoefficient between each of those and each parameter trans-formed to the real scale (Bakker et al 2009 Hostetler et al2013) Partial rank correlation coefficients quantify the sensi-tivity of the probability of quasi‐extinction to input variables(ie survival and reproductive parameters) with higher absolutevalues indicating stronger influence of that variable on quasi‐extinction probabilities

Genetic ManagementOne of our goals was to estimate the benefits (improvements ingenetic variation demographic parameters and populationgrowth and persistence parameters) and costs (financial costs ofcapture transportation quarantine release and monitoring ofintroduced panthers) of genetic management To achieve thisobjective we extended the IBM to include a genetic componentTo simulate individual‐ and population‐level heterozygosity overtime we assigned each panther alleles for 16 microsatellite locibased on the empirical allele frequency in the population (esti-mated from genetic samples collected during 2008ndash2015) whenwe initialized the model (Pierson et al 2015) At each time stepwe simulated Mendelian inheritance for kittens produced suchthat each kitten randomly inherited alleles from its mother andfrom a male panther randomly chosen to have putatively siredthe litter We did not explicitly force inbreeding to occur butthe probability of inbreeding naturally increased as the popula-tion size decreasedIntrogression strategiesmdashWe modeled genetic introgression as a

result of the release of 2‐year‐old female pumas from Texas intothe simulated Florida panther population For the geneticintrogression implemented in 1995 Texas females between 15and 25 years of age were released because females at that agehave lower mortality rates (Benson et al 2011) higherreproductive potential and also because it is much easier todetect with certainty the successful reproduction by females thanby males The ability to more closely monitor reproduction andthe birth of kittens is an important consideration in reducing theprobability of a genomic sweep of Texas alleles into the Floridapanther population Wildlife managers removed the last 2surviving Texas females from the wild population in Florida in2003 to reduce the possibility of genomic sweep To simulatethis strategy in the model we removed any Texas females thatwere still alive 5 years after each introgression event Weassigned Texas females alleles based on the allele frequency ofthe Texas population (based on genotypes from 48 samplescollected in west Texas that was inclusive of the pumas releasedin South Florida in 1995) when they entered the Florida pantherpopulation We could not estimate demographic rates separatelyfor the released Texas females because of small sample size (only8 females were released in 1995) therefore we assumed that

van de Kerk et al bull Florida Panther Population and Genetic Viability 15

survival and reproductive rates and the effects of covariates onthese rates for the released Texas females were identical to thoseof female Florida panthersThe impact of genetic introgression depends on the frequency

of introgression events and the number of individuals introducedat each attempt Thus we defined introgression strategies basedon the number of Texas females to be released (0 5 10 or 15)and the interval between genetic introgressions (10 20 40 and80 yr) for a total of 13 strategies We simulated each manage-ment strategy with density dependence estimated using theMPC and MVM abundance indices

We sampled 1000 parameter sets and for each of these setswe ran all introgression strategies 1000 times for 200 years Weapplied the same parametric uncertainty and environmentalstochasticity across all management scenarios to allow for bettercomparison of introgression strategies At each time step werecorded the projected population size and average individualheterozygosity We calculated the heterozygosity for each in-dividual at birth based on the alleles it inherited This individualheterozygosity then informed the survival rate of that individualbased on the empirically estimated relationship between in-dividual heterozygosity and age‐specific survival rates Hetero-zygosity did not change throughout an individualrsquos lifetime butits effect on survival changed as individuals aged because theeffect of individual heterozygosity on survival rate differedamong age classes Survival rates of genetically sampled kittenswere biased high because some kittens were sampled later in lifewhen they had already survived for some time To correct forthis bias we adjusted kitten survival rates predicted by theheterozygosity model proportionally From the population tra-jectories we estimated the probability of quasi‐extinction (de-fined as the probability that the population will fall below 10 or30 individuals or that individuals of only 1 sex will remain)

Cost‐benefit analysismdashExperts estimated financial costs ofintrogression based on their experience with the geneticintrogression executed in 1995 and we adjusted estimates tocurrent costs to account for inflation (R T McBride RancherrsquosSupply Incorporated M Cunningham and D P OnoratoFlorida Fish and Wildlife Conservation Commission personalcommunication) Costs included capturing and transportingfemale pumas from the Texas source population caring for thepumas while they were in quarantine and post‐introgressionmonitoring To compare the financial costs of the differentscenarios we also calculated the average costs incurred over

Table 2 Sample size (n) number of alleles (Na) allelic richness (AR) observedheterozygosity (Ho) and expected heterozygosity (He) at 16 microsatellite loci inradio‐collared Florida panthers sampled from 1981 to 2013 in South FloridaUSA We calculated these values from a subset (n= 137) of the samples used inthe analyses and they include only adult and subadult panthers We excludedkittens because they can result in an overrepresentation of certain alleles

Locus n Na AR Ho He

FCA090 129 5 50 0333 0401FCA133 136 3 30 0618 0608FCA243 136 5 50 0419 0487F124 137 7 69 0708 0729F37 129 4 40 0395 0397FCA075 131 5 50 0534 0598FCA559 133 5 50 0496 0503FCA057 134 6 59 0679 0624FCA081 134 7 69 0209 0206FCA566 130 6 60 0462 0475F42 133 10 100 0737 0766FCA043 136 5 49 0596 0588FCA161 132 6 60 0500 0515FCA293 132 4 40 0614 0624FCA369 131 6 60 0573 0567FCA668 133 7 70 0406 0462Mean 1329 57 57 0517 0535

Figure 4 Proportional membership (q) of radio‐collared Florida panthers (n= 218) and pumas from Texas USA (n= 7) in 2 clusters identified by STRUCTUREEach individual animal is represented by a separate vertical bar Yellow indicates canonical panther ancestry and blue represents admixed ancestry The admixedancestry of the Texas females and F1 generation panthers that were radiocollared are highlighted red and green respectively to demarcate those groups The pre‐introgression period is inclusive of panthers radiocollared from 10 February 1981 to 6 March 1996 The post‐introgression era inclusive of the F1 panthers includesanimals radiocollared from 4 March 1997 to 18 February 2015 in South Florida USA

16 Wildlife Monographs bull 203

100 years assuming that the population would be managedaccording to a particular strategy Our goal with theseexploratory analyses was to evaluate the relative costs andbenefits of alternative management scenarios

RESULTS

Genetic VariationWe assessed genetic variation at 16 microsatellite loci for 137DNA samples collected from adult and subadult panthers(1981ndash2013) that were captured and subsequently radiocollaredThe number of alleles at these loci ranged from 3ndash10 and allelicrichness averaged 57 (Table 2) Average expected hetero-zygosity for this sample was 0535 and average observed het-erozygosity was 0517 (Table 2)We determined individual heterozygosity (1 minusHL) and an-

cestry for the complete sample of panther genotypes (n= 503)collected from 1981 to 2015 for use as covariates in subsequentanalyses Average individual heterozygosity was 0559 andranged from 0ndash0980 Our STRUCTURE analysis providedstrong support for K= 2 clusters based on the probability of thedata (log Pr(D)|K) and ΔK in STRUCTURE HARVESTERBased on this result we categorized the ancestry of each pantheras either canonical (n= 123) or admixed (n= 380) using valuesof proportional membership (q Fig 4) The average individualheterozygosity of canonical panthers was 0386plusmn 0012 (SE) foradmixed panthers it was 0615plusmn 0007

Survival and Cause‐Specific Mortality RatesOf the 395 kittens that were PIT‐tagged and released 55 werelater encountered alive 15 were recovered dead and 85 werepart of failed litters (Table 1) We radio‐tracked 209 subadult oradult panthers for a total of 244195 panther‐days and docu-mented 126 mortalities (Table 1)Survival depended strongly on age and kittens had the lowest

overall rate (032plusmn 009 Fig 5A Table 3) The model‐averagedkitten survival was 047plusmn 009 when we included only geneti-cally sampled individuals in the analysis (Table 3) as statedabove this estimate is biased high because many kittenswere genetically sampled when they were recaptured alive orrecovered dead at older ages We found no evidence for sex‐specific differences in kitten survival but females survived betterthan males in all older age classes For females subadults hadthe highest survival rates followed by prime adults and oldadults For males prime adults had the highest survival ratefollowed by subadults and old adults (Fig 5A and Table 3)For panthers of all ages there was substantial evidence that

ancestry affected survival with F1 admixed panthers of all ageshaving the highest rates and canonical individuals having thelowest (Fig 5B) The combined AIC weights of models thatincluded an ancestry covariate were 0880 for kittens and 0992for older panthers The survival rates of subadult prime‐adultand old‐adult panthers in the period immediately after the in-trogression (1995ndash2004) were similar to those in more recentyears (2005ndash2013 Fig 5C)After genetic introgression individual panther heterozygosity

increased (Fig 6) and varied among ancestry categories individualheterozygosity was the highest for F1 panthers (078plusmn 001)

A

B

C

Figure 5 Overall age‐ and sex‐specific survival rates (plusmnSE A) age‐ and sex‐specific survival rates (plusmnSE) for Florida panthers of canonical F1 admixed andother admixed ancestry (B) and age‐ and sex‐specific survival estimates (plusmnSE)for subadult and adult Florida panthers in the period immediately post‐introgression (1995ndash2004) and more recently (2005ndash2013 C) in SouthFlorida USA

Table 3 Model‐averaged estimates (plusmnSE) of demographic parameters for theFlorida panther obtained using data collected during 1981ndash2013 in SouthFlorida USA

Parameter Sex Age class Estimate

Survival Both Kitten 032plusmn 009a

Female Subadult 097plusmn 002Prime adult 086+ 003Old adult 078plusmn 009

Male Subadult 066plusmn 006Prime adult 077plusmn 005Old adult 065plusmn 010

Probability of reproduction Female Subadult 035plusmn 008Prime adult 050plusmn 005Old adult 025plusmn 006

Kittens produced annually Female Subadult 280plusmn 005Prime adult 267plusmn 001Old adult 228plusmn 013

a Overall kitten survival (based on all kittens in our data set including thosethat were not genetically sampled) Estimate of survival of geneticallysampled kittens was 047plusmn 009 this estimate is biased high because manykittens were genetically sampled when they were recaptured alive or re-covered dead at older ages

van de Kerk et al bull Florida Panther Population and Genetic Viability 17

followed by those for non‐F1 admixed (062plusmn 001) and canonical(039plusmn 001) panthers Individual heterozygosity positively affectedsurvival rates of kittens (slope parameter η= 1455 95CI= 0505ndash2405) Individual heterozygosity also positively af-fected survival of subadult and adult panthers but the evidence forthis effect was weaker because the confidence interval for the ha-zard ratio overlapped 10 (hazard ratio exp(ɩ)= 0311 95CI= 0092ndash1054 Table 4 and Fig 7)The MPC increased after genetic introgression (Fig 2) This

abundance index was also included in top‐ranking modelsindicating that population density affected survival (Table 4)Abundance reduced the survival of kittens (slope parameterη= minus0035 95 CI= minus0049 to minus0021) evidence was weak forthe effect of abundance on survival of subadult and adult panthers(hazard ratio exp(ɩ)= 1002 95 CI= 0995ndash1010 Fig 7) Re-gression coefficients associated with the effect of abundance andindividual heterozygosity on demographic parameters are presentedin Table B1 (available online in Supporting Information)The most frequently observed causes of death for radio‐

collared panthers were vehicle collision and intraspecific ag-gression Cause‐specific mortality analyses revealed thatmortality rates among possible causes were generally similar forthe 2 sexes but that males were more likely to die from in-traspecific aggression than were females (Fig 8A) Male mor-tality from intraspecific aggression was higher for subadult andold adult panthers than for prime adults (Fig 8B) For bothmales and females canonical panthers were more likely to diefrom intraspecific aggression than were admixed panthers(Fig 8C)

Reproductive ParametersOf 101 radio‐collared females 67 reproduced at least once duringour monitoring we used these individuals to assess the annualprobability of reproduction The top‐ranked model for the annual

probability of reproduction included age effect only suggesting thatadding ancestry or abundance did not improve model parsimony(Table 4) Model‐averaged annual probabilities of reproductiondiffered between female subadults (035plusmn 008) prime adults(050plusmn 005) and older adults (025plusmn 006)The most parsimonious model for the number of kittens

produced annually by females that produced at least 1 litter (ieconditional on reproduction) included effects of age ancestryand abundance a model that included only the effect of age wasalmost equally supported (Table 4) The model‐averagednumber of kittens produced annually conditional on re-production was 280plusmn 075 for subadults 267plusmn 043 for primeadults and 228plusmn 083 for old adults Confidence intervals forthe slope parameter (β) overlapped 0 for both abundance(β= 0010 95 CI= minus0001 to 0021) and ancestry (β= minus073795 CI= minus1637 to 0162)

Population Dynamics and PersistenceDeterministic and stochastic demographymdashThe deterministic (λ)

and stochastic (λs) annual population growth rate calculated usingthe matrix population model were 106 (5th and 95th percentiles099ndash114) and 104 (072ndash141) respectively The generation timewas 47 years A female starting in age class one (kitten) wasexpected to live another 58 years and produce 10 kittens onaverage during her lifetime Overall λs was proportionately(elasticity) most sensitive to changes in female prime adultsurvival on the absolute scale (sensitivity) however it was mostsensitive to changes in kitten survival (Fig 9) Populationtrajectories probabilities of quasi‐extinction and times untilquasi‐extinction estimated using the matrix population modeland IBM for comparable scenarios were nearly identical for acritical quasi‐extinction threshold of 10 panthers (Figs 10 and 11)and for a critical quasi‐extinction threshold of 30 panthers(Appendix E available online in Supporting Information)Minimum population count scenariomdashUnder the MPC scenario

(ie density dependence estimated using the minimumpopulation count data) with a critical threshold of 10panthers the population size reached 99 (72ndash104) and 100(77ndash105) at 200 years (Fig 10) as predicted by the IBM and thematrix model respectively The cumulative quasi‐extinctionprobabilities within 100 years based on the MPC scenario were15 (IBM 0ndash48) and 30 (matrix model 0ndash76) Theseprobabilities increased to 25 (IBM 0ndash114) and 38 (matrixmodel 0ndash154) by year 200 (Fig 10) The mean times untilquasi‐extinction were 15 years (IBM 0ndash67) and 15 years (matrixmodel 0ndash72) conditioned on going quasi‐extinct within 100years Expected times until quasi‐extinction conditioned onquasi‐extinction within 200 years were higher 34 years (IBM0ndash137) and 32 years (matrix model 0ndash134 Fig 10)Motor vehicle mortality scenariomdashUnder the MVM scenario

(ie density dependence estimated using estimates of pantherpopulation size from McClintock et al 2015) with a criticalthreshold of 10 panthers the projected population reached 188(142ndash217) and 190 (143ndash219) at 200 years as predicted by theIBM and the matrix model respectively (Fig 11) Thecumulative quasi‐extinction probabilities within 100 years were14 (IBM 0ndash08) and 13 (matrix model 0ndash06) Theseprobabilities increased to 20 (IBM 0ndash17) and 19 (matrix

000

025

050

075

100

1995 2000 2005 2010Year of birth

Indi

vidu

al h

eter

ozyg

osity

Figure 6 Temporal changes in the individual heterozygosity of Floridapanthers born during the period 1995ndash2013 in South Florida USA Pointsare measurements solid line is linear regression shaded area is 95 confidenceinterval of slope Slope= 0006plusmn 0001 R2= 009

18 Wildlife Monographs bull 203

model 0ndash16) within year 200 (Fig 11) The mean times until quasi‐extinction based on the MVM scenario were 5 years (IBM 0ndash61)and 6 years (matrix model 0ndash64) conditioned on going quasi‐extinctwithin 100 years Expected times until quasi‐extinction conditionedon quasi‐extinction within 200 years were 11 years (IBM 0ndash113)and 11 years (matrix model 0ndash112 Fig 11)

Sensitivity of extinction probability to demographic parametersmdashThe partial rank correlation coefficients between quasi‐extinctionprobabilities and demographic parameters were negative anincrease in any of the demographic parameters decreased thequasi‐extinction probability Probabilities of quasi‐extinction within200 years were most sensitive to changes in kitten survival for bothscenarios (Fig 12) followed by survival of subadult females in theMVM scenarios and survival of prime adult females in the MPCscenario The probability of reproduction of prime adult femaleshad the third‐largest partial rank correlation coefficient with quasi‐extinction probability for both scenarios

Benefits and Costs of Genetic ManagementMinimum population count scenariomdashThe MPC scenario with

a critical threshold of 10 panthers predicted that withoutmanagement intervention average individual heterozygositywould decrease reaching 057 (049ndash064) at 100 years and053 (042ndash062) at 200 years (Fig 13) The population woulddecrease slightly reaching an average of 79 (41ndash99) at 100 yearsand 76 (43ndash98) at 200 years (Fig 14) The probabilities of quasi‐extinction under the no‐intervention MPC scenario when weconsidered the effects of genetic erosion were 13 (0ndash99) at 100years and 23 (0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with

the translocation of 5 Texas females average individualheterozygosity under the MPC scenario increased and thenstabilized at 068 (064ndash072) at 100 years and remained atthat level at 200 years (Fig 13) The population reached anaverage of 88 (62ndash104) at 100 years and stayed the same at200 years (Fig 14) The probabilities of quasi‐extinctionunder this introgression strategy were 6 (0ndash53) within 100years and 7 (0ndash82) within 200 years (Fig 15) Results ofanalyses using a critical threshold of 30 panthers revealed asimilar pattern of relative benefits However the prob-abilities of quasi‐extinction were substantially higher (ge45)across all scenarios (Appendix E)

Motor vehicle mortality scenariomdashWithout managementintervention the average individual heterozygosity predictedby the MVM scenario and a critical threshold of 10 panthersdecreased to 060 (046ndash069) at 100 years and to 059 (039ndash070) at 200 years (Fig 13) The population increased slightlyand reached an equilibrium at 154 individuals (46ndash217) at 100years and 157 (57ndash218) at 200 years (Fig 14) The probabilitiesof quasi‐extinction were 17 (0ndash100) at 100 years and 22(0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with 5

female pumas from Texas average individual heterozygosityincreased slightly reaching 067 (060ndash073) at 100 years and068 (059ndash075) at 200 years (Fig 13) Under this strategy thepopulation reached an average of 164 individuals (44ndash226) at

Table 4 Model comparison results testing for the effects of several covariateson survival of all kittens survival of genetically sampled kittens survival ofsubadult and adult Florida panthers probability of reproduction and kittensproduced annually for Florida panthers sampled from 1981ndash2013 in SouthFlorida USA

Model Parameters ΔAICa Weight

Kitten survival (all data)Base+ F1b+ abundancec 10 000 0988Base+ abundance 9 1018 0006Base+ F1 9 1035 0005Base 8 2711 lt0001Base+ sexd 9 2912 lt0001Base+ litter sizee 9 2914 lt0001

Kitten survival (genetically sampled kittens only)Base+ F1+ abundance 10 000 0541Base+ F1CanAdmf+ abundance 11 099 0329Base+Hetg+ abundance 10 310 0115Base+CanAdmh+ abundance 10 791 0010Base+ abundance 9 944 0005Base+ F1 9 1638 lt0001Base+ F1CanAdm 10 1837 lt0001Base+Het 9 2872 lt0001Base 8 3296 lt0001Base+CanAdm 9 3348 lt0001

Subadult and adult survivalBase+ F1CanAdm+ abundance 7 000 0360Base+ F1 5 053 0276Base+ F1CanAdm 6 105 0213Base+ F1+ abundance 6 222 0119Base+CanAdm+ abundance 6 575 0020Base+CanAdm 5 908 0004Base+Het 5 964 0003Base+Het+ abundance 6 984 0003Base 4 1118 0001Base+ abundance 5 1278 0001

Probability of reproductionAge 3 000 0242Age+CanAdm 4 012 0228Age+ abundance 4 173 0102Age+ F1 4 190 0094Age+CanAdm+ abundance 5 216 0082Age+Het 4 219 0081Age+ F1CanAdm 5 232 0076Age+ F1+ abundance 5 372 0038Age+Het+ abundance 5 399 0033Age+ F1CanAdm+ abundance 6 444 0026

Kittens produced annuallyAge+ F1+ abundance 5 000 0194Age 3 060 0144Age+ abundance 4 061 0143Age+ F1 4 097 0120Age+CanAdm 4 139 0097Age+ F1CanAdm+ abundance 6 196 0073Age+ F1CanAdm 5 231 0061Age+CanAdm+ abundance 5 238 0059Age+Het+ abundance 5 250 0056Age+Het 4 259 0053

a Akaikersquos information criterion (AIC) for kitten survival models was adjustedfor small sample size and overdispersion (QAICc)

b F1 divides Florida panthers into 2 groups F1 admixed and all other panthersc Index of Florida panther abundanced Sex of an individual Florida panther (male or female)e Size of the litter in which a Florida panther kitten was bornf F1CanAdm divides panthers into 3 groups F1 admixed canonical andother admixed panthers

g Het is the individual heterozygosityh CanAdm divides panthers into 2 groups canonical and admixed (includingF1 admixed) panthers

van de Kerk et al bull Florida Panther Population and Genetic Viability 19

100 years and 170 (67ndash231) at 200 years based on the MVMscenario (Fig 14) The probabilities of quasi‐extinction underthis introgression strategy were 10 (0ndash99) at 100 years and12 (0ndash99) at 200 years (Fig 15)The projected population size and average individual hetero-

zygosity reached equilibrium levels with periodic fluctuations inthese values when introgression was implemented (Figs 16ndash17)Genetic introgression decreased the time until quasi‐extinctionacross all scenarios (Fig 18) Results of analyses using a criticalthreshold of 30 panthers revealed a similar pattern of relative ben-efits However the probabilities of quasi‐extinction were sub-stantially higher (ge45) across all scenarios (Appendix E)

Addition of pumas from Texas increased both the populationsize and average individual heterozygosity consequentlyintrogression caused periodic increases in average individualheterozygosity and population size at the interval at which theintrogression occurred Whereas average individual hetero-zygosity gradually decreased following introgression densitydependence caused the population size to fluctuate especiallywhen a larger number of pumas from Texas were released Thepositive effect of genetic introgression initiatives on quasi‐extinction probabilities decreased as the time interval betweenthese events increased More frequent genetic introgressions ledto greater increases in average individual heterozygosity and

Old adult

Prime adult

Subadult

Kitten

025 050 075

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Individual heterozygosity

Ann

ual s

urvi

val r

ate

Old adult

Prime adult

Subadult

Kitten

40 60 80 100 120

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Abundance index

Ann

ual s

urvi

val r

ate

Kittens Males Females

Figure 7 The effect of individual heterozygosity (left) and the abundance index (right) on Florida panther age‐ and sex‐specific survival estimates (shaded area isSE) in South Florida USA 1981ndash2013 Abundance index data are from McBride et al (2008) and R T McBride and C McBride (Rancherrsquos Supply Incorporatedunpublished report)

20 Wildlife Monographs bull 203

equilibrium population size while reducing the probability ofquasi‐extinction Comparatively performing genetic introgres-sion less often but with more individuals per release was lesseffective at improving the long‐term prospects for the pantherpopulation (Figs 13ndash15)

Financial cost and persistence benefitsmdashThe total estimated costof 1 genetic introgression was $40000 $80000 or $120000for releasing 5 10 or 15 pumas from Texas respectively(Table 5) The total cost incurred over 100 years for eachstrategy ranged from $50000 to $12 million (Table 6)The most expensive strategy involved the introduction of15 pumas every 10 years this strategy reduced the probability

of quasi‐extinction by 58 and 40 under the MPC and MVMscenarios respectively (Table 6) Introducing 5 pumas every80 years cost the least but improvements in populationpersistence under this strategy were insubstantial (Table 6)Releasing more panthers more frequently caused increasedpopulation fluctuations but did not necessarily reduce the quasi‐extinction probability (Figs 14 and 15 Table 6)

DISCUSSION

The duration (34 yr of data) and sample sizes associated with theFlorida Panther Project allowed us to obtain a comprehensiveunderstanding of genetic variation demographic parameters

A

B

C

Figure 8 Cause‐specific mortality rates (plusmnSE) for male and female Florida panthers (A) subadult prime‐adult and old‐adult Florida panthers of both sexes (B)and canonical and admixed Florida panthers of both sexes (C) in South Florida USA 1981ndash2013 Causes of mortality are vehicle collision intraspecific aggression(ISA) other causes (known causes such as diseases injuries and infections unrelated to the first 2 causes) and unknown cause (mortalities for which we could notassign a cause)

van de Kerk et al bull Florida Panther Population and Genetic Viability 21

probability of population persistence and the duration of thepositive impacts of genetic restoration on this endangered po-pulation The Florida panther population was in dire straits inthe early 1990s and our results clearly demonstrated that the

implementation of the introgression project in 1995 subse-quently accrued a series of genetic and demographic improve-ments that have led to the change from a declining population toone that is growing and expanding its current breeding range

Sensitivity Elasticity

0 1 2 3 00 02 04

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

Para

met

er

Figure 9 Sensitivity and elasticity (proportional sensitivity) of stochastic population growth rate (λs) of the Florida panther to vital demographic parameters in SouthFlorida USA 1981ndash2013 Horizontal lines represent 5th and 95th percentiles

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

70

80

90

100

110

0

5

10

15

0

50

100

Years

StatisticMeanMedian

Figure 10 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the minimum population count(MPC) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

125

150

175

200

00

05

10

15

20

0

30

60

90

Years

StatisticMeanMedian

Figure 11 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the motor vehicle mortality(MVM) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

22 Wildlife Monographs bull 203

MPC MVM

minus08 minus06 minus04 minus02 00 minus08 minus06 minus04 minus02 00

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Male subadult survival

Male prime adult survival

Male old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

PRCC

Para

met

er

Figure 12 Partial rank correlation coefficient (PRCC) between quasi‐extinction probabilities and the demographic parameters of the Florida panther for theminimum population count (MPC) and motor vehicle mortality (MVM) scenarios Error bars indicate standard deviation Based on data collected in South FloridaUSA 1981ndash2013

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 13 Average projected individual heterozygosities in the Florida panther population over 200 years without genetic management intervention and with eachof the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include theintroduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years Average individual heterozygosity peaks when pumas are added to the Floridapanther population Based on data collected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 23

Our research has further validated the success of genetic in-trogression by quantifying the reduction in the probability ofextinction of the panther population because of this manage-ment initiative These findings all bode well for continuedprogress toward recovery That said the Florida panther po-pulation remains small and isolated from other populations ofpuma from western North America thereby denoting that theeventual impact of genetic drift and inbreeding may negativelyaffect the population in the future Realizing this will continueto be an issue that managers will have to contemplate wemodeled the impact of varied genetic management scenarios todetermine the frequency of introgression along with the numberof panthers that should be released during each initiative tominimize cost and maximize benefits to the population Theseresults will prove invaluable to managers attempting to conservethe Florida panther

Genetic Variation Demographic Parametersand Persistence of Heterotic Benefits from GeneticIntrogressionOur measure of individual heterozygosity (1 minusHL) was higheron average for the admixed (0615) panthers than for those ofcanonical ancestry (0386) Levels of individual heterozygosityfor admixed panthers were comparable to levels observed in asample of pumas from west Texas (x plusmn SE= 0695plusmn 0019n= 41) which further demonstrates the improvement in levelsof genetic variation in the Florida population since 1995 The

correlation between HL and several demographic parametershas been noted in wild populations including cheetahs (Terrellet al 2016) and several species of birds such as European shags(Phalacrocorax aristotelis male and female survival probabilityVelando et al 2015) and Egyptian vulture (Neophron percnop-terus age at recruitment Agudo et al 2012) If we focus only onpanthers captured and radiocollared from 2012ndash2015 (FP211ndashFP240) all of which had estimated birth years ge2005 (ge10 yrpost‐introgression) those panthers had an average individualheterozygosity level of 0618plusmn 0029 highlighting the elevatedlevels of genetic variation that continue to persist in cohorts ofpanthers 5 generations after the release of female pumas fromTexasThe proportional membership values (q) from our

STRUCTURE analysis allowed us to delineate 2 clusters ofancestry for Florida panthers (Fig 4) During the pre‐in-trogression era the population was dominated by panthers thatretained a high percentage of the canonical ancestry which wasaffected by inbreeding depression The post‐introgression eraancestry values are comprised of many admixed individuals inessence demonstrating the success of the introgression in termsof increasing genetic variation in the population and mi-micking gene flow that historically occurred with panthers andother populations of pumas (Onorato et al 2010) A somewhatanalogous situation although abetted by natural immigrationwas evident in the ancestral variation in a small population ofpuma intersected by a major freeway in California USA In

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

75

80

85

90

95

100

130

150

170

190

Years

Popu

latio

n si

ze

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 14 Average projected population sizes in the Florida panther population over 200 years without intervention and with each of the genetic introgressionstrategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include the introduction of 5 10 or 15pumas from Texas USA every 10 20 40 or 80 years Peaks occur when pumas are added to the Florida panther population Based on data collected in SouthFlorida USA 1981ndash2013

24 Wildlife Monographs bull 203

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

25

50

75

100

0

25

50

75

100

Introgression interval (yrs)

PQ

E (

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 15 Average probability of quasi‐extinction (PQE) of the Florida panther population within 100 years and within 200 years without genetic managementintervention (N) and with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenariosIntrogression strategies include the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Errorbars indicate 5th and 95th percentiles Based on data collected in South Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

50

60

70

80

90

100

110

50

100

150

200

Years

Popu

latio

n si

ze

Figure 16 Average projected Florida panther population trajectories under a geneticintrogression strategy involving the release of 5 female pumas from Texas USA every20 years for the minimum population count (MPC) and motor vehicle mortality(MVM) scenarios Shaded area indicates 5th and 95th percentiles and isrepresentative of confidence intervals for other scenarios Based on data collected inSouth Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Figure 17 Average projected Florida panther population‐level heterozygositiesunder a genetic introgression strategy involving the release of 5 female pumas fromTexas USA every 20 years for the minimum population count (MPC) and motorvehicle mortality (MVM) scenarios Shaded area bounds the 5th and 95th percentilesand is representative of confidence intervals for other scenarios Based on datacollected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 25

this case the migration of just a single male across the freewayto the south resulted in an increase in the number of in-dividuals with admixed ancestry and substantially improvedgenetic diversity of the subpopulation south of the freeway(Riley et al 2014)Our estimate of overall kitten survival (0321plusmn 0056) was

similar to that reported by Hostetler et al (2010) who used13 years of post‐introgression data (0323plusmn 0071) These valuesare lower than kitten survival estimates derived from severalpuma populations in western North America (044ndash091 Loganand Sweanor 2001 Lambert et al 2006 Laundre et al 2007Robinson et al 2008 Ruth et al 2011) When we included onlydata for individuals that had been genetically sampled our es-timate was 15 higher The proportion of admixed kittens thatwere genetically sampled increased as the population expandedwhich could have resulted in higher estimates of kitten survivalbecause admixed kittens survive better than canonical kittensKitten survival was strongly influenced by genetic ancestry withsurvival rates of F1 kittens being almost twice that of canonicalkittens (Fig 5B) Consistent with the findings of Hostetler et al(2010) increases in heterozygosity coincided with increasedkitten survival whereas population density negatively affectedkitten survival Population density often has a strong negativeimpact on survival of young age classes due to infanticide andcannibalism which has been reported from several carnivore

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

50

100

150

0

50

100

150

Introgression interval (yrs)

TQE

(yrs

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 18 Average times until quasi‐extinction (TQE) for the Florida panther population within 100 years and within 200 years without genetic management interventionand with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategiesinclude the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Error bars indicate 5th and 95thpercentiles Based on data collected in South Florida USA 1981ndash2013

Table 5 Average cost (2017 US dollars) of introgression strategies employedover 100 years that involve the introduction of 5 10 or 15 pumas from TexasUSA into the Florida panther population in South Florida USA at an in-creasingly higher frequency Costs of capture (including transportation) quar-antine and post‐introgression monitoring were estimated by Roy McBrideMark Cunningham and Dave Onorato respectively

Frequency Component 5 pumas 10 pumas 15 pumas

Once Capture 25000 50000 75000

Quarantine 5000 10000 15000

Monitoring 10000 20000 30000

Total 40000 80000 120000

Every 80 years Capture 31250 62500 93750

Quarantine 6250 12500 18750

Monitoring 12500 25000 37500

Total 50000 100000 150000

Every 40 years Capture 62500 125000 187500

Quarantine 12500 25000 37500

Monitoring 25000 50000 75000

Total 100000 200000 300000

Every 20 years Capture 125000 250000 375000

Quarantine 25000 50000 75000

Monitoring 50000 100000 150000

Total 200000 400000 600000

Every 10 years Capture 250000 500000 750000

Quarantine 50000 100000 150000

Monitoring 100000 200000 300000

Total 400000 800000 1200000

26 Wildlife Monographs bull 203

species including polar bears (Ursus maritimus Derocher andWiig 1999) black bears (Ursus americanus Garrison et al 2007)and pumas (Logan and Sweanor 2001)Our estimates of sex‐ and age‐specific survival rates of subadult

and adult panthers (Table 3) and the effects of covariates on theserates were similar to those reported by Benson et al (2011) derivedfrom data collected through 2006 Our survival rates for males(065ndash077) were lower than females (078ndash097) for all 3 ageclasses (subadult prime adult and old adult) which is the typicaltrend observed in most puma populations (eg Ruth et al 19982011) However an unhunted puma population occupying afragmented urban landscape in southern California exhibited lowersurvival (0586 females 0525 males Vickers et al 2015) perhapsas a result of severe habitat fragmentation and the lack of extensiveswaths of public land protected in perpetuity Most populations ofpuma in western North America are typically subject to variedlevels of management via regulated hunting which often is biasedtoward the take of males (Cooley et al 2009 Ruth et al 2011)Consequently annual survival estimates from hunted populationsin northeast Washington (0679ndash0733 females 0341 malesRobinson et al 2008) and southeastern Arizona (0685 females0584 males Cunningham et al 2001) were generally lower thanthose we estimated for Florida panthersCause‐specific mortality analysis showed that male panthers

experienced a substantially greater mortality risk from in-traspecific aggression This differs from the pattern observedduring a long‐term study in Arizona where females were at ahigher risk of intraspecific mortality than males (46 of maleand 53 of female mortality Logan and Sweanor 2001)Mortality associated with intraspecific strife has been docu-mented in hunted and unhunted populations (this study Loganand Sweanor 2001 Stoner et al 2010) and its root cause hasbeen difficult to determine (eg population density and preypopulation trends) That being the case finding ways to reducewhat is often a major source of mortality for puma populationscontinues to pose a challenge to managersCanonical panthers were substantially more likely to die from

intraspecific aggression than were admixed panthers This mayat least partly explain the difference in survival between admixed

and canonical panthers Canonical panthers are known to bemore likely to suffer from varied correlates of inbreeding de-pression than admixed animals including atrial septal defects(Johnson et al 2010) and compromised immune systems(Roelke et al 1993) These factors have the potential to affectfitness and perhaps dominance of individuals when engaged inan intraspecific encounter A somewhat similar scenario wasobserved by Hogg et al (2006) in a population of bighorn sheepthat were part of an introgression project Admixed males in thispopulation had a greater probability of paternity than males thatwere less outbred This included coursing males with higherlevels of admixture being markedly more successful at fightingmales that were defending females in estrous (Hogg et al 2006)Heterosis resulting from genetic introgression is expected to be

the strongest in F1 individuals (Shull 1908 Crow 1948 Burkeand Arnold 2001 Waller 2015) and to progressively decline insubsequent generations (Pickup et al 2013 Frankham 2015)Given that admixed (especially F1) panthers survived better thancanonical panthers and that individual heterozygosity improvedsurvival of both sexes and all age classes our data provide evi-dence for heterotic advantage whereby individuals of mixedancestry had higher fitness (Shull 1908 Crow 1948) Our resultsare consistent with findings of other studies that involved in-tentional genetic introgression for conservation purposes Forexample Hogg et al (2006) detected substantial improvementsin survival and reproduction of introgressed bighorn sheep andMadsen et al (1999 2004) reported a dramatic increase in thepopulation of adders after genetic introgressionKnowledge of the persistence of benefits to a population ac-

crued via genetic introgression is essential for determiningwhen additional introgression may be warranted There is adepauperate amount of information regarding this topic and theidentification of factors that may influence persistence of thebenefits of genetic introgression (Tallmon et al 2004 Hedricket al 2014 Whiteley et al 2015) For example in minks(Neovison vison) Thirstrup et al (2014) found that outcrossingled to larger litters in F2 and F3 but this benefit disappeared insubsequent generations In a population of gray wolves Hedricket al (2014) suggested that benefits of genetic introgression may

Table 6 Costs and benefits accumulated over 100 years for genetic introgression at different intervals and with different numbers of pumas from Texas USAintroduced into the Florida panther population in South Florida USA Costs (2017 US dollars) include capture and transportation quarantine and post‐introgression monitoring Benefits are represented as average probability of quasi‐extinction (PQE and 5th and 95th percentiles) and changes (Δ) therein under thescenario using the minimum population count (McBride et al 2008 MPC) and the scenario using the motor vehicle mortality population estimates (McClintocket al 2015 MVM) The table is sorted by percent change in probability of quasi‐extinction under the MPC scenario

Interval (yr) Pumas Cost MPC PQE () ΔMPC PQE () MVM PQE () ΔMVM PQE ()

ndash 0 0 13 (0ndash99) 0 17 (0ndash100) 080 10 100000 12 (0ndash99) 10 (0ndash58) 16 (0ndash100) 5 (0ndash38)80 15 150000 12 (0ndash99) 13 (0ndash65) 15 (0ndash100) 8 (0ndash56)80 5 50000 12 (0ndash99) 14 (0ndash73) 15 (0ndash99) 9 (0ndash41)40 15 300000 10 (0ndash98) 26 (0ndash104) 14 (0ndash99) 13 (0ndash60)40 10 200000 9 (0ndash94) 35 (0ndash183) 13 (0ndash99) 21 (0ndash95)40 5 100000 8 (0ndash89) 39 (0ndash211) 12 (0ndash99) 26 (0ndash124)20 15 600000 8 (0ndash88) 42 (0ndash257) 13 (0ndash99) 24 (0ndash123)20 10 400000 6 (0ndash67) 54 (0ndash318) 10 (0ndash99) 37 (0ndash217)10 15 1200000 6 (0ndash53) 58 (0ndash372) 10 (0ndash99) 40 (0ndash208)20 5 200000 5 (0ndash50) 59 (0ndash338) 9 (0ndash99) 44 (0ndash352)10 10 800000 4 (0ndash16) 69 (0ndash489) 8 (0ndash92) 55 (0ndash401)10 5 400000 4 (0ndash7) 73 (0ndash502) 6 (0ndash71) 63 (0ndash503)

van de Kerk et al bull Florida Panther Population and Genetic Viability 27

wane after 2 or 3 generations The WrightndashFisher model ofgenetic drift predicts a geometric decline in expected hetero-zygosity at the rate of (1 minus 12Ne) per generation where Ne is theeffective population size (Hartl and Clark 1997 Frankham et al2014) Thus it seems logical to assume that the duration of thebenefits of introgression is limited especially in a species per-sisting in a small population such as Florida panthersWe expected Florida panther survival in the most recent years

(2005ndash2013) post‐introgression to differ from that observed in thedecade following the introgression in 1995 because pantherabundance and heterozygosity factors known to influence panthersurvival (Hostetler et al 2010 Benson et al 2011) have changed(Figs 2 and 6) We found that subadult and adult survival rates inrecent years (2005ndash2013) were similar to those in the period im-mediately following introgression (1995ndash2004 Fig 5C) overallkitten survival was similar to estimates of Hostetler et al (2010) Infact the pattern of covariate effects on Florida panther survivalreported by Benson et al (2011) and Hostetler et al (2010) hasbarely changed The negative effects of population density andpositive effects of heterozygosity may counterbalance survival ratesreducing population fluctuations Our result that benefits of geneticrestoration have remained in the population nearly for 5 genera-tions post‐intervention was surprising but also encouraging Pos-sible explanations for this observation may include 1) genetic ero-sion occurs at slower rate than previously thought 2) introducing 5migrants in 1 genetic intervention may have beneficial effects si-milar to those from 1 migrant per generation over multiple gen-erations or 3) other and as yet unknown factors may reduce therate of genetic erosion and subsequent demographic effects Adensity‐dependent effect on kitten survival could also explain whynon‐F1 admixed kittens did not survive substantially better thandid canonical kittens most kittens sampled from 2005ndash2013 wereadmixed and their survival was lower possibly because of a density‐dependent effect as the Florida panther population continuouslygrew during that period Trinkel et al (2010) also found thatpopulation density and inbreeding coefficient interacted to affectdemographic parameters and growth rate of an African lion po-pulation Likewise population density interacted with winterweather to affect the survival and population growth rates in Soaysheep (Ovis aries Milner et al 1999 Coulson et al 2001)

Population Dynamics and PersistenceThe finite deterministic and stochastic growth rates were gt10reflecting that much of the data used in this study were collectedfollowing genetic introgression in 1995 during which time thepopulation increased substantially Because our demographicparameter estimates did not change markedly in comparison tothose of Hostetler et al (2013) the similarity between thepopulation growth estimates from these studies was expectedBut these estimates reflect population growth during thedata‐collection period and do not predict future growth In factestimates of population size reported by McClintock et al(2015) suggest that population growth may already be slowing(Fig 2) A similar pattern was observed in an introduced po-pulation of African lion which increased steadily from 1995until 2001 then fluctuated around the presumed carrying ca-pacity thereafter (Trinkel et al 2010) Several other African lionpopulations that experienced various degrees of recovery

subsequently became relatively stable or exhibited decliningtrends (Bauer et al 2015)Our estimates of quasi‐extinction probabilities were lower than

those reported by Hostetler et al (2013) using a 2‐sex age‐structured matrix population model Lower probabilities ofquasi‐extinction than those reported by the earlier study forcomparable scenarios were likely a consequence of the fact thatthe estimates of abundance (McBride et al 2008) and thus thedensity‐dependent effects on survival of kittens and old panthers(gt10 yr) have changed in recent years Additionally these earlyestimates of the probability of quasi‐extinction and of time toquasi‐extinction did not consider genetic erosion that will in-evitably occur without management interventionUnder the MVM scenario the equilibrium population size was

twice that attained under the MPC scenario The MVM popula-tion estimates are approximately twice the MPCs (Fig 2) as aresult the simulated population under the MVM scenario achieveda higher equilibrium size Probability of quasi‐extinction under theMVM scenario was generally greater than that under the MPCscenario This result is a consequence of the fact that the estimateddensity dependence on kitten survival based on the MPC scenario isstronger than that for theMVM scenario (regression coefficients forabundance for the MPC scenario= minus0068 vs minus0002 for theMVM scenario Appendix B Table B1) Consequently simulatedpopulations under the MPC scenario have a stronger tendency tomove toward the equilibrium population size which inherentlyreduces the quasi‐extinction probability (eg Royama 1992)As expected for long‐lived mammals (Heppell et al 2000 Oli

and Dobson 2003) the population growth rate was proportio-nately most sensitive to changes in survival of prime adultsfollowed by that in younger age classes Global sensitivity ana-lysis in contrast showed that under all scenarios extinctionparameters were particularly sensitive to changes in survival ofFlorida panther kittens This result is a consequence of the highsensitivity of the population growth rate to kitten survival(Fig 9) and a larger standard error of kitten survival (Table 3)Results of our sensitivity analyses indicate that future researchshould focus on acquiring additional information on early lifestages to improve the accuracy and precision of estimates ofkitten survival Our results suggest that demographic variables(or their environmental drivers) that strongly influence extinc-tion risks are not necessarily those with the largest elasticityvalues A study of island fox (Bakker et al 2009) found thatextinction risk was strongly influenced by survival modifierparameters that had low elasticity values

Genetic Management When and How to InterveneThe use of genetic introgression as a management tool has al-ways been controversial and the probable effectiveness it mighthave in preventing the imminent demise of the panther popu-lation was initially debated (Creel 2006 Maehr et al 2006Pimm et al 2006) These debates notwithstanding the pantherpopulation had suffered from genetic morphological and bio-medical correlates of inbreeding before this management in-tervention (Johnson et al 2010 Onorato et al 2010) whereasthe post‐introgression period has been characterized by a sub-stantial reduction in the biomedical correlates of inbreeding andimprovements in demographic vigor and abundance (McBride

28 Wildlife Monographs bull 203

et al 2008 Hostetler et al 2010 2013 Johnson et al 2010Benson et al 2011 McClintock et al 2015) Nevertheless thepopulation remains small and isolated habitat is still being lostand there is no current possibility of natural gene flow betweenthis and other North American puma populations thus therecurrence of inbreeding‐related problems and subsequent po-pulation declines are inevitable The question therefore is notwhether genetic management of the Florida panther populationis needed but when and how it should be implementedWithout genetic introgression probabilities of quasi‐extinc-

tion were substantially greater than those from models thatincorporated periodic genetic introgression events (Fig 15)highlighting the importance of genetic management in smallisolated populations When 5 female pumas are added to theFlorida panther population every 10 years genetic variationincreased substantially under the MPC and MVM scenariosThe IBM predicted that the equillibrium population size wouldbe substantially larger when 5 pumas were released into thepopulation every 10 years than populations without intervention(ie no introgression) Releasing 15 Texas females did not af-ford substantially greater benefit to the equilibrium populationsize than did releasing only 5 Texas females Instead addingmore individuals caused increasingly large fluctuations in po-pulation size due to the numerical increases and density de-pendence (Fig 14) possibly diminishing the benefits associatedwith increased genetic variationCompared with the no‐intervention scenario (ie no genetic

introgression implemented) the introduction of 5 female pumasevery 10 years reduced quasi‐extinction probabilities from 13to 4 and from 17 to 6 for the MPC and MVM scenariosrespectively The benefit of genetic‐introgression strategies de-creased as the interval between successive management inter-ventions increased and no benefit was evident once the intervalreached 80 years (Fig 15) Introgression reduced the averagetime until quasi‐extinction (Fig 18) This somewhat counter‐intuitive result can be explained by the fact that repeated geneticintrogression makes the population more robust over timewhich will reduce the probability of extinction Consequentlyfew simulated population trajectories that fall below the criticalthreshold do so early reducing the mean time to extinctionAlso density dependence has a stabilizing effect on populations(Royama 1992) populations tend toward equilibrium popula-tion sizes which reduces the probability over time of popula-tions falling below quasi‐extinction threshold However timeuntil quasi‐extinction must be interpreted in conjunction withthe probability of quasi‐extinction which is very low for allscenarios with genetic introgression (Fig 15)Cost is a significant impediment to the implementation of a

genetic introgression program because captures relocations andmonitoring efforts before and after introgression are expensive(Edmands 2007 Frankham 2010b 2015 2016) Costs may beacceptable to society if the management actions result in sub-stantial fitness benefits especially if the species of concern ispopular and charismatic (Frankham 2015) We estimated the100‐year cost of introgression strategies modeled here to rangefrom $50000 to $12 million More expensive strategies gen-erally led to larger decreases in the probability of quasi‐extinc-tion but cheaper strategies also substantially improved

population viability (ie reduced quasi‐extinction probabilityTable 6) One of the cheaper strategies (5 pumas released every20 years) which cost an estimated $200000 over 100 yearsreduced the probability of quasi‐extinction by 59 and 44 forthe MPC and MVM scenarios respectively But these pre-liminary cost estimates are based on expenses incurred duringthe 1995 introgression and include costs of capture quarantinetransportation release and post‐release monitoring of femalesonly Although the cost for future genetic interventions wouldlikely be higher than those reported here the relative differencesin cost between alternative intervention strategies should remainapproximately the sameOur ability to simulate genetics and link it to the viability of

the Florida panther population is based on the empirically es-timated relationship between individual heterozygosity andsurvival Such heterozygosity and fitness correlations have beenstudied for decades (David 1998) but have only recently beenused to predict population viability (Benson et al 2016) noneto our knowledge have incorporated cost into their modelingefforts The mechanisms underlying heterozygosity and fitnesscorrelations remain unclear (David 1998 Szulkin et al 2010)and their significance as a proxy for inbreeding depression hasbeen debated (Balloux et al 2004 Miller and Coltman 2014)Nevertheless evidence continues to mount demonstratingpositive relationships between heterozygosity and survival(Coulson et al 1998ab Hansson et al 2004 Silva et al 2005Acevedo‐Whitehouse et al 2006 Jensen et al 2007 Velandoet al 2015) and between heterozygosity and fecundity (Seddonet al 2004 Charpentier et al 2005 Agudo et al 2012 Velandoet al 2015) Although the reported correlations are generallyweak their consequences can be substantial if population growthand persistence parameters are highly sensitive to the affectedvital rates (Velando et al 2015) Compared with scenarios thatignore genetics incorporating the effects of inbreeding on sur-vival increased quasi‐extinction probabilities within a 100‐yeartimeframe from 15 to 13 and from 14 to 17 for theMPC and MVM scenarios respectively These results high-light the importance of incorporating genetics when analyzingthe viability of small isolated populationsAlthough conservation biologists have recognized the im-

portance of genetics in population viability many still fail toincorporate genetic factors into PVA models (Reed et al 2002)Explicit consideration of genetics in PVAs requires long‐termgenetic and demographic data This permits the assessment ofthe functional relationship between genetic variability and de-mographic parameters Population viability analysis softwarepackages such as VORTEX (Lacy 1993 2000) offer a powerfulframework for individual‐based simulations but they often donot adequately capture a speciesrsquo life history or genetics Basedon a thorough review of the importance of genetic factors inwildlife conservation Frankham et al (2014) concluded thatgenetic factors can strongly influence the dynamics and persis-tence of small andor fragmented populations They furthernoted that ldquoMost PVA‐based risk assessments ignore or in-adequately model genetic factorsrdquo and recommended that ldquoPVAshould routinely include realistic inbreeding depressionhelliprdquo(Frankham et al 201457) Our custom‐built IBM permitted usto develop a model that adequately captured the Florida panther

van de Kerk et al bull Florida Panther Population and Genetic Viability 29

life history and we could parameterize the model with our long‐term genetic and demographic dataThe explicit consideration of the effects of inbreeding on

Florida panther population dynamics and persistence representsan important step forward Nonetheless our model remainsimperfect First we neglected the possible effects of habitat lossdetrimental anthropogenic activities climate change the prob-ability of immigration of pumas from western populationsdisease or catastrophes on demographic rates and populationviability Although immigration from puma populations outsideFlorida is possible its probability is very low given the extensivechallenges the anthropogenically altered landscape would posefor such a migrant to make it successfully to South Florida Weomitted mutations from our model because we have no in-formation on mutation rates though we expect that they are lowbased on values reported for other mammals (Kumar and Sub-ramanian 2002) Finally we only considered possible effects ofinbreeding on age‐specific survival rates because there was noevidence that heterozygosity affected female reproduction Lossof heterozygosity can negatively affect reproduction becauseinbred male panthers can suffer from cryptorchidism which canadversely affect their fecundity However given the polygynousmating system we expect the Florida panther population growthis primarily female‐limitedAlthough it is generally accepted that genetic introgression

only temporarily relieves a population from inbreeding depres-sion we know of no cases in which it has been implementedmore than once to conserve a threatened wildlife populationOur study provides an example of how demographic and mo-lecular data can be used within an IBM framework to evaluatethe efficacy of alternative genetic management strategies via theFlorida panther as a case study Our model can be adjusted forand applied to other species and offers great potential as a toolfor genetic management of small isolated wildlife populations

MANAGEMENT IMPLICATIONS

Our results and those of Hostetler et al (2013) and Johnsonet al (2010) provide persuasive evidence that the genetic in-tervention implemented in 1995 prevented the demise of theFlorida panther and restored demographic vigor to the popu-lation The Florida panther population remains small and iso-lated from other puma populations the closest puma populationis located gt1000 km away in Texas and the intervening land-scape is highly developed and fragmented with many interstate(and other high‐traffic) highways and major urban centersTheory suggests that 1 migrant per generation is sufficient tominimize the loss of genetic diversity (eg Mills and Allendorf1996) However the possibility of successful migration of apuma to South Florida followed by successful mating every ~5years is very low considering the distance between Texas andFlorida (Hedrick 1995) and the many risks and barriers thatdispersing pumas face (Beier 1995 Maehr et al 2002 Thatcheret al 2006) Consequently the pantherrsquos long‐term persistencewill likely depend on subsequent timely genetic introgressionsgiven the near‐impossibility of natural gene flow from otherpuma populations To that end our study offers considerableinsights into benefits of different genetic management strategiesand associated costs

Without genetic management the Florida panther populationfaces a substantial risk of quasi‐extinction within 100 years (10ndash46 depending on quasi‐extinction threshold and scenarios)Twenty‐three years have passed since genetic introgression wasimplemented and the population appears healthy as indicatedby measures of genetic variation increases in the populationsize and improvements in age‐specific survival rates Ourfindings suggest that releasing more pumas more frequently doesnot necessarily improve persistence probability because such astrategy would cause larger fluctuations in population size(Fig 14) which negatively affects persistence probability Thusextinction risks faced by inbred populations of large carnivorescan be substantially reduced by releasing approximately 5 im-migrants every 2 decades We suggest that such a managementstrategy be followed only in conjunction with continued long‐term monitoring of the population Our analyses demonstratethat population growth and persistence are highly sensitive tokitten and female (adult and subadult) survival Continuing tocollect data on these demographic variables along with bio-metric and genetic sampling will remain important to pantherrecovery and vigilance in identifying issues associated with in-breeding depression The collection of these monitoring datashould allow managers to detect any demographic or geneticchanges in a timely fashion and respond with genetic in-trogression or other management actions if warrantedRecovery objectives as defined by the USFWS in its current

recovery plan (USFWS 2008) delineate the need for additionalpopulations in the historical range to downlist or delist theFlorida panther If the only extant population of Floridapanthers continued to grow it could serve as a source ofpanthers to be translocated to suitable habitat to seed addi-tional populations Maintaining current information on thegenetic health of the population and precise estimates of itssize will be important in assessing whether it can withstand theremoval of a subset of animals for translocation Although the2017 documentation of a female panther with kittens north ofthe current breeding range is encouraging in terms of thepotential for population expansionmdashgiven that no female hadbeen recorded north of the Caloosahatchee River since 1972mdashthe re‐establishment of separate new populations will mostlikely require translocations by wildlife managers and a lengthysociopolitical processConservation of threatened or endangered species such as the

Florida panther is inherently challenging For panthers and otherlarge carnivores that require vast extents of natural habitat topersist habitat is typically the most limiting factor that will de-termine whether recovery efforts succeed Although geneticmanagement invariably plays a key role in the long‐term persis-tence of the panther population even a healthy population caneventually succumb to the impacts of habitat loss The humanpopulation of Florida continues to be one of the fastest‐growingin the nation the United States Census Bureau currently ranksFlorida as the third most populous Therefore habitat loss isgoing to continue to be a key issue for panthers Diligence towardpreserving panther habitat in its historical range via conservationeasements or other means and trying to keep such areas inter-connected via corridors is a goal stakeholders such as governmentagencies sportsmen non‐governmental organizations ranchers

30 Wildlife Monographs bull 203

and other private landowners must work on collaboratively toachieve

ACKNOWLEDGMENTS

We thank D Shindle D Jennings T OrsquoMeara D Land andC Belden for valuable advice throughout the project We thankS Bass M Criffield M Cunningham D Giardina D JansenA Johnson D Land M Lotz C McBride Rocky McBrideRowdy McBride Roy McBride L Oberhofer S Schulze staffsat Everglades National Park and Big Cypress National Preserveand others for assistance with fieldwork We also thank JAustin M Conroy J Hines J Laake S Mills J Nichols JHines M Sunquist J Fryxell and T Coulson for advice andassistance regarding data analysis and panther biology TheMuseum of Texas Tech University Natural Science ResearchLaboratory provided samples from pumas from west Texas forcomparative genetic analyses M Ben‐David D Land WMcCown E Hellgren and 4 anonymous reviewers providedmany helpful comments on the manuscript We thank NereaUbierna Lopez for completing the Spanish translation of theabstract We are grateful to the Department of ResearchComputing University of Florida for excellent high‐perfor-mance computing services We would like to thank US Fishand Wildlife Service Florida Fish and Wildlife ConservationCommission (FWC) US National Park Service QuantitativeSpatial Ecology Evolution and Environment (QSE3) In-tegrative Graduate Education and Research Traineeship(IGERT) program funded by the National Science Foundationand the University of Florida for financially supporting thisstudy Funding for panther research conducted by the FWC issupported predominantly via donations accrued with vehicleregistrations for ldquoProtect the Pantherrdquo license plates

LITERATURE CITEDAcevedo‐Whitehouse K T R Spraker E Lyons S R Melin F Gulland RL Delong and W Amos 2006 Contrasting effects of heterozygosity onsurvival and hookworm resistance in California sea lion pups MolecularEcology 151973ndash1982

Adams J R L M Vucetich P W Hedrick R O Peterson and J AVucetich 2011 Genomic sweep and potential genetic rescue during limitingenvironmental conditions in an isolated wolf population Proceedings of theRoyal Society B Biological Sciences 2783336ndash3344

Agresti A 2013 Categorical data analysis Third edition John Wiley amp SonsHoboken New Jersey USA

Agudo R M Carrete M Alcaide C Rico F Hiraldo and J A Donaacutezar2012 Genetic diversity at neutral and adaptive loci determines individualfitness in a long‐lived territorial bird Proceedings of the Royal Society BBiological Sciences 2793241ndash3249

Albeke S E N P Nibbelink and M Ben‐David 2015 Modeling behavior bycoastal river otter (Lontra canadensis) in response to prey availability in PrinceWilliam Sound Alaska a spatially‐explicit individual‐based approach PLOSOne 10e0126208

Alho J S K Vaumllimaumlki and J Merilauml 2010 Rhh an R extension for estimatingmultilocus heterozygosity and heterozygosity‐heterozygosity correlationMolecular Ecology Resources 10720ndash722

Alvarez K 1993 Twilight of the panther Myakka River Publishing SarasotaFlorida USA

Aparicio J M J Ortego and P J Cordero 2006 What should we weigh toestimate heterozygosity alleles or loci Molecular Ecology 154659ndash4665

Bakker V J D F Doak G W Roemer D K Garcelon T J Coonan S AMorrison C Lynch K Ralls and R Shaw 2009 Incorporating ecologicaldrivers and uncertainty into a demographic population viability analysis for theisland fox Ecological Monographs 7977ndash108

Balloux F W Amos and T Coulson 2004 Does heterozygosity estimateinbreeding in real populations Molecular Ecology 133021ndash3031

Barone M A M E Roelke J Howard J L Brown A E Anderson and DE Wildt 1994 Reproductive characteristics of male Florida panthersmdashcomparative studies from Florida Texas Colorado Latin‐America andNorth‐American Zoos Journal of Mammalogy 75150ndash162

Bateson Z W P O Dunn S D Hull A E Henschen J A Johnson and LA Whittingham 2014 Genetic restoration of a threatened population ofgreater prairie‐chickens Biological Conservation 17412ndash19

Bauer H G Chapron K Nowell P Henschel P Funston L T B HunterD W Macdonald and C Packer 2015 Lion (Panthera leo) populations aredeclining rapidly across Africa except in intensively managed areas Pro-ceedings of National Academy of Sciences of the United States of America11214894ndash14899

Beier P 1995 Dispersal of juvenile cougars in fragmented habitat Journal ofWildlife Management 59228ndash237

Benson J F J A Hostetler D P Onorato W E Johnson M E Roelke SJ OrsquoBrien D Jansen and M K Oli 2011 Intentional genetic introgressioninfluences survival of adults and subadults in a small inbred felid populationJournal of Animal Ecology 80958ndash967

Benson J F P J Mahoney J A Sikich L E K Serieys J P Pollinger H BErnest and S P D Riley 2016 Interactions between demography geneticsand landscape connectivity increase extinction probability for a small popu-lation of large carnivores in a major metropolitan area Proceedings of theRoyal Society B Biological Sciences 28320160957

Beverly F 1874 The Florida panther Forest and Stream 3290Boyce M S C V Haridas C T Lee and the NCEAS Stochastic Demo-graphy Working Group 2006 Demography in an increasingly variable worldTrends in Ecology amp Evolution 21141ndash148

Brook B W J J OrsquoGrady A P Chapman M A Burgman H R Akcakayaand R Frankham 2000 Predictive accuracy of population viability analysis inconservation biology Nature 404385ndash387

Brys R and H Jacquemyn 2016 Severe outbreeding and inbreeding de-pression maintain mating system differentiation in Epipactis (Orchidaceae)Journal of Evolutionary Biology 29352ndash359

Burke J M and M L Arnold 2001 Genetics and the fitness of hybridsAnnual Review of Genetics 3531ndash52

Burnham K P 1993 A theory for combined analysis of ring recovery andrecapture data Pages 199ndash213 in J‐D Lebreton and P M North editorsMarked individuals in the study of bird population Springer‐Verlag NewYork New York USA

Burnham K P and D R Anderson 2002 Model selection and multimodelinference a practical information‐theoretic approach Second edition SpringerVerlag New York New York USA

Caswell H 2001 Matrix population models construction analysis and in-terpretation Sinauer Associates Sunderland Massachusetts USA

Charpentier M J M Setchell F Prugnolle L A Knapp E J Wickings PPeignot and M Hossaert‐McKey 2005 Genetic diversity and reproductivesuccess in mandrills (Mandrillus sphinx) Proceedings of the NationalAcademy of Sciences of the United States of America 10216723ndash16728

Cooch E and G C White editors 2018 Program MARK a gentle in-troduction lthttpwwwphidotorgsoftwaremarkdocsbookgt Accessed 7Apr 2016

Cooley H S R B Wielgus G M Koehler H S Robinson and B TMaletzke 2009 Does hunting regulate cougar populations A test of thecompensatory mortality hypothesis Ecology 902913ndash2921

Coulson T E A Catchpole S D Albon B J Morgan J M Pemberton TH Clutton‐Brock M J Crawley and B T Grenfell 2001 Age sex densitywinter weather and population crashes in Soay sheep Science 2921528ndash1531

Coulson T J Pemberton S Albon M Beaumont T Marshall J Slate FGuinness and T Clutton‐Brock 1998a Microsatellites reveal heterosis in reddeer Proceedings of the Royal Society B Biological Sciences 265489ndash495

Coulson T N S D Albon J M Pemberton J Slate F E Guinness and TH Clutton‐Brock 1998b Genotype by environment interactions in wintersurvival in red deer Journal of Animal Ecology 67434ndash445

Cox D 1972 Regression models and life‐tables Journal of the Royal StatisticalSociety Series B Statistical Methodology 34187ndash220

Creel S 2006 Recovery of the Florida panthermdashgenetic rescue demographic rescueor both Response to Pimm et al (2006) Animal Conservation 9125ndash126

Crnokrak P and D A Roff 1999 Inbreeding depression in the wild Heredity83260ndash270

Crow J 1948 Alternative hypotheses of hybrid vigor Genetics 33477ndash487

van de Kerk et al bull Florida Panther Population and Genetic Viability 31

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 9: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

when it is more informative (ie has more alleles) and has aneven distribution of allele frequencies For our demographicanalyses we calculated individual heterozygosity for each in-dividual panther as 1 minusHL As do most indices of individualheterozygosity HL takes into account the average allele fre-quency in the population (Aparicio et al 2006) Thus duringsimulations the allele frequency in the population changesthroughout an individualrsquos lifetime which causes its hetero-zygosity also to change This is not biologically plausible becausegenetic properties are retained throughout an individualrsquos life-time Therefore for use in model simulations we simply cal-culated heterozygosity for each individual as the proportion ofheterozygous lociWe also used genotype data to implement a Bayesian clus-

tering analysis in Program STRUCTURE (Pritchard et al2000) to infer ancestral clusters of panthers This method uses aMarkov chain Monte Carlo (MCMC) method to determine thenumber of genetic clusters (K) in the sample while also pro-viding an individualrsquos percentage of ancestry (q values) allocatedto each cluster Runs for this analysis included a 100000MCMC burn‐in period followed by 500000 MCMC iterationsfor 1ndash10 ancestral genetic clusters (K= 1 to 10) with 10 re-petitions for each K To determine the most likely number of Kgenetic clusters we used the logarithm of the probability of thedata (log Pr(D)∣K Pritchard et al 2000) and estimates of ΔK(Evanno et al 2005) in Program STRUCTURE HAR-VESTER (Earl and VonHoldt 2011) We then used q valuesprovided in runs for the best K to assign individual panthers aseither canonical (in most cases ge90 pre‐introgression ancestrysee Johnson et al 2010) or admixed (lt90 pre‐introgressionancestry) We recognized admixed panthers that were the off-spring of matings between a Texas female puma and a canonicalmale panther as F1 admixed panthers for analyses

Demographic ParametersSurvival of kittensmdashMost kittens PIT‐tagged at den sites were

never encountered again A small proportion of the PIT‐taggedkittens were recovered dead or were recaptured as subadults oradults and radiocollared so that their fate could be monitoredWe also identified instances of complete litter failure when afemale died within 9 months after giving birth (the minimumage of independence for kittens) or when a female wasdocumented to have another litter less than 12 months aftergiving birth (the minimum age of independence plus a 3‐monthgestation period Hostetler et al 2010) Thus our data setconsisted of 1) live recaptures and subsequent radio‐tracking ofpanthers of various ages that had been PIT‐tagged as kittens 2)recovery of dead panthers of various ages that had been PIT‐tagged as kitten 3) live captures and subsequent radio‐trackingof other panthers and 4) instances of complete litter failure(Table 1) We analyzed these data using Burnhamrsquos live‐recapture dead‐recovery modeling framework (Burnham 1993Williams et al 2002) to estimate survival of kittens (0ndash1‐yearold) and to test covariate effects on this parameter We includeddata on individuals encountered dead or alive at an older age inour analyses because their encounter histories also providedinformation on kitten survival We used a live‐dead data inputformat coded with a 1‐year time step (Cooch and White 2018)

All known litter failures occurred within the first year of thekittensrsquo lives so we treated all litter failures as recoveries withinthe first year We treated panthers that would have died if wehad not removed them to captivity as having died on the date ofremoval Data preparation and analytical approach are describedin more detail by Hostetler et al (2010)In addition to survival probability (S) the Burnham model

incorporates parameters for recapture probability (p) recoveryprobability (r) and site fidelity (f) We set r and p to 10 forradio‐tracked individuals because their status was known eachyear We also fixed f at 10 for all individuals because the re-capture and recovery areas were the same and encompassed theentire range of the Florida panther Based on findings of Hos-tetler et al (2010) we used a base model that 1) allowed survivalto differ between kittens and older panthers 2) allowed survivalto differ between sexes and among age classes (females ages 1and 2 ge3 males ages 1 2 and 3 ge4) 3) constrained recaptureprobability to be the same for all uncollared panthers and 4)allowed recovery rates to differ between kittens and uncollaredolder panthers Even with our additional data this model re-mained the best fit for a range of models for recapture recoveryand survival of kittens subadults and adults of various age‐classcategories (Hostetler et al 2010)Subadult and adult survivalmdashTo estimate survival for panthers

ge1 year of age we analyzed radio‐tracking data using a Coxproportional hazard‐modeling framework (Cox 1972 Therneauand Grambsch 2000) We followed procedures outlined byBenson et al (2011) for data preparation and analysis Brieflywe right‐censored panthers on the last day of observation beforetheir collar failed When an individual aged into a new age classwhile being tracked we created 2 data entries 1 ending the dayit entered the new age class the other starting on that day Weused the FlemingndashHarrington method to generate survivalestimates from the Cox analysis and the Huber sandwichmethod to estimate robust standard errors (Therneau andGrambsch 2000 Benson et al 2011)We considered 3 age classes subadults (1ndash25 yr for females and

1ndash35 yr for males) prime adults (25ndash10 yr for females and35ndash10 yr for males) and old adults (gt10 yr old for both sexes)after Benson et al (2011) We estimated overall survival using abase model that allowed survival to differ between subadultfemales subadult males prime adult females prime adult malesand old adults of either sex (old age was additive with sex other-wise age and sex were interactive) This model provided the best fitto the data relative to a range of models considering differences insurvival between prime‐adult and older‐adult age classes both ad-ditively and interactively with sex (Benson et al 2011)To examine whether survival rates observed in recent years

differed from those observed immediately following introgres-sion we also estimated age‐specific survival rates for 2 nearlyequal periods of time immediately post‐introgression (1995ndash2004) and the more recent period (2005ndash2013) For theseanalyses we separated the radio‐tracking data into the 2 periodsand estimated survival rates for each period separately using thebase model Similar analyses for kitten survival were not possiblebecause of data limitationsProbability of reproductionmdashWe used telemetry data obtained

from radio‐tracked females that reproduced at least once to

van de Kerk et al bull Florida Panther Population and Genetic Viability 11

estimate annual probability of reproduction of subadult prime‐adult and old‐adult females using complementary logndashlogregression (Agresti 2013) following methods described indetail by Hostetler et al (2012) Briefly we modeled theprobability of a female panther giving birth (b) as a function ofcovariates as

γ= minus [minus ( )]b z1 exp exp i i

where zi is a row vector of covariates (see below for covariateinformation) for female panther i and γ is a column vector ofmodel coefficients (Appendix B Table B2) To account fordifferences in observation time we included log(m) as an offsetin the model where m is the number of months a femalepanther was radio‐tracked (Hostetler et al 2012)

Reproductive outputmdashTo estimate the number of kittensproduced by reproductive females we used cumulative logitregression (Min and Agresti 2005 Agresti 2013) Weconsidered a model that includes J categories for the numberof offspring with j denoting the number of offspring ( j= 1 2hellip J and J= 6) provided that a female reproduces Weconsidered J= 6 which is greater than the largest litter sizeobserved in our study (4 kittens) because females can produce 2or more litters per year if litters fail We modeled the probabilitythat a female i produces a litter of size at most j (Yi) as

le⎧

⎨⎩

δ( ) = == hellip minus

=

( )+ θ βminus +Y jj J

j JPr

1 2 1

1i ij e

1

1 xj i

where θj is the intercept for the annual number of kittens pro-duced by a female i being at most j xi is a row vector of cov-ariates for female panther i (see below) and β is a column vectorof model coefficients The probability that Yi will be exactly j isgiven by

⎧⎨⎩

πδ

δ δ( = ) = =

=

minus = hellip( minus )Y j

jj J

Pr 1

2 i ij

i

ij i j

1

1

Finally the expected annual number of kittens produced byfemale i (νi) is given by

sumν π==

j i

j

J

ij

1

Additional details regarding the estimation and modelingof the annual number of kittens produced can be found inHostetler et al (2012)

CovariatesmdashIn addition to sex and age we tested for the effect ofabundance genetic ancestry and individual heterozygosity on theaforementioned demographic parameters (model coefficients forkitten and adult survival are given by η and ι respectively AppendixB Table B2) We used a minimum population count (MPC) basedon radio‐tracking data and field evidence of subadult and adult (ieindependent age) panthers as an index of abundance (McBride et al2008) For these analyses we did not use the population sizeestimates reported by McClintock et al (2015) because unlike theminimum counts they did not cover the entire study period

(McBride et al 2008 R T McBride and C McBride RancherrsquosSupply Incorporated unpublished report Fig 2)To test if demographic parameters differed depending on an-

cestry we considered 3 ancestry models dividing the panthers into1) F1 admixed and all other panthers 2) canonical and admixed(including F1 admixed) panthers and 3) F1 admixed canonicaland other admixed panthers We quantified genetic diversity usingindividual heterozygosity as described previously We tested for theeffect of genetic covariates using a subset of kittens that were ge-netically sampled We calculated model‐averaged estimates ofmodel parameters on the scale in which they were estimated(Appendix B available online in Supporting Information)

Cause‐specific mortalitymdashWe performed cause‐specificmortality analysis only on dead radio‐collared panthers becausedetection rates of uncollared panther mortalities are highlycorrelated with the cause of death (eg vehicle collision)Following Benson et al (2011) we attributed each mortality to1 of 4 causes 1) vehicle collision 2) intraspecific aggression 3)other causes (including known causes such as diseases injuriesand infections unrelated to the first 2 causes) and 4) unknown(ie mortalities for which we could not assign a cause) Wethen used the non‐parametric cumulative incidence functionestimator a generalization of the staggered‐entry KaplanndashMeiermethod of survival estimation to estimate cause‐specificmortality rates for male and female panthers in different ageclasses (Pollock et al 1989 Heisey and Patterson 2006 Bensonet al 2011) We compared cause‐specific mortality rates betweensexes age classes and ancestries

Statistical inferencemdashWe used an information‐theoreticapproach (Akaikersquos information criterion [AIC]) for modelselection and statistical inference (Burnham and Anderson 2002)For the analysis of kitten survival we adjusted the AIC foroverdispersion and small sample size (QAICc details in Hostetleret al 2010) We performed all statistical analyses in Program R (RCore Team 2016) We used Burnhamrsquos live recapturendashdead

0

100

200

300

1985 1990 1995 2000 2005 2010Year

Indi

vidu

als Method

MPC

MVM

Figure 2 The Florida panther minimum population count (MPC) and populationsize estimated using motor vehicle collision mortalities (MVM) during our studyperiod Minimum population counts are from McBride et al (2008) and R TMcBride and C McBride (Rancherrsquos Supply Incorporated unpublished report)Estimates based on MVM are from McClintock et al (2015) The solid vertical lineindicates the year of genetic introgression

12 Wildlife Monographs bull 203

recovery model (Burnham 1993) to estimate and model kittensurvival using the RMark package version 2113 (Laake 2013) asan interface for Program MARK (White and Burnham 1999)We implemented the Cox proportional hazard model for esti-

mation and modeling of survival of subadults and adults using the Rpackage SURVIVAL (Therneau and Grambsch 2000) We per-formed cause‐specific mortality analyses using an R version of S‐PLUS code provided by Heisey and Patterson (2006)To account for model selection uncertainty we calculated

model‐averaged parameter estimates across all models using theR package AICcmodavg (Mazerolle 2017) using AIC weights asmodel weights (Burnham and Anderson 2002) We used modelswithout the categorical ancestry variables for model averagingand estimated parameters based on the average value for thecontinuous covariates (individual heterozygosity and abundanceindex) Because only a subset of the kittens was geneticallysampled we estimated 2 kitten survival rates First we estimatedkitten survival based on the data set that included all kittens(overall kitten survival) Second we estimated kitten survivalusing a subset of the data that only included kittens that weregenetically sampled (survival of genetically sampled kittens)

Matrix Population ModelsWe investigated Florida panther population dynamics andpersistence using 2 complementary modeling frameworks ma-trix population models (Caswell 2001) and IBMs (Grimm andRailsback 2005 McLane et al 2011 Railsback and Grimm2011 Albeke et al 2015) Matrix population models offer aflexible and powerful framework for investigating the dynamicsand persistence of age‐ or stage‐structured populations (egCaswell 2001 Robinson et al 2008 Kohira et al 2009 Hunteret al 2010 Hostetler et al 2012) With a fully developed theorybehind them these models also serve as a check for resultsobtained from simulation‐based models such as IBMs We usedan age‐structured‐matrix population model for deterministic andstochastic demographic analyses and for preliminary investiga-tions of Florida panther population viability (Caswell 2001Morris and Doak 2002)For deterministic and stochastic demographic analyses we

parameterized a female‐only 19 times 19 age‐specific populationprojection matrix of the form

⎢⎢⎢⎢⎢

⋯ ⋯

⋯ ⋯

⋱ ⋱ ⋮

⋮ ⋱ ⋱ ⋱ ⋮

⎥⎥⎥⎥⎥

( )=t

F F F

P

P

P

A0 0

0

0 0 0

t t t

t

t

t

1 2 19

1

2

18

where Fi and Pi are the age‐specific fertility and survival ratesrespectively and ( )tA is a time‐varying (or stochastic) popula-tion projection matrix We assumed that the longevity and ageof last reproduction of female Florida panthers was 185 yearswe based this assumption on the observation that the oldestfemale in our study was 186 years old Florida panthers re-produce year‐round thus we used birth‐flow methods to esti-mate Fi and Pi values following Caswell (2001) and Hostetleret al (2013) We estimated Pi as

= +P S S i i i 1

where Si is the age‐specific survival probability We estimated Fi as

⎛⎝

⎞⎠

=+ +F S

m Pm

2i k

i i i 1

where Sk is the kitten survival probability and mi is the age‐specific fecundity rate which was estimated as

ν=m b f i i i k

where bi is the breeding probability for a female in age class iduring time step t νi is the annual number of kittens producedby a breeding female in age class i and fk is the proportionfemale kittens at birth (assumed to be 05)We estimated the finite deterministic (λ) and stochastic annual

population growth rate (λs) and stochastic sensitivities and elasti-cities following methods described in detail by Caswell (2001) Toestimate λs we assumed identically and independently distributedenvironments (Caswell 2001 Morris and Doak 2002 Haridas andTuljapurkar 2005) and used a parametric bootstrapping approachto obtain a sequence of demographic parameters for 50000 ma-trices We then estimated log(λs) from this sequence of matrices(Tuljapurkar 1990 Caswell 2001) and exponentiated it to obtain λsWe calculated the sensitivity and elasticity of λs to changes in lower‐level vital demographic rates as per Caswell (2001) and Haridas andTuljapurkar (2005)For population projection and population viability analysis (PVA)

simulations we expanded the population projection matrix into a34 times 34 2‐sex age‐structured matrix to include female and maleFlorida panthers (Caswell 2001 Hostetler et al 2013) Malescontribute to the population only through survival in this modelingframework We assumed maximum longevity for males to be 145years of age because the oldest male in our study was 144 years oldThe population projection matrix was of the following form

⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢

⋯ ⋯ ⋯ ⋯

⋯ ⋯ ⋯ ⋯

⋱ ⋱ ⋮ ⋮ ⋱ ⋱ ⋮

⋮ ⋱ ⋱ ⋱ ⋮ ⋮ ⋱ ⋱ ⋮

⋯ ⋯ ⋯

⋯ ⋯ ⋯ ⋯

⋯ ⋯ ⋱ ⋱ ⋮

⋮ ⋮ ⋱ ⋱ ⋮ ⋱ ⋱ ⋮

⋯ ⋯ ⋯

⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥

( )=

( ) ( ) ( )

( )

( )

( )

( ) ( ) ( )

( )

( )

t

F N F N F N

P N

P N

P N

M N M N M N

Q N

Q N

A n

0 0

0 0 0 0

0

0 0 0 0 0

0 0

0 0 0

00 0 0 0 0

t t t

t

t

t

t t t

t

t

1 2 19

1

2

18

1 2 19

1

14

where Pi and Qi are age‐specific survival rates of female and maleFlorida panthers respectively and Fi and Mi are the rates atwhich female and male kittens are produced by females of ageclass i respectively Here the population projection matrix isboth time‐varying and density‐dependent (Caswell 2001)We incorporated the influence of parameter uncertainty en-

vironmental stochasticity and density dependence on Floridapanther population dynamics and persistence following the ap-proach outlined by Hostetler et al (2013) we used the sameapproach in the IBM analysis which is described below UnlikeIBMs matrix models do not intrinsically include demographicstochasticity so we explicitly included the influence of

van de Kerk et al bull Florida Panther Population and Genetic Viability 13

demographic stochasticity by simulating the fates of individualsas described in Caswell (2001)Population projections using matrix models necessitate a

vector of initial sex‐ and age‐specific abundance Because suchdata are currently unavailable for the Florida panther we esti-mated it based on the stable sex and age distribution and MPCin 2013 Using point estimates of the demographic parameterswe estimated the stable sex and age distribution for the 2‐sexdeterministic and density‐independent population projectionmatrix We then multiplied the stable sex and age distributionby the MPC in 2013 to get the starting population vector n(0)We projected the population over time as n(t+ 1)=A(tn)n(t)where A(tn) indicates the time‐specific density‐dependentpopulation projection matrix (Caswell 2001)We had 2 indices of abundance available the MPC (McBride

et al 2008) and population size estimated using motor vehiclecollision mortalities (MVM McClintock et al 2015) These 2indices are substantially different because MPC does not ac-count for imperfect detection whereas MVM accounts forimperfect detection and provides an estimate of population size(Fig 2) Therefore we examined 2 scenarios one using densitydependence based on MPC and the other using density de-pendence based on MVM We ran 1000 bootstraps of para-meter values and 1000 simulations per bootstrap for each sce-nario We projected population trajectories over 200 years andthen estimated population growth rates probabilities ofquasi‐extinction and time until quasi‐extinction and comparedthese with the results obtained from the IBM (see below) Weran scenarios with a quasi‐extinction threshold of 10 and 30panthers We used the mean of probabilities of quasi‐extinction(PQE) across simulations as the point estimate for PQE and the5th and 95th percentiles across simulations as the 90 con-fidence interval Because the confidence intervals werequantile‐based but the point estimates were not it is possiblewith skewed distributions for the point estimates to be outsidethe confidence interval We implemented the matrix

population models in the R computing environment (R CoreTeam 2016)

IBMs and PVABecause of well‐developed theory ease of implementation andthe modeling flexibility that they offer matrix populationmodels have become the model of choice for investigating thedynamics and persistence of age‐ and stage‐structured popula-tions (Caswell 2001) However it is difficult to incorporateattributes of individuals such as genetics and behavior usingmatrix models Quantifying the effects of genetic erosion on theFlorida panther population dynamics and persistence andevaluating benefits and costs of alternative genetic managementscenarios were important goals of our study Because IBMs relyon a bottom‐up approach that begins by explicitly consideringindividuals (McLane et al 2011 Railsback and Grimm 2011Albeke et al 2015) it is possible to represent genetic attributesof each individual and to track genetic variation over time Weused IBMs as the primary modeling framework in this studybecause they allowed for explicit consideration of genetics bothat individual and population levels and population viabilityassessment under alternative genetic management scenariosWe developed an IBM (Grimm 1999 Grimm and Railsback

2005 McLane et al 2011 Railsback and Grimm 2011 Albekeet al 2015) tailored to the life history of the Florida panther(Fig 3) We simulated every individual from birth until deathbased on a set of rules describing fates (eg birth and death) andattributes (eg genetics) of individuals at each annual time stepAs in other IBMs population‐level processes (eg populationsize and genetic variation) emerged from fates of and interac-tions among individuals (Grimm 1999 Railsback and Grimm2011) We developed an overview design concepts and details(ODD) protocol (Grimm et al 2006 2010) to describe ourIBM (Appendix C available online in Supporting Information)and provide pseudocodes for individual‐based simulations(Appendix D available online in Supporting Information)

For each panther at time = t

Kitten (lt1 yr)

Reproduce

Survive

Kitten(s) born

Kitten

Sex and age class

Subadult male

Prime adult male

Old adult male

Subadult female

Prime adult female

Old adult female

Age one year

No Male

Female

Yes

No

Yes

t=t+1

Yes

lt35 yrs

35-10 yrs

gt10 yrs

lt25 yrs

25-10 yrs

gt10 yrs

Figure 3 Schematic representation of Florida panther life history We tailored the individual‐based population model to represent the life‐history patterndepicted here

14 Wildlife Monographs bull 203

Dynamics and persistence of populations are influenced bymany factors such as demographic and environmental sto-chasticity and density dependence these effects and parametricuncertainty must be considered in population models whenpossible (Caswell 2001 Boyce et al 2006 Bakker et al 2009Hostetler et al 2013) Because by definition IBMs simulate thelife history of individuals in a population they inherently in-corporate the effect of demographic stochasticity Our analysesrevealed that survival of kittens subadults and adults and theprobability of reproduction varied over time so we incorporatedenvironmental stochasticity in the model for these variablesfollowing Hostetler et al (2013) Likewise we found evidenceof density‐dependent effects on kitten survival Because theMPC might be an underestimate of population size (McBrideet al 2008) we performed simulations using both the MPC andMVM abundance indicesTo account for model selection and parametric uncertainty we

used a Monte Carlo simulation approach For each bootstrapsample we selected a model based on the AIC (or QAICc)weights We then sampled the intercept and slope for kittensurvival (on the logit scale) subadult and adult survival (on thelogit scale with log‐hazard effect sizes) and probability of re-production (on the complementary logndashlog scale Appendix B)from multivariate normal distributions with mean vectors equalto the estimated parameters and variance‐covariance matricesequal to the sampling variance‐covariance matrices We thentransformed the results to nominal scale and converted the es-timates to age‐specific survival and reproduction parameters asdescribed in Hostetler et al (2013)We ran 1000 simulations for 1000 parametric bootstraps for

200 years for the MPC and the MVM scenario for both thestochastic 2‐sex matrix‐population model and the IBM Wetracked the population size at each time step and estimatedquasi‐extinction probabilities and times based on the populationtrajectory We considered the population to be quasi‐extinct ifindividuals of only 1 sex remained or if the population size fellbelow critical thresholds set at 10 and 30 individuals We alsoestimated expected time to quasi‐extinction for both thresholdsdefined as the mean and median time until a population be-comes quasi‐extinct conditioned on quasi‐extinction We im-plemented the IBM using the RNetLogo package version 10ndash2(Thiele et al 2012 Thiele 2014) as an interface for the IBMplatform NetLogo (Wilensky 1999)

Sensitivity AnalysisAn important component of demographic analysis is sensitivityanalysis which is the quantification of the sensitivity of amodelrsquos outcome to changes in the input parameters (Caswell2001 Bakker et al 2009 Thiele et al 2014) Using an IBM weperformed global sensitivity analyses for 2 (ie MPC andMVM) density‐dependent scenarios to assess how sensitivequasi‐extinction times and probabilities were to changes (anduncertainties) in the estimated demographic rates We usedLatin hypercube sampling to sample parameters from the entireparameter space (Marino et al 2008 Thiele et al 2014) Wesampled intercept and slope for kitten survival (on logit scale)subadult and adult survival (on logit scale with log‐hazard effectsizes) and probability of reproduction (on complementary

logndashlog scale) from normal distributions We sampled theprobability distribution for the number of kittens producedannually (on the real scale) from a Dirichlet distribution themultivariate generalization of the beta distribution (Kotz et al2000) We sampled 1000 different parameter sets and ran 1000simulations for each scenario We calculated the probability ofquasi‐extinction within 200 years for each density‐dependencescenario (MPC or MVM) using a quasi‐extinction threshold of10 individuals and then calculated the partial rank correlationcoefficient between each of those and each parameter trans-formed to the real scale (Bakker et al 2009 Hostetler et al2013) Partial rank correlation coefficients quantify the sensi-tivity of the probability of quasi‐extinction to input variables(ie survival and reproductive parameters) with higher absolutevalues indicating stronger influence of that variable on quasi‐extinction probabilities

Genetic ManagementOne of our goals was to estimate the benefits (improvements ingenetic variation demographic parameters and populationgrowth and persistence parameters) and costs (financial costs ofcapture transportation quarantine release and monitoring ofintroduced panthers) of genetic management To achieve thisobjective we extended the IBM to include a genetic componentTo simulate individual‐ and population‐level heterozygosity overtime we assigned each panther alleles for 16 microsatellite locibased on the empirical allele frequency in the population (esti-mated from genetic samples collected during 2008ndash2015) whenwe initialized the model (Pierson et al 2015) At each time stepwe simulated Mendelian inheritance for kittens produced suchthat each kitten randomly inherited alleles from its mother andfrom a male panther randomly chosen to have putatively siredthe litter We did not explicitly force inbreeding to occur butthe probability of inbreeding naturally increased as the popula-tion size decreasedIntrogression strategiesmdashWe modeled genetic introgression as a

result of the release of 2‐year‐old female pumas from Texas intothe simulated Florida panther population For the geneticintrogression implemented in 1995 Texas females between 15and 25 years of age were released because females at that agehave lower mortality rates (Benson et al 2011) higherreproductive potential and also because it is much easier todetect with certainty the successful reproduction by females thanby males The ability to more closely monitor reproduction andthe birth of kittens is an important consideration in reducing theprobability of a genomic sweep of Texas alleles into the Floridapanther population Wildlife managers removed the last 2surviving Texas females from the wild population in Florida in2003 to reduce the possibility of genomic sweep To simulatethis strategy in the model we removed any Texas females thatwere still alive 5 years after each introgression event Weassigned Texas females alleles based on the allele frequency ofthe Texas population (based on genotypes from 48 samplescollected in west Texas that was inclusive of the pumas releasedin South Florida in 1995) when they entered the Florida pantherpopulation We could not estimate demographic rates separatelyfor the released Texas females because of small sample size (only8 females were released in 1995) therefore we assumed that

van de Kerk et al bull Florida Panther Population and Genetic Viability 15

survival and reproductive rates and the effects of covariates onthese rates for the released Texas females were identical to thoseof female Florida panthersThe impact of genetic introgression depends on the frequency

of introgression events and the number of individuals introducedat each attempt Thus we defined introgression strategies basedon the number of Texas females to be released (0 5 10 or 15)and the interval between genetic introgressions (10 20 40 and80 yr) for a total of 13 strategies We simulated each manage-ment strategy with density dependence estimated using theMPC and MVM abundance indices

We sampled 1000 parameter sets and for each of these setswe ran all introgression strategies 1000 times for 200 years Weapplied the same parametric uncertainty and environmentalstochasticity across all management scenarios to allow for bettercomparison of introgression strategies At each time step werecorded the projected population size and average individualheterozygosity We calculated the heterozygosity for each in-dividual at birth based on the alleles it inherited This individualheterozygosity then informed the survival rate of that individualbased on the empirically estimated relationship between in-dividual heterozygosity and age‐specific survival rates Hetero-zygosity did not change throughout an individualrsquos lifetime butits effect on survival changed as individuals aged because theeffect of individual heterozygosity on survival rate differedamong age classes Survival rates of genetically sampled kittenswere biased high because some kittens were sampled later in lifewhen they had already survived for some time To correct forthis bias we adjusted kitten survival rates predicted by theheterozygosity model proportionally From the population tra-jectories we estimated the probability of quasi‐extinction (de-fined as the probability that the population will fall below 10 or30 individuals or that individuals of only 1 sex will remain)

Cost‐benefit analysismdashExperts estimated financial costs ofintrogression based on their experience with the geneticintrogression executed in 1995 and we adjusted estimates tocurrent costs to account for inflation (R T McBride RancherrsquosSupply Incorporated M Cunningham and D P OnoratoFlorida Fish and Wildlife Conservation Commission personalcommunication) Costs included capturing and transportingfemale pumas from the Texas source population caring for thepumas while they were in quarantine and post‐introgressionmonitoring To compare the financial costs of the differentscenarios we also calculated the average costs incurred over

Table 2 Sample size (n) number of alleles (Na) allelic richness (AR) observedheterozygosity (Ho) and expected heterozygosity (He) at 16 microsatellite loci inradio‐collared Florida panthers sampled from 1981 to 2013 in South FloridaUSA We calculated these values from a subset (n= 137) of the samples used inthe analyses and they include only adult and subadult panthers We excludedkittens because they can result in an overrepresentation of certain alleles

Locus n Na AR Ho He

FCA090 129 5 50 0333 0401FCA133 136 3 30 0618 0608FCA243 136 5 50 0419 0487F124 137 7 69 0708 0729F37 129 4 40 0395 0397FCA075 131 5 50 0534 0598FCA559 133 5 50 0496 0503FCA057 134 6 59 0679 0624FCA081 134 7 69 0209 0206FCA566 130 6 60 0462 0475F42 133 10 100 0737 0766FCA043 136 5 49 0596 0588FCA161 132 6 60 0500 0515FCA293 132 4 40 0614 0624FCA369 131 6 60 0573 0567FCA668 133 7 70 0406 0462Mean 1329 57 57 0517 0535

Figure 4 Proportional membership (q) of radio‐collared Florida panthers (n= 218) and pumas from Texas USA (n= 7) in 2 clusters identified by STRUCTUREEach individual animal is represented by a separate vertical bar Yellow indicates canonical panther ancestry and blue represents admixed ancestry The admixedancestry of the Texas females and F1 generation panthers that were radiocollared are highlighted red and green respectively to demarcate those groups The pre‐introgression period is inclusive of panthers radiocollared from 10 February 1981 to 6 March 1996 The post‐introgression era inclusive of the F1 panthers includesanimals radiocollared from 4 March 1997 to 18 February 2015 in South Florida USA

16 Wildlife Monographs bull 203

100 years assuming that the population would be managedaccording to a particular strategy Our goal with theseexploratory analyses was to evaluate the relative costs andbenefits of alternative management scenarios

RESULTS

Genetic VariationWe assessed genetic variation at 16 microsatellite loci for 137DNA samples collected from adult and subadult panthers(1981ndash2013) that were captured and subsequently radiocollaredThe number of alleles at these loci ranged from 3ndash10 and allelicrichness averaged 57 (Table 2) Average expected hetero-zygosity for this sample was 0535 and average observed het-erozygosity was 0517 (Table 2)We determined individual heterozygosity (1 minusHL) and an-

cestry for the complete sample of panther genotypes (n= 503)collected from 1981 to 2015 for use as covariates in subsequentanalyses Average individual heterozygosity was 0559 andranged from 0ndash0980 Our STRUCTURE analysis providedstrong support for K= 2 clusters based on the probability of thedata (log Pr(D)|K) and ΔK in STRUCTURE HARVESTERBased on this result we categorized the ancestry of each pantheras either canonical (n= 123) or admixed (n= 380) using valuesof proportional membership (q Fig 4) The average individualheterozygosity of canonical panthers was 0386plusmn 0012 (SE) foradmixed panthers it was 0615plusmn 0007

Survival and Cause‐Specific Mortality RatesOf the 395 kittens that were PIT‐tagged and released 55 werelater encountered alive 15 were recovered dead and 85 werepart of failed litters (Table 1) We radio‐tracked 209 subadult oradult panthers for a total of 244195 panther‐days and docu-mented 126 mortalities (Table 1)Survival depended strongly on age and kittens had the lowest

overall rate (032plusmn 009 Fig 5A Table 3) The model‐averagedkitten survival was 047plusmn 009 when we included only geneti-cally sampled individuals in the analysis (Table 3) as statedabove this estimate is biased high because many kittenswere genetically sampled when they were recaptured alive orrecovered dead at older ages We found no evidence for sex‐specific differences in kitten survival but females survived betterthan males in all older age classes For females subadults hadthe highest survival rates followed by prime adults and oldadults For males prime adults had the highest survival ratefollowed by subadults and old adults (Fig 5A and Table 3)For panthers of all ages there was substantial evidence that

ancestry affected survival with F1 admixed panthers of all ageshaving the highest rates and canonical individuals having thelowest (Fig 5B) The combined AIC weights of models thatincluded an ancestry covariate were 0880 for kittens and 0992for older panthers The survival rates of subadult prime‐adultand old‐adult panthers in the period immediately after the in-trogression (1995ndash2004) were similar to those in more recentyears (2005ndash2013 Fig 5C)After genetic introgression individual panther heterozygosity

increased (Fig 6) and varied among ancestry categories individualheterozygosity was the highest for F1 panthers (078plusmn 001)

A

B

C

Figure 5 Overall age‐ and sex‐specific survival rates (plusmnSE A) age‐ and sex‐specific survival rates (plusmnSE) for Florida panthers of canonical F1 admixed andother admixed ancestry (B) and age‐ and sex‐specific survival estimates (plusmnSE)for subadult and adult Florida panthers in the period immediately post‐introgression (1995ndash2004) and more recently (2005ndash2013 C) in SouthFlorida USA

Table 3 Model‐averaged estimates (plusmnSE) of demographic parameters for theFlorida panther obtained using data collected during 1981ndash2013 in SouthFlorida USA

Parameter Sex Age class Estimate

Survival Both Kitten 032plusmn 009a

Female Subadult 097plusmn 002Prime adult 086+ 003Old adult 078plusmn 009

Male Subadult 066plusmn 006Prime adult 077plusmn 005Old adult 065plusmn 010

Probability of reproduction Female Subadult 035plusmn 008Prime adult 050plusmn 005Old adult 025plusmn 006

Kittens produced annually Female Subadult 280plusmn 005Prime adult 267plusmn 001Old adult 228plusmn 013

a Overall kitten survival (based on all kittens in our data set including thosethat were not genetically sampled) Estimate of survival of geneticallysampled kittens was 047plusmn 009 this estimate is biased high because manykittens were genetically sampled when they were recaptured alive or re-covered dead at older ages

van de Kerk et al bull Florida Panther Population and Genetic Viability 17

followed by those for non‐F1 admixed (062plusmn 001) and canonical(039plusmn 001) panthers Individual heterozygosity positively affectedsurvival rates of kittens (slope parameter η= 1455 95CI= 0505ndash2405) Individual heterozygosity also positively af-fected survival of subadult and adult panthers but the evidence forthis effect was weaker because the confidence interval for the ha-zard ratio overlapped 10 (hazard ratio exp(ɩ)= 0311 95CI= 0092ndash1054 Table 4 and Fig 7)The MPC increased after genetic introgression (Fig 2) This

abundance index was also included in top‐ranking modelsindicating that population density affected survival (Table 4)Abundance reduced the survival of kittens (slope parameterη= minus0035 95 CI= minus0049 to minus0021) evidence was weak forthe effect of abundance on survival of subadult and adult panthers(hazard ratio exp(ɩ)= 1002 95 CI= 0995ndash1010 Fig 7) Re-gression coefficients associated with the effect of abundance andindividual heterozygosity on demographic parameters are presentedin Table B1 (available online in Supporting Information)The most frequently observed causes of death for radio‐

collared panthers were vehicle collision and intraspecific ag-gression Cause‐specific mortality analyses revealed thatmortality rates among possible causes were generally similar forthe 2 sexes but that males were more likely to die from in-traspecific aggression than were females (Fig 8A) Male mor-tality from intraspecific aggression was higher for subadult andold adult panthers than for prime adults (Fig 8B) For bothmales and females canonical panthers were more likely to diefrom intraspecific aggression than were admixed panthers(Fig 8C)

Reproductive ParametersOf 101 radio‐collared females 67 reproduced at least once duringour monitoring we used these individuals to assess the annualprobability of reproduction The top‐ranked model for the annual

probability of reproduction included age effect only suggesting thatadding ancestry or abundance did not improve model parsimony(Table 4) Model‐averaged annual probabilities of reproductiondiffered between female subadults (035plusmn 008) prime adults(050plusmn 005) and older adults (025plusmn 006)The most parsimonious model for the number of kittens

produced annually by females that produced at least 1 litter (ieconditional on reproduction) included effects of age ancestryand abundance a model that included only the effect of age wasalmost equally supported (Table 4) The model‐averagednumber of kittens produced annually conditional on re-production was 280plusmn 075 for subadults 267plusmn 043 for primeadults and 228plusmn 083 for old adults Confidence intervals forthe slope parameter (β) overlapped 0 for both abundance(β= 0010 95 CI= minus0001 to 0021) and ancestry (β= minus073795 CI= minus1637 to 0162)

Population Dynamics and PersistenceDeterministic and stochastic demographymdashThe deterministic (λ)

and stochastic (λs) annual population growth rate calculated usingthe matrix population model were 106 (5th and 95th percentiles099ndash114) and 104 (072ndash141) respectively The generation timewas 47 years A female starting in age class one (kitten) wasexpected to live another 58 years and produce 10 kittens onaverage during her lifetime Overall λs was proportionately(elasticity) most sensitive to changes in female prime adultsurvival on the absolute scale (sensitivity) however it was mostsensitive to changes in kitten survival (Fig 9) Populationtrajectories probabilities of quasi‐extinction and times untilquasi‐extinction estimated using the matrix population modeland IBM for comparable scenarios were nearly identical for acritical quasi‐extinction threshold of 10 panthers (Figs 10 and 11)and for a critical quasi‐extinction threshold of 30 panthers(Appendix E available online in Supporting Information)Minimum population count scenariomdashUnder the MPC scenario

(ie density dependence estimated using the minimumpopulation count data) with a critical threshold of 10panthers the population size reached 99 (72ndash104) and 100(77ndash105) at 200 years (Fig 10) as predicted by the IBM and thematrix model respectively The cumulative quasi‐extinctionprobabilities within 100 years based on the MPC scenario were15 (IBM 0ndash48) and 30 (matrix model 0ndash76) Theseprobabilities increased to 25 (IBM 0ndash114) and 38 (matrixmodel 0ndash154) by year 200 (Fig 10) The mean times untilquasi‐extinction were 15 years (IBM 0ndash67) and 15 years (matrixmodel 0ndash72) conditioned on going quasi‐extinct within 100years Expected times until quasi‐extinction conditioned onquasi‐extinction within 200 years were higher 34 years (IBM0ndash137) and 32 years (matrix model 0ndash134 Fig 10)Motor vehicle mortality scenariomdashUnder the MVM scenario

(ie density dependence estimated using estimates of pantherpopulation size from McClintock et al 2015) with a criticalthreshold of 10 panthers the projected population reached 188(142ndash217) and 190 (143ndash219) at 200 years as predicted by theIBM and the matrix model respectively (Fig 11) Thecumulative quasi‐extinction probabilities within 100 years were14 (IBM 0ndash08) and 13 (matrix model 0ndash06) Theseprobabilities increased to 20 (IBM 0ndash17) and 19 (matrix

000

025

050

075

100

1995 2000 2005 2010Year of birth

Indi

vidu

al h

eter

ozyg

osity

Figure 6 Temporal changes in the individual heterozygosity of Floridapanthers born during the period 1995ndash2013 in South Florida USA Pointsare measurements solid line is linear regression shaded area is 95 confidenceinterval of slope Slope= 0006plusmn 0001 R2= 009

18 Wildlife Monographs bull 203

model 0ndash16) within year 200 (Fig 11) The mean times until quasi‐extinction based on the MVM scenario were 5 years (IBM 0ndash61)and 6 years (matrix model 0ndash64) conditioned on going quasi‐extinctwithin 100 years Expected times until quasi‐extinction conditionedon quasi‐extinction within 200 years were 11 years (IBM 0ndash113)and 11 years (matrix model 0ndash112 Fig 11)

Sensitivity of extinction probability to demographic parametersmdashThe partial rank correlation coefficients between quasi‐extinctionprobabilities and demographic parameters were negative anincrease in any of the demographic parameters decreased thequasi‐extinction probability Probabilities of quasi‐extinction within200 years were most sensitive to changes in kitten survival for bothscenarios (Fig 12) followed by survival of subadult females in theMVM scenarios and survival of prime adult females in the MPCscenario The probability of reproduction of prime adult femaleshad the third‐largest partial rank correlation coefficient with quasi‐extinction probability for both scenarios

Benefits and Costs of Genetic ManagementMinimum population count scenariomdashThe MPC scenario with

a critical threshold of 10 panthers predicted that withoutmanagement intervention average individual heterozygositywould decrease reaching 057 (049ndash064) at 100 years and053 (042ndash062) at 200 years (Fig 13) The population woulddecrease slightly reaching an average of 79 (41ndash99) at 100 yearsand 76 (43ndash98) at 200 years (Fig 14) The probabilities of quasi‐extinction under the no‐intervention MPC scenario when weconsidered the effects of genetic erosion were 13 (0ndash99) at 100years and 23 (0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with

the translocation of 5 Texas females average individualheterozygosity under the MPC scenario increased and thenstabilized at 068 (064ndash072) at 100 years and remained atthat level at 200 years (Fig 13) The population reached anaverage of 88 (62ndash104) at 100 years and stayed the same at200 years (Fig 14) The probabilities of quasi‐extinctionunder this introgression strategy were 6 (0ndash53) within 100years and 7 (0ndash82) within 200 years (Fig 15) Results ofanalyses using a critical threshold of 30 panthers revealed asimilar pattern of relative benefits However the prob-abilities of quasi‐extinction were substantially higher (ge45)across all scenarios (Appendix E)

Motor vehicle mortality scenariomdashWithout managementintervention the average individual heterozygosity predictedby the MVM scenario and a critical threshold of 10 panthersdecreased to 060 (046ndash069) at 100 years and to 059 (039ndash070) at 200 years (Fig 13) The population increased slightlyand reached an equilibrium at 154 individuals (46ndash217) at 100years and 157 (57ndash218) at 200 years (Fig 14) The probabilitiesof quasi‐extinction were 17 (0ndash100) at 100 years and 22(0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with 5

female pumas from Texas average individual heterozygosityincreased slightly reaching 067 (060ndash073) at 100 years and068 (059ndash075) at 200 years (Fig 13) Under this strategy thepopulation reached an average of 164 individuals (44ndash226) at

Table 4 Model comparison results testing for the effects of several covariateson survival of all kittens survival of genetically sampled kittens survival ofsubadult and adult Florida panthers probability of reproduction and kittensproduced annually for Florida panthers sampled from 1981ndash2013 in SouthFlorida USA

Model Parameters ΔAICa Weight

Kitten survival (all data)Base+ F1b+ abundancec 10 000 0988Base+ abundance 9 1018 0006Base+ F1 9 1035 0005Base 8 2711 lt0001Base+ sexd 9 2912 lt0001Base+ litter sizee 9 2914 lt0001

Kitten survival (genetically sampled kittens only)Base+ F1+ abundance 10 000 0541Base+ F1CanAdmf+ abundance 11 099 0329Base+Hetg+ abundance 10 310 0115Base+CanAdmh+ abundance 10 791 0010Base+ abundance 9 944 0005Base+ F1 9 1638 lt0001Base+ F1CanAdm 10 1837 lt0001Base+Het 9 2872 lt0001Base 8 3296 lt0001Base+CanAdm 9 3348 lt0001

Subadult and adult survivalBase+ F1CanAdm+ abundance 7 000 0360Base+ F1 5 053 0276Base+ F1CanAdm 6 105 0213Base+ F1+ abundance 6 222 0119Base+CanAdm+ abundance 6 575 0020Base+CanAdm 5 908 0004Base+Het 5 964 0003Base+Het+ abundance 6 984 0003Base 4 1118 0001Base+ abundance 5 1278 0001

Probability of reproductionAge 3 000 0242Age+CanAdm 4 012 0228Age+ abundance 4 173 0102Age+ F1 4 190 0094Age+CanAdm+ abundance 5 216 0082Age+Het 4 219 0081Age+ F1CanAdm 5 232 0076Age+ F1+ abundance 5 372 0038Age+Het+ abundance 5 399 0033Age+ F1CanAdm+ abundance 6 444 0026

Kittens produced annuallyAge+ F1+ abundance 5 000 0194Age 3 060 0144Age+ abundance 4 061 0143Age+ F1 4 097 0120Age+CanAdm 4 139 0097Age+ F1CanAdm+ abundance 6 196 0073Age+ F1CanAdm 5 231 0061Age+CanAdm+ abundance 5 238 0059Age+Het+ abundance 5 250 0056Age+Het 4 259 0053

a Akaikersquos information criterion (AIC) for kitten survival models was adjustedfor small sample size and overdispersion (QAICc)

b F1 divides Florida panthers into 2 groups F1 admixed and all other panthersc Index of Florida panther abundanced Sex of an individual Florida panther (male or female)e Size of the litter in which a Florida panther kitten was bornf F1CanAdm divides panthers into 3 groups F1 admixed canonical andother admixed panthers

g Het is the individual heterozygosityh CanAdm divides panthers into 2 groups canonical and admixed (includingF1 admixed) panthers

van de Kerk et al bull Florida Panther Population and Genetic Viability 19

100 years and 170 (67ndash231) at 200 years based on the MVMscenario (Fig 14) The probabilities of quasi‐extinction underthis introgression strategy were 10 (0ndash99) at 100 years and12 (0ndash99) at 200 years (Fig 15)The projected population size and average individual hetero-

zygosity reached equilibrium levels with periodic fluctuations inthese values when introgression was implemented (Figs 16ndash17)Genetic introgression decreased the time until quasi‐extinctionacross all scenarios (Fig 18) Results of analyses using a criticalthreshold of 30 panthers revealed a similar pattern of relative ben-efits However the probabilities of quasi‐extinction were sub-stantially higher (ge45) across all scenarios (Appendix E)

Addition of pumas from Texas increased both the populationsize and average individual heterozygosity consequentlyintrogression caused periodic increases in average individualheterozygosity and population size at the interval at which theintrogression occurred Whereas average individual hetero-zygosity gradually decreased following introgression densitydependence caused the population size to fluctuate especiallywhen a larger number of pumas from Texas were released Thepositive effect of genetic introgression initiatives on quasi‐extinction probabilities decreased as the time interval betweenthese events increased More frequent genetic introgressions ledto greater increases in average individual heterozygosity and

Old adult

Prime adult

Subadult

Kitten

025 050 075

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Individual heterozygosity

Ann

ual s

urvi

val r

ate

Old adult

Prime adult

Subadult

Kitten

40 60 80 100 120

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Abundance index

Ann

ual s

urvi

val r

ate

Kittens Males Females

Figure 7 The effect of individual heterozygosity (left) and the abundance index (right) on Florida panther age‐ and sex‐specific survival estimates (shaded area isSE) in South Florida USA 1981ndash2013 Abundance index data are from McBride et al (2008) and R T McBride and C McBride (Rancherrsquos Supply Incorporatedunpublished report)

20 Wildlife Monographs bull 203

equilibrium population size while reducing the probability ofquasi‐extinction Comparatively performing genetic introgres-sion less often but with more individuals per release was lesseffective at improving the long‐term prospects for the pantherpopulation (Figs 13ndash15)

Financial cost and persistence benefitsmdashThe total estimated costof 1 genetic introgression was $40000 $80000 or $120000for releasing 5 10 or 15 pumas from Texas respectively(Table 5) The total cost incurred over 100 years for eachstrategy ranged from $50000 to $12 million (Table 6)The most expensive strategy involved the introduction of15 pumas every 10 years this strategy reduced the probability

of quasi‐extinction by 58 and 40 under the MPC and MVMscenarios respectively (Table 6) Introducing 5 pumas every80 years cost the least but improvements in populationpersistence under this strategy were insubstantial (Table 6)Releasing more panthers more frequently caused increasedpopulation fluctuations but did not necessarily reduce the quasi‐extinction probability (Figs 14 and 15 Table 6)

DISCUSSION

The duration (34 yr of data) and sample sizes associated with theFlorida Panther Project allowed us to obtain a comprehensiveunderstanding of genetic variation demographic parameters

A

B

C

Figure 8 Cause‐specific mortality rates (plusmnSE) for male and female Florida panthers (A) subadult prime‐adult and old‐adult Florida panthers of both sexes (B)and canonical and admixed Florida panthers of both sexes (C) in South Florida USA 1981ndash2013 Causes of mortality are vehicle collision intraspecific aggression(ISA) other causes (known causes such as diseases injuries and infections unrelated to the first 2 causes) and unknown cause (mortalities for which we could notassign a cause)

van de Kerk et al bull Florida Panther Population and Genetic Viability 21

probability of population persistence and the duration of thepositive impacts of genetic restoration on this endangered po-pulation The Florida panther population was in dire straits inthe early 1990s and our results clearly demonstrated that the

implementation of the introgression project in 1995 subse-quently accrued a series of genetic and demographic improve-ments that have led to the change from a declining population toone that is growing and expanding its current breeding range

Sensitivity Elasticity

0 1 2 3 00 02 04

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

Para

met

er

Figure 9 Sensitivity and elasticity (proportional sensitivity) of stochastic population growth rate (λs) of the Florida panther to vital demographic parameters in SouthFlorida USA 1981ndash2013 Horizontal lines represent 5th and 95th percentiles

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

70

80

90

100

110

0

5

10

15

0

50

100

Years

StatisticMeanMedian

Figure 10 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the minimum population count(MPC) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

125

150

175

200

00

05

10

15

20

0

30

60

90

Years

StatisticMeanMedian

Figure 11 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the motor vehicle mortality(MVM) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

22 Wildlife Monographs bull 203

MPC MVM

minus08 minus06 minus04 minus02 00 minus08 minus06 minus04 minus02 00

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Male subadult survival

Male prime adult survival

Male old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

PRCC

Para

met

er

Figure 12 Partial rank correlation coefficient (PRCC) between quasi‐extinction probabilities and the demographic parameters of the Florida panther for theminimum population count (MPC) and motor vehicle mortality (MVM) scenarios Error bars indicate standard deviation Based on data collected in South FloridaUSA 1981ndash2013

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 13 Average projected individual heterozygosities in the Florida panther population over 200 years without genetic management intervention and with eachof the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include theintroduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years Average individual heterozygosity peaks when pumas are added to the Floridapanther population Based on data collected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 23

Our research has further validated the success of genetic in-trogression by quantifying the reduction in the probability ofextinction of the panther population because of this manage-ment initiative These findings all bode well for continuedprogress toward recovery That said the Florida panther po-pulation remains small and isolated from other populations ofpuma from western North America thereby denoting that theeventual impact of genetic drift and inbreeding may negativelyaffect the population in the future Realizing this will continueto be an issue that managers will have to contemplate wemodeled the impact of varied genetic management scenarios todetermine the frequency of introgression along with the numberof panthers that should be released during each initiative tominimize cost and maximize benefits to the population Theseresults will prove invaluable to managers attempting to conservethe Florida panther

Genetic Variation Demographic Parametersand Persistence of Heterotic Benefits from GeneticIntrogressionOur measure of individual heterozygosity (1 minusHL) was higheron average for the admixed (0615) panthers than for those ofcanonical ancestry (0386) Levels of individual heterozygosityfor admixed panthers were comparable to levels observed in asample of pumas from west Texas (x plusmn SE= 0695plusmn 0019n= 41) which further demonstrates the improvement in levelsof genetic variation in the Florida population since 1995 The

correlation between HL and several demographic parametershas been noted in wild populations including cheetahs (Terrellet al 2016) and several species of birds such as European shags(Phalacrocorax aristotelis male and female survival probabilityVelando et al 2015) and Egyptian vulture (Neophron percnop-terus age at recruitment Agudo et al 2012) If we focus only onpanthers captured and radiocollared from 2012ndash2015 (FP211ndashFP240) all of which had estimated birth years ge2005 (ge10 yrpost‐introgression) those panthers had an average individualheterozygosity level of 0618plusmn 0029 highlighting the elevatedlevels of genetic variation that continue to persist in cohorts ofpanthers 5 generations after the release of female pumas fromTexasThe proportional membership values (q) from our

STRUCTURE analysis allowed us to delineate 2 clusters ofancestry for Florida panthers (Fig 4) During the pre‐in-trogression era the population was dominated by panthers thatretained a high percentage of the canonical ancestry which wasaffected by inbreeding depression The post‐introgression eraancestry values are comprised of many admixed individuals inessence demonstrating the success of the introgression in termsof increasing genetic variation in the population and mi-micking gene flow that historically occurred with panthers andother populations of pumas (Onorato et al 2010) A somewhatanalogous situation although abetted by natural immigrationwas evident in the ancestral variation in a small population ofpuma intersected by a major freeway in California USA In

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

75

80

85

90

95

100

130

150

170

190

Years

Popu

latio

n si

ze

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 14 Average projected population sizes in the Florida panther population over 200 years without intervention and with each of the genetic introgressionstrategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include the introduction of 5 10 or 15pumas from Texas USA every 10 20 40 or 80 years Peaks occur when pumas are added to the Florida panther population Based on data collected in SouthFlorida USA 1981ndash2013

24 Wildlife Monographs bull 203

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

25

50

75

100

0

25

50

75

100

Introgression interval (yrs)

PQ

E (

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 15 Average probability of quasi‐extinction (PQE) of the Florida panther population within 100 years and within 200 years without genetic managementintervention (N) and with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenariosIntrogression strategies include the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Errorbars indicate 5th and 95th percentiles Based on data collected in South Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

50

60

70

80

90

100

110

50

100

150

200

Years

Popu

latio

n si

ze

Figure 16 Average projected Florida panther population trajectories under a geneticintrogression strategy involving the release of 5 female pumas from Texas USA every20 years for the minimum population count (MPC) and motor vehicle mortality(MVM) scenarios Shaded area indicates 5th and 95th percentiles and isrepresentative of confidence intervals for other scenarios Based on data collected inSouth Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Figure 17 Average projected Florida panther population‐level heterozygositiesunder a genetic introgression strategy involving the release of 5 female pumas fromTexas USA every 20 years for the minimum population count (MPC) and motorvehicle mortality (MVM) scenarios Shaded area bounds the 5th and 95th percentilesand is representative of confidence intervals for other scenarios Based on datacollected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 25

this case the migration of just a single male across the freewayto the south resulted in an increase in the number of in-dividuals with admixed ancestry and substantially improvedgenetic diversity of the subpopulation south of the freeway(Riley et al 2014)Our estimate of overall kitten survival (0321plusmn 0056) was

similar to that reported by Hostetler et al (2010) who used13 years of post‐introgression data (0323plusmn 0071) These valuesare lower than kitten survival estimates derived from severalpuma populations in western North America (044ndash091 Loganand Sweanor 2001 Lambert et al 2006 Laundre et al 2007Robinson et al 2008 Ruth et al 2011) When we included onlydata for individuals that had been genetically sampled our es-timate was 15 higher The proportion of admixed kittens thatwere genetically sampled increased as the population expandedwhich could have resulted in higher estimates of kitten survivalbecause admixed kittens survive better than canonical kittensKitten survival was strongly influenced by genetic ancestry withsurvival rates of F1 kittens being almost twice that of canonicalkittens (Fig 5B) Consistent with the findings of Hostetler et al(2010) increases in heterozygosity coincided with increasedkitten survival whereas population density negatively affectedkitten survival Population density often has a strong negativeimpact on survival of young age classes due to infanticide andcannibalism which has been reported from several carnivore

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

50

100

150

0

50

100

150

Introgression interval (yrs)

TQE

(yrs

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 18 Average times until quasi‐extinction (TQE) for the Florida panther population within 100 years and within 200 years without genetic management interventionand with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategiesinclude the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Error bars indicate 5th and 95thpercentiles Based on data collected in South Florida USA 1981ndash2013

Table 5 Average cost (2017 US dollars) of introgression strategies employedover 100 years that involve the introduction of 5 10 or 15 pumas from TexasUSA into the Florida panther population in South Florida USA at an in-creasingly higher frequency Costs of capture (including transportation) quar-antine and post‐introgression monitoring were estimated by Roy McBrideMark Cunningham and Dave Onorato respectively

Frequency Component 5 pumas 10 pumas 15 pumas

Once Capture 25000 50000 75000

Quarantine 5000 10000 15000

Monitoring 10000 20000 30000

Total 40000 80000 120000

Every 80 years Capture 31250 62500 93750

Quarantine 6250 12500 18750

Monitoring 12500 25000 37500

Total 50000 100000 150000

Every 40 years Capture 62500 125000 187500

Quarantine 12500 25000 37500

Monitoring 25000 50000 75000

Total 100000 200000 300000

Every 20 years Capture 125000 250000 375000

Quarantine 25000 50000 75000

Monitoring 50000 100000 150000

Total 200000 400000 600000

Every 10 years Capture 250000 500000 750000

Quarantine 50000 100000 150000

Monitoring 100000 200000 300000

Total 400000 800000 1200000

26 Wildlife Monographs bull 203

species including polar bears (Ursus maritimus Derocher andWiig 1999) black bears (Ursus americanus Garrison et al 2007)and pumas (Logan and Sweanor 2001)Our estimates of sex‐ and age‐specific survival rates of subadult

and adult panthers (Table 3) and the effects of covariates on theserates were similar to those reported by Benson et al (2011) derivedfrom data collected through 2006 Our survival rates for males(065ndash077) were lower than females (078ndash097) for all 3 ageclasses (subadult prime adult and old adult) which is the typicaltrend observed in most puma populations (eg Ruth et al 19982011) However an unhunted puma population occupying afragmented urban landscape in southern California exhibited lowersurvival (0586 females 0525 males Vickers et al 2015) perhapsas a result of severe habitat fragmentation and the lack of extensiveswaths of public land protected in perpetuity Most populations ofpuma in western North America are typically subject to variedlevels of management via regulated hunting which often is biasedtoward the take of males (Cooley et al 2009 Ruth et al 2011)Consequently annual survival estimates from hunted populationsin northeast Washington (0679ndash0733 females 0341 malesRobinson et al 2008) and southeastern Arizona (0685 females0584 males Cunningham et al 2001) were generally lower thanthose we estimated for Florida panthersCause‐specific mortality analysis showed that male panthers

experienced a substantially greater mortality risk from in-traspecific aggression This differs from the pattern observedduring a long‐term study in Arizona where females were at ahigher risk of intraspecific mortality than males (46 of maleand 53 of female mortality Logan and Sweanor 2001)Mortality associated with intraspecific strife has been docu-mented in hunted and unhunted populations (this study Loganand Sweanor 2001 Stoner et al 2010) and its root cause hasbeen difficult to determine (eg population density and preypopulation trends) That being the case finding ways to reducewhat is often a major source of mortality for puma populationscontinues to pose a challenge to managersCanonical panthers were substantially more likely to die from

intraspecific aggression than were admixed panthers This mayat least partly explain the difference in survival between admixed

and canonical panthers Canonical panthers are known to bemore likely to suffer from varied correlates of inbreeding de-pression than admixed animals including atrial septal defects(Johnson et al 2010) and compromised immune systems(Roelke et al 1993) These factors have the potential to affectfitness and perhaps dominance of individuals when engaged inan intraspecific encounter A somewhat similar scenario wasobserved by Hogg et al (2006) in a population of bighorn sheepthat were part of an introgression project Admixed males in thispopulation had a greater probability of paternity than males thatwere less outbred This included coursing males with higherlevels of admixture being markedly more successful at fightingmales that were defending females in estrous (Hogg et al 2006)Heterosis resulting from genetic introgression is expected to be

the strongest in F1 individuals (Shull 1908 Crow 1948 Burkeand Arnold 2001 Waller 2015) and to progressively decline insubsequent generations (Pickup et al 2013 Frankham 2015)Given that admixed (especially F1) panthers survived better thancanonical panthers and that individual heterozygosity improvedsurvival of both sexes and all age classes our data provide evi-dence for heterotic advantage whereby individuals of mixedancestry had higher fitness (Shull 1908 Crow 1948) Our resultsare consistent with findings of other studies that involved in-tentional genetic introgression for conservation purposes Forexample Hogg et al (2006) detected substantial improvementsin survival and reproduction of introgressed bighorn sheep andMadsen et al (1999 2004) reported a dramatic increase in thepopulation of adders after genetic introgressionKnowledge of the persistence of benefits to a population ac-

crued via genetic introgression is essential for determiningwhen additional introgression may be warranted There is adepauperate amount of information regarding this topic and theidentification of factors that may influence persistence of thebenefits of genetic introgression (Tallmon et al 2004 Hedricket al 2014 Whiteley et al 2015) For example in minks(Neovison vison) Thirstrup et al (2014) found that outcrossingled to larger litters in F2 and F3 but this benefit disappeared insubsequent generations In a population of gray wolves Hedricket al (2014) suggested that benefits of genetic introgression may

Table 6 Costs and benefits accumulated over 100 years for genetic introgression at different intervals and with different numbers of pumas from Texas USAintroduced into the Florida panther population in South Florida USA Costs (2017 US dollars) include capture and transportation quarantine and post‐introgression monitoring Benefits are represented as average probability of quasi‐extinction (PQE and 5th and 95th percentiles) and changes (Δ) therein under thescenario using the minimum population count (McBride et al 2008 MPC) and the scenario using the motor vehicle mortality population estimates (McClintocket al 2015 MVM) The table is sorted by percent change in probability of quasi‐extinction under the MPC scenario

Interval (yr) Pumas Cost MPC PQE () ΔMPC PQE () MVM PQE () ΔMVM PQE ()

ndash 0 0 13 (0ndash99) 0 17 (0ndash100) 080 10 100000 12 (0ndash99) 10 (0ndash58) 16 (0ndash100) 5 (0ndash38)80 15 150000 12 (0ndash99) 13 (0ndash65) 15 (0ndash100) 8 (0ndash56)80 5 50000 12 (0ndash99) 14 (0ndash73) 15 (0ndash99) 9 (0ndash41)40 15 300000 10 (0ndash98) 26 (0ndash104) 14 (0ndash99) 13 (0ndash60)40 10 200000 9 (0ndash94) 35 (0ndash183) 13 (0ndash99) 21 (0ndash95)40 5 100000 8 (0ndash89) 39 (0ndash211) 12 (0ndash99) 26 (0ndash124)20 15 600000 8 (0ndash88) 42 (0ndash257) 13 (0ndash99) 24 (0ndash123)20 10 400000 6 (0ndash67) 54 (0ndash318) 10 (0ndash99) 37 (0ndash217)10 15 1200000 6 (0ndash53) 58 (0ndash372) 10 (0ndash99) 40 (0ndash208)20 5 200000 5 (0ndash50) 59 (0ndash338) 9 (0ndash99) 44 (0ndash352)10 10 800000 4 (0ndash16) 69 (0ndash489) 8 (0ndash92) 55 (0ndash401)10 5 400000 4 (0ndash7) 73 (0ndash502) 6 (0ndash71) 63 (0ndash503)

van de Kerk et al bull Florida Panther Population and Genetic Viability 27

wane after 2 or 3 generations The WrightndashFisher model ofgenetic drift predicts a geometric decline in expected hetero-zygosity at the rate of (1 minus 12Ne) per generation where Ne is theeffective population size (Hartl and Clark 1997 Frankham et al2014) Thus it seems logical to assume that the duration of thebenefits of introgression is limited especially in a species per-sisting in a small population such as Florida panthersWe expected Florida panther survival in the most recent years

(2005ndash2013) post‐introgression to differ from that observed in thedecade following the introgression in 1995 because pantherabundance and heterozygosity factors known to influence panthersurvival (Hostetler et al 2010 Benson et al 2011) have changed(Figs 2 and 6) We found that subadult and adult survival rates inrecent years (2005ndash2013) were similar to those in the period im-mediately following introgression (1995ndash2004 Fig 5C) overallkitten survival was similar to estimates of Hostetler et al (2010) Infact the pattern of covariate effects on Florida panther survivalreported by Benson et al (2011) and Hostetler et al (2010) hasbarely changed The negative effects of population density andpositive effects of heterozygosity may counterbalance survival ratesreducing population fluctuations Our result that benefits of geneticrestoration have remained in the population nearly for 5 genera-tions post‐intervention was surprising but also encouraging Pos-sible explanations for this observation may include 1) genetic ero-sion occurs at slower rate than previously thought 2) introducing 5migrants in 1 genetic intervention may have beneficial effects si-milar to those from 1 migrant per generation over multiple gen-erations or 3) other and as yet unknown factors may reduce therate of genetic erosion and subsequent demographic effects Adensity‐dependent effect on kitten survival could also explain whynon‐F1 admixed kittens did not survive substantially better thandid canonical kittens most kittens sampled from 2005ndash2013 wereadmixed and their survival was lower possibly because of a density‐dependent effect as the Florida panther population continuouslygrew during that period Trinkel et al (2010) also found thatpopulation density and inbreeding coefficient interacted to affectdemographic parameters and growth rate of an African lion po-pulation Likewise population density interacted with winterweather to affect the survival and population growth rates in Soaysheep (Ovis aries Milner et al 1999 Coulson et al 2001)

Population Dynamics and PersistenceThe finite deterministic and stochastic growth rates were gt10reflecting that much of the data used in this study were collectedfollowing genetic introgression in 1995 during which time thepopulation increased substantially Because our demographicparameter estimates did not change markedly in comparison tothose of Hostetler et al (2013) the similarity between thepopulation growth estimates from these studies was expectedBut these estimates reflect population growth during thedata‐collection period and do not predict future growth In factestimates of population size reported by McClintock et al(2015) suggest that population growth may already be slowing(Fig 2) A similar pattern was observed in an introduced po-pulation of African lion which increased steadily from 1995until 2001 then fluctuated around the presumed carrying ca-pacity thereafter (Trinkel et al 2010) Several other African lionpopulations that experienced various degrees of recovery

subsequently became relatively stable or exhibited decliningtrends (Bauer et al 2015)Our estimates of quasi‐extinction probabilities were lower than

those reported by Hostetler et al (2013) using a 2‐sex age‐structured matrix population model Lower probabilities ofquasi‐extinction than those reported by the earlier study forcomparable scenarios were likely a consequence of the fact thatthe estimates of abundance (McBride et al 2008) and thus thedensity‐dependent effects on survival of kittens and old panthers(gt10 yr) have changed in recent years Additionally these earlyestimates of the probability of quasi‐extinction and of time toquasi‐extinction did not consider genetic erosion that will in-evitably occur without management interventionUnder the MVM scenario the equilibrium population size was

twice that attained under the MPC scenario The MVM popula-tion estimates are approximately twice the MPCs (Fig 2) as aresult the simulated population under the MVM scenario achieveda higher equilibrium size Probability of quasi‐extinction under theMVM scenario was generally greater than that under the MPCscenario This result is a consequence of the fact that the estimateddensity dependence on kitten survival based on the MPC scenario isstronger than that for theMVM scenario (regression coefficients forabundance for the MPC scenario= minus0068 vs minus0002 for theMVM scenario Appendix B Table B1) Consequently simulatedpopulations under the MPC scenario have a stronger tendency tomove toward the equilibrium population size which inherentlyreduces the quasi‐extinction probability (eg Royama 1992)As expected for long‐lived mammals (Heppell et al 2000 Oli

and Dobson 2003) the population growth rate was proportio-nately most sensitive to changes in survival of prime adultsfollowed by that in younger age classes Global sensitivity ana-lysis in contrast showed that under all scenarios extinctionparameters were particularly sensitive to changes in survival ofFlorida panther kittens This result is a consequence of the highsensitivity of the population growth rate to kitten survival(Fig 9) and a larger standard error of kitten survival (Table 3)Results of our sensitivity analyses indicate that future researchshould focus on acquiring additional information on early lifestages to improve the accuracy and precision of estimates ofkitten survival Our results suggest that demographic variables(or their environmental drivers) that strongly influence extinc-tion risks are not necessarily those with the largest elasticityvalues A study of island fox (Bakker et al 2009) found thatextinction risk was strongly influenced by survival modifierparameters that had low elasticity values

Genetic Management When and How to InterveneThe use of genetic introgression as a management tool has al-ways been controversial and the probable effectiveness it mighthave in preventing the imminent demise of the panther popu-lation was initially debated (Creel 2006 Maehr et al 2006Pimm et al 2006) These debates notwithstanding the pantherpopulation had suffered from genetic morphological and bio-medical correlates of inbreeding before this management in-tervention (Johnson et al 2010 Onorato et al 2010) whereasthe post‐introgression period has been characterized by a sub-stantial reduction in the biomedical correlates of inbreeding andimprovements in demographic vigor and abundance (McBride

28 Wildlife Monographs bull 203

et al 2008 Hostetler et al 2010 2013 Johnson et al 2010Benson et al 2011 McClintock et al 2015) Nevertheless thepopulation remains small and isolated habitat is still being lostand there is no current possibility of natural gene flow betweenthis and other North American puma populations thus therecurrence of inbreeding‐related problems and subsequent po-pulation declines are inevitable The question therefore is notwhether genetic management of the Florida panther populationis needed but when and how it should be implementedWithout genetic introgression probabilities of quasi‐extinc-

tion were substantially greater than those from models thatincorporated periodic genetic introgression events (Fig 15)highlighting the importance of genetic management in smallisolated populations When 5 female pumas are added to theFlorida panther population every 10 years genetic variationincreased substantially under the MPC and MVM scenariosThe IBM predicted that the equillibrium population size wouldbe substantially larger when 5 pumas were released into thepopulation every 10 years than populations without intervention(ie no introgression) Releasing 15 Texas females did not af-ford substantially greater benefit to the equilibrium populationsize than did releasing only 5 Texas females Instead addingmore individuals caused increasingly large fluctuations in po-pulation size due to the numerical increases and density de-pendence (Fig 14) possibly diminishing the benefits associatedwith increased genetic variationCompared with the no‐intervention scenario (ie no genetic

introgression implemented) the introduction of 5 female pumasevery 10 years reduced quasi‐extinction probabilities from 13to 4 and from 17 to 6 for the MPC and MVM scenariosrespectively The benefit of genetic‐introgression strategies de-creased as the interval between successive management inter-ventions increased and no benefit was evident once the intervalreached 80 years (Fig 15) Introgression reduced the averagetime until quasi‐extinction (Fig 18) This somewhat counter‐intuitive result can be explained by the fact that repeated geneticintrogression makes the population more robust over timewhich will reduce the probability of extinction Consequentlyfew simulated population trajectories that fall below the criticalthreshold do so early reducing the mean time to extinctionAlso density dependence has a stabilizing effect on populations(Royama 1992) populations tend toward equilibrium popula-tion sizes which reduces the probability over time of popula-tions falling below quasi‐extinction threshold However timeuntil quasi‐extinction must be interpreted in conjunction withthe probability of quasi‐extinction which is very low for allscenarios with genetic introgression (Fig 15)Cost is a significant impediment to the implementation of a

genetic introgression program because captures relocations andmonitoring efforts before and after introgression are expensive(Edmands 2007 Frankham 2010b 2015 2016) Costs may beacceptable to society if the management actions result in sub-stantial fitness benefits especially if the species of concern ispopular and charismatic (Frankham 2015) We estimated the100‐year cost of introgression strategies modeled here to rangefrom $50000 to $12 million More expensive strategies gen-erally led to larger decreases in the probability of quasi‐extinc-tion but cheaper strategies also substantially improved

population viability (ie reduced quasi‐extinction probabilityTable 6) One of the cheaper strategies (5 pumas released every20 years) which cost an estimated $200000 over 100 yearsreduced the probability of quasi‐extinction by 59 and 44 forthe MPC and MVM scenarios respectively But these pre-liminary cost estimates are based on expenses incurred duringthe 1995 introgression and include costs of capture quarantinetransportation release and post‐release monitoring of femalesonly Although the cost for future genetic interventions wouldlikely be higher than those reported here the relative differencesin cost between alternative intervention strategies should remainapproximately the sameOur ability to simulate genetics and link it to the viability of

the Florida panther population is based on the empirically es-timated relationship between individual heterozygosity andsurvival Such heterozygosity and fitness correlations have beenstudied for decades (David 1998) but have only recently beenused to predict population viability (Benson et al 2016) noneto our knowledge have incorporated cost into their modelingefforts The mechanisms underlying heterozygosity and fitnesscorrelations remain unclear (David 1998 Szulkin et al 2010)and their significance as a proxy for inbreeding depression hasbeen debated (Balloux et al 2004 Miller and Coltman 2014)Nevertheless evidence continues to mount demonstratingpositive relationships between heterozygosity and survival(Coulson et al 1998ab Hansson et al 2004 Silva et al 2005Acevedo‐Whitehouse et al 2006 Jensen et al 2007 Velandoet al 2015) and between heterozygosity and fecundity (Seddonet al 2004 Charpentier et al 2005 Agudo et al 2012 Velandoet al 2015) Although the reported correlations are generallyweak their consequences can be substantial if population growthand persistence parameters are highly sensitive to the affectedvital rates (Velando et al 2015) Compared with scenarios thatignore genetics incorporating the effects of inbreeding on sur-vival increased quasi‐extinction probabilities within a 100‐yeartimeframe from 15 to 13 and from 14 to 17 for theMPC and MVM scenarios respectively These results high-light the importance of incorporating genetics when analyzingthe viability of small isolated populationsAlthough conservation biologists have recognized the im-

portance of genetics in population viability many still fail toincorporate genetic factors into PVA models (Reed et al 2002)Explicit consideration of genetics in PVAs requires long‐termgenetic and demographic data This permits the assessment ofthe functional relationship between genetic variability and de-mographic parameters Population viability analysis softwarepackages such as VORTEX (Lacy 1993 2000) offer a powerfulframework for individual‐based simulations but they often donot adequately capture a speciesrsquo life history or genetics Basedon a thorough review of the importance of genetic factors inwildlife conservation Frankham et al (2014) concluded thatgenetic factors can strongly influence the dynamics and persis-tence of small andor fragmented populations They furthernoted that ldquoMost PVA‐based risk assessments ignore or in-adequately model genetic factorsrdquo and recommended that ldquoPVAshould routinely include realistic inbreeding depressionhelliprdquo(Frankham et al 201457) Our custom‐built IBM permitted usto develop a model that adequately captured the Florida panther

van de Kerk et al bull Florida Panther Population and Genetic Viability 29

life history and we could parameterize the model with our long‐term genetic and demographic dataThe explicit consideration of the effects of inbreeding on

Florida panther population dynamics and persistence representsan important step forward Nonetheless our model remainsimperfect First we neglected the possible effects of habitat lossdetrimental anthropogenic activities climate change the prob-ability of immigration of pumas from western populationsdisease or catastrophes on demographic rates and populationviability Although immigration from puma populations outsideFlorida is possible its probability is very low given the extensivechallenges the anthropogenically altered landscape would posefor such a migrant to make it successfully to South Florida Weomitted mutations from our model because we have no in-formation on mutation rates though we expect that they are lowbased on values reported for other mammals (Kumar and Sub-ramanian 2002) Finally we only considered possible effects ofinbreeding on age‐specific survival rates because there was noevidence that heterozygosity affected female reproduction Lossof heterozygosity can negatively affect reproduction becauseinbred male panthers can suffer from cryptorchidism which canadversely affect their fecundity However given the polygynousmating system we expect the Florida panther population growthis primarily female‐limitedAlthough it is generally accepted that genetic introgression

only temporarily relieves a population from inbreeding depres-sion we know of no cases in which it has been implementedmore than once to conserve a threatened wildlife populationOur study provides an example of how demographic and mo-lecular data can be used within an IBM framework to evaluatethe efficacy of alternative genetic management strategies via theFlorida panther as a case study Our model can be adjusted forand applied to other species and offers great potential as a toolfor genetic management of small isolated wildlife populations

MANAGEMENT IMPLICATIONS

Our results and those of Hostetler et al (2013) and Johnsonet al (2010) provide persuasive evidence that the genetic in-tervention implemented in 1995 prevented the demise of theFlorida panther and restored demographic vigor to the popu-lation The Florida panther population remains small and iso-lated from other puma populations the closest puma populationis located gt1000 km away in Texas and the intervening land-scape is highly developed and fragmented with many interstate(and other high‐traffic) highways and major urban centersTheory suggests that 1 migrant per generation is sufficient tominimize the loss of genetic diversity (eg Mills and Allendorf1996) However the possibility of successful migration of apuma to South Florida followed by successful mating every ~5years is very low considering the distance between Texas andFlorida (Hedrick 1995) and the many risks and barriers thatdispersing pumas face (Beier 1995 Maehr et al 2002 Thatcheret al 2006) Consequently the pantherrsquos long‐term persistencewill likely depend on subsequent timely genetic introgressionsgiven the near‐impossibility of natural gene flow from otherpuma populations To that end our study offers considerableinsights into benefits of different genetic management strategiesand associated costs

Without genetic management the Florida panther populationfaces a substantial risk of quasi‐extinction within 100 years (10ndash46 depending on quasi‐extinction threshold and scenarios)Twenty‐three years have passed since genetic introgression wasimplemented and the population appears healthy as indicatedby measures of genetic variation increases in the populationsize and improvements in age‐specific survival rates Ourfindings suggest that releasing more pumas more frequently doesnot necessarily improve persistence probability because such astrategy would cause larger fluctuations in population size(Fig 14) which negatively affects persistence probability Thusextinction risks faced by inbred populations of large carnivorescan be substantially reduced by releasing approximately 5 im-migrants every 2 decades We suggest that such a managementstrategy be followed only in conjunction with continued long‐term monitoring of the population Our analyses demonstratethat population growth and persistence are highly sensitive tokitten and female (adult and subadult) survival Continuing tocollect data on these demographic variables along with bio-metric and genetic sampling will remain important to pantherrecovery and vigilance in identifying issues associated with in-breeding depression The collection of these monitoring datashould allow managers to detect any demographic or geneticchanges in a timely fashion and respond with genetic in-trogression or other management actions if warrantedRecovery objectives as defined by the USFWS in its current

recovery plan (USFWS 2008) delineate the need for additionalpopulations in the historical range to downlist or delist theFlorida panther If the only extant population of Floridapanthers continued to grow it could serve as a source ofpanthers to be translocated to suitable habitat to seed addi-tional populations Maintaining current information on thegenetic health of the population and precise estimates of itssize will be important in assessing whether it can withstand theremoval of a subset of animals for translocation Although the2017 documentation of a female panther with kittens north ofthe current breeding range is encouraging in terms of thepotential for population expansionmdashgiven that no female hadbeen recorded north of the Caloosahatchee River since 1972mdashthe re‐establishment of separate new populations will mostlikely require translocations by wildlife managers and a lengthysociopolitical processConservation of threatened or endangered species such as the

Florida panther is inherently challenging For panthers and otherlarge carnivores that require vast extents of natural habitat topersist habitat is typically the most limiting factor that will de-termine whether recovery efforts succeed Although geneticmanagement invariably plays a key role in the long‐term persis-tence of the panther population even a healthy population caneventually succumb to the impacts of habitat loss The humanpopulation of Florida continues to be one of the fastest‐growingin the nation the United States Census Bureau currently ranksFlorida as the third most populous Therefore habitat loss isgoing to continue to be a key issue for panthers Diligence towardpreserving panther habitat in its historical range via conservationeasements or other means and trying to keep such areas inter-connected via corridors is a goal stakeholders such as governmentagencies sportsmen non‐governmental organizations ranchers

30 Wildlife Monographs bull 203

and other private landowners must work on collaboratively toachieve

ACKNOWLEDGMENTS

We thank D Shindle D Jennings T OrsquoMeara D Land andC Belden for valuable advice throughout the project We thankS Bass M Criffield M Cunningham D Giardina D JansenA Johnson D Land M Lotz C McBride Rocky McBrideRowdy McBride Roy McBride L Oberhofer S Schulze staffsat Everglades National Park and Big Cypress National Preserveand others for assistance with fieldwork We also thank JAustin M Conroy J Hines J Laake S Mills J Nichols JHines M Sunquist J Fryxell and T Coulson for advice andassistance regarding data analysis and panther biology TheMuseum of Texas Tech University Natural Science ResearchLaboratory provided samples from pumas from west Texas forcomparative genetic analyses M Ben‐David D Land WMcCown E Hellgren and 4 anonymous reviewers providedmany helpful comments on the manuscript We thank NereaUbierna Lopez for completing the Spanish translation of theabstract We are grateful to the Department of ResearchComputing University of Florida for excellent high‐perfor-mance computing services We would like to thank US Fishand Wildlife Service Florida Fish and Wildlife ConservationCommission (FWC) US National Park Service QuantitativeSpatial Ecology Evolution and Environment (QSE3) In-tegrative Graduate Education and Research Traineeship(IGERT) program funded by the National Science Foundationand the University of Florida for financially supporting thisstudy Funding for panther research conducted by the FWC issupported predominantly via donations accrued with vehicleregistrations for ldquoProtect the Pantherrdquo license plates

LITERATURE CITEDAcevedo‐Whitehouse K T R Spraker E Lyons S R Melin F Gulland RL Delong and W Amos 2006 Contrasting effects of heterozygosity onsurvival and hookworm resistance in California sea lion pups MolecularEcology 151973ndash1982

Adams J R L M Vucetich P W Hedrick R O Peterson and J AVucetich 2011 Genomic sweep and potential genetic rescue during limitingenvironmental conditions in an isolated wolf population Proceedings of theRoyal Society B Biological Sciences 2783336ndash3344

Agresti A 2013 Categorical data analysis Third edition John Wiley amp SonsHoboken New Jersey USA

Agudo R M Carrete M Alcaide C Rico F Hiraldo and J A Donaacutezar2012 Genetic diversity at neutral and adaptive loci determines individualfitness in a long‐lived territorial bird Proceedings of the Royal Society BBiological Sciences 2793241ndash3249

Albeke S E N P Nibbelink and M Ben‐David 2015 Modeling behavior bycoastal river otter (Lontra canadensis) in response to prey availability in PrinceWilliam Sound Alaska a spatially‐explicit individual‐based approach PLOSOne 10e0126208

Alho J S K Vaumllimaumlki and J Merilauml 2010 Rhh an R extension for estimatingmultilocus heterozygosity and heterozygosity‐heterozygosity correlationMolecular Ecology Resources 10720ndash722

Alvarez K 1993 Twilight of the panther Myakka River Publishing SarasotaFlorida USA

Aparicio J M J Ortego and P J Cordero 2006 What should we weigh toestimate heterozygosity alleles or loci Molecular Ecology 154659ndash4665

Bakker V J D F Doak G W Roemer D K Garcelon T J Coonan S AMorrison C Lynch K Ralls and R Shaw 2009 Incorporating ecologicaldrivers and uncertainty into a demographic population viability analysis for theisland fox Ecological Monographs 7977ndash108

Balloux F W Amos and T Coulson 2004 Does heterozygosity estimateinbreeding in real populations Molecular Ecology 133021ndash3031

Barone M A M E Roelke J Howard J L Brown A E Anderson and DE Wildt 1994 Reproductive characteristics of male Florida panthersmdashcomparative studies from Florida Texas Colorado Latin‐America andNorth‐American Zoos Journal of Mammalogy 75150ndash162

Bateson Z W P O Dunn S D Hull A E Henschen J A Johnson and LA Whittingham 2014 Genetic restoration of a threatened population ofgreater prairie‐chickens Biological Conservation 17412ndash19

Bauer H G Chapron K Nowell P Henschel P Funston L T B HunterD W Macdonald and C Packer 2015 Lion (Panthera leo) populations aredeclining rapidly across Africa except in intensively managed areas Pro-ceedings of National Academy of Sciences of the United States of America11214894ndash14899

Beier P 1995 Dispersal of juvenile cougars in fragmented habitat Journal ofWildlife Management 59228ndash237

Benson J F J A Hostetler D P Onorato W E Johnson M E Roelke SJ OrsquoBrien D Jansen and M K Oli 2011 Intentional genetic introgressioninfluences survival of adults and subadults in a small inbred felid populationJournal of Animal Ecology 80958ndash967

Benson J F P J Mahoney J A Sikich L E K Serieys J P Pollinger H BErnest and S P D Riley 2016 Interactions between demography geneticsand landscape connectivity increase extinction probability for a small popu-lation of large carnivores in a major metropolitan area Proceedings of theRoyal Society B Biological Sciences 28320160957

Beverly F 1874 The Florida panther Forest and Stream 3290Boyce M S C V Haridas C T Lee and the NCEAS Stochastic Demo-graphy Working Group 2006 Demography in an increasingly variable worldTrends in Ecology amp Evolution 21141ndash148

Brook B W J J OrsquoGrady A P Chapman M A Burgman H R Akcakayaand R Frankham 2000 Predictive accuracy of population viability analysis inconservation biology Nature 404385ndash387

Brys R and H Jacquemyn 2016 Severe outbreeding and inbreeding de-pression maintain mating system differentiation in Epipactis (Orchidaceae)Journal of Evolutionary Biology 29352ndash359

Burke J M and M L Arnold 2001 Genetics and the fitness of hybridsAnnual Review of Genetics 3531ndash52

Burnham K P 1993 A theory for combined analysis of ring recovery andrecapture data Pages 199ndash213 in J‐D Lebreton and P M North editorsMarked individuals in the study of bird population Springer‐Verlag NewYork New York USA

Burnham K P and D R Anderson 2002 Model selection and multimodelinference a practical information‐theoretic approach Second edition SpringerVerlag New York New York USA

Caswell H 2001 Matrix population models construction analysis and in-terpretation Sinauer Associates Sunderland Massachusetts USA

Charpentier M J M Setchell F Prugnolle L A Knapp E J Wickings PPeignot and M Hossaert‐McKey 2005 Genetic diversity and reproductivesuccess in mandrills (Mandrillus sphinx) Proceedings of the NationalAcademy of Sciences of the United States of America 10216723ndash16728

Cooch E and G C White editors 2018 Program MARK a gentle in-troduction lthttpwwwphidotorgsoftwaremarkdocsbookgt Accessed 7Apr 2016

Cooley H S R B Wielgus G M Koehler H S Robinson and B TMaletzke 2009 Does hunting regulate cougar populations A test of thecompensatory mortality hypothesis Ecology 902913ndash2921

Coulson T E A Catchpole S D Albon B J Morgan J M Pemberton TH Clutton‐Brock M J Crawley and B T Grenfell 2001 Age sex densitywinter weather and population crashes in Soay sheep Science 2921528ndash1531

Coulson T J Pemberton S Albon M Beaumont T Marshall J Slate FGuinness and T Clutton‐Brock 1998a Microsatellites reveal heterosis in reddeer Proceedings of the Royal Society B Biological Sciences 265489ndash495

Coulson T N S D Albon J M Pemberton J Slate F E Guinness and TH Clutton‐Brock 1998b Genotype by environment interactions in wintersurvival in red deer Journal of Animal Ecology 67434ndash445

Cox D 1972 Regression models and life‐tables Journal of the Royal StatisticalSociety Series B Statistical Methodology 34187ndash220

Creel S 2006 Recovery of the Florida panthermdashgenetic rescue demographic rescueor both Response to Pimm et al (2006) Animal Conservation 9125ndash126

Crnokrak P and D A Roff 1999 Inbreeding depression in the wild Heredity83260ndash270

Crow J 1948 Alternative hypotheses of hybrid vigor Genetics 33477ndash487

van de Kerk et al bull Florida Panther Population and Genetic Viability 31

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 10: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

estimate annual probability of reproduction of subadult prime‐adult and old‐adult females using complementary logndashlogregression (Agresti 2013) following methods described indetail by Hostetler et al (2012) Briefly we modeled theprobability of a female panther giving birth (b) as a function ofcovariates as

γ= minus [minus ( )]b z1 exp exp i i

where zi is a row vector of covariates (see below for covariateinformation) for female panther i and γ is a column vector ofmodel coefficients (Appendix B Table B2) To account fordifferences in observation time we included log(m) as an offsetin the model where m is the number of months a femalepanther was radio‐tracked (Hostetler et al 2012)

Reproductive outputmdashTo estimate the number of kittensproduced by reproductive females we used cumulative logitregression (Min and Agresti 2005 Agresti 2013) Weconsidered a model that includes J categories for the numberof offspring with j denoting the number of offspring ( j= 1 2hellip J and J= 6) provided that a female reproduces Weconsidered J= 6 which is greater than the largest litter sizeobserved in our study (4 kittens) because females can produce 2or more litters per year if litters fail We modeled the probabilitythat a female i produces a litter of size at most j (Yi) as

le⎧

⎨⎩

δ( ) = == hellip minus

=

( )+ θ βminus +Y jj J

j JPr

1 2 1

1i ij e

1

1 xj i

where θj is the intercept for the annual number of kittens pro-duced by a female i being at most j xi is a row vector of cov-ariates for female panther i (see below) and β is a column vectorof model coefficients The probability that Yi will be exactly j isgiven by

⎧⎨⎩

πδ

δ δ( = ) = =

=

minus = hellip( minus )Y j

jj J

Pr 1

2 i ij

i

ij i j

1

1

Finally the expected annual number of kittens produced byfemale i (νi) is given by

sumν π==

j i

j

J

ij

1

Additional details regarding the estimation and modelingof the annual number of kittens produced can be found inHostetler et al (2012)

CovariatesmdashIn addition to sex and age we tested for the effect ofabundance genetic ancestry and individual heterozygosity on theaforementioned demographic parameters (model coefficients forkitten and adult survival are given by η and ι respectively AppendixB Table B2) We used a minimum population count (MPC) basedon radio‐tracking data and field evidence of subadult and adult (ieindependent age) panthers as an index of abundance (McBride et al2008) For these analyses we did not use the population sizeestimates reported by McClintock et al (2015) because unlike theminimum counts they did not cover the entire study period

(McBride et al 2008 R T McBride and C McBride RancherrsquosSupply Incorporated unpublished report Fig 2)To test if demographic parameters differed depending on an-

cestry we considered 3 ancestry models dividing the panthers into1) F1 admixed and all other panthers 2) canonical and admixed(including F1 admixed) panthers and 3) F1 admixed canonicaland other admixed panthers We quantified genetic diversity usingindividual heterozygosity as described previously We tested for theeffect of genetic covariates using a subset of kittens that were ge-netically sampled We calculated model‐averaged estimates ofmodel parameters on the scale in which they were estimated(Appendix B available online in Supporting Information)

Cause‐specific mortalitymdashWe performed cause‐specificmortality analysis only on dead radio‐collared panthers becausedetection rates of uncollared panther mortalities are highlycorrelated with the cause of death (eg vehicle collision)Following Benson et al (2011) we attributed each mortality to1 of 4 causes 1) vehicle collision 2) intraspecific aggression 3)other causes (including known causes such as diseases injuriesand infections unrelated to the first 2 causes) and 4) unknown(ie mortalities for which we could not assign a cause) Wethen used the non‐parametric cumulative incidence functionestimator a generalization of the staggered‐entry KaplanndashMeiermethod of survival estimation to estimate cause‐specificmortality rates for male and female panthers in different ageclasses (Pollock et al 1989 Heisey and Patterson 2006 Bensonet al 2011) We compared cause‐specific mortality rates betweensexes age classes and ancestries

Statistical inferencemdashWe used an information‐theoreticapproach (Akaikersquos information criterion [AIC]) for modelselection and statistical inference (Burnham and Anderson 2002)For the analysis of kitten survival we adjusted the AIC foroverdispersion and small sample size (QAICc details in Hostetleret al 2010) We performed all statistical analyses in Program R (RCore Team 2016) We used Burnhamrsquos live recapturendashdead

0

100

200

300

1985 1990 1995 2000 2005 2010Year

Indi

vidu

als Method

MPC

MVM

Figure 2 The Florida panther minimum population count (MPC) and populationsize estimated using motor vehicle collision mortalities (MVM) during our studyperiod Minimum population counts are from McBride et al (2008) and R TMcBride and C McBride (Rancherrsquos Supply Incorporated unpublished report)Estimates based on MVM are from McClintock et al (2015) The solid vertical lineindicates the year of genetic introgression

12 Wildlife Monographs bull 203

recovery model (Burnham 1993) to estimate and model kittensurvival using the RMark package version 2113 (Laake 2013) asan interface for Program MARK (White and Burnham 1999)We implemented the Cox proportional hazard model for esti-

mation and modeling of survival of subadults and adults using the Rpackage SURVIVAL (Therneau and Grambsch 2000) We per-formed cause‐specific mortality analyses using an R version of S‐PLUS code provided by Heisey and Patterson (2006)To account for model selection uncertainty we calculated

model‐averaged parameter estimates across all models using theR package AICcmodavg (Mazerolle 2017) using AIC weights asmodel weights (Burnham and Anderson 2002) We used modelswithout the categorical ancestry variables for model averagingand estimated parameters based on the average value for thecontinuous covariates (individual heterozygosity and abundanceindex) Because only a subset of the kittens was geneticallysampled we estimated 2 kitten survival rates First we estimatedkitten survival based on the data set that included all kittens(overall kitten survival) Second we estimated kitten survivalusing a subset of the data that only included kittens that weregenetically sampled (survival of genetically sampled kittens)

Matrix Population ModelsWe investigated Florida panther population dynamics andpersistence using 2 complementary modeling frameworks ma-trix population models (Caswell 2001) and IBMs (Grimm andRailsback 2005 McLane et al 2011 Railsback and Grimm2011 Albeke et al 2015) Matrix population models offer aflexible and powerful framework for investigating the dynamicsand persistence of age‐ or stage‐structured populations (egCaswell 2001 Robinson et al 2008 Kohira et al 2009 Hunteret al 2010 Hostetler et al 2012) With a fully developed theorybehind them these models also serve as a check for resultsobtained from simulation‐based models such as IBMs We usedan age‐structured‐matrix population model for deterministic andstochastic demographic analyses and for preliminary investiga-tions of Florida panther population viability (Caswell 2001Morris and Doak 2002)For deterministic and stochastic demographic analyses we

parameterized a female‐only 19 times 19 age‐specific populationprojection matrix of the form

⎢⎢⎢⎢⎢

⋯ ⋯

⋯ ⋯

⋱ ⋱ ⋮

⋮ ⋱ ⋱ ⋱ ⋮

⎥⎥⎥⎥⎥

( )=t

F F F

P

P

P

A0 0

0

0 0 0

t t t

t

t

t

1 2 19

1

2

18

where Fi and Pi are the age‐specific fertility and survival ratesrespectively and ( )tA is a time‐varying (or stochastic) popula-tion projection matrix We assumed that the longevity and ageof last reproduction of female Florida panthers was 185 yearswe based this assumption on the observation that the oldestfemale in our study was 186 years old Florida panthers re-produce year‐round thus we used birth‐flow methods to esti-mate Fi and Pi values following Caswell (2001) and Hostetleret al (2013) We estimated Pi as

= +P S S i i i 1

where Si is the age‐specific survival probability We estimated Fi as

⎛⎝

⎞⎠

=+ +F S

m Pm

2i k

i i i 1

where Sk is the kitten survival probability and mi is the age‐specific fecundity rate which was estimated as

ν=m b f i i i k

where bi is the breeding probability for a female in age class iduring time step t νi is the annual number of kittens producedby a breeding female in age class i and fk is the proportionfemale kittens at birth (assumed to be 05)We estimated the finite deterministic (λ) and stochastic annual

population growth rate (λs) and stochastic sensitivities and elasti-cities following methods described in detail by Caswell (2001) Toestimate λs we assumed identically and independently distributedenvironments (Caswell 2001 Morris and Doak 2002 Haridas andTuljapurkar 2005) and used a parametric bootstrapping approachto obtain a sequence of demographic parameters for 50000 ma-trices We then estimated log(λs) from this sequence of matrices(Tuljapurkar 1990 Caswell 2001) and exponentiated it to obtain λsWe calculated the sensitivity and elasticity of λs to changes in lower‐level vital demographic rates as per Caswell (2001) and Haridas andTuljapurkar (2005)For population projection and population viability analysis (PVA)

simulations we expanded the population projection matrix into a34 times 34 2‐sex age‐structured matrix to include female and maleFlorida panthers (Caswell 2001 Hostetler et al 2013) Malescontribute to the population only through survival in this modelingframework We assumed maximum longevity for males to be 145years of age because the oldest male in our study was 144 years oldThe population projection matrix was of the following form

⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢

⋯ ⋯ ⋯ ⋯

⋯ ⋯ ⋯ ⋯

⋱ ⋱ ⋮ ⋮ ⋱ ⋱ ⋮

⋮ ⋱ ⋱ ⋱ ⋮ ⋮ ⋱ ⋱ ⋮

⋯ ⋯ ⋯

⋯ ⋯ ⋯ ⋯

⋯ ⋯ ⋱ ⋱ ⋮

⋮ ⋮ ⋱ ⋱ ⋮ ⋱ ⋱ ⋮

⋯ ⋯ ⋯

⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥

( )=

( ) ( ) ( )

( )

( )

( )

( ) ( ) ( )

( )

( )

t

F N F N F N

P N

P N

P N

M N M N M N

Q N

Q N

A n

0 0

0 0 0 0

0

0 0 0 0 0

0 0

0 0 0

00 0 0 0 0

t t t

t

t

t

t t t

t

t

1 2 19

1

2

18

1 2 19

1

14

where Pi and Qi are age‐specific survival rates of female and maleFlorida panthers respectively and Fi and Mi are the rates atwhich female and male kittens are produced by females of ageclass i respectively Here the population projection matrix isboth time‐varying and density‐dependent (Caswell 2001)We incorporated the influence of parameter uncertainty en-

vironmental stochasticity and density dependence on Floridapanther population dynamics and persistence following the ap-proach outlined by Hostetler et al (2013) we used the sameapproach in the IBM analysis which is described below UnlikeIBMs matrix models do not intrinsically include demographicstochasticity so we explicitly included the influence of

van de Kerk et al bull Florida Panther Population and Genetic Viability 13

demographic stochasticity by simulating the fates of individualsas described in Caswell (2001)Population projections using matrix models necessitate a

vector of initial sex‐ and age‐specific abundance Because suchdata are currently unavailable for the Florida panther we esti-mated it based on the stable sex and age distribution and MPCin 2013 Using point estimates of the demographic parameterswe estimated the stable sex and age distribution for the 2‐sexdeterministic and density‐independent population projectionmatrix We then multiplied the stable sex and age distributionby the MPC in 2013 to get the starting population vector n(0)We projected the population over time as n(t+ 1)=A(tn)n(t)where A(tn) indicates the time‐specific density‐dependentpopulation projection matrix (Caswell 2001)We had 2 indices of abundance available the MPC (McBride

et al 2008) and population size estimated using motor vehiclecollision mortalities (MVM McClintock et al 2015) These 2indices are substantially different because MPC does not ac-count for imperfect detection whereas MVM accounts forimperfect detection and provides an estimate of population size(Fig 2) Therefore we examined 2 scenarios one using densitydependence based on MPC and the other using density de-pendence based on MVM We ran 1000 bootstraps of para-meter values and 1000 simulations per bootstrap for each sce-nario We projected population trajectories over 200 years andthen estimated population growth rates probabilities ofquasi‐extinction and time until quasi‐extinction and comparedthese with the results obtained from the IBM (see below) Weran scenarios with a quasi‐extinction threshold of 10 and 30panthers We used the mean of probabilities of quasi‐extinction(PQE) across simulations as the point estimate for PQE and the5th and 95th percentiles across simulations as the 90 con-fidence interval Because the confidence intervals werequantile‐based but the point estimates were not it is possiblewith skewed distributions for the point estimates to be outsidethe confidence interval We implemented the matrix

population models in the R computing environment (R CoreTeam 2016)

IBMs and PVABecause of well‐developed theory ease of implementation andthe modeling flexibility that they offer matrix populationmodels have become the model of choice for investigating thedynamics and persistence of age‐ and stage‐structured popula-tions (Caswell 2001) However it is difficult to incorporateattributes of individuals such as genetics and behavior usingmatrix models Quantifying the effects of genetic erosion on theFlorida panther population dynamics and persistence andevaluating benefits and costs of alternative genetic managementscenarios were important goals of our study Because IBMs relyon a bottom‐up approach that begins by explicitly consideringindividuals (McLane et al 2011 Railsback and Grimm 2011Albeke et al 2015) it is possible to represent genetic attributesof each individual and to track genetic variation over time Weused IBMs as the primary modeling framework in this studybecause they allowed for explicit consideration of genetics bothat individual and population levels and population viabilityassessment under alternative genetic management scenariosWe developed an IBM (Grimm 1999 Grimm and Railsback

2005 McLane et al 2011 Railsback and Grimm 2011 Albekeet al 2015) tailored to the life history of the Florida panther(Fig 3) We simulated every individual from birth until deathbased on a set of rules describing fates (eg birth and death) andattributes (eg genetics) of individuals at each annual time stepAs in other IBMs population‐level processes (eg populationsize and genetic variation) emerged from fates of and interac-tions among individuals (Grimm 1999 Railsback and Grimm2011) We developed an overview design concepts and details(ODD) protocol (Grimm et al 2006 2010) to describe ourIBM (Appendix C available online in Supporting Information)and provide pseudocodes for individual‐based simulations(Appendix D available online in Supporting Information)

For each panther at time = t

Kitten (lt1 yr)

Reproduce

Survive

Kitten(s) born

Kitten

Sex and age class

Subadult male

Prime adult male

Old adult male

Subadult female

Prime adult female

Old adult female

Age one year

No Male

Female

Yes

No

Yes

t=t+1

Yes

lt35 yrs

35-10 yrs

gt10 yrs

lt25 yrs

25-10 yrs

gt10 yrs

Figure 3 Schematic representation of Florida panther life history We tailored the individual‐based population model to represent the life‐history patterndepicted here

14 Wildlife Monographs bull 203

Dynamics and persistence of populations are influenced bymany factors such as demographic and environmental sto-chasticity and density dependence these effects and parametricuncertainty must be considered in population models whenpossible (Caswell 2001 Boyce et al 2006 Bakker et al 2009Hostetler et al 2013) Because by definition IBMs simulate thelife history of individuals in a population they inherently in-corporate the effect of demographic stochasticity Our analysesrevealed that survival of kittens subadults and adults and theprobability of reproduction varied over time so we incorporatedenvironmental stochasticity in the model for these variablesfollowing Hostetler et al (2013) Likewise we found evidenceof density‐dependent effects on kitten survival Because theMPC might be an underestimate of population size (McBrideet al 2008) we performed simulations using both the MPC andMVM abundance indicesTo account for model selection and parametric uncertainty we

used a Monte Carlo simulation approach For each bootstrapsample we selected a model based on the AIC (or QAICc)weights We then sampled the intercept and slope for kittensurvival (on the logit scale) subadult and adult survival (on thelogit scale with log‐hazard effect sizes) and probability of re-production (on the complementary logndashlog scale Appendix B)from multivariate normal distributions with mean vectors equalto the estimated parameters and variance‐covariance matricesequal to the sampling variance‐covariance matrices We thentransformed the results to nominal scale and converted the es-timates to age‐specific survival and reproduction parameters asdescribed in Hostetler et al (2013)We ran 1000 simulations for 1000 parametric bootstraps for

200 years for the MPC and the MVM scenario for both thestochastic 2‐sex matrix‐population model and the IBM Wetracked the population size at each time step and estimatedquasi‐extinction probabilities and times based on the populationtrajectory We considered the population to be quasi‐extinct ifindividuals of only 1 sex remained or if the population size fellbelow critical thresholds set at 10 and 30 individuals We alsoestimated expected time to quasi‐extinction for both thresholdsdefined as the mean and median time until a population be-comes quasi‐extinct conditioned on quasi‐extinction We im-plemented the IBM using the RNetLogo package version 10ndash2(Thiele et al 2012 Thiele 2014) as an interface for the IBMplatform NetLogo (Wilensky 1999)

Sensitivity AnalysisAn important component of demographic analysis is sensitivityanalysis which is the quantification of the sensitivity of amodelrsquos outcome to changes in the input parameters (Caswell2001 Bakker et al 2009 Thiele et al 2014) Using an IBM weperformed global sensitivity analyses for 2 (ie MPC andMVM) density‐dependent scenarios to assess how sensitivequasi‐extinction times and probabilities were to changes (anduncertainties) in the estimated demographic rates We usedLatin hypercube sampling to sample parameters from the entireparameter space (Marino et al 2008 Thiele et al 2014) Wesampled intercept and slope for kitten survival (on logit scale)subadult and adult survival (on logit scale with log‐hazard effectsizes) and probability of reproduction (on complementary

logndashlog scale) from normal distributions We sampled theprobability distribution for the number of kittens producedannually (on the real scale) from a Dirichlet distribution themultivariate generalization of the beta distribution (Kotz et al2000) We sampled 1000 different parameter sets and ran 1000simulations for each scenario We calculated the probability ofquasi‐extinction within 200 years for each density‐dependencescenario (MPC or MVM) using a quasi‐extinction threshold of10 individuals and then calculated the partial rank correlationcoefficient between each of those and each parameter trans-formed to the real scale (Bakker et al 2009 Hostetler et al2013) Partial rank correlation coefficients quantify the sensi-tivity of the probability of quasi‐extinction to input variables(ie survival and reproductive parameters) with higher absolutevalues indicating stronger influence of that variable on quasi‐extinction probabilities

Genetic ManagementOne of our goals was to estimate the benefits (improvements ingenetic variation demographic parameters and populationgrowth and persistence parameters) and costs (financial costs ofcapture transportation quarantine release and monitoring ofintroduced panthers) of genetic management To achieve thisobjective we extended the IBM to include a genetic componentTo simulate individual‐ and population‐level heterozygosity overtime we assigned each panther alleles for 16 microsatellite locibased on the empirical allele frequency in the population (esti-mated from genetic samples collected during 2008ndash2015) whenwe initialized the model (Pierson et al 2015) At each time stepwe simulated Mendelian inheritance for kittens produced suchthat each kitten randomly inherited alleles from its mother andfrom a male panther randomly chosen to have putatively siredthe litter We did not explicitly force inbreeding to occur butthe probability of inbreeding naturally increased as the popula-tion size decreasedIntrogression strategiesmdashWe modeled genetic introgression as a

result of the release of 2‐year‐old female pumas from Texas intothe simulated Florida panther population For the geneticintrogression implemented in 1995 Texas females between 15and 25 years of age were released because females at that agehave lower mortality rates (Benson et al 2011) higherreproductive potential and also because it is much easier todetect with certainty the successful reproduction by females thanby males The ability to more closely monitor reproduction andthe birth of kittens is an important consideration in reducing theprobability of a genomic sweep of Texas alleles into the Floridapanther population Wildlife managers removed the last 2surviving Texas females from the wild population in Florida in2003 to reduce the possibility of genomic sweep To simulatethis strategy in the model we removed any Texas females thatwere still alive 5 years after each introgression event Weassigned Texas females alleles based on the allele frequency ofthe Texas population (based on genotypes from 48 samplescollected in west Texas that was inclusive of the pumas releasedin South Florida in 1995) when they entered the Florida pantherpopulation We could not estimate demographic rates separatelyfor the released Texas females because of small sample size (only8 females were released in 1995) therefore we assumed that

van de Kerk et al bull Florida Panther Population and Genetic Viability 15

survival and reproductive rates and the effects of covariates onthese rates for the released Texas females were identical to thoseof female Florida panthersThe impact of genetic introgression depends on the frequency

of introgression events and the number of individuals introducedat each attempt Thus we defined introgression strategies basedon the number of Texas females to be released (0 5 10 or 15)and the interval between genetic introgressions (10 20 40 and80 yr) for a total of 13 strategies We simulated each manage-ment strategy with density dependence estimated using theMPC and MVM abundance indices

We sampled 1000 parameter sets and for each of these setswe ran all introgression strategies 1000 times for 200 years Weapplied the same parametric uncertainty and environmentalstochasticity across all management scenarios to allow for bettercomparison of introgression strategies At each time step werecorded the projected population size and average individualheterozygosity We calculated the heterozygosity for each in-dividual at birth based on the alleles it inherited This individualheterozygosity then informed the survival rate of that individualbased on the empirically estimated relationship between in-dividual heterozygosity and age‐specific survival rates Hetero-zygosity did not change throughout an individualrsquos lifetime butits effect on survival changed as individuals aged because theeffect of individual heterozygosity on survival rate differedamong age classes Survival rates of genetically sampled kittenswere biased high because some kittens were sampled later in lifewhen they had already survived for some time To correct forthis bias we adjusted kitten survival rates predicted by theheterozygosity model proportionally From the population tra-jectories we estimated the probability of quasi‐extinction (de-fined as the probability that the population will fall below 10 or30 individuals or that individuals of only 1 sex will remain)

Cost‐benefit analysismdashExperts estimated financial costs ofintrogression based on their experience with the geneticintrogression executed in 1995 and we adjusted estimates tocurrent costs to account for inflation (R T McBride RancherrsquosSupply Incorporated M Cunningham and D P OnoratoFlorida Fish and Wildlife Conservation Commission personalcommunication) Costs included capturing and transportingfemale pumas from the Texas source population caring for thepumas while they were in quarantine and post‐introgressionmonitoring To compare the financial costs of the differentscenarios we also calculated the average costs incurred over

Table 2 Sample size (n) number of alleles (Na) allelic richness (AR) observedheterozygosity (Ho) and expected heterozygosity (He) at 16 microsatellite loci inradio‐collared Florida panthers sampled from 1981 to 2013 in South FloridaUSA We calculated these values from a subset (n= 137) of the samples used inthe analyses and they include only adult and subadult panthers We excludedkittens because they can result in an overrepresentation of certain alleles

Locus n Na AR Ho He

FCA090 129 5 50 0333 0401FCA133 136 3 30 0618 0608FCA243 136 5 50 0419 0487F124 137 7 69 0708 0729F37 129 4 40 0395 0397FCA075 131 5 50 0534 0598FCA559 133 5 50 0496 0503FCA057 134 6 59 0679 0624FCA081 134 7 69 0209 0206FCA566 130 6 60 0462 0475F42 133 10 100 0737 0766FCA043 136 5 49 0596 0588FCA161 132 6 60 0500 0515FCA293 132 4 40 0614 0624FCA369 131 6 60 0573 0567FCA668 133 7 70 0406 0462Mean 1329 57 57 0517 0535

Figure 4 Proportional membership (q) of radio‐collared Florida panthers (n= 218) and pumas from Texas USA (n= 7) in 2 clusters identified by STRUCTUREEach individual animal is represented by a separate vertical bar Yellow indicates canonical panther ancestry and blue represents admixed ancestry The admixedancestry of the Texas females and F1 generation panthers that were radiocollared are highlighted red and green respectively to demarcate those groups The pre‐introgression period is inclusive of panthers radiocollared from 10 February 1981 to 6 March 1996 The post‐introgression era inclusive of the F1 panthers includesanimals radiocollared from 4 March 1997 to 18 February 2015 in South Florida USA

16 Wildlife Monographs bull 203

100 years assuming that the population would be managedaccording to a particular strategy Our goal with theseexploratory analyses was to evaluate the relative costs andbenefits of alternative management scenarios

RESULTS

Genetic VariationWe assessed genetic variation at 16 microsatellite loci for 137DNA samples collected from adult and subadult panthers(1981ndash2013) that were captured and subsequently radiocollaredThe number of alleles at these loci ranged from 3ndash10 and allelicrichness averaged 57 (Table 2) Average expected hetero-zygosity for this sample was 0535 and average observed het-erozygosity was 0517 (Table 2)We determined individual heterozygosity (1 minusHL) and an-

cestry for the complete sample of panther genotypes (n= 503)collected from 1981 to 2015 for use as covariates in subsequentanalyses Average individual heterozygosity was 0559 andranged from 0ndash0980 Our STRUCTURE analysis providedstrong support for K= 2 clusters based on the probability of thedata (log Pr(D)|K) and ΔK in STRUCTURE HARVESTERBased on this result we categorized the ancestry of each pantheras either canonical (n= 123) or admixed (n= 380) using valuesof proportional membership (q Fig 4) The average individualheterozygosity of canonical panthers was 0386plusmn 0012 (SE) foradmixed panthers it was 0615plusmn 0007

Survival and Cause‐Specific Mortality RatesOf the 395 kittens that were PIT‐tagged and released 55 werelater encountered alive 15 were recovered dead and 85 werepart of failed litters (Table 1) We radio‐tracked 209 subadult oradult panthers for a total of 244195 panther‐days and docu-mented 126 mortalities (Table 1)Survival depended strongly on age and kittens had the lowest

overall rate (032plusmn 009 Fig 5A Table 3) The model‐averagedkitten survival was 047plusmn 009 when we included only geneti-cally sampled individuals in the analysis (Table 3) as statedabove this estimate is biased high because many kittenswere genetically sampled when they were recaptured alive orrecovered dead at older ages We found no evidence for sex‐specific differences in kitten survival but females survived betterthan males in all older age classes For females subadults hadthe highest survival rates followed by prime adults and oldadults For males prime adults had the highest survival ratefollowed by subadults and old adults (Fig 5A and Table 3)For panthers of all ages there was substantial evidence that

ancestry affected survival with F1 admixed panthers of all ageshaving the highest rates and canonical individuals having thelowest (Fig 5B) The combined AIC weights of models thatincluded an ancestry covariate were 0880 for kittens and 0992for older panthers The survival rates of subadult prime‐adultand old‐adult panthers in the period immediately after the in-trogression (1995ndash2004) were similar to those in more recentyears (2005ndash2013 Fig 5C)After genetic introgression individual panther heterozygosity

increased (Fig 6) and varied among ancestry categories individualheterozygosity was the highest for F1 panthers (078plusmn 001)

A

B

C

Figure 5 Overall age‐ and sex‐specific survival rates (plusmnSE A) age‐ and sex‐specific survival rates (plusmnSE) for Florida panthers of canonical F1 admixed andother admixed ancestry (B) and age‐ and sex‐specific survival estimates (plusmnSE)for subadult and adult Florida panthers in the period immediately post‐introgression (1995ndash2004) and more recently (2005ndash2013 C) in SouthFlorida USA

Table 3 Model‐averaged estimates (plusmnSE) of demographic parameters for theFlorida panther obtained using data collected during 1981ndash2013 in SouthFlorida USA

Parameter Sex Age class Estimate

Survival Both Kitten 032plusmn 009a

Female Subadult 097plusmn 002Prime adult 086+ 003Old adult 078plusmn 009

Male Subadult 066plusmn 006Prime adult 077plusmn 005Old adult 065plusmn 010

Probability of reproduction Female Subadult 035plusmn 008Prime adult 050plusmn 005Old adult 025plusmn 006

Kittens produced annually Female Subadult 280plusmn 005Prime adult 267plusmn 001Old adult 228plusmn 013

a Overall kitten survival (based on all kittens in our data set including thosethat were not genetically sampled) Estimate of survival of geneticallysampled kittens was 047plusmn 009 this estimate is biased high because manykittens were genetically sampled when they were recaptured alive or re-covered dead at older ages

van de Kerk et al bull Florida Panther Population and Genetic Viability 17

followed by those for non‐F1 admixed (062plusmn 001) and canonical(039plusmn 001) panthers Individual heterozygosity positively affectedsurvival rates of kittens (slope parameter η= 1455 95CI= 0505ndash2405) Individual heterozygosity also positively af-fected survival of subadult and adult panthers but the evidence forthis effect was weaker because the confidence interval for the ha-zard ratio overlapped 10 (hazard ratio exp(ɩ)= 0311 95CI= 0092ndash1054 Table 4 and Fig 7)The MPC increased after genetic introgression (Fig 2) This

abundance index was also included in top‐ranking modelsindicating that population density affected survival (Table 4)Abundance reduced the survival of kittens (slope parameterη= minus0035 95 CI= minus0049 to minus0021) evidence was weak forthe effect of abundance on survival of subadult and adult panthers(hazard ratio exp(ɩ)= 1002 95 CI= 0995ndash1010 Fig 7) Re-gression coefficients associated with the effect of abundance andindividual heterozygosity on demographic parameters are presentedin Table B1 (available online in Supporting Information)The most frequently observed causes of death for radio‐

collared panthers were vehicle collision and intraspecific ag-gression Cause‐specific mortality analyses revealed thatmortality rates among possible causes were generally similar forthe 2 sexes but that males were more likely to die from in-traspecific aggression than were females (Fig 8A) Male mor-tality from intraspecific aggression was higher for subadult andold adult panthers than for prime adults (Fig 8B) For bothmales and females canonical panthers were more likely to diefrom intraspecific aggression than were admixed panthers(Fig 8C)

Reproductive ParametersOf 101 radio‐collared females 67 reproduced at least once duringour monitoring we used these individuals to assess the annualprobability of reproduction The top‐ranked model for the annual

probability of reproduction included age effect only suggesting thatadding ancestry or abundance did not improve model parsimony(Table 4) Model‐averaged annual probabilities of reproductiondiffered between female subadults (035plusmn 008) prime adults(050plusmn 005) and older adults (025plusmn 006)The most parsimonious model for the number of kittens

produced annually by females that produced at least 1 litter (ieconditional on reproduction) included effects of age ancestryand abundance a model that included only the effect of age wasalmost equally supported (Table 4) The model‐averagednumber of kittens produced annually conditional on re-production was 280plusmn 075 for subadults 267plusmn 043 for primeadults and 228plusmn 083 for old adults Confidence intervals forthe slope parameter (β) overlapped 0 for both abundance(β= 0010 95 CI= minus0001 to 0021) and ancestry (β= minus073795 CI= minus1637 to 0162)

Population Dynamics and PersistenceDeterministic and stochastic demographymdashThe deterministic (λ)

and stochastic (λs) annual population growth rate calculated usingthe matrix population model were 106 (5th and 95th percentiles099ndash114) and 104 (072ndash141) respectively The generation timewas 47 years A female starting in age class one (kitten) wasexpected to live another 58 years and produce 10 kittens onaverage during her lifetime Overall λs was proportionately(elasticity) most sensitive to changes in female prime adultsurvival on the absolute scale (sensitivity) however it was mostsensitive to changes in kitten survival (Fig 9) Populationtrajectories probabilities of quasi‐extinction and times untilquasi‐extinction estimated using the matrix population modeland IBM for comparable scenarios were nearly identical for acritical quasi‐extinction threshold of 10 panthers (Figs 10 and 11)and for a critical quasi‐extinction threshold of 30 panthers(Appendix E available online in Supporting Information)Minimum population count scenariomdashUnder the MPC scenario

(ie density dependence estimated using the minimumpopulation count data) with a critical threshold of 10panthers the population size reached 99 (72ndash104) and 100(77ndash105) at 200 years (Fig 10) as predicted by the IBM and thematrix model respectively The cumulative quasi‐extinctionprobabilities within 100 years based on the MPC scenario were15 (IBM 0ndash48) and 30 (matrix model 0ndash76) Theseprobabilities increased to 25 (IBM 0ndash114) and 38 (matrixmodel 0ndash154) by year 200 (Fig 10) The mean times untilquasi‐extinction were 15 years (IBM 0ndash67) and 15 years (matrixmodel 0ndash72) conditioned on going quasi‐extinct within 100years Expected times until quasi‐extinction conditioned onquasi‐extinction within 200 years were higher 34 years (IBM0ndash137) and 32 years (matrix model 0ndash134 Fig 10)Motor vehicle mortality scenariomdashUnder the MVM scenario

(ie density dependence estimated using estimates of pantherpopulation size from McClintock et al 2015) with a criticalthreshold of 10 panthers the projected population reached 188(142ndash217) and 190 (143ndash219) at 200 years as predicted by theIBM and the matrix model respectively (Fig 11) Thecumulative quasi‐extinction probabilities within 100 years were14 (IBM 0ndash08) and 13 (matrix model 0ndash06) Theseprobabilities increased to 20 (IBM 0ndash17) and 19 (matrix

000

025

050

075

100

1995 2000 2005 2010Year of birth

Indi

vidu

al h

eter

ozyg

osity

Figure 6 Temporal changes in the individual heterozygosity of Floridapanthers born during the period 1995ndash2013 in South Florida USA Pointsare measurements solid line is linear regression shaded area is 95 confidenceinterval of slope Slope= 0006plusmn 0001 R2= 009

18 Wildlife Monographs bull 203

model 0ndash16) within year 200 (Fig 11) The mean times until quasi‐extinction based on the MVM scenario were 5 years (IBM 0ndash61)and 6 years (matrix model 0ndash64) conditioned on going quasi‐extinctwithin 100 years Expected times until quasi‐extinction conditionedon quasi‐extinction within 200 years were 11 years (IBM 0ndash113)and 11 years (matrix model 0ndash112 Fig 11)

Sensitivity of extinction probability to demographic parametersmdashThe partial rank correlation coefficients between quasi‐extinctionprobabilities and demographic parameters were negative anincrease in any of the demographic parameters decreased thequasi‐extinction probability Probabilities of quasi‐extinction within200 years were most sensitive to changes in kitten survival for bothscenarios (Fig 12) followed by survival of subadult females in theMVM scenarios and survival of prime adult females in the MPCscenario The probability of reproduction of prime adult femaleshad the third‐largest partial rank correlation coefficient with quasi‐extinction probability for both scenarios

Benefits and Costs of Genetic ManagementMinimum population count scenariomdashThe MPC scenario with

a critical threshold of 10 panthers predicted that withoutmanagement intervention average individual heterozygositywould decrease reaching 057 (049ndash064) at 100 years and053 (042ndash062) at 200 years (Fig 13) The population woulddecrease slightly reaching an average of 79 (41ndash99) at 100 yearsand 76 (43ndash98) at 200 years (Fig 14) The probabilities of quasi‐extinction under the no‐intervention MPC scenario when weconsidered the effects of genetic erosion were 13 (0ndash99) at 100years and 23 (0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with

the translocation of 5 Texas females average individualheterozygosity under the MPC scenario increased and thenstabilized at 068 (064ndash072) at 100 years and remained atthat level at 200 years (Fig 13) The population reached anaverage of 88 (62ndash104) at 100 years and stayed the same at200 years (Fig 14) The probabilities of quasi‐extinctionunder this introgression strategy were 6 (0ndash53) within 100years and 7 (0ndash82) within 200 years (Fig 15) Results ofanalyses using a critical threshold of 30 panthers revealed asimilar pattern of relative benefits However the prob-abilities of quasi‐extinction were substantially higher (ge45)across all scenarios (Appendix E)

Motor vehicle mortality scenariomdashWithout managementintervention the average individual heterozygosity predictedby the MVM scenario and a critical threshold of 10 panthersdecreased to 060 (046ndash069) at 100 years and to 059 (039ndash070) at 200 years (Fig 13) The population increased slightlyand reached an equilibrium at 154 individuals (46ndash217) at 100years and 157 (57ndash218) at 200 years (Fig 14) The probabilitiesof quasi‐extinction were 17 (0ndash100) at 100 years and 22(0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with 5

female pumas from Texas average individual heterozygosityincreased slightly reaching 067 (060ndash073) at 100 years and068 (059ndash075) at 200 years (Fig 13) Under this strategy thepopulation reached an average of 164 individuals (44ndash226) at

Table 4 Model comparison results testing for the effects of several covariateson survival of all kittens survival of genetically sampled kittens survival ofsubadult and adult Florida panthers probability of reproduction and kittensproduced annually for Florida panthers sampled from 1981ndash2013 in SouthFlorida USA

Model Parameters ΔAICa Weight

Kitten survival (all data)Base+ F1b+ abundancec 10 000 0988Base+ abundance 9 1018 0006Base+ F1 9 1035 0005Base 8 2711 lt0001Base+ sexd 9 2912 lt0001Base+ litter sizee 9 2914 lt0001

Kitten survival (genetically sampled kittens only)Base+ F1+ abundance 10 000 0541Base+ F1CanAdmf+ abundance 11 099 0329Base+Hetg+ abundance 10 310 0115Base+CanAdmh+ abundance 10 791 0010Base+ abundance 9 944 0005Base+ F1 9 1638 lt0001Base+ F1CanAdm 10 1837 lt0001Base+Het 9 2872 lt0001Base 8 3296 lt0001Base+CanAdm 9 3348 lt0001

Subadult and adult survivalBase+ F1CanAdm+ abundance 7 000 0360Base+ F1 5 053 0276Base+ F1CanAdm 6 105 0213Base+ F1+ abundance 6 222 0119Base+CanAdm+ abundance 6 575 0020Base+CanAdm 5 908 0004Base+Het 5 964 0003Base+Het+ abundance 6 984 0003Base 4 1118 0001Base+ abundance 5 1278 0001

Probability of reproductionAge 3 000 0242Age+CanAdm 4 012 0228Age+ abundance 4 173 0102Age+ F1 4 190 0094Age+CanAdm+ abundance 5 216 0082Age+Het 4 219 0081Age+ F1CanAdm 5 232 0076Age+ F1+ abundance 5 372 0038Age+Het+ abundance 5 399 0033Age+ F1CanAdm+ abundance 6 444 0026

Kittens produced annuallyAge+ F1+ abundance 5 000 0194Age 3 060 0144Age+ abundance 4 061 0143Age+ F1 4 097 0120Age+CanAdm 4 139 0097Age+ F1CanAdm+ abundance 6 196 0073Age+ F1CanAdm 5 231 0061Age+CanAdm+ abundance 5 238 0059Age+Het+ abundance 5 250 0056Age+Het 4 259 0053

a Akaikersquos information criterion (AIC) for kitten survival models was adjustedfor small sample size and overdispersion (QAICc)

b F1 divides Florida panthers into 2 groups F1 admixed and all other panthersc Index of Florida panther abundanced Sex of an individual Florida panther (male or female)e Size of the litter in which a Florida panther kitten was bornf F1CanAdm divides panthers into 3 groups F1 admixed canonical andother admixed panthers

g Het is the individual heterozygosityh CanAdm divides panthers into 2 groups canonical and admixed (includingF1 admixed) panthers

van de Kerk et al bull Florida Panther Population and Genetic Viability 19

100 years and 170 (67ndash231) at 200 years based on the MVMscenario (Fig 14) The probabilities of quasi‐extinction underthis introgression strategy were 10 (0ndash99) at 100 years and12 (0ndash99) at 200 years (Fig 15)The projected population size and average individual hetero-

zygosity reached equilibrium levels with periodic fluctuations inthese values when introgression was implemented (Figs 16ndash17)Genetic introgression decreased the time until quasi‐extinctionacross all scenarios (Fig 18) Results of analyses using a criticalthreshold of 30 panthers revealed a similar pattern of relative ben-efits However the probabilities of quasi‐extinction were sub-stantially higher (ge45) across all scenarios (Appendix E)

Addition of pumas from Texas increased both the populationsize and average individual heterozygosity consequentlyintrogression caused periodic increases in average individualheterozygosity and population size at the interval at which theintrogression occurred Whereas average individual hetero-zygosity gradually decreased following introgression densitydependence caused the population size to fluctuate especiallywhen a larger number of pumas from Texas were released Thepositive effect of genetic introgression initiatives on quasi‐extinction probabilities decreased as the time interval betweenthese events increased More frequent genetic introgressions ledto greater increases in average individual heterozygosity and

Old adult

Prime adult

Subadult

Kitten

025 050 075

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Individual heterozygosity

Ann

ual s

urvi

val r

ate

Old adult

Prime adult

Subadult

Kitten

40 60 80 100 120

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Abundance index

Ann

ual s

urvi

val r

ate

Kittens Males Females

Figure 7 The effect of individual heterozygosity (left) and the abundance index (right) on Florida panther age‐ and sex‐specific survival estimates (shaded area isSE) in South Florida USA 1981ndash2013 Abundance index data are from McBride et al (2008) and R T McBride and C McBride (Rancherrsquos Supply Incorporatedunpublished report)

20 Wildlife Monographs bull 203

equilibrium population size while reducing the probability ofquasi‐extinction Comparatively performing genetic introgres-sion less often but with more individuals per release was lesseffective at improving the long‐term prospects for the pantherpopulation (Figs 13ndash15)

Financial cost and persistence benefitsmdashThe total estimated costof 1 genetic introgression was $40000 $80000 or $120000for releasing 5 10 or 15 pumas from Texas respectively(Table 5) The total cost incurred over 100 years for eachstrategy ranged from $50000 to $12 million (Table 6)The most expensive strategy involved the introduction of15 pumas every 10 years this strategy reduced the probability

of quasi‐extinction by 58 and 40 under the MPC and MVMscenarios respectively (Table 6) Introducing 5 pumas every80 years cost the least but improvements in populationpersistence under this strategy were insubstantial (Table 6)Releasing more panthers more frequently caused increasedpopulation fluctuations but did not necessarily reduce the quasi‐extinction probability (Figs 14 and 15 Table 6)

DISCUSSION

The duration (34 yr of data) and sample sizes associated with theFlorida Panther Project allowed us to obtain a comprehensiveunderstanding of genetic variation demographic parameters

A

B

C

Figure 8 Cause‐specific mortality rates (plusmnSE) for male and female Florida panthers (A) subadult prime‐adult and old‐adult Florida panthers of both sexes (B)and canonical and admixed Florida panthers of both sexes (C) in South Florida USA 1981ndash2013 Causes of mortality are vehicle collision intraspecific aggression(ISA) other causes (known causes such as diseases injuries and infections unrelated to the first 2 causes) and unknown cause (mortalities for which we could notassign a cause)

van de Kerk et al bull Florida Panther Population and Genetic Viability 21

probability of population persistence and the duration of thepositive impacts of genetic restoration on this endangered po-pulation The Florida panther population was in dire straits inthe early 1990s and our results clearly demonstrated that the

implementation of the introgression project in 1995 subse-quently accrued a series of genetic and demographic improve-ments that have led to the change from a declining population toone that is growing and expanding its current breeding range

Sensitivity Elasticity

0 1 2 3 00 02 04

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

Para

met

er

Figure 9 Sensitivity and elasticity (proportional sensitivity) of stochastic population growth rate (λs) of the Florida panther to vital demographic parameters in SouthFlorida USA 1981ndash2013 Horizontal lines represent 5th and 95th percentiles

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

70

80

90

100

110

0

5

10

15

0

50

100

Years

StatisticMeanMedian

Figure 10 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the minimum population count(MPC) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

125

150

175

200

00

05

10

15

20

0

30

60

90

Years

StatisticMeanMedian

Figure 11 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the motor vehicle mortality(MVM) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

22 Wildlife Monographs bull 203

MPC MVM

minus08 minus06 minus04 minus02 00 minus08 minus06 minus04 minus02 00

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Male subadult survival

Male prime adult survival

Male old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

PRCC

Para

met

er

Figure 12 Partial rank correlation coefficient (PRCC) between quasi‐extinction probabilities and the demographic parameters of the Florida panther for theminimum population count (MPC) and motor vehicle mortality (MVM) scenarios Error bars indicate standard deviation Based on data collected in South FloridaUSA 1981ndash2013

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 13 Average projected individual heterozygosities in the Florida panther population over 200 years without genetic management intervention and with eachof the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include theintroduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years Average individual heterozygosity peaks when pumas are added to the Floridapanther population Based on data collected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 23

Our research has further validated the success of genetic in-trogression by quantifying the reduction in the probability ofextinction of the panther population because of this manage-ment initiative These findings all bode well for continuedprogress toward recovery That said the Florida panther po-pulation remains small and isolated from other populations ofpuma from western North America thereby denoting that theeventual impact of genetic drift and inbreeding may negativelyaffect the population in the future Realizing this will continueto be an issue that managers will have to contemplate wemodeled the impact of varied genetic management scenarios todetermine the frequency of introgression along with the numberof panthers that should be released during each initiative tominimize cost and maximize benefits to the population Theseresults will prove invaluable to managers attempting to conservethe Florida panther

Genetic Variation Demographic Parametersand Persistence of Heterotic Benefits from GeneticIntrogressionOur measure of individual heterozygosity (1 minusHL) was higheron average for the admixed (0615) panthers than for those ofcanonical ancestry (0386) Levels of individual heterozygosityfor admixed panthers were comparable to levels observed in asample of pumas from west Texas (x plusmn SE= 0695plusmn 0019n= 41) which further demonstrates the improvement in levelsof genetic variation in the Florida population since 1995 The

correlation between HL and several demographic parametershas been noted in wild populations including cheetahs (Terrellet al 2016) and several species of birds such as European shags(Phalacrocorax aristotelis male and female survival probabilityVelando et al 2015) and Egyptian vulture (Neophron percnop-terus age at recruitment Agudo et al 2012) If we focus only onpanthers captured and radiocollared from 2012ndash2015 (FP211ndashFP240) all of which had estimated birth years ge2005 (ge10 yrpost‐introgression) those panthers had an average individualheterozygosity level of 0618plusmn 0029 highlighting the elevatedlevels of genetic variation that continue to persist in cohorts ofpanthers 5 generations after the release of female pumas fromTexasThe proportional membership values (q) from our

STRUCTURE analysis allowed us to delineate 2 clusters ofancestry for Florida panthers (Fig 4) During the pre‐in-trogression era the population was dominated by panthers thatretained a high percentage of the canonical ancestry which wasaffected by inbreeding depression The post‐introgression eraancestry values are comprised of many admixed individuals inessence demonstrating the success of the introgression in termsof increasing genetic variation in the population and mi-micking gene flow that historically occurred with panthers andother populations of pumas (Onorato et al 2010) A somewhatanalogous situation although abetted by natural immigrationwas evident in the ancestral variation in a small population ofpuma intersected by a major freeway in California USA In

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

75

80

85

90

95

100

130

150

170

190

Years

Popu

latio

n si

ze

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 14 Average projected population sizes in the Florida panther population over 200 years without intervention and with each of the genetic introgressionstrategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include the introduction of 5 10 or 15pumas from Texas USA every 10 20 40 or 80 years Peaks occur when pumas are added to the Florida panther population Based on data collected in SouthFlorida USA 1981ndash2013

24 Wildlife Monographs bull 203

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

25

50

75

100

0

25

50

75

100

Introgression interval (yrs)

PQ

E (

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 15 Average probability of quasi‐extinction (PQE) of the Florida panther population within 100 years and within 200 years without genetic managementintervention (N) and with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenariosIntrogression strategies include the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Errorbars indicate 5th and 95th percentiles Based on data collected in South Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

50

60

70

80

90

100

110

50

100

150

200

Years

Popu

latio

n si

ze

Figure 16 Average projected Florida panther population trajectories under a geneticintrogression strategy involving the release of 5 female pumas from Texas USA every20 years for the minimum population count (MPC) and motor vehicle mortality(MVM) scenarios Shaded area indicates 5th and 95th percentiles and isrepresentative of confidence intervals for other scenarios Based on data collected inSouth Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Figure 17 Average projected Florida panther population‐level heterozygositiesunder a genetic introgression strategy involving the release of 5 female pumas fromTexas USA every 20 years for the minimum population count (MPC) and motorvehicle mortality (MVM) scenarios Shaded area bounds the 5th and 95th percentilesand is representative of confidence intervals for other scenarios Based on datacollected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 25

this case the migration of just a single male across the freewayto the south resulted in an increase in the number of in-dividuals with admixed ancestry and substantially improvedgenetic diversity of the subpopulation south of the freeway(Riley et al 2014)Our estimate of overall kitten survival (0321plusmn 0056) was

similar to that reported by Hostetler et al (2010) who used13 years of post‐introgression data (0323plusmn 0071) These valuesare lower than kitten survival estimates derived from severalpuma populations in western North America (044ndash091 Loganand Sweanor 2001 Lambert et al 2006 Laundre et al 2007Robinson et al 2008 Ruth et al 2011) When we included onlydata for individuals that had been genetically sampled our es-timate was 15 higher The proportion of admixed kittens thatwere genetically sampled increased as the population expandedwhich could have resulted in higher estimates of kitten survivalbecause admixed kittens survive better than canonical kittensKitten survival was strongly influenced by genetic ancestry withsurvival rates of F1 kittens being almost twice that of canonicalkittens (Fig 5B) Consistent with the findings of Hostetler et al(2010) increases in heterozygosity coincided with increasedkitten survival whereas population density negatively affectedkitten survival Population density often has a strong negativeimpact on survival of young age classes due to infanticide andcannibalism which has been reported from several carnivore

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

50

100

150

0

50

100

150

Introgression interval (yrs)

TQE

(yrs

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 18 Average times until quasi‐extinction (TQE) for the Florida panther population within 100 years and within 200 years without genetic management interventionand with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategiesinclude the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Error bars indicate 5th and 95thpercentiles Based on data collected in South Florida USA 1981ndash2013

Table 5 Average cost (2017 US dollars) of introgression strategies employedover 100 years that involve the introduction of 5 10 or 15 pumas from TexasUSA into the Florida panther population in South Florida USA at an in-creasingly higher frequency Costs of capture (including transportation) quar-antine and post‐introgression monitoring were estimated by Roy McBrideMark Cunningham and Dave Onorato respectively

Frequency Component 5 pumas 10 pumas 15 pumas

Once Capture 25000 50000 75000

Quarantine 5000 10000 15000

Monitoring 10000 20000 30000

Total 40000 80000 120000

Every 80 years Capture 31250 62500 93750

Quarantine 6250 12500 18750

Monitoring 12500 25000 37500

Total 50000 100000 150000

Every 40 years Capture 62500 125000 187500

Quarantine 12500 25000 37500

Monitoring 25000 50000 75000

Total 100000 200000 300000

Every 20 years Capture 125000 250000 375000

Quarantine 25000 50000 75000

Monitoring 50000 100000 150000

Total 200000 400000 600000

Every 10 years Capture 250000 500000 750000

Quarantine 50000 100000 150000

Monitoring 100000 200000 300000

Total 400000 800000 1200000

26 Wildlife Monographs bull 203

species including polar bears (Ursus maritimus Derocher andWiig 1999) black bears (Ursus americanus Garrison et al 2007)and pumas (Logan and Sweanor 2001)Our estimates of sex‐ and age‐specific survival rates of subadult

and adult panthers (Table 3) and the effects of covariates on theserates were similar to those reported by Benson et al (2011) derivedfrom data collected through 2006 Our survival rates for males(065ndash077) were lower than females (078ndash097) for all 3 ageclasses (subadult prime adult and old adult) which is the typicaltrend observed in most puma populations (eg Ruth et al 19982011) However an unhunted puma population occupying afragmented urban landscape in southern California exhibited lowersurvival (0586 females 0525 males Vickers et al 2015) perhapsas a result of severe habitat fragmentation and the lack of extensiveswaths of public land protected in perpetuity Most populations ofpuma in western North America are typically subject to variedlevels of management via regulated hunting which often is biasedtoward the take of males (Cooley et al 2009 Ruth et al 2011)Consequently annual survival estimates from hunted populationsin northeast Washington (0679ndash0733 females 0341 malesRobinson et al 2008) and southeastern Arizona (0685 females0584 males Cunningham et al 2001) were generally lower thanthose we estimated for Florida panthersCause‐specific mortality analysis showed that male panthers

experienced a substantially greater mortality risk from in-traspecific aggression This differs from the pattern observedduring a long‐term study in Arizona where females were at ahigher risk of intraspecific mortality than males (46 of maleand 53 of female mortality Logan and Sweanor 2001)Mortality associated with intraspecific strife has been docu-mented in hunted and unhunted populations (this study Loganand Sweanor 2001 Stoner et al 2010) and its root cause hasbeen difficult to determine (eg population density and preypopulation trends) That being the case finding ways to reducewhat is often a major source of mortality for puma populationscontinues to pose a challenge to managersCanonical panthers were substantially more likely to die from

intraspecific aggression than were admixed panthers This mayat least partly explain the difference in survival between admixed

and canonical panthers Canonical panthers are known to bemore likely to suffer from varied correlates of inbreeding de-pression than admixed animals including atrial septal defects(Johnson et al 2010) and compromised immune systems(Roelke et al 1993) These factors have the potential to affectfitness and perhaps dominance of individuals when engaged inan intraspecific encounter A somewhat similar scenario wasobserved by Hogg et al (2006) in a population of bighorn sheepthat were part of an introgression project Admixed males in thispopulation had a greater probability of paternity than males thatwere less outbred This included coursing males with higherlevels of admixture being markedly more successful at fightingmales that were defending females in estrous (Hogg et al 2006)Heterosis resulting from genetic introgression is expected to be

the strongest in F1 individuals (Shull 1908 Crow 1948 Burkeand Arnold 2001 Waller 2015) and to progressively decline insubsequent generations (Pickup et al 2013 Frankham 2015)Given that admixed (especially F1) panthers survived better thancanonical panthers and that individual heterozygosity improvedsurvival of both sexes and all age classes our data provide evi-dence for heterotic advantage whereby individuals of mixedancestry had higher fitness (Shull 1908 Crow 1948) Our resultsare consistent with findings of other studies that involved in-tentional genetic introgression for conservation purposes Forexample Hogg et al (2006) detected substantial improvementsin survival and reproduction of introgressed bighorn sheep andMadsen et al (1999 2004) reported a dramatic increase in thepopulation of adders after genetic introgressionKnowledge of the persistence of benefits to a population ac-

crued via genetic introgression is essential for determiningwhen additional introgression may be warranted There is adepauperate amount of information regarding this topic and theidentification of factors that may influence persistence of thebenefits of genetic introgression (Tallmon et al 2004 Hedricket al 2014 Whiteley et al 2015) For example in minks(Neovison vison) Thirstrup et al (2014) found that outcrossingled to larger litters in F2 and F3 but this benefit disappeared insubsequent generations In a population of gray wolves Hedricket al (2014) suggested that benefits of genetic introgression may

Table 6 Costs and benefits accumulated over 100 years for genetic introgression at different intervals and with different numbers of pumas from Texas USAintroduced into the Florida panther population in South Florida USA Costs (2017 US dollars) include capture and transportation quarantine and post‐introgression monitoring Benefits are represented as average probability of quasi‐extinction (PQE and 5th and 95th percentiles) and changes (Δ) therein under thescenario using the minimum population count (McBride et al 2008 MPC) and the scenario using the motor vehicle mortality population estimates (McClintocket al 2015 MVM) The table is sorted by percent change in probability of quasi‐extinction under the MPC scenario

Interval (yr) Pumas Cost MPC PQE () ΔMPC PQE () MVM PQE () ΔMVM PQE ()

ndash 0 0 13 (0ndash99) 0 17 (0ndash100) 080 10 100000 12 (0ndash99) 10 (0ndash58) 16 (0ndash100) 5 (0ndash38)80 15 150000 12 (0ndash99) 13 (0ndash65) 15 (0ndash100) 8 (0ndash56)80 5 50000 12 (0ndash99) 14 (0ndash73) 15 (0ndash99) 9 (0ndash41)40 15 300000 10 (0ndash98) 26 (0ndash104) 14 (0ndash99) 13 (0ndash60)40 10 200000 9 (0ndash94) 35 (0ndash183) 13 (0ndash99) 21 (0ndash95)40 5 100000 8 (0ndash89) 39 (0ndash211) 12 (0ndash99) 26 (0ndash124)20 15 600000 8 (0ndash88) 42 (0ndash257) 13 (0ndash99) 24 (0ndash123)20 10 400000 6 (0ndash67) 54 (0ndash318) 10 (0ndash99) 37 (0ndash217)10 15 1200000 6 (0ndash53) 58 (0ndash372) 10 (0ndash99) 40 (0ndash208)20 5 200000 5 (0ndash50) 59 (0ndash338) 9 (0ndash99) 44 (0ndash352)10 10 800000 4 (0ndash16) 69 (0ndash489) 8 (0ndash92) 55 (0ndash401)10 5 400000 4 (0ndash7) 73 (0ndash502) 6 (0ndash71) 63 (0ndash503)

van de Kerk et al bull Florida Panther Population and Genetic Viability 27

wane after 2 or 3 generations The WrightndashFisher model ofgenetic drift predicts a geometric decline in expected hetero-zygosity at the rate of (1 minus 12Ne) per generation where Ne is theeffective population size (Hartl and Clark 1997 Frankham et al2014) Thus it seems logical to assume that the duration of thebenefits of introgression is limited especially in a species per-sisting in a small population such as Florida panthersWe expected Florida panther survival in the most recent years

(2005ndash2013) post‐introgression to differ from that observed in thedecade following the introgression in 1995 because pantherabundance and heterozygosity factors known to influence panthersurvival (Hostetler et al 2010 Benson et al 2011) have changed(Figs 2 and 6) We found that subadult and adult survival rates inrecent years (2005ndash2013) were similar to those in the period im-mediately following introgression (1995ndash2004 Fig 5C) overallkitten survival was similar to estimates of Hostetler et al (2010) Infact the pattern of covariate effects on Florida panther survivalreported by Benson et al (2011) and Hostetler et al (2010) hasbarely changed The negative effects of population density andpositive effects of heterozygosity may counterbalance survival ratesreducing population fluctuations Our result that benefits of geneticrestoration have remained in the population nearly for 5 genera-tions post‐intervention was surprising but also encouraging Pos-sible explanations for this observation may include 1) genetic ero-sion occurs at slower rate than previously thought 2) introducing 5migrants in 1 genetic intervention may have beneficial effects si-milar to those from 1 migrant per generation over multiple gen-erations or 3) other and as yet unknown factors may reduce therate of genetic erosion and subsequent demographic effects Adensity‐dependent effect on kitten survival could also explain whynon‐F1 admixed kittens did not survive substantially better thandid canonical kittens most kittens sampled from 2005ndash2013 wereadmixed and their survival was lower possibly because of a density‐dependent effect as the Florida panther population continuouslygrew during that period Trinkel et al (2010) also found thatpopulation density and inbreeding coefficient interacted to affectdemographic parameters and growth rate of an African lion po-pulation Likewise population density interacted with winterweather to affect the survival and population growth rates in Soaysheep (Ovis aries Milner et al 1999 Coulson et al 2001)

Population Dynamics and PersistenceThe finite deterministic and stochastic growth rates were gt10reflecting that much of the data used in this study were collectedfollowing genetic introgression in 1995 during which time thepopulation increased substantially Because our demographicparameter estimates did not change markedly in comparison tothose of Hostetler et al (2013) the similarity between thepopulation growth estimates from these studies was expectedBut these estimates reflect population growth during thedata‐collection period and do not predict future growth In factestimates of population size reported by McClintock et al(2015) suggest that population growth may already be slowing(Fig 2) A similar pattern was observed in an introduced po-pulation of African lion which increased steadily from 1995until 2001 then fluctuated around the presumed carrying ca-pacity thereafter (Trinkel et al 2010) Several other African lionpopulations that experienced various degrees of recovery

subsequently became relatively stable or exhibited decliningtrends (Bauer et al 2015)Our estimates of quasi‐extinction probabilities were lower than

those reported by Hostetler et al (2013) using a 2‐sex age‐structured matrix population model Lower probabilities ofquasi‐extinction than those reported by the earlier study forcomparable scenarios were likely a consequence of the fact thatthe estimates of abundance (McBride et al 2008) and thus thedensity‐dependent effects on survival of kittens and old panthers(gt10 yr) have changed in recent years Additionally these earlyestimates of the probability of quasi‐extinction and of time toquasi‐extinction did not consider genetic erosion that will in-evitably occur without management interventionUnder the MVM scenario the equilibrium population size was

twice that attained under the MPC scenario The MVM popula-tion estimates are approximately twice the MPCs (Fig 2) as aresult the simulated population under the MVM scenario achieveda higher equilibrium size Probability of quasi‐extinction under theMVM scenario was generally greater than that under the MPCscenario This result is a consequence of the fact that the estimateddensity dependence on kitten survival based on the MPC scenario isstronger than that for theMVM scenario (regression coefficients forabundance for the MPC scenario= minus0068 vs minus0002 for theMVM scenario Appendix B Table B1) Consequently simulatedpopulations under the MPC scenario have a stronger tendency tomove toward the equilibrium population size which inherentlyreduces the quasi‐extinction probability (eg Royama 1992)As expected for long‐lived mammals (Heppell et al 2000 Oli

and Dobson 2003) the population growth rate was proportio-nately most sensitive to changes in survival of prime adultsfollowed by that in younger age classes Global sensitivity ana-lysis in contrast showed that under all scenarios extinctionparameters were particularly sensitive to changes in survival ofFlorida panther kittens This result is a consequence of the highsensitivity of the population growth rate to kitten survival(Fig 9) and a larger standard error of kitten survival (Table 3)Results of our sensitivity analyses indicate that future researchshould focus on acquiring additional information on early lifestages to improve the accuracy and precision of estimates ofkitten survival Our results suggest that demographic variables(or their environmental drivers) that strongly influence extinc-tion risks are not necessarily those with the largest elasticityvalues A study of island fox (Bakker et al 2009) found thatextinction risk was strongly influenced by survival modifierparameters that had low elasticity values

Genetic Management When and How to InterveneThe use of genetic introgression as a management tool has al-ways been controversial and the probable effectiveness it mighthave in preventing the imminent demise of the panther popu-lation was initially debated (Creel 2006 Maehr et al 2006Pimm et al 2006) These debates notwithstanding the pantherpopulation had suffered from genetic morphological and bio-medical correlates of inbreeding before this management in-tervention (Johnson et al 2010 Onorato et al 2010) whereasthe post‐introgression period has been characterized by a sub-stantial reduction in the biomedical correlates of inbreeding andimprovements in demographic vigor and abundance (McBride

28 Wildlife Monographs bull 203

et al 2008 Hostetler et al 2010 2013 Johnson et al 2010Benson et al 2011 McClintock et al 2015) Nevertheless thepopulation remains small and isolated habitat is still being lostand there is no current possibility of natural gene flow betweenthis and other North American puma populations thus therecurrence of inbreeding‐related problems and subsequent po-pulation declines are inevitable The question therefore is notwhether genetic management of the Florida panther populationis needed but when and how it should be implementedWithout genetic introgression probabilities of quasi‐extinc-

tion were substantially greater than those from models thatincorporated periodic genetic introgression events (Fig 15)highlighting the importance of genetic management in smallisolated populations When 5 female pumas are added to theFlorida panther population every 10 years genetic variationincreased substantially under the MPC and MVM scenariosThe IBM predicted that the equillibrium population size wouldbe substantially larger when 5 pumas were released into thepopulation every 10 years than populations without intervention(ie no introgression) Releasing 15 Texas females did not af-ford substantially greater benefit to the equilibrium populationsize than did releasing only 5 Texas females Instead addingmore individuals caused increasingly large fluctuations in po-pulation size due to the numerical increases and density de-pendence (Fig 14) possibly diminishing the benefits associatedwith increased genetic variationCompared with the no‐intervention scenario (ie no genetic

introgression implemented) the introduction of 5 female pumasevery 10 years reduced quasi‐extinction probabilities from 13to 4 and from 17 to 6 for the MPC and MVM scenariosrespectively The benefit of genetic‐introgression strategies de-creased as the interval between successive management inter-ventions increased and no benefit was evident once the intervalreached 80 years (Fig 15) Introgression reduced the averagetime until quasi‐extinction (Fig 18) This somewhat counter‐intuitive result can be explained by the fact that repeated geneticintrogression makes the population more robust over timewhich will reduce the probability of extinction Consequentlyfew simulated population trajectories that fall below the criticalthreshold do so early reducing the mean time to extinctionAlso density dependence has a stabilizing effect on populations(Royama 1992) populations tend toward equilibrium popula-tion sizes which reduces the probability over time of popula-tions falling below quasi‐extinction threshold However timeuntil quasi‐extinction must be interpreted in conjunction withthe probability of quasi‐extinction which is very low for allscenarios with genetic introgression (Fig 15)Cost is a significant impediment to the implementation of a

genetic introgression program because captures relocations andmonitoring efforts before and after introgression are expensive(Edmands 2007 Frankham 2010b 2015 2016) Costs may beacceptable to society if the management actions result in sub-stantial fitness benefits especially if the species of concern ispopular and charismatic (Frankham 2015) We estimated the100‐year cost of introgression strategies modeled here to rangefrom $50000 to $12 million More expensive strategies gen-erally led to larger decreases in the probability of quasi‐extinc-tion but cheaper strategies also substantially improved

population viability (ie reduced quasi‐extinction probabilityTable 6) One of the cheaper strategies (5 pumas released every20 years) which cost an estimated $200000 over 100 yearsreduced the probability of quasi‐extinction by 59 and 44 forthe MPC and MVM scenarios respectively But these pre-liminary cost estimates are based on expenses incurred duringthe 1995 introgression and include costs of capture quarantinetransportation release and post‐release monitoring of femalesonly Although the cost for future genetic interventions wouldlikely be higher than those reported here the relative differencesin cost between alternative intervention strategies should remainapproximately the sameOur ability to simulate genetics and link it to the viability of

the Florida panther population is based on the empirically es-timated relationship between individual heterozygosity andsurvival Such heterozygosity and fitness correlations have beenstudied for decades (David 1998) but have only recently beenused to predict population viability (Benson et al 2016) noneto our knowledge have incorporated cost into their modelingefforts The mechanisms underlying heterozygosity and fitnesscorrelations remain unclear (David 1998 Szulkin et al 2010)and their significance as a proxy for inbreeding depression hasbeen debated (Balloux et al 2004 Miller and Coltman 2014)Nevertheless evidence continues to mount demonstratingpositive relationships between heterozygosity and survival(Coulson et al 1998ab Hansson et al 2004 Silva et al 2005Acevedo‐Whitehouse et al 2006 Jensen et al 2007 Velandoet al 2015) and between heterozygosity and fecundity (Seddonet al 2004 Charpentier et al 2005 Agudo et al 2012 Velandoet al 2015) Although the reported correlations are generallyweak their consequences can be substantial if population growthand persistence parameters are highly sensitive to the affectedvital rates (Velando et al 2015) Compared with scenarios thatignore genetics incorporating the effects of inbreeding on sur-vival increased quasi‐extinction probabilities within a 100‐yeartimeframe from 15 to 13 and from 14 to 17 for theMPC and MVM scenarios respectively These results high-light the importance of incorporating genetics when analyzingthe viability of small isolated populationsAlthough conservation biologists have recognized the im-

portance of genetics in population viability many still fail toincorporate genetic factors into PVA models (Reed et al 2002)Explicit consideration of genetics in PVAs requires long‐termgenetic and demographic data This permits the assessment ofthe functional relationship between genetic variability and de-mographic parameters Population viability analysis softwarepackages such as VORTEX (Lacy 1993 2000) offer a powerfulframework for individual‐based simulations but they often donot adequately capture a speciesrsquo life history or genetics Basedon a thorough review of the importance of genetic factors inwildlife conservation Frankham et al (2014) concluded thatgenetic factors can strongly influence the dynamics and persis-tence of small andor fragmented populations They furthernoted that ldquoMost PVA‐based risk assessments ignore or in-adequately model genetic factorsrdquo and recommended that ldquoPVAshould routinely include realistic inbreeding depressionhelliprdquo(Frankham et al 201457) Our custom‐built IBM permitted usto develop a model that adequately captured the Florida panther

van de Kerk et al bull Florida Panther Population and Genetic Viability 29

life history and we could parameterize the model with our long‐term genetic and demographic dataThe explicit consideration of the effects of inbreeding on

Florida panther population dynamics and persistence representsan important step forward Nonetheless our model remainsimperfect First we neglected the possible effects of habitat lossdetrimental anthropogenic activities climate change the prob-ability of immigration of pumas from western populationsdisease or catastrophes on demographic rates and populationviability Although immigration from puma populations outsideFlorida is possible its probability is very low given the extensivechallenges the anthropogenically altered landscape would posefor such a migrant to make it successfully to South Florida Weomitted mutations from our model because we have no in-formation on mutation rates though we expect that they are lowbased on values reported for other mammals (Kumar and Sub-ramanian 2002) Finally we only considered possible effects ofinbreeding on age‐specific survival rates because there was noevidence that heterozygosity affected female reproduction Lossof heterozygosity can negatively affect reproduction becauseinbred male panthers can suffer from cryptorchidism which canadversely affect their fecundity However given the polygynousmating system we expect the Florida panther population growthis primarily female‐limitedAlthough it is generally accepted that genetic introgression

only temporarily relieves a population from inbreeding depres-sion we know of no cases in which it has been implementedmore than once to conserve a threatened wildlife populationOur study provides an example of how demographic and mo-lecular data can be used within an IBM framework to evaluatethe efficacy of alternative genetic management strategies via theFlorida panther as a case study Our model can be adjusted forand applied to other species and offers great potential as a toolfor genetic management of small isolated wildlife populations

MANAGEMENT IMPLICATIONS

Our results and those of Hostetler et al (2013) and Johnsonet al (2010) provide persuasive evidence that the genetic in-tervention implemented in 1995 prevented the demise of theFlorida panther and restored demographic vigor to the popu-lation The Florida panther population remains small and iso-lated from other puma populations the closest puma populationis located gt1000 km away in Texas and the intervening land-scape is highly developed and fragmented with many interstate(and other high‐traffic) highways and major urban centersTheory suggests that 1 migrant per generation is sufficient tominimize the loss of genetic diversity (eg Mills and Allendorf1996) However the possibility of successful migration of apuma to South Florida followed by successful mating every ~5years is very low considering the distance between Texas andFlorida (Hedrick 1995) and the many risks and barriers thatdispersing pumas face (Beier 1995 Maehr et al 2002 Thatcheret al 2006) Consequently the pantherrsquos long‐term persistencewill likely depend on subsequent timely genetic introgressionsgiven the near‐impossibility of natural gene flow from otherpuma populations To that end our study offers considerableinsights into benefits of different genetic management strategiesand associated costs

Without genetic management the Florida panther populationfaces a substantial risk of quasi‐extinction within 100 years (10ndash46 depending on quasi‐extinction threshold and scenarios)Twenty‐three years have passed since genetic introgression wasimplemented and the population appears healthy as indicatedby measures of genetic variation increases in the populationsize and improvements in age‐specific survival rates Ourfindings suggest that releasing more pumas more frequently doesnot necessarily improve persistence probability because such astrategy would cause larger fluctuations in population size(Fig 14) which negatively affects persistence probability Thusextinction risks faced by inbred populations of large carnivorescan be substantially reduced by releasing approximately 5 im-migrants every 2 decades We suggest that such a managementstrategy be followed only in conjunction with continued long‐term monitoring of the population Our analyses demonstratethat population growth and persistence are highly sensitive tokitten and female (adult and subadult) survival Continuing tocollect data on these demographic variables along with bio-metric and genetic sampling will remain important to pantherrecovery and vigilance in identifying issues associated with in-breeding depression The collection of these monitoring datashould allow managers to detect any demographic or geneticchanges in a timely fashion and respond with genetic in-trogression or other management actions if warrantedRecovery objectives as defined by the USFWS in its current

recovery plan (USFWS 2008) delineate the need for additionalpopulations in the historical range to downlist or delist theFlorida panther If the only extant population of Floridapanthers continued to grow it could serve as a source ofpanthers to be translocated to suitable habitat to seed addi-tional populations Maintaining current information on thegenetic health of the population and precise estimates of itssize will be important in assessing whether it can withstand theremoval of a subset of animals for translocation Although the2017 documentation of a female panther with kittens north ofthe current breeding range is encouraging in terms of thepotential for population expansionmdashgiven that no female hadbeen recorded north of the Caloosahatchee River since 1972mdashthe re‐establishment of separate new populations will mostlikely require translocations by wildlife managers and a lengthysociopolitical processConservation of threatened or endangered species such as the

Florida panther is inherently challenging For panthers and otherlarge carnivores that require vast extents of natural habitat topersist habitat is typically the most limiting factor that will de-termine whether recovery efforts succeed Although geneticmanagement invariably plays a key role in the long‐term persis-tence of the panther population even a healthy population caneventually succumb to the impacts of habitat loss The humanpopulation of Florida continues to be one of the fastest‐growingin the nation the United States Census Bureau currently ranksFlorida as the third most populous Therefore habitat loss isgoing to continue to be a key issue for panthers Diligence towardpreserving panther habitat in its historical range via conservationeasements or other means and trying to keep such areas inter-connected via corridors is a goal stakeholders such as governmentagencies sportsmen non‐governmental organizations ranchers

30 Wildlife Monographs bull 203

and other private landowners must work on collaboratively toachieve

ACKNOWLEDGMENTS

We thank D Shindle D Jennings T OrsquoMeara D Land andC Belden for valuable advice throughout the project We thankS Bass M Criffield M Cunningham D Giardina D JansenA Johnson D Land M Lotz C McBride Rocky McBrideRowdy McBride Roy McBride L Oberhofer S Schulze staffsat Everglades National Park and Big Cypress National Preserveand others for assistance with fieldwork We also thank JAustin M Conroy J Hines J Laake S Mills J Nichols JHines M Sunquist J Fryxell and T Coulson for advice andassistance regarding data analysis and panther biology TheMuseum of Texas Tech University Natural Science ResearchLaboratory provided samples from pumas from west Texas forcomparative genetic analyses M Ben‐David D Land WMcCown E Hellgren and 4 anonymous reviewers providedmany helpful comments on the manuscript We thank NereaUbierna Lopez for completing the Spanish translation of theabstract We are grateful to the Department of ResearchComputing University of Florida for excellent high‐perfor-mance computing services We would like to thank US Fishand Wildlife Service Florida Fish and Wildlife ConservationCommission (FWC) US National Park Service QuantitativeSpatial Ecology Evolution and Environment (QSE3) In-tegrative Graduate Education and Research Traineeship(IGERT) program funded by the National Science Foundationand the University of Florida for financially supporting thisstudy Funding for panther research conducted by the FWC issupported predominantly via donations accrued with vehicleregistrations for ldquoProtect the Pantherrdquo license plates

LITERATURE CITEDAcevedo‐Whitehouse K T R Spraker E Lyons S R Melin F Gulland RL Delong and W Amos 2006 Contrasting effects of heterozygosity onsurvival and hookworm resistance in California sea lion pups MolecularEcology 151973ndash1982

Adams J R L M Vucetich P W Hedrick R O Peterson and J AVucetich 2011 Genomic sweep and potential genetic rescue during limitingenvironmental conditions in an isolated wolf population Proceedings of theRoyal Society B Biological Sciences 2783336ndash3344

Agresti A 2013 Categorical data analysis Third edition John Wiley amp SonsHoboken New Jersey USA

Agudo R M Carrete M Alcaide C Rico F Hiraldo and J A Donaacutezar2012 Genetic diversity at neutral and adaptive loci determines individualfitness in a long‐lived territorial bird Proceedings of the Royal Society BBiological Sciences 2793241ndash3249

Albeke S E N P Nibbelink and M Ben‐David 2015 Modeling behavior bycoastal river otter (Lontra canadensis) in response to prey availability in PrinceWilliam Sound Alaska a spatially‐explicit individual‐based approach PLOSOne 10e0126208

Alho J S K Vaumllimaumlki and J Merilauml 2010 Rhh an R extension for estimatingmultilocus heterozygosity and heterozygosity‐heterozygosity correlationMolecular Ecology Resources 10720ndash722

Alvarez K 1993 Twilight of the panther Myakka River Publishing SarasotaFlorida USA

Aparicio J M J Ortego and P J Cordero 2006 What should we weigh toestimate heterozygosity alleles or loci Molecular Ecology 154659ndash4665

Bakker V J D F Doak G W Roemer D K Garcelon T J Coonan S AMorrison C Lynch K Ralls and R Shaw 2009 Incorporating ecologicaldrivers and uncertainty into a demographic population viability analysis for theisland fox Ecological Monographs 7977ndash108

Balloux F W Amos and T Coulson 2004 Does heterozygosity estimateinbreeding in real populations Molecular Ecology 133021ndash3031

Barone M A M E Roelke J Howard J L Brown A E Anderson and DE Wildt 1994 Reproductive characteristics of male Florida panthersmdashcomparative studies from Florida Texas Colorado Latin‐America andNorth‐American Zoos Journal of Mammalogy 75150ndash162

Bateson Z W P O Dunn S D Hull A E Henschen J A Johnson and LA Whittingham 2014 Genetic restoration of a threatened population ofgreater prairie‐chickens Biological Conservation 17412ndash19

Bauer H G Chapron K Nowell P Henschel P Funston L T B HunterD W Macdonald and C Packer 2015 Lion (Panthera leo) populations aredeclining rapidly across Africa except in intensively managed areas Pro-ceedings of National Academy of Sciences of the United States of America11214894ndash14899

Beier P 1995 Dispersal of juvenile cougars in fragmented habitat Journal ofWildlife Management 59228ndash237

Benson J F J A Hostetler D P Onorato W E Johnson M E Roelke SJ OrsquoBrien D Jansen and M K Oli 2011 Intentional genetic introgressioninfluences survival of adults and subadults in a small inbred felid populationJournal of Animal Ecology 80958ndash967

Benson J F P J Mahoney J A Sikich L E K Serieys J P Pollinger H BErnest and S P D Riley 2016 Interactions between demography geneticsand landscape connectivity increase extinction probability for a small popu-lation of large carnivores in a major metropolitan area Proceedings of theRoyal Society B Biological Sciences 28320160957

Beverly F 1874 The Florida panther Forest and Stream 3290Boyce M S C V Haridas C T Lee and the NCEAS Stochastic Demo-graphy Working Group 2006 Demography in an increasingly variable worldTrends in Ecology amp Evolution 21141ndash148

Brook B W J J OrsquoGrady A P Chapman M A Burgman H R Akcakayaand R Frankham 2000 Predictive accuracy of population viability analysis inconservation biology Nature 404385ndash387

Brys R and H Jacquemyn 2016 Severe outbreeding and inbreeding de-pression maintain mating system differentiation in Epipactis (Orchidaceae)Journal of Evolutionary Biology 29352ndash359

Burke J M and M L Arnold 2001 Genetics and the fitness of hybridsAnnual Review of Genetics 3531ndash52

Burnham K P 1993 A theory for combined analysis of ring recovery andrecapture data Pages 199ndash213 in J‐D Lebreton and P M North editorsMarked individuals in the study of bird population Springer‐Verlag NewYork New York USA

Burnham K P and D R Anderson 2002 Model selection and multimodelinference a practical information‐theoretic approach Second edition SpringerVerlag New York New York USA

Caswell H 2001 Matrix population models construction analysis and in-terpretation Sinauer Associates Sunderland Massachusetts USA

Charpentier M J M Setchell F Prugnolle L A Knapp E J Wickings PPeignot and M Hossaert‐McKey 2005 Genetic diversity and reproductivesuccess in mandrills (Mandrillus sphinx) Proceedings of the NationalAcademy of Sciences of the United States of America 10216723ndash16728

Cooch E and G C White editors 2018 Program MARK a gentle in-troduction lthttpwwwphidotorgsoftwaremarkdocsbookgt Accessed 7Apr 2016

Cooley H S R B Wielgus G M Koehler H S Robinson and B TMaletzke 2009 Does hunting regulate cougar populations A test of thecompensatory mortality hypothesis Ecology 902913ndash2921

Coulson T E A Catchpole S D Albon B J Morgan J M Pemberton TH Clutton‐Brock M J Crawley and B T Grenfell 2001 Age sex densitywinter weather and population crashes in Soay sheep Science 2921528ndash1531

Coulson T J Pemberton S Albon M Beaumont T Marshall J Slate FGuinness and T Clutton‐Brock 1998a Microsatellites reveal heterosis in reddeer Proceedings of the Royal Society B Biological Sciences 265489ndash495

Coulson T N S D Albon J M Pemberton J Slate F E Guinness and TH Clutton‐Brock 1998b Genotype by environment interactions in wintersurvival in red deer Journal of Animal Ecology 67434ndash445

Cox D 1972 Regression models and life‐tables Journal of the Royal StatisticalSociety Series B Statistical Methodology 34187ndash220

Creel S 2006 Recovery of the Florida panthermdashgenetic rescue demographic rescueor both Response to Pimm et al (2006) Animal Conservation 9125ndash126

Crnokrak P and D A Roff 1999 Inbreeding depression in the wild Heredity83260ndash270

Crow J 1948 Alternative hypotheses of hybrid vigor Genetics 33477ndash487

van de Kerk et al bull Florida Panther Population and Genetic Viability 31

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 11: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

recovery model (Burnham 1993) to estimate and model kittensurvival using the RMark package version 2113 (Laake 2013) asan interface for Program MARK (White and Burnham 1999)We implemented the Cox proportional hazard model for esti-

mation and modeling of survival of subadults and adults using the Rpackage SURVIVAL (Therneau and Grambsch 2000) We per-formed cause‐specific mortality analyses using an R version of S‐PLUS code provided by Heisey and Patterson (2006)To account for model selection uncertainty we calculated

model‐averaged parameter estimates across all models using theR package AICcmodavg (Mazerolle 2017) using AIC weights asmodel weights (Burnham and Anderson 2002) We used modelswithout the categorical ancestry variables for model averagingand estimated parameters based on the average value for thecontinuous covariates (individual heterozygosity and abundanceindex) Because only a subset of the kittens was geneticallysampled we estimated 2 kitten survival rates First we estimatedkitten survival based on the data set that included all kittens(overall kitten survival) Second we estimated kitten survivalusing a subset of the data that only included kittens that weregenetically sampled (survival of genetically sampled kittens)

Matrix Population ModelsWe investigated Florida panther population dynamics andpersistence using 2 complementary modeling frameworks ma-trix population models (Caswell 2001) and IBMs (Grimm andRailsback 2005 McLane et al 2011 Railsback and Grimm2011 Albeke et al 2015) Matrix population models offer aflexible and powerful framework for investigating the dynamicsand persistence of age‐ or stage‐structured populations (egCaswell 2001 Robinson et al 2008 Kohira et al 2009 Hunteret al 2010 Hostetler et al 2012) With a fully developed theorybehind them these models also serve as a check for resultsobtained from simulation‐based models such as IBMs We usedan age‐structured‐matrix population model for deterministic andstochastic demographic analyses and for preliminary investiga-tions of Florida panther population viability (Caswell 2001Morris and Doak 2002)For deterministic and stochastic demographic analyses we

parameterized a female‐only 19 times 19 age‐specific populationprojection matrix of the form

⎢⎢⎢⎢⎢

⋯ ⋯

⋯ ⋯

⋱ ⋱ ⋮

⋮ ⋱ ⋱ ⋱ ⋮

⎥⎥⎥⎥⎥

( )=t

F F F

P

P

P

A0 0

0

0 0 0

t t t

t

t

t

1 2 19

1

2

18

where Fi and Pi are the age‐specific fertility and survival ratesrespectively and ( )tA is a time‐varying (or stochastic) popula-tion projection matrix We assumed that the longevity and ageof last reproduction of female Florida panthers was 185 yearswe based this assumption on the observation that the oldestfemale in our study was 186 years old Florida panthers re-produce year‐round thus we used birth‐flow methods to esti-mate Fi and Pi values following Caswell (2001) and Hostetleret al (2013) We estimated Pi as

= +P S S i i i 1

where Si is the age‐specific survival probability We estimated Fi as

⎛⎝

⎞⎠

=+ +F S

m Pm

2i k

i i i 1

where Sk is the kitten survival probability and mi is the age‐specific fecundity rate which was estimated as

ν=m b f i i i k

where bi is the breeding probability for a female in age class iduring time step t νi is the annual number of kittens producedby a breeding female in age class i and fk is the proportionfemale kittens at birth (assumed to be 05)We estimated the finite deterministic (λ) and stochastic annual

population growth rate (λs) and stochastic sensitivities and elasti-cities following methods described in detail by Caswell (2001) Toestimate λs we assumed identically and independently distributedenvironments (Caswell 2001 Morris and Doak 2002 Haridas andTuljapurkar 2005) and used a parametric bootstrapping approachto obtain a sequence of demographic parameters for 50000 ma-trices We then estimated log(λs) from this sequence of matrices(Tuljapurkar 1990 Caswell 2001) and exponentiated it to obtain λsWe calculated the sensitivity and elasticity of λs to changes in lower‐level vital demographic rates as per Caswell (2001) and Haridas andTuljapurkar (2005)For population projection and population viability analysis (PVA)

simulations we expanded the population projection matrix into a34 times 34 2‐sex age‐structured matrix to include female and maleFlorida panthers (Caswell 2001 Hostetler et al 2013) Malescontribute to the population only through survival in this modelingframework We assumed maximum longevity for males to be 145years of age because the oldest male in our study was 144 years oldThe population projection matrix was of the following form

⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢

⋯ ⋯ ⋯ ⋯

⋯ ⋯ ⋯ ⋯

⋱ ⋱ ⋮ ⋮ ⋱ ⋱ ⋮

⋮ ⋱ ⋱ ⋱ ⋮ ⋮ ⋱ ⋱ ⋮

⋯ ⋯ ⋯

⋯ ⋯ ⋯ ⋯

⋯ ⋯ ⋱ ⋱ ⋮

⋮ ⋮ ⋱ ⋱ ⋮ ⋱ ⋱ ⋮

⋯ ⋯ ⋯

⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥

( )=

( ) ( ) ( )

( )

( )

( )

( ) ( ) ( )

( )

( )

t

F N F N F N

P N

P N

P N

M N M N M N

Q N

Q N

A n

0 0

0 0 0 0

0

0 0 0 0 0

0 0

0 0 0

00 0 0 0 0

t t t

t

t

t

t t t

t

t

1 2 19

1

2

18

1 2 19

1

14

where Pi and Qi are age‐specific survival rates of female and maleFlorida panthers respectively and Fi and Mi are the rates atwhich female and male kittens are produced by females of ageclass i respectively Here the population projection matrix isboth time‐varying and density‐dependent (Caswell 2001)We incorporated the influence of parameter uncertainty en-

vironmental stochasticity and density dependence on Floridapanther population dynamics and persistence following the ap-proach outlined by Hostetler et al (2013) we used the sameapproach in the IBM analysis which is described below UnlikeIBMs matrix models do not intrinsically include demographicstochasticity so we explicitly included the influence of

van de Kerk et al bull Florida Panther Population and Genetic Viability 13

demographic stochasticity by simulating the fates of individualsas described in Caswell (2001)Population projections using matrix models necessitate a

vector of initial sex‐ and age‐specific abundance Because suchdata are currently unavailable for the Florida panther we esti-mated it based on the stable sex and age distribution and MPCin 2013 Using point estimates of the demographic parameterswe estimated the stable sex and age distribution for the 2‐sexdeterministic and density‐independent population projectionmatrix We then multiplied the stable sex and age distributionby the MPC in 2013 to get the starting population vector n(0)We projected the population over time as n(t+ 1)=A(tn)n(t)where A(tn) indicates the time‐specific density‐dependentpopulation projection matrix (Caswell 2001)We had 2 indices of abundance available the MPC (McBride

et al 2008) and population size estimated using motor vehiclecollision mortalities (MVM McClintock et al 2015) These 2indices are substantially different because MPC does not ac-count for imperfect detection whereas MVM accounts forimperfect detection and provides an estimate of population size(Fig 2) Therefore we examined 2 scenarios one using densitydependence based on MPC and the other using density de-pendence based on MVM We ran 1000 bootstraps of para-meter values and 1000 simulations per bootstrap for each sce-nario We projected population trajectories over 200 years andthen estimated population growth rates probabilities ofquasi‐extinction and time until quasi‐extinction and comparedthese with the results obtained from the IBM (see below) Weran scenarios with a quasi‐extinction threshold of 10 and 30panthers We used the mean of probabilities of quasi‐extinction(PQE) across simulations as the point estimate for PQE and the5th and 95th percentiles across simulations as the 90 con-fidence interval Because the confidence intervals werequantile‐based but the point estimates were not it is possiblewith skewed distributions for the point estimates to be outsidethe confidence interval We implemented the matrix

population models in the R computing environment (R CoreTeam 2016)

IBMs and PVABecause of well‐developed theory ease of implementation andthe modeling flexibility that they offer matrix populationmodels have become the model of choice for investigating thedynamics and persistence of age‐ and stage‐structured popula-tions (Caswell 2001) However it is difficult to incorporateattributes of individuals such as genetics and behavior usingmatrix models Quantifying the effects of genetic erosion on theFlorida panther population dynamics and persistence andevaluating benefits and costs of alternative genetic managementscenarios were important goals of our study Because IBMs relyon a bottom‐up approach that begins by explicitly consideringindividuals (McLane et al 2011 Railsback and Grimm 2011Albeke et al 2015) it is possible to represent genetic attributesof each individual and to track genetic variation over time Weused IBMs as the primary modeling framework in this studybecause they allowed for explicit consideration of genetics bothat individual and population levels and population viabilityassessment under alternative genetic management scenariosWe developed an IBM (Grimm 1999 Grimm and Railsback

2005 McLane et al 2011 Railsback and Grimm 2011 Albekeet al 2015) tailored to the life history of the Florida panther(Fig 3) We simulated every individual from birth until deathbased on a set of rules describing fates (eg birth and death) andattributes (eg genetics) of individuals at each annual time stepAs in other IBMs population‐level processes (eg populationsize and genetic variation) emerged from fates of and interac-tions among individuals (Grimm 1999 Railsback and Grimm2011) We developed an overview design concepts and details(ODD) protocol (Grimm et al 2006 2010) to describe ourIBM (Appendix C available online in Supporting Information)and provide pseudocodes for individual‐based simulations(Appendix D available online in Supporting Information)

For each panther at time = t

Kitten (lt1 yr)

Reproduce

Survive

Kitten(s) born

Kitten

Sex and age class

Subadult male

Prime adult male

Old adult male

Subadult female

Prime adult female

Old adult female

Age one year

No Male

Female

Yes

No

Yes

t=t+1

Yes

lt35 yrs

35-10 yrs

gt10 yrs

lt25 yrs

25-10 yrs

gt10 yrs

Figure 3 Schematic representation of Florida panther life history We tailored the individual‐based population model to represent the life‐history patterndepicted here

14 Wildlife Monographs bull 203

Dynamics and persistence of populations are influenced bymany factors such as demographic and environmental sto-chasticity and density dependence these effects and parametricuncertainty must be considered in population models whenpossible (Caswell 2001 Boyce et al 2006 Bakker et al 2009Hostetler et al 2013) Because by definition IBMs simulate thelife history of individuals in a population they inherently in-corporate the effect of demographic stochasticity Our analysesrevealed that survival of kittens subadults and adults and theprobability of reproduction varied over time so we incorporatedenvironmental stochasticity in the model for these variablesfollowing Hostetler et al (2013) Likewise we found evidenceof density‐dependent effects on kitten survival Because theMPC might be an underestimate of population size (McBrideet al 2008) we performed simulations using both the MPC andMVM abundance indicesTo account for model selection and parametric uncertainty we

used a Monte Carlo simulation approach For each bootstrapsample we selected a model based on the AIC (or QAICc)weights We then sampled the intercept and slope for kittensurvival (on the logit scale) subadult and adult survival (on thelogit scale with log‐hazard effect sizes) and probability of re-production (on the complementary logndashlog scale Appendix B)from multivariate normal distributions with mean vectors equalto the estimated parameters and variance‐covariance matricesequal to the sampling variance‐covariance matrices We thentransformed the results to nominal scale and converted the es-timates to age‐specific survival and reproduction parameters asdescribed in Hostetler et al (2013)We ran 1000 simulations for 1000 parametric bootstraps for

200 years for the MPC and the MVM scenario for both thestochastic 2‐sex matrix‐population model and the IBM Wetracked the population size at each time step and estimatedquasi‐extinction probabilities and times based on the populationtrajectory We considered the population to be quasi‐extinct ifindividuals of only 1 sex remained or if the population size fellbelow critical thresholds set at 10 and 30 individuals We alsoestimated expected time to quasi‐extinction for both thresholdsdefined as the mean and median time until a population be-comes quasi‐extinct conditioned on quasi‐extinction We im-plemented the IBM using the RNetLogo package version 10ndash2(Thiele et al 2012 Thiele 2014) as an interface for the IBMplatform NetLogo (Wilensky 1999)

Sensitivity AnalysisAn important component of demographic analysis is sensitivityanalysis which is the quantification of the sensitivity of amodelrsquos outcome to changes in the input parameters (Caswell2001 Bakker et al 2009 Thiele et al 2014) Using an IBM weperformed global sensitivity analyses for 2 (ie MPC andMVM) density‐dependent scenarios to assess how sensitivequasi‐extinction times and probabilities were to changes (anduncertainties) in the estimated demographic rates We usedLatin hypercube sampling to sample parameters from the entireparameter space (Marino et al 2008 Thiele et al 2014) Wesampled intercept and slope for kitten survival (on logit scale)subadult and adult survival (on logit scale with log‐hazard effectsizes) and probability of reproduction (on complementary

logndashlog scale) from normal distributions We sampled theprobability distribution for the number of kittens producedannually (on the real scale) from a Dirichlet distribution themultivariate generalization of the beta distribution (Kotz et al2000) We sampled 1000 different parameter sets and ran 1000simulations for each scenario We calculated the probability ofquasi‐extinction within 200 years for each density‐dependencescenario (MPC or MVM) using a quasi‐extinction threshold of10 individuals and then calculated the partial rank correlationcoefficient between each of those and each parameter trans-formed to the real scale (Bakker et al 2009 Hostetler et al2013) Partial rank correlation coefficients quantify the sensi-tivity of the probability of quasi‐extinction to input variables(ie survival and reproductive parameters) with higher absolutevalues indicating stronger influence of that variable on quasi‐extinction probabilities

Genetic ManagementOne of our goals was to estimate the benefits (improvements ingenetic variation demographic parameters and populationgrowth and persistence parameters) and costs (financial costs ofcapture transportation quarantine release and monitoring ofintroduced panthers) of genetic management To achieve thisobjective we extended the IBM to include a genetic componentTo simulate individual‐ and population‐level heterozygosity overtime we assigned each panther alleles for 16 microsatellite locibased on the empirical allele frequency in the population (esti-mated from genetic samples collected during 2008ndash2015) whenwe initialized the model (Pierson et al 2015) At each time stepwe simulated Mendelian inheritance for kittens produced suchthat each kitten randomly inherited alleles from its mother andfrom a male panther randomly chosen to have putatively siredthe litter We did not explicitly force inbreeding to occur butthe probability of inbreeding naturally increased as the popula-tion size decreasedIntrogression strategiesmdashWe modeled genetic introgression as a

result of the release of 2‐year‐old female pumas from Texas intothe simulated Florida panther population For the geneticintrogression implemented in 1995 Texas females between 15and 25 years of age were released because females at that agehave lower mortality rates (Benson et al 2011) higherreproductive potential and also because it is much easier todetect with certainty the successful reproduction by females thanby males The ability to more closely monitor reproduction andthe birth of kittens is an important consideration in reducing theprobability of a genomic sweep of Texas alleles into the Floridapanther population Wildlife managers removed the last 2surviving Texas females from the wild population in Florida in2003 to reduce the possibility of genomic sweep To simulatethis strategy in the model we removed any Texas females thatwere still alive 5 years after each introgression event Weassigned Texas females alleles based on the allele frequency ofthe Texas population (based on genotypes from 48 samplescollected in west Texas that was inclusive of the pumas releasedin South Florida in 1995) when they entered the Florida pantherpopulation We could not estimate demographic rates separatelyfor the released Texas females because of small sample size (only8 females were released in 1995) therefore we assumed that

van de Kerk et al bull Florida Panther Population and Genetic Viability 15

survival and reproductive rates and the effects of covariates onthese rates for the released Texas females were identical to thoseof female Florida panthersThe impact of genetic introgression depends on the frequency

of introgression events and the number of individuals introducedat each attempt Thus we defined introgression strategies basedon the number of Texas females to be released (0 5 10 or 15)and the interval between genetic introgressions (10 20 40 and80 yr) for a total of 13 strategies We simulated each manage-ment strategy with density dependence estimated using theMPC and MVM abundance indices

We sampled 1000 parameter sets and for each of these setswe ran all introgression strategies 1000 times for 200 years Weapplied the same parametric uncertainty and environmentalstochasticity across all management scenarios to allow for bettercomparison of introgression strategies At each time step werecorded the projected population size and average individualheterozygosity We calculated the heterozygosity for each in-dividual at birth based on the alleles it inherited This individualheterozygosity then informed the survival rate of that individualbased on the empirically estimated relationship between in-dividual heterozygosity and age‐specific survival rates Hetero-zygosity did not change throughout an individualrsquos lifetime butits effect on survival changed as individuals aged because theeffect of individual heterozygosity on survival rate differedamong age classes Survival rates of genetically sampled kittenswere biased high because some kittens were sampled later in lifewhen they had already survived for some time To correct forthis bias we adjusted kitten survival rates predicted by theheterozygosity model proportionally From the population tra-jectories we estimated the probability of quasi‐extinction (de-fined as the probability that the population will fall below 10 or30 individuals or that individuals of only 1 sex will remain)

Cost‐benefit analysismdashExperts estimated financial costs ofintrogression based on their experience with the geneticintrogression executed in 1995 and we adjusted estimates tocurrent costs to account for inflation (R T McBride RancherrsquosSupply Incorporated M Cunningham and D P OnoratoFlorida Fish and Wildlife Conservation Commission personalcommunication) Costs included capturing and transportingfemale pumas from the Texas source population caring for thepumas while they were in quarantine and post‐introgressionmonitoring To compare the financial costs of the differentscenarios we also calculated the average costs incurred over

Table 2 Sample size (n) number of alleles (Na) allelic richness (AR) observedheterozygosity (Ho) and expected heterozygosity (He) at 16 microsatellite loci inradio‐collared Florida panthers sampled from 1981 to 2013 in South FloridaUSA We calculated these values from a subset (n= 137) of the samples used inthe analyses and they include only adult and subadult panthers We excludedkittens because they can result in an overrepresentation of certain alleles

Locus n Na AR Ho He

FCA090 129 5 50 0333 0401FCA133 136 3 30 0618 0608FCA243 136 5 50 0419 0487F124 137 7 69 0708 0729F37 129 4 40 0395 0397FCA075 131 5 50 0534 0598FCA559 133 5 50 0496 0503FCA057 134 6 59 0679 0624FCA081 134 7 69 0209 0206FCA566 130 6 60 0462 0475F42 133 10 100 0737 0766FCA043 136 5 49 0596 0588FCA161 132 6 60 0500 0515FCA293 132 4 40 0614 0624FCA369 131 6 60 0573 0567FCA668 133 7 70 0406 0462Mean 1329 57 57 0517 0535

Figure 4 Proportional membership (q) of radio‐collared Florida panthers (n= 218) and pumas from Texas USA (n= 7) in 2 clusters identified by STRUCTUREEach individual animal is represented by a separate vertical bar Yellow indicates canonical panther ancestry and blue represents admixed ancestry The admixedancestry of the Texas females and F1 generation panthers that were radiocollared are highlighted red and green respectively to demarcate those groups The pre‐introgression period is inclusive of panthers radiocollared from 10 February 1981 to 6 March 1996 The post‐introgression era inclusive of the F1 panthers includesanimals radiocollared from 4 March 1997 to 18 February 2015 in South Florida USA

16 Wildlife Monographs bull 203

100 years assuming that the population would be managedaccording to a particular strategy Our goal with theseexploratory analyses was to evaluate the relative costs andbenefits of alternative management scenarios

RESULTS

Genetic VariationWe assessed genetic variation at 16 microsatellite loci for 137DNA samples collected from adult and subadult panthers(1981ndash2013) that were captured and subsequently radiocollaredThe number of alleles at these loci ranged from 3ndash10 and allelicrichness averaged 57 (Table 2) Average expected hetero-zygosity for this sample was 0535 and average observed het-erozygosity was 0517 (Table 2)We determined individual heterozygosity (1 minusHL) and an-

cestry for the complete sample of panther genotypes (n= 503)collected from 1981 to 2015 for use as covariates in subsequentanalyses Average individual heterozygosity was 0559 andranged from 0ndash0980 Our STRUCTURE analysis providedstrong support for K= 2 clusters based on the probability of thedata (log Pr(D)|K) and ΔK in STRUCTURE HARVESTERBased on this result we categorized the ancestry of each pantheras either canonical (n= 123) or admixed (n= 380) using valuesof proportional membership (q Fig 4) The average individualheterozygosity of canonical panthers was 0386plusmn 0012 (SE) foradmixed panthers it was 0615plusmn 0007

Survival and Cause‐Specific Mortality RatesOf the 395 kittens that were PIT‐tagged and released 55 werelater encountered alive 15 were recovered dead and 85 werepart of failed litters (Table 1) We radio‐tracked 209 subadult oradult panthers for a total of 244195 panther‐days and docu-mented 126 mortalities (Table 1)Survival depended strongly on age and kittens had the lowest

overall rate (032plusmn 009 Fig 5A Table 3) The model‐averagedkitten survival was 047plusmn 009 when we included only geneti-cally sampled individuals in the analysis (Table 3) as statedabove this estimate is biased high because many kittenswere genetically sampled when they were recaptured alive orrecovered dead at older ages We found no evidence for sex‐specific differences in kitten survival but females survived betterthan males in all older age classes For females subadults hadthe highest survival rates followed by prime adults and oldadults For males prime adults had the highest survival ratefollowed by subadults and old adults (Fig 5A and Table 3)For panthers of all ages there was substantial evidence that

ancestry affected survival with F1 admixed panthers of all ageshaving the highest rates and canonical individuals having thelowest (Fig 5B) The combined AIC weights of models thatincluded an ancestry covariate were 0880 for kittens and 0992for older panthers The survival rates of subadult prime‐adultand old‐adult panthers in the period immediately after the in-trogression (1995ndash2004) were similar to those in more recentyears (2005ndash2013 Fig 5C)After genetic introgression individual panther heterozygosity

increased (Fig 6) and varied among ancestry categories individualheterozygosity was the highest for F1 panthers (078plusmn 001)

A

B

C

Figure 5 Overall age‐ and sex‐specific survival rates (plusmnSE A) age‐ and sex‐specific survival rates (plusmnSE) for Florida panthers of canonical F1 admixed andother admixed ancestry (B) and age‐ and sex‐specific survival estimates (plusmnSE)for subadult and adult Florida panthers in the period immediately post‐introgression (1995ndash2004) and more recently (2005ndash2013 C) in SouthFlorida USA

Table 3 Model‐averaged estimates (plusmnSE) of demographic parameters for theFlorida panther obtained using data collected during 1981ndash2013 in SouthFlorida USA

Parameter Sex Age class Estimate

Survival Both Kitten 032plusmn 009a

Female Subadult 097plusmn 002Prime adult 086+ 003Old adult 078plusmn 009

Male Subadult 066plusmn 006Prime adult 077plusmn 005Old adult 065plusmn 010

Probability of reproduction Female Subadult 035plusmn 008Prime adult 050plusmn 005Old adult 025plusmn 006

Kittens produced annually Female Subadult 280plusmn 005Prime adult 267plusmn 001Old adult 228plusmn 013

a Overall kitten survival (based on all kittens in our data set including thosethat were not genetically sampled) Estimate of survival of geneticallysampled kittens was 047plusmn 009 this estimate is biased high because manykittens were genetically sampled when they were recaptured alive or re-covered dead at older ages

van de Kerk et al bull Florida Panther Population and Genetic Viability 17

followed by those for non‐F1 admixed (062plusmn 001) and canonical(039plusmn 001) panthers Individual heterozygosity positively affectedsurvival rates of kittens (slope parameter η= 1455 95CI= 0505ndash2405) Individual heterozygosity also positively af-fected survival of subadult and adult panthers but the evidence forthis effect was weaker because the confidence interval for the ha-zard ratio overlapped 10 (hazard ratio exp(ɩ)= 0311 95CI= 0092ndash1054 Table 4 and Fig 7)The MPC increased after genetic introgression (Fig 2) This

abundance index was also included in top‐ranking modelsindicating that population density affected survival (Table 4)Abundance reduced the survival of kittens (slope parameterη= minus0035 95 CI= minus0049 to minus0021) evidence was weak forthe effect of abundance on survival of subadult and adult panthers(hazard ratio exp(ɩ)= 1002 95 CI= 0995ndash1010 Fig 7) Re-gression coefficients associated with the effect of abundance andindividual heterozygosity on demographic parameters are presentedin Table B1 (available online in Supporting Information)The most frequently observed causes of death for radio‐

collared panthers were vehicle collision and intraspecific ag-gression Cause‐specific mortality analyses revealed thatmortality rates among possible causes were generally similar forthe 2 sexes but that males were more likely to die from in-traspecific aggression than were females (Fig 8A) Male mor-tality from intraspecific aggression was higher for subadult andold adult panthers than for prime adults (Fig 8B) For bothmales and females canonical panthers were more likely to diefrom intraspecific aggression than were admixed panthers(Fig 8C)

Reproductive ParametersOf 101 radio‐collared females 67 reproduced at least once duringour monitoring we used these individuals to assess the annualprobability of reproduction The top‐ranked model for the annual

probability of reproduction included age effect only suggesting thatadding ancestry or abundance did not improve model parsimony(Table 4) Model‐averaged annual probabilities of reproductiondiffered between female subadults (035plusmn 008) prime adults(050plusmn 005) and older adults (025plusmn 006)The most parsimonious model for the number of kittens

produced annually by females that produced at least 1 litter (ieconditional on reproduction) included effects of age ancestryand abundance a model that included only the effect of age wasalmost equally supported (Table 4) The model‐averagednumber of kittens produced annually conditional on re-production was 280plusmn 075 for subadults 267plusmn 043 for primeadults and 228plusmn 083 for old adults Confidence intervals forthe slope parameter (β) overlapped 0 for both abundance(β= 0010 95 CI= minus0001 to 0021) and ancestry (β= minus073795 CI= minus1637 to 0162)

Population Dynamics and PersistenceDeterministic and stochastic demographymdashThe deterministic (λ)

and stochastic (λs) annual population growth rate calculated usingthe matrix population model were 106 (5th and 95th percentiles099ndash114) and 104 (072ndash141) respectively The generation timewas 47 years A female starting in age class one (kitten) wasexpected to live another 58 years and produce 10 kittens onaverage during her lifetime Overall λs was proportionately(elasticity) most sensitive to changes in female prime adultsurvival on the absolute scale (sensitivity) however it was mostsensitive to changes in kitten survival (Fig 9) Populationtrajectories probabilities of quasi‐extinction and times untilquasi‐extinction estimated using the matrix population modeland IBM for comparable scenarios were nearly identical for acritical quasi‐extinction threshold of 10 panthers (Figs 10 and 11)and for a critical quasi‐extinction threshold of 30 panthers(Appendix E available online in Supporting Information)Minimum population count scenariomdashUnder the MPC scenario

(ie density dependence estimated using the minimumpopulation count data) with a critical threshold of 10panthers the population size reached 99 (72ndash104) and 100(77ndash105) at 200 years (Fig 10) as predicted by the IBM and thematrix model respectively The cumulative quasi‐extinctionprobabilities within 100 years based on the MPC scenario were15 (IBM 0ndash48) and 30 (matrix model 0ndash76) Theseprobabilities increased to 25 (IBM 0ndash114) and 38 (matrixmodel 0ndash154) by year 200 (Fig 10) The mean times untilquasi‐extinction were 15 years (IBM 0ndash67) and 15 years (matrixmodel 0ndash72) conditioned on going quasi‐extinct within 100years Expected times until quasi‐extinction conditioned onquasi‐extinction within 200 years were higher 34 years (IBM0ndash137) and 32 years (matrix model 0ndash134 Fig 10)Motor vehicle mortality scenariomdashUnder the MVM scenario

(ie density dependence estimated using estimates of pantherpopulation size from McClintock et al 2015) with a criticalthreshold of 10 panthers the projected population reached 188(142ndash217) and 190 (143ndash219) at 200 years as predicted by theIBM and the matrix model respectively (Fig 11) Thecumulative quasi‐extinction probabilities within 100 years were14 (IBM 0ndash08) and 13 (matrix model 0ndash06) Theseprobabilities increased to 20 (IBM 0ndash17) and 19 (matrix

000

025

050

075

100

1995 2000 2005 2010Year of birth

Indi

vidu

al h

eter

ozyg

osity

Figure 6 Temporal changes in the individual heterozygosity of Floridapanthers born during the period 1995ndash2013 in South Florida USA Pointsare measurements solid line is linear regression shaded area is 95 confidenceinterval of slope Slope= 0006plusmn 0001 R2= 009

18 Wildlife Monographs bull 203

model 0ndash16) within year 200 (Fig 11) The mean times until quasi‐extinction based on the MVM scenario were 5 years (IBM 0ndash61)and 6 years (matrix model 0ndash64) conditioned on going quasi‐extinctwithin 100 years Expected times until quasi‐extinction conditionedon quasi‐extinction within 200 years were 11 years (IBM 0ndash113)and 11 years (matrix model 0ndash112 Fig 11)

Sensitivity of extinction probability to demographic parametersmdashThe partial rank correlation coefficients between quasi‐extinctionprobabilities and demographic parameters were negative anincrease in any of the demographic parameters decreased thequasi‐extinction probability Probabilities of quasi‐extinction within200 years were most sensitive to changes in kitten survival for bothscenarios (Fig 12) followed by survival of subadult females in theMVM scenarios and survival of prime adult females in the MPCscenario The probability of reproduction of prime adult femaleshad the third‐largest partial rank correlation coefficient with quasi‐extinction probability for both scenarios

Benefits and Costs of Genetic ManagementMinimum population count scenariomdashThe MPC scenario with

a critical threshold of 10 panthers predicted that withoutmanagement intervention average individual heterozygositywould decrease reaching 057 (049ndash064) at 100 years and053 (042ndash062) at 200 years (Fig 13) The population woulddecrease slightly reaching an average of 79 (41ndash99) at 100 yearsand 76 (43ndash98) at 200 years (Fig 14) The probabilities of quasi‐extinction under the no‐intervention MPC scenario when weconsidered the effects of genetic erosion were 13 (0ndash99) at 100years and 23 (0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with

the translocation of 5 Texas females average individualheterozygosity under the MPC scenario increased and thenstabilized at 068 (064ndash072) at 100 years and remained atthat level at 200 years (Fig 13) The population reached anaverage of 88 (62ndash104) at 100 years and stayed the same at200 years (Fig 14) The probabilities of quasi‐extinctionunder this introgression strategy were 6 (0ndash53) within 100years and 7 (0ndash82) within 200 years (Fig 15) Results ofanalyses using a critical threshold of 30 panthers revealed asimilar pattern of relative benefits However the prob-abilities of quasi‐extinction were substantially higher (ge45)across all scenarios (Appendix E)

Motor vehicle mortality scenariomdashWithout managementintervention the average individual heterozygosity predictedby the MVM scenario and a critical threshold of 10 panthersdecreased to 060 (046ndash069) at 100 years and to 059 (039ndash070) at 200 years (Fig 13) The population increased slightlyand reached an equilibrium at 154 individuals (46ndash217) at 100years and 157 (57ndash218) at 200 years (Fig 14) The probabilitiesof quasi‐extinction were 17 (0ndash100) at 100 years and 22(0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with 5

female pumas from Texas average individual heterozygosityincreased slightly reaching 067 (060ndash073) at 100 years and068 (059ndash075) at 200 years (Fig 13) Under this strategy thepopulation reached an average of 164 individuals (44ndash226) at

Table 4 Model comparison results testing for the effects of several covariateson survival of all kittens survival of genetically sampled kittens survival ofsubadult and adult Florida panthers probability of reproduction and kittensproduced annually for Florida panthers sampled from 1981ndash2013 in SouthFlorida USA

Model Parameters ΔAICa Weight

Kitten survival (all data)Base+ F1b+ abundancec 10 000 0988Base+ abundance 9 1018 0006Base+ F1 9 1035 0005Base 8 2711 lt0001Base+ sexd 9 2912 lt0001Base+ litter sizee 9 2914 lt0001

Kitten survival (genetically sampled kittens only)Base+ F1+ abundance 10 000 0541Base+ F1CanAdmf+ abundance 11 099 0329Base+Hetg+ abundance 10 310 0115Base+CanAdmh+ abundance 10 791 0010Base+ abundance 9 944 0005Base+ F1 9 1638 lt0001Base+ F1CanAdm 10 1837 lt0001Base+Het 9 2872 lt0001Base 8 3296 lt0001Base+CanAdm 9 3348 lt0001

Subadult and adult survivalBase+ F1CanAdm+ abundance 7 000 0360Base+ F1 5 053 0276Base+ F1CanAdm 6 105 0213Base+ F1+ abundance 6 222 0119Base+CanAdm+ abundance 6 575 0020Base+CanAdm 5 908 0004Base+Het 5 964 0003Base+Het+ abundance 6 984 0003Base 4 1118 0001Base+ abundance 5 1278 0001

Probability of reproductionAge 3 000 0242Age+CanAdm 4 012 0228Age+ abundance 4 173 0102Age+ F1 4 190 0094Age+CanAdm+ abundance 5 216 0082Age+Het 4 219 0081Age+ F1CanAdm 5 232 0076Age+ F1+ abundance 5 372 0038Age+Het+ abundance 5 399 0033Age+ F1CanAdm+ abundance 6 444 0026

Kittens produced annuallyAge+ F1+ abundance 5 000 0194Age 3 060 0144Age+ abundance 4 061 0143Age+ F1 4 097 0120Age+CanAdm 4 139 0097Age+ F1CanAdm+ abundance 6 196 0073Age+ F1CanAdm 5 231 0061Age+CanAdm+ abundance 5 238 0059Age+Het+ abundance 5 250 0056Age+Het 4 259 0053

a Akaikersquos information criterion (AIC) for kitten survival models was adjustedfor small sample size and overdispersion (QAICc)

b F1 divides Florida panthers into 2 groups F1 admixed and all other panthersc Index of Florida panther abundanced Sex of an individual Florida panther (male or female)e Size of the litter in which a Florida panther kitten was bornf F1CanAdm divides panthers into 3 groups F1 admixed canonical andother admixed panthers

g Het is the individual heterozygosityh CanAdm divides panthers into 2 groups canonical and admixed (includingF1 admixed) panthers

van de Kerk et al bull Florida Panther Population and Genetic Viability 19

100 years and 170 (67ndash231) at 200 years based on the MVMscenario (Fig 14) The probabilities of quasi‐extinction underthis introgression strategy were 10 (0ndash99) at 100 years and12 (0ndash99) at 200 years (Fig 15)The projected population size and average individual hetero-

zygosity reached equilibrium levels with periodic fluctuations inthese values when introgression was implemented (Figs 16ndash17)Genetic introgression decreased the time until quasi‐extinctionacross all scenarios (Fig 18) Results of analyses using a criticalthreshold of 30 panthers revealed a similar pattern of relative ben-efits However the probabilities of quasi‐extinction were sub-stantially higher (ge45) across all scenarios (Appendix E)

Addition of pumas from Texas increased both the populationsize and average individual heterozygosity consequentlyintrogression caused periodic increases in average individualheterozygosity and population size at the interval at which theintrogression occurred Whereas average individual hetero-zygosity gradually decreased following introgression densitydependence caused the population size to fluctuate especiallywhen a larger number of pumas from Texas were released Thepositive effect of genetic introgression initiatives on quasi‐extinction probabilities decreased as the time interval betweenthese events increased More frequent genetic introgressions ledto greater increases in average individual heterozygosity and

Old adult

Prime adult

Subadult

Kitten

025 050 075

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Individual heterozygosity

Ann

ual s

urvi

val r

ate

Old adult

Prime adult

Subadult

Kitten

40 60 80 100 120

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Abundance index

Ann

ual s

urvi

val r

ate

Kittens Males Females

Figure 7 The effect of individual heterozygosity (left) and the abundance index (right) on Florida panther age‐ and sex‐specific survival estimates (shaded area isSE) in South Florida USA 1981ndash2013 Abundance index data are from McBride et al (2008) and R T McBride and C McBride (Rancherrsquos Supply Incorporatedunpublished report)

20 Wildlife Monographs bull 203

equilibrium population size while reducing the probability ofquasi‐extinction Comparatively performing genetic introgres-sion less often but with more individuals per release was lesseffective at improving the long‐term prospects for the pantherpopulation (Figs 13ndash15)

Financial cost and persistence benefitsmdashThe total estimated costof 1 genetic introgression was $40000 $80000 or $120000for releasing 5 10 or 15 pumas from Texas respectively(Table 5) The total cost incurred over 100 years for eachstrategy ranged from $50000 to $12 million (Table 6)The most expensive strategy involved the introduction of15 pumas every 10 years this strategy reduced the probability

of quasi‐extinction by 58 and 40 under the MPC and MVMscenarios respectively (Table 6) Introducing 5 pumas every80 years cost the least but improvements in populationpersistence under this strategy were insubstantial (Table 6)Releasing more panthers more frequently caused increasedpopulation fluctuations but did not necessarily reduce the quasi‐extinction probability (Figs 14 and 15 Table 6)

DISCUSSION

The duration (34 yr of data) and sample sizes associated with theFlorida Panther Project allowed us to obtain a comprehensiveunderstanding of genetic variation demographic parameters

A

B

C

Figure 8 Cause‐specific mortality rates (plusmnSE) for male and female Florida panthers (A) subadult prime‐adult and old‐adult Florida panthers of both sexes (B)and canonical and admixed Florida panthers of both sexes (C) in South Florida USA 1981ndash2013 Causes of mortality are vehicle collision intraspecific aggression(ISA) other causes (known causes such as diseases injuries and infections unrelated to the first 2 causes) and unknown cause (mortalities for which we could notassign a cause)

van de Kerk et al bull Florida Panther Population and Genetic Viability 21

probability of population persistence and the duration of thepositive impacts of genetic restoration on this endangered po-pulation The Florida panther population was in dire straits inthe early 1990s and our results clearly demonstrated that the

implementation of the introgression project in 1995 subse-quently accrued a series of genetic and demographic improve-ments that have led to the change from a declining population toone that is growing and expanding its current breeding range

Sensitivity Elasticity

0 1 2 3 00 02 04

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

Para

met

er

Figure 9 Sensitivity and elasticity (proportional sensitivity) of stochastic population growth rate (λs) of the Florida panther to vital demographic parameters in SouthFlorida USA 1981ndash2013 Horizontal lines represent 5th and 95th percentiles

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

70

80

90

100

110

0

5

10

15

0

50

100

Years

StatisticMeanMedian

Figure 10 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the minimum population count(MPC) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

125

150

175

200

00

05

10

15

20

0

30

60

90

Years

StatisticMeanMedian

Figure 11 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the motor vehicle mortality(MVM) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

22 Wildlife Monographs bull 203

MPC MVM

minus08 minus06 minus04 minus02 00 minus08 minus06 minus04 minus02 00

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Male subadult survival

Male prime adult survival

Male old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

PRCC

Para

met

er

Figure 12 Partial rank correlation coefficient (PRCC) between quasi‐extinction probabilities and the demographic parameters of the Florida panther for theminimum population count (MPC) and motor vehicle mortality (MVM) scenarios Error bars indicate standard deviation Based on data collected in South FloridaUSA 1981ndash2013

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 13 Average projected individual heterozygosities in the Florida panther population over 200 years without genetic management intervention and with eachof the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include theintroduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years Average individual heterozygosity peaks when pumas are added to the Floridapanther population Based on data collected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 23

Our research has further validated the success of genetic in-trogression by quantifying the reduction in the probability ofextinction of the panther population because of this manage-ment initiative These findings all bode well for continuedprogress toward recovery That said the Florida panther po-pulation remains small and isolated from other populations ofpuma from western North America thereby denoting that theeventual impact of genetic drift and inbreeding may negativelyaffect the population in the future Realizing this will continueto be an issue that managers will have to contemplate wemodeled the impact of varied genetic management scenarios todetermine the frequency of introgression along with the numberof panthers that should be released during each initiative tominimize cost and maximize benefits to the population Theseresults will prove invaluable to managers attempting to conservethe Florida panther

Genetic Variation Demographic Parametersand Persistence of Heterotic Benefits from GeneticIntrogressionOur measure of individual heterozygosity (1 minusHL) was higheron average for the admixed (0615) panthers than for those ofcanonical ancestry (0386) Levels of individual heterozygosityfor admixed panthers were comparable to levels observed in asample of pumas from west Texas (x plusmn SE= 0695plusmn 0019n= 41) which further demonstrates the improvement in levelsof genetic variation in the Florida population since 1995 The

correlation between HL and several demographic parametershas been noted in wild populations including cheetahs (Terrellet al 2016) and several species of birds such as European shags(Phalacrocorax aristotelis male and female survival probabilityVelando et al 2015) and Egyptian vulture (Neophron percnop-terus age at recruitment Agudo et al 2012) If we focus only onpanthers captured and radiocollared from 2012ndash2015 (FP211ndashFP240) all of which had estimated birth years ge2005 (ge10 yrpost‐introgression) those panthers had an average individualheterozygosity level of 0618plusmn 0029 highlighting the elevatedlevels of genetic variation that continue to persist in cohorts ofpanthers 5 generations after the release of female pumas fromTexasThe proportional membership values (q) from our

STRUCTURE analysis allowed us to delineate 2 clusters ofancestry for Florida panthers (Fig 4) During the pre‐in-trogression era the population was dominated by panthers thatretained a high percentage of the canonical ancestry which wasaffected by inbreeding depression The post‐introgression eraancestry values are comprised of many admixed individuals inessence demonstrating the success of the introgression in termsof increasing genetic variation in the population and mi-micking gene flow that historically occurred with panthers andother populations of pumas (Onorato et al 2010) A somewhatanalogous situation although abetted by natural immigrationwas evident in the ancestral variation in a small population ofpuma intersected by a major freeway in California USA In

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

75

80

85

90

95

100

130

150

170

190

Years

Popu

latio

n si

ze

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 14 Average projected population sizes in the Florida panther population over 200 years without intervention and with each of the genetic introgressionstrategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include the introduction of 5 10 or 15pumas from Texas USA every 10 20 40 or 80 years Peaks occur when pumas are added to the Florida panther population Based on data collected in SouthFlorida USA 1981ndash2013

24 Wildlife Monographs bull 203

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

25

50

75

100

0

25

50

75

100

Introgression interval (yrs)

PQ

E (

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 15 Average probability of quasi‐extinction (PQE) of the Florida panther population within 100 years and within 200 years without genetic managementintervention (N) and with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenariosIntrogression strategies include the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Errorbars indicate 5th and 95th percentiles Based on data collected in South Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

50

60

70

80

90

100

110

50

100

150

200

Years

Popu

latio

n si

ze

Figure 16 Average projected Florida panther population trajectories under a geneticintrogression strategy involving the release of 5 female pumas from Texas USA every20 years for the minimum population count (MPC) and motor vehicle mortality(MVM) scenarios Shaded area indicates 5th and 95th percentiles and isrepresentative of confidence intervals for other scenarios Based on data collected inSouth Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Figure 17 Average projected Florida panther population‐level heterozygositiesunder a genetic introgression strategy involving the release of 5 female pumas fromTexas USA every 20 years for the minimum population count (MPC) and motorvehicle mortality (MVM) scenarios Shaded area bounds the 5th and 95th percentilesand is representative of confidence intervals for other scenarios Based on datacollected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 25

this case the migration of just a single male across the freewayto the south resulted in an increase in the number of in-dividuals with admixed ancestry and substantially improvedgenetic diversity of the subpopulation south of the freeway(Riley et al 2014)Our estimate of overall kitten survival (0321plusmn 0056) was

similar to that reported by Hostetler et al (2010) who used13 years of post‐introgression data (0323plusmn 0071) These valuesare lower than kitten survival estimates derived from severalpuma populations in western North America (044ndash091 Loganand Sweanor 2001 Lambert et al 2006 Laundre et al 2007Robinson et al 2008 Ruth et al 2011) When we included onlydata for individuals that had been genetically sampled our es-timate was 15 higher The proportion of admixed kittens thatwere genetically sampled increased as the population expandedwhich could have resulted in higher estimates of kitten survivalbecause admixed kittens survive better than canonical kittensKitten survival was strongly influenced by genetic ancestry withsurvival rates of F1 kittens being almost twice that of canonicalkittens (Fig 5B) Consistent with the findings of Hostetler et al(2010) increases in heterozygosity coincided with increasedkitten survival whereas population density negatively affectedkitten survival Population density often has a strong negativeimpact on survival of young age classes due to infanticide andcannibalism which has been reported from several carnivore

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

50

100

150

0

50

100

150

Introgression interval (yrs)

TQE

(yrs

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 18 Average times until quasi‐extinction (TQE) for the Florida panther population within 100 years and within 200 years without genetic management interventionand with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategiesinclude the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Error bars indicate 5th and 95thpercentiles Based on data collected in South Florida USA 1981ndash2013

Table 5 Average cost (2017 US dollars) of introgression strategies employedover 100 years that involve the introduction of 5 10 or 15 pumas from TexasUSA into the Florida panther population in South Florida USA at an in-creasingly higher frequency Costs of capture (including transportation) quar-antine and post‐introgression monitoring were estimated by Roy McBrideMark Cunningham and Dave Onorato respectively

Frequency Component 5 pumas 10 pumas 15 pumas

Once Capture 25000 50000 75000

Quarantine 5000 10000 15000

Monitoring 10000 20000 30000

Total 40000 80000 120000

Every 80 years Capture 31250 62500 93750

Quarantine 6250 12500 18750

Monitoring 12500 25000 37500

Total 50000 100000 150000

Every 40 years Capture 62500 125000 187500

Quarantine 12500 25000 37500

Monitoring 25000 50000 75000

Total 100000 200000 300000

Every 20 years Capture 125000 250000 375000

Quarantine 25000 50000 75000

Monitoring 50000 100000 150000

Total 200000 400000 600000

Every 10 years Capture 250000 500000 750000

Quarantine 50000 100000 150000

Monitoring 100000 200000 300000

Total 400000 800000 1200000

26 Wildlife Monographs bull 203

species including polar bears (Ursus maritimus Derocher andWiig 1999) black bears (Ursus americanus Garrison et al 2007)and pumas (Logan and Sweanor 2001)Our estimates of sex‐ and age‐specific survival rates of subadult

and adult panthers (Table 3) and the effects of covariates on theserates were similar to those reported by Benson et al (2011) derivedfrom data collected through 2006 Our survival rates for males(065ndash077) were lower than females (078ndash097) for all 3 ageclasses (subadult prime adult and old adult) which is the typicaltrend observed in most puma populations (eg Ruth et al 19982011) However an unhunted puma population occupying afragmented urban landscape in southern California exhibited lowersurvival (0586 females 0525 males Vickers et al 2015) perhapsas a result of severe habitat fragmentation and the lack of extensiveswaths of public land protected in perpetuity Most populations ofpuma in western North America are typically subject to variedlevels of management via regulated hunting which often is biasedtoward the take of males (Cooley et al 2009 Ruth et al 2011)Consequently annual survival estimates from hunted populationsin northeast Washington (0679ndash0733 females 0341 malesRobinson et al 2008) and southeastern Arizona (0685 females0584 males Cunningham et al 2001) were generally lower thanthose we estimated for Florida panthersCause‐specific mortality analysis showed that male panthers

experienced a substantially greater mortality risk from in-traspecific aggression This differs from the pattern observedduring a long‐term study in Arizona where females were at ahigher risk of intraspecific mortality than males (46 of maleand 53 of female mortality Logan and Sweanor 2001)Mortality associated with intraspecific strife has been docu-mented in hunted and unhunted populations (this study Loganand Sweanor 2001 Stoner et al 2010) and its root cause hasbeen difficult to determine (eg population density and preypopulation trends) That being the case finding ways to reducewhat is often a major source of mortality for puma populationscontinues to pose a challenge to managersCanonical panthers were substantially more likely to die from

intraspecific aggression than were admixed panthers This mayat least partly explain the difference in survival between admixed

and canonical panthers Canonical panthers are known to bemore likely to suffer from varied correlates of inbreeding de-pression than admixed animals including atrial septal defects(Johnson et al 2010) and compromised immune systems(Roelke et al 1993) These factors have the potential to affectfitness and perhaps dominance of individuals when engaged inan intraspecific encounter A somewhat similar scenario wasobserved by Hogg et al (2006) in a population of bighorn sheepthat were part of an introgression project Admixed males in thispopulation had a greater probability of paternity than males thatwere less outbred This included coursing males with higherlevels of admixture being markedly more successful at fightingmales that were defending females in estrous (Hogg et al 2006)Heterosis resulting from genetic introgression is expected to be

the strongest in F1 individuals (Shull 1908 Crow 1948 Burkeand Arnold 2001 Waller 2015) and to progressively decline insubsequent generations (Pickup et al 2013 Frankham 2015)Given that admixed (especially F1) panthers survived better thancanonical panthers and that individual heterozygosity improvedsurvival of both sexes and all age classes our data provide evi-dence for heterotic advantage whereby individuals of mixedancestry had higher fitness (Shull 1908 Crow 1948) Our resultsare consistent with findings of other studies that involved in-tentional genetic introgression for conservation purposes Forexample Hogg et al (2006) detected substantial improvementsin survival and reproduction of introgressed bighorn sheep andMadsen et al (1999 2004) reported a dramatic increase in thepopulation of adders after genetic introgressionKnowledge of the persistence of benefits to a population ac-

crued via genetic introgression is essential for determiningwhen additional introgression may be warranted There is adepauperate amount of information regarding this topic and theidentification of factors that may influence persistence of thebenefits of genetic introgression (Tallmon et al 2004 Hedricket al 2014 Whiteley et al 2015) For example in minks(Neovison vison) Thirstrup et al (2014) found that outcrossingled to larger litters in F2 and F3 but this benefit disappeared insubsequent generations In a population of gray wolves Hedricket al (2014) suggested that benefits of genetic introgression may

Table 6 Costs and benefits accumulated over 100 years for genetic introgression at different intervals and with different numbers of pumas from Texas USAintroduced into the Florida panther population in South Florida USA Costs (2017 US dollars) include capture and transportation quarantine and post‐introgression monitoring Benefits are represented as average probability of quasi‐extinction (PQE and 5th and 95th percentiles) and changes (Δ) therein under thescenario using the minimum population count (McBride et al 2008 MPC) and the scenario using the motor vehicle mortality population estimates (McClintocket al 2015 MVM) The table is sorted by percent change in probability of quasi‐extinction under the MPC scenario

Interval (yr) Pumas Cost MPC PQE () ΔMPC PQE () MVM PQE () ΔMVM PQE ()

ndash 0 0 13 (0ndash99) 0 17 (0ndash100) 080 10 100000 12 (0ndash99) 10 (0ndash58) 16 (0ndash100) 5 (0ndash38)80 15 150000 12 (0ndash99) 13 (0ndash65) 15 (0ndash100) 8 (0ndash56)80 5 50000 12 (0ndash99) 14 (0ndash73) 15 (0ndash99) 9 (0ndash41)40 15 300000 10 (0ndash98) 26 (0ndash104) 14 (0ndash99) 13 (0ndash60)40 10 200000 9 (0ndash94) 35 (0ndash183) 13 (0ndash99) 21 (0ndash95)40 5 100000 8 (0ndash89) 39 (0ndash211) 12 (0ndash99) 26 (0ndash124)20 15 600000 8 (0ndash88) 42 (0ndash257) 13 (0ndash99) 24 (0ndash123)20 10 400000 6 (0ndash67) 54 (0ndash318) 10 (0ndash99) 37 (0ndash217)10 15 1200000 6 (0ndash53) 58 (0ndash372) 10 (0ndash99) 40 (0ndash208)20 5 200000 5 (0ndash50) 59 (0ndash338) 9 (0ndash99) 44 (0ndash352)10 10 800000 4 (0ndash16) 69 (0ndash489) 8 (0ndash92) 55 (0ndash401)10 5 400000 4 (0ndash7) 73 (0ndash502) 6 (0ndash71) 63 (0ndash503)

van de Kerk et al bull Florida Panther Population and Genetic Viability 27

wane after 2 or 3 generations The WrightndashFisher model ofgenetic drift predicts a geometric decline in expected hetero-zygosity at the rate of (1 minus 12Ne) per generation where Ne is theeffective population size (Hartl and Clark 1997 Frankham et al2014) Thus it seems logical to assume that the duration of thebenefits of introgression is limited especially in a species per-sisting in a small population such as Florida panthersWe expected Florida panther survival in the most recent years

(2005ndash2013) post‐introgression to differ from that observed in thedecade following the introgression in 1995 because pantherabundance and heterozygosity factors known to influence panthersurvival (Hostetler et al 2010 Benson et al 2011) have changed(Figs 2 and 6) We found that subadult and adult survival rates inrecent years (2005ndash2013) were similar to those in the period im-mediately following introgression (1995ndash2004 Fig 5C) overallkitten survival was similar to estimates of Hostetler et al (2010) Infact the pattern of covariate effects on Florida panther survivalreported by Benson et al (2011) and Hostetler et al (2010) hasbarely changed The negative effects of population density andpositive effects of heterozygosity may counterbalance survival ratesreducing population fluctuations Our result that benefits of geneticrestoration have remained in the population nearly for 5 genera-tions post‐intervention was surprising but also encouraging Pos-sible explanations for this observation may include 1) genetic ero-sion occurs at slower rate than previously thought 2) introducing 5migrants in 1 genetic intervention may have beneficial effects si-milar to those from 1 migrant per generation over multiple gen-erations or 3) other and as yet unknown factors may reduce therate of genetic erosion and subsequent demographic effects Adensity‐dependent effect on kitten survival could also explain whynon‐F1 admixed kittens did not survive substantially better thandid canonical kittens most kittens sampled from 2005ndash2013 wereadmixed and their survival was lower possibly because of a density‐dependent effect as the Florida panther population continuouslygrew during that period Trinkel et al (2010) also found thatpopulation density and inbreeding coefficient interacted to affectdemographic parameters and growth rate of an African lion po-pulation Likewise population density interacted with winterweather to affect the survival and population growth rates in Soaysheep (Ovis aries Milner et al 1999 Coulson et al 2001)

Population Dynamics and PersistenceThe finite deterministic and stochastic growth rates were gt10reflecting that much of the data used in this study were collectedfollowing genetic introgression in 1995 during which time thepopulation increased substantially Because our demographicparameter estimates did not change markedly in comparison tothose of Hostetler et al (2013) the similarity between thepopulation growth estimates from these studies was expectedBut these estimates reflect population growth during thedata‐collection period and do not predict future growth In factestimates of population size reported by McClintock et al(2015) suggest that population growth may already be slowing(Fig 2) A similar pattern was observed in an introduced po-pulation of African lion which increased steadily from 1995until 2001 then fluctuated around the presumed carrying ca-pacity thereafter (Trinkel et al 2010) Several other African lionpopulations that experienced various degrees of recovery

subsequently became relatively stable or exhibited decliningtrends (Bauer et al 2015)Our estimates of quasi‐extinction probabilities were lower than

those reported by Hostetler et al (2013) using a 2‐sex age‐structured matrix population model Lower probabilities ofquasi‐extinction than those reported by the earlier study forcomparable scenarios were likely a consequence of the fact thatthe estimates of abundance (McBride et al 2008) and thus thedensity‐dependent effects on survival of kittens and old panthers(gt10 yr) have changed in recent years Additionally these earlyestimates of the probability of quasi‐extinction and of time toquasi‐extinction did not consider genetic erosion that will in-evitably occur without management interventionUnder the MVM scenario the equilibrium population size was

twice that attained under the MPC scenario The MVM popula-tion estimates are approximately twice the MPCs (Fig 2) as aresult the simulated population under the MVM scenario achieveda higher equilibrium size Probability of quasi‐extinction under theMVM scenario was generally greater than that under the MPCscenario This result is a consequence of the fact that the estimateddensity dependence on kitten survival based on the MPC scenario isstronger than that for theMVM scenario (regression coefficients forabundance for the MPC scenario= minus0068 vs minus0002 for theMVM scenario Appendix B Table B1) Consequently simulatedpopulations under the MPC scenario have a stronger tendency tomove toward the equilibrium population size which inherentlyreduces the quasi‐extinction probability (eg Royama 1992)As expected for long‐lived mammals (Heppell et al 2000 Oli

and Dobson 2003) the population growth rate was proportio-nately most sensitive to changes in survival of prime adultsfollowed by that in younger age classes Global sensitivity ana-lysis in contrast showed that under all scenarios extinctionparameters were particularly sensitive to changes in survival ofFlorida panther kittens This result is a consequence of the highsensitivity of the population growth rate to kitten survival(Fig 9) and a larger standard error of kitten survival (Table 3)Results of our sensitivity analyses indicate that future researchshould focus on acquiring additional information on early lifestages to improve the accuracy and precision of estimates ofkitten survival Our results suggest that demographic variables(or their environmental drivers) that strongly influence extinc-tion risks are not necessarily those with the largest elasticityvalues A study of island fox (Bakker et al 2009) found thatextinction risk was strongly influenced by survival modifierparameters that had low elasticity values

Genetic Management When and How to InterveneThe use of genetic introgression as a management tool has al-ways been controversial and the probable effectiveness it mighthave in preventing the imminent demise of the panther popu-lation was initially debated (Creel 2006 Maehr et al 2006Pimm et al 2006) These debates notwithstanding the pantherpopulation had suffered from genetic morphological and bio-medical correlates of inbreeding before this management in-tervention (Johnson et al 2010 Onorato et al 2010) whereasthe post‐introgression period has been characterized by a sub-stantial reduction in the biomedical correlates of inbreeding andimprovements in demographic vigor and abundance (McBride

28 Wildlife Monographs bull 203

et al 2008 Hostetler et al 2010 2013 Johnson et al 2010Benson et al 2011 McClintock et al 2015) Nevertheless thepopulation remains small and isolated habitat is still being lostand there is no current possibility of natural gene flow betweenthis and other North American puma populations thus therecurrence of inbreeding‐related problems and subsequent po-pulation declines are inevitable The question therefore is notwhether genetic management of the Florida panther populationis needed but when and how it should be implementedWithout genetic introgression probabilities of quasi‐extinc-

tion were substantially greater than those from models thatincorporated periodic genetic introgression events (Fig 15)highlighting the importance of genetic management in smallisolated populations When 5 female pumas are added to theFlorida panther population every 10 years genetic variationincreased substantially under the MPC and MVM scenariosThe IBM predicted that the equillibrium population size wouldbe substantially larger when 5 pumas were released into thepopulation every 10 years than populations without intervention(ie no introgression) Releasing 15 Texas females did not af-ford substantially greater benefit to the equilibrium populationsize than did releasing only 5 Texas females Instead addingmore individuals caused increasingly large fluctuations in po-pulation size due to the numerical increases and density de-pendence (Fig 14) possibly diminishing the benefits associatedwith increased genetic variationCompared with the no‐intervention scenario (ie no genetic

introgression implemented) the introduction of 5 female pumasevery 10 years reduced quasi‐extinction probabilities from 13to 4 and from 17 to 6 for the MPC and MVM scenariosrespectively The benefit of genetic‐introgression strategies de-creased as the interval between successive management inter-ventions increased and no benefit was evident once the intervalreached 80 years (Fig 15) Introgression reduced the averagetime until quasi‐extinction (Fig 18) This somewhat counter‐intuitive result can be explained by the fact that repeated geneticintrogression makes the population more robust over timewhich will reduce the probability of extinction Consequentlyfew simulated population trajectories that fall below the criticalthreshold do so early reducing the mean time to extinctionAlso density dependence has a stabilizing effect on populations(Royama 1992) populations tend toward equilibrium popula-tion sizes which reduces the probability over time of popula-tions falling below quasi‐extinction threshold However timeuntil quasi‐extinction must be interpreted in conjunction withthe probability of quasi‐extinction which is very low for allscenarios with genetic introgression (Fig 15)Cost is a significant impediment to the implementation of a

genetic introgression program because captures relocations andmonitoring efforts before and after introgression are expensive(Edmands 2007 Frankham 2010b 2015 2016) Costs may beacceptable to society if the management actions result in sub-stantial fitness benefits especially if the species of concern ispopular and charismatic (Frankham 2015) We estimated the100‐year cost of introgression strategies modeled here to rangefrom $50000 to $12 million More expensive strategies gen-erally led to larger decreases in the probability of quasi‐extinc-tion but cheaper strategies also substantially improved

population viability (ie reduced quasi‐extinction probabilityTable 6) One of the cheaper strategies (5 pumas released every20 years) which cost an estimated $200000 over 100 yearsreduced the probability of quasi‐extinction by 59 and 44 forthe MPC and MVM scenarios respectively But these pre-liminary cost estimates are based on expenses incurred duringthe 1995 introgression and include costs of capture quarantinetransportation release and post‐release monitoring of femalesonly Although the cost for future genetic interventions wouldlikely be higher than those reported here the relative differencesin cost between alternative intervention strategies should remainapproximately the sameOur ability to simulate genetics and link it to the viability of

the Florida panther population is based on the empirically es-timated relationship between individual heterozygosity andsurvival Such heterozygosity and fitness correlations have beenstudied for decades (David 1998) but have only recently beenused to predict population viability (Benson et al 2016) noneto our knowledge have incorporated cost into their modelingefforts The mechanisms underlying heterozygosity and fitnesscorrelations remain unclear (David 1998 Szulkin et al 2010)and their significance as a proxy for inbreeding depression hasbeen debated (Balloux et al 2004 Miller and Coltman 2014)Nevertheless evidence continues to mount demonstratingpositive relationships between heterozygosity and survival(Coulson et al 1998ab Hansson et al 2004 Silva et al 2005Acevedo‐Whitehouse et al 2006 Jensen et al 2007 Velandoet al 2015) and between heterozygosity and fecundity (Seddonet al 2004 Charpentier et al 2005 Agudo et al 2012 Velandoet al 2015) Although the reported correlations are generallyweak their consequences can be substantial if population growthand persistence parameters are highly sensitive to the affectedvital rates (Velando et al 2015) Compared with scenarios thatignore genetics incorporating the effects of inbreeding on sur-vival increased quasi‐extinction probabilities within a 100‐yeartimeframe from 15 to 13 and from 14 to 17 for theMPC and MVM scenarios respectively These results high-light the importance of incorporating genetics when analyzingthe viability of small isolated populationsAlthough conservation biologists have recognized the im-

portance of genetics in population viability many still fail toincorporate genetic factors into PVA models (Reed et al 2002)Explicit consideration of genetics in PVAs requires long‐termgenetic and demographic data This permits the assessment ofthe functional relationship between genetic variability and de-mographic parameters Population viability analysis softwarepackages such as VORTEX (Lacy 1993 2000) offer a powerfulframework for individual‐based simulations but they often donot adequately capture a speciesrsquo life history or genetics Basedon a thorough review of the importance of genetic factors inwildlife conservation Frankham et al (2014) concluded thatgenetic factors can strongly influence the dynamics and persis-tence of small andor fragmented populations They furthernoted that ldquoMost PVA‐based risk assessments ignore or in-adequately model genetic factorsrdquo and recommended that ldquoPVAshould routinely include realistic inbreeding depressionhelliprdquo(Frankham et al 201457) Our custom‐built IBM permitted usto develop a model that adequately captured the Florida panther

van de Kerk et al bull Florida Panther Population and Genetic Viability 29

life history and we could parameterize the model with our long‐term genetic and demographic dataThe explicit consideration of the effects of inbreeding on

Florida panther population dynamics and persistence representsan important step forward Nonetheless our model remainsimperfect First we neglected the possible effects of habitat lossdetrimental anthropogenic activities climate change the prob-ability of immigration of pumas from western populationsdisease or catastrophes on demographic rates and populationviability Although immigration from puma populations outsideFlorida is possible its probability is very low given the extensivechallenges the anthropogenically altered landscape would posefor such a migrant to make it successfully to South Florida Weomitted mutations from our model because we have no in-formation on mutation rates though we expect that they are lowbased on values reported for other mammals (Kumar and Sub-ramanian 2002) Finally we only considered possible effects ofinbreeding on age‐specific survival rates because there was noevidence that heterozygosity affected female reproduction Lossof heterozygosity can negatively affect reproduction becauseinbred male panthers can suffer from cryptorchidism which canadversely affect their fecundity However given the polygynousmating system we expect the Florida panther population growthis primarily female‐limitedAlthough it is generally accepted that genetic introgression

only temporarily relieves a population from inbreeding depres-sion we know of no cases in which it has been implementedmore than once to conserve a threatened wildlife populationOur study provides an example of how demographic and mo-lecular data can be used within an IBM framework to evaluatethe efficacy of alternative genetic management strategies via theFlorida panther as a case study Our model can be adjusted forand applied to other species and offers great potential as a toolfor genetic management of small isolated wildlife populations

MANAGEMENT IMPLICATIONS

Our results and those of Hostetler et al (2013) and Johnsonet al (2010) provide persuasive evidence that the genetic in-tervention implemented in 1995 prevented the demise of theFlorida panther and restored demographic vigor to the popu-lation The Florida panther population remains small and iso-lated from other puma populations the closest puma populationis located gt1000 km away in Texas and the intervening land-scape is highly developed and fragmented with many interstate(and other high‐traffic) highways and major urban centersTheory suggests that 1 migrant per generation is sufficient tominimize the loss of genetic diversity (eg Mills and Allendorf1996) However the possibility of successful migration of apuma to South Florida followed by successful mating every ~5years is very low considering the distance between Texas andFlorida (Hedrick 1995) and the many risks and barriers thatdispersing pumas face (Beier 1995 Maehr et al 2002 Thatcheret al 2006) Consequently the pantherrsquos long‐term persistencewill likely depend on subsequent timely genetic introgressionsgiven the near‐impossibility of natural gene flow from otherpuma populations To that end our study offers considerableinsights into benefits of different genetic management strategiesand associated costs

Without genetic management the Florida panther populationfaces a substantial risk of quasi‐extinction within 100 years (10ndash46 depending on quasi‐extinction threshold and scenarios)Twenty‐three years have passed since genetic introgression wasimplemented and the population appears healthy as indicatedby measures of genetic variation increases in the populationsize and improvements in age‐specific survival rates Ourfindings suggest that releasing more pumas more frequently doesnot necessarily improve persistence probability because such astrategy would cause larger fluctuations in population size(Fig 14) which negatively affects persistence probability Thusextinction risks faced by inbred populations of large carnivorescan be substantially reduced by releasing approximately 5 im-migrants every 2 decades We suggest that such a managementstrategy be followed only in conjunction with continued long‐term monitoring of the population Our analyses demonstratethat population growth and persistence are highly sensitive tokitten and female (adult and subadult) survival Continuing tocollect data on these demographic variables along with bio-metric and genetic sampling will remain important to pantherrecovery and vigilance in identifying issues associated with in-breeding depression The collection of these monitoring datashould allow managers to detect any demographic or geneticchanges in a timely fashion and respond with genetic in-trogression or other management actions if warrantedRecovery objectives as defined by the USFWS in its current

recovery plan (USFWS 2008) delineate the need for additionalpopulations in the historical range to downlist or delist theFlorida panther If the only extant population of Floridapanthers continued to grow it could serve as a source ofpanthers to be translocated to suitable habitat to seed addi-tional populations Maintaining current information on thegenetic health of the population and precise estimates of itssize will be important in assessing whether it can withstand theremoval of a subset of animals for translocation Although the2017 documentation of a female panther with kittens north ofthe current breeding range is encouraging in terms of thepotential for population expansionmdashgiven that no female hadbeen recorded north of the Caloosahatchee River since 1972mdashthe re‐establishment of separate new populations will mostlikely require translocations by wildlife managers and a lengthysociopolitical processConservation of threatened or endangered species such as the

Florida panther is inherently challenging For panthers and otherlarge carnivores that require vast extents of natural habitat topersist habitat is typically the most limiting factor that will de-termine whether recovery efforts succeed Although geneticmanagement invariably plays a key role in the long‐term persis-tence of the panther population even a healthy population caneventually succumb to the impacts of habitat loss The humanpopulation of Florida continues to be one of the fastest‐growingin the nation the United States Census Bureau currently ranksFlorida as the third most populous Therefore habitat loss isgoing to continue to be a key issue for panthers Diligence towardpreserving panther habitat in its historical range via conservationeasements or other means and trying to keep such areas inter-connected via corridors is a goal stakeholders such as governmentagencies sportsmen non‐governmental organizations ranchers

30 Wildlife Monographs bull 203

and other private landowners must work on collaboratively toachieve

ACKNOWLEDGMENTS

We thank D Shindle D Jennings T OrsquoMeara D Land andC Belden for valuable advice throughout the project We thankS Bass M Criffield M Cunningham D Giardina D JansenA Johnson D Land M Lotz C McBride Rocky McBrideRowdy McBride Roy McBride L Oberhofer S Schulze staffsat Everglades National Park and Big Cypress National Preserveand others for assistance with fieldwork We also thank JAustin M Conroy J Hines J Laake S Mills J Nichols JHines M Sunquist J Fryxell and T Coulson for advice andassistance regarding data analysis and panther biology TheMuseum of Texas Tech University Natural Science ResearchLaboratory provided samples from pumas from west Texas forcomparative genetic analyses M Ben‐David D Land WMcCown E Hellgren and 4 anonymous reviewers providedmany helpful comments on the manuscript We thank NereaUbierna Lopez for completing the Spanish translation of theabstract We are grateful to the Department of ResearchComputing University of Florida for excellent high‐perfor-mance computing services We would like to thank US Fishand Wildlife Service Florida Fish and Wildlife ConservationCommission (FWC) US National Park Service QuantitativeSpatial Ecology Evolution and Environment (QSE3) In-tegrative Graduate Education and Research Traineeship(IGERT) program funded by the National Science Foundationand the University of Florida for financially supporting thisstudy Funding for panther research conducted by the FWC issupported predominantly via donations accrued with vehicleregistrations for ldquoProtect the Pantherrdquo license plates

LITERATURE CITEDAcevedo‐Whitehouse K T R Spraker E Lyons S R Melin F Gulland RL Delong and W Amos 2006 Contrasting effects of heterozygosity onsurvival and hookworm resistance in California sea lion pups MolecularEcology 151973ndash1982

Adams J R L M Vucetich P W Hedrick R O Peterson and J AVucetich 2011 Genomic sweep and potential genetic rescue during limitingenvironmental conditions in an isolated wolf population Proceedings of theRoyal Society B Biological Sciences 2783336ndash3344

Agresti A 2013 Categorical data analysis Third edition John Wiley amp SonsHoboken New Jersey USA

Agudo R M Carrete M Alcaide C Rico F Hiraldo and J A Donaacutezar2012 Genetic diversity at neutral and adaptive loci determines individualfitness in a long‐lived territorial bird Proceedings of the Royal Society BBiological Sciences 2793241ndash3249

Albeke S E N P Nibbelink and M Ben‐David 2015 Modeling behavior bycoastal river otter (Lontra canadensis) in response to prey availability in PrinceWilliam Sound Alaska a spatially‐explicit individual‐based approach PLOSOne 10e0126208

Alho J S K Vaumllimaumlki and J Merilauml 2010 Rhh an R extension for estimatingmultilocus heterozygosity and heterozygosity‐heterozygosity correlationMolecular Ecology Resources 10720ndash722

Alvarez K 1993 Twilight of the panther Myakka River Publishing SarasotaFlorida USA

Aparicio J M J Ortego and P J Cordero 2006 What should we weigh toestimate heterozygosity alleles or loci Molecular Ecology 154659ndash4665

Bakker V J D F Doak G W Roemer D K Garcelon T J Coonan S AMorrison C Lynch K Ralls and R Shaw 2009 Incorporating ecologicaldrivers and uncertainty into a demographic population viability analysis for theisland fox Ecological Monographs 7977ndash108

Balloux F W Amos and T Coulson 2004 Does heterozygosity estimateinbreeding in real populations Molecular Ecology 133021ndash3031

Barone M A M E Roelke J Howard J L Brown A E Anderson and DE Wildt 1994 Reproductive characteristics of male Florida panthersmdashcomparative studies from Florida Texas Colorado Latin‐America andNorth‐American Zoos Journal of Mammalogy 75150ndash162

Bateson Z W P O Dunn S D Hull A E Henschen J A Johnson and LA Whittingham 2014 Genetic restoration of a threatened population ofgreater prairie‐chickens Biological Conservation 17412ndash19

Bauer H G Chapron K Nowell P Henschel P Funston L T B HunterD W Macdonald and C Packer 2015 Lion (Panthera leo) populations aredeclining rapidly across Africa except in intensively managed areas Pro-ceedings of National Academy of Sciences of the United States of America11214894ndash14899

Beier P 1995 Dispersal of juvenile cougars in fragmented habitat Journal ofWildlife Management 59228ndash237

Benson J F J A Hostetler D P Onorato W E Johnson M E Roelke SJ OrsquoBrien D Jansen and M K Oli 2011 Intentional genetic introgressioninfluences survival of adults and subadults in a small inbred felid populationJournal of Animal Ecology 80958ndash967

Benson J F P J Mahoney J A Sikich L E K Serieys J P Pollinger H BErnest and S P D Riley 2016 Interactions between demography geneticsand landscape connectivity increase extinction probability for a small popu-lation of large carnivores in a major metropolitan area Proceedings of theRoyal Society B Biological Sciences 28320160957

Beverly F 1874 The Florida panther Forest and Stream 3290Boyce M S C V Haridas C T Lee and the NCEAS Stochastic Demo-graphy Working Group 2006 Demography in an increasingly variable worldTrends in Ecology amp Evolution 21141ndash148

Brook B W J J OrsquoGrady A P Chapman M A Burgman H R Akcakayaand R Frankham 2000 Predictive accuracy of population viability analysis inconservation biology Nature 404385ndash387

Brys R and H Jacquemyn 2016 Severe outbreeding and inbreeding de-pression maintain mating system differentiation in Epipactis (Orchidaceae)Journal of Evolutionary Biology 29352ndash359

Burke J M and M L Arnold 2001 Genetics and the fitness of hybridsAnnual Review of Genetics 3531ndash52

Burnham K P 1993 A theory for combined analysis of ring recovery andrecapture data Pages 199ndash213 in J‐D Lebreton and P M North editorsMarked individuals in the study of bird population Springer‐Verlag NewYork New York USA

Burnham K P and D R Anderson 2002 Model selection and multimodelinference a practical information‐theoretic approach Second edition SpringerVerlag New York New York USA

Caswell H 2001 Matrix population models construction analysis and in-terpretation Sinauer Associates Sunderland Massachusetts USA

Charpentier M J M Setchell F Prugnolle L A Knapp E J Wickings PPeignot and M Hossaert‐McKey 2005 Genetic diversity and reproductivesuccess in mandrills (Mandrillus sphinx) Proceedings of the NationalAcademy of Sciences of the United States of America 10216723ndash16728

Cooch E and G C White editors 2018 Program MARK a gentle in-troduction lthttpwwwphidotorgsoftwaremarkdocsbookgt Accessed 7Apr 2016

Cooley H S R B Wielgus G M Koehler H S Robinson and B TMaletzke 2009 Does hunting regulate cougar populations A test of thecompensatory mortality hypothesis Ecology 902913ndash2921

Coulson T E A Catchpole S D Albon B J Morgan J M Pemberton TH Clutton‐Brock M J Crawley and B T Grenfell 2001 Age sex densitywinter weather and population crashes in Soay sheep Science 2921528ndash1531

Coulson T J Pemberton S Albon M Beaumont T Marshall J Slate FGuinness and T Clutton‐Brock 1998a Microsatellites reveal heterosis in reddeer Proceedings of the Royal Society B Biological Sciences 265489ndash495

Coulson T N S D Albon J M Pemberton J Slate F E Guinness and TH Clutton‐Brock 1998b Genotype by environment interactions in wintersurvival in red deer Journal of Animal Ecology 67434ndash445

Cox D 1972 Regression models and life‐tables Journal of the Royal StatisticalSociety Series B Statistical Methodology 34187ndash220

Creel S 2006 Recovery of the Florida panthermdashgenetic rescue demographic rescueor both Response to Pimm et al (2006) Animal Conservation 9125ndash126

Crnokrak P and D A Roff 1999 Inbreeding depression in the wild Heredity83260ndash270

Crow J 1948 Alternative hypotheses of hybrid vigor Genetics 33477ndash487

van de Kerk et al bull Florida Panther Population and Genetic Viability 31

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 12: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

demographic stochasticity by simulating the fates of individualsas described in Caswell (2001)Population projections using matrix models necessitate a

vector of initial sex‐ and age‐specific abundance Because suchdata are currently unavailable for the Florida panther we esti-mated it based on the stable sex and age distribution and MPCin 2013 Using point estimates of the demographic parameterswe estimated the stable sex and age distribution for the 2‐sexdeterministic and density‐independent population projectionmatrix We then multiplied the stable sex and age distributionby the MPC in 2013 to get the starting population vector n(0)We projected the population over time as n(t+ 1)=A(tn)n(t)where A(tn) indicates the time‐specific density‐dependentpopulation projection matrix (Caswell 2001)We had 2 indices of abundance available the MPC (McBride

et al 2008) and population size estimated using motor vehiclecollision mortalities (MVM McClintock et al 2015) These 2indices are substantially different because MPC does not ac-count for imperfect detection whereas MVM accounts forimperfect detection and provides an estimate of population size(Fig 2) Therefore we examined 2 scenarios one using densitydependence based on MPC and the other using density de-pendence based on MVM We ran 1000 bootstraps of para-meter values and 1000 simulations per bootstrap for each sce-nario We projected population trajectories over 200 years andthen estimated population growth rates probabilities ofquasi‐extinction and time until quasi‐extinction and comparedthese with the results obtained from the IBM (see below) Weran scenarios with a quasi‐extinction threshold of 10 and 30panthers We used the mean of probabilities of quasi‐extinction(PQE) across simulations as the point estimate for PQE and the5th and 95th percentiles across simulations as the 90 con-fidence interval Because the confidence intervals werequantile‐based but the point estimates were not it is possiblewith skewed distributions for the point estimates to be outsidethe confidence interval We implemented the matrix

population models in the R computing environment (R CoreTeam 2016)

IBMs and PVABecause of well‐developed theory ease of implementation andthe modeling flexibility that they offer matrix populationmodels have become the model of choice for investigating thedynamics and persistence of age‐ and stage‐structured popula-tions (Caswell 2001) However it is difficult to incorporateattributes of individuals such as genetics and behavior usingmatrix models Quantifying the effects of genetic erosion on theFlorida panther population dynamics and persistence andevaluating benefits and costs of alternative genetic managementscenarios were important goals of our study Because IBMs relyon a bottom‐up approach that begins by explicitly consideringindividuals (McLane et al 2011 Railsback and Grimm 2011Albeke et al 2015) it is possible to represent genetic attributesof each individual and to track genetic variation over time Weused IBMs as the primary modeling framework in this studybecause they allowed for explicit consideration of genetics bothat individual and population levels and population viabilityassessment under alternative genetic management scenariosWe developed an IBM (Grimm 1999 Grimm and Railsback

2005 McLane et al 2011 Railsback and Grimm 2011 Albekeet al 2015) tailored to the life history of the Florida panther(Fig 3) We simulated every individual from birth until deathbased on a set of rules describing fates (eg birth and death) andattributes (eg genetics) of individuals at each annual time stepAs in other IBMs population‐level processes (eg populationsize and genetic variation) emerged from fates of and interac-tions among individuals (Grimm 1999 Railsback and Grimm2011) We developed an overview design concepts and details(ODD) protocol (Grimm et al 2006 2010) to describe ourIBM (Appendix C available online in Supporting Information)and provide pseudocodes for individual‐based simulations(Appendix D available online in Supporting Information)

For each panther at time = t

Kitten (lt1 yr)

Reproduce

Survive

Kitten(s) born

Kitten

Sex and age class

Subadult male

Prime adult male

Old adult male

Subadult female

Prime adult female

Old adult female

Age one year

No Male

Female

Yes

No

Yes

t=t+1

Yes

lt35 yrs

35-10 yrs

gt10 yrs

lt25 yrs

25-10 yrs

gt10 yrs

Figure 3 Schematic representation of Florida panther life history We tailored the individual‐based population model to represent the life‐history patterndepicted here

14 Wildlife Monographs bull 203

Dynamics and persistence of populations are influenced bymany factors such as demographic and environmental sto-chasticity and density dependence these effects and parametricuncertainty must be considered in population models whenpossible (Caswell 2001 Boyce et al 2006 Bakker et al 2009Hostetler et al 2013) Because by definition IBMs simulate thelife history of individuals in a population they inherently in-corporate the effect of demographic stochasticity Our analysesrevealed that survival of kittens subadults and adults and theprobability of reproduction varied over time so we incorporatedenvironmental stochasticity in the model for these variablesfollowing Hostetler et al (2013) Likewise we found evidenceof density‐dependent effects on kitten survival Because theMPC might be an underestimate of population size (McBrideet al 2008) we performed simulations using both the MPC andMVM abundance indicesTo account for model selection and parametric uncertainty we

used a Monte Carlo simulation approach For each bootstrapsample we selected a model based on the AIC (or QAICc)weights We then sampled the intercept and slope for kittensurvival (on the logit scale) subadult and adult survival (on thelogit scale with log‐hazard effect sizes) and probability of re-production (on the complementary logndashlog scale Appendix B)from multivariate normal distributions with mean vectors equalto the estimated parameters and variance‐covariance matricesequal to the sampling variance‐covariance matrices We thentransformed the results to nominal scale and converted the es-timates to age‐specific survival and reproduction parameters asdescribed in Hostetler et al (2013)We ran 1000 simulations for 1000 parametric bootstraps for

200 years for the MPC and the MVM scenario for both thestochastic 2‐sex matrix‐population model and the IBM Wetracked the population size at each time step and estimatedquasi‐extinction probabilities and times based on the populationtrajectory We considered the population to be quasi‐extinct ifindividuals of only 1 sex remained or if the population size fellbelow critical thresholds set at 10 and 30 individuals We alsoestimated expected time to quasi‐extinction for both thresholdsdefined as the mean and median time until a population be-comes quasi‐extinct conditioned on quasi‐extinction We im-plemented the IBM using the RNetLogo package version 10ndash2(Thiele et al 2012 Thiele 2014) as an interface for the IBMplatform NetLogo (Wilensky 1999)

Sensitivity AnalysisAn important component of demographic analysis is sensitivityanalysis which is the quantification of the sensitivity of amodelrsquos outcome to changes in the input parameters (Caswell2001 Bakker et al 2009 Thiele et al 2014) Using an IBM weperformed global sensitivity analyses for 2 (ie MPC andMVM) density‐dependent scenarios to assess how sensitivequasi‐extinction times and probabilities were to changes (anduncertainties) in the estimated demographic rates We usedLatin hypercube sampling to sample parameters from the entireparameter space (Marino et al 2008 Thiele et al 2014) Wesampled intercept and slope for kitten survival (on logit scale)subadult and adult survival (on logit scale with log‐hazard effectsizes) and probability of reproduction (on complementary

logndashlog scale) from normal distributions We sampled theprobability distribution for the number of kittens producedannually (on the real scale) from a Dirichlet distribution themultivariate generalization of the beta distribution (Kotz et al2000) We sampled 1000 different parameter sets and ran 1000simulations for each scenario We calculated the probability ofquasi‐extinction within 200 years for each density‐dependencescenario (MPC or MVM) using a quasi‐extinction threshold of10 individuals and then calculated the partial rank correlationcoefficient between each of those and each parameter trans-formed to the real scale (Bakker et al 2009 Hostetler et al2013) Partial rank correlation coefficients quantify the sensi-tivity of the probability of quasi‐extinction to input variables(ie survival and reproductive parameters) with higher absolutevalues indicating stronger influence of that variable on quasi‐extinction probabilities

Genetic ManagementOne of our goals was to estimate the benefits (improvements ingenetic variation demographic parameters and populationgrowth and persistence parameters) and costs (financial costs ofcapture transportation quarantine release and monitoring ofintroduced panthers) of genetic management To achieve thisobjective we extended the IBM to include a genetic componentTo simulate individual‐ and population‐level heterozygosity overtime we assigned each panther alleles for 16 microsatellite locibased on the empirical allele frequency in the population (esti-mated from genetic samples collected during 2008ndash2015) whenwe initialized the model (Pierson et al 2015) At each time stepwe simulated Mendelian inheritance for kittens produced suchthat each kitten randomly inherited alleles from its mother andfrom a male panther randomly chosen to have putatively siredthe litter We did not explicitly force inbreeding to occur butthe probability of inbreeding naturally increased as the popula-tion size decreasedIntrogression strategiesmdashWe modeled genetic introgression as a

result of the release of 2‐year‐old female pumas from Texas intothe simulated Florida panther population For the geneticintrogression implemented in 1995 Texas females between 15and 25 years of age were released because females at that agehave lower mortality rates (Benson et al 2011) higherreproductive potential and also because it is much easier todetect with certainty the successful reproduction by females thanby males The ability to more closely monitor reproduction andthe birth of kittens is an important consideration in reducing theprobability of a genomic sweep of Texas alleles into the Floridapanther population Wildlife managers removed the last 2surviving Texas females from the wild population in Florida in2003 to reduce the possibility of genomic sweep To simulatethis strategy in the model we removed any Texas females thatwere still alive 5 years after each introgression event Weassigned Texas females alleles based on the allele frequency ofthe Texas population (based on genotypes from 48 samplescollected in west Texas that was inclusive of the pumas releasedin South Florida in 1995) when they entered the Florida pantherpopulation We could not estimate demographic rates separatelyfor the released Texas females because of small sample size (only8 females were released in 1995) therefore we assumed that

van de Kerk et al bull Florida Panther Population and Genetic Viability 15

survival and reproductive rates and the effects of covariates onthese rates for the released Texas females were identical to thoseof female Florida panthersThe impact of genetic introgression depends on the frequency

of introgression events and the number of individuals introducedat each attempt Thus we defined introgression strategies basedon the number of Texas females to be released (0 5 10 or 15)and the interval between genetic introgressions (10 20 40 and80 yr) for a total of 13 strategies We simulated each manage-ment strategy with density dependence estimated using theMPC and MVM abundance indices

We sampled 1000 parameter sets and for each of these setswe ran all introgression strategies 1000 times for 200 years Weapplied the same parametric uncertainty and environmentalstochasticity across all management scenarios to allow for bettercomparison of introgression strategies At each time step werecorded the projected population size and average individualheterozygosity We calculated the heterozygosity for each in-dividual at birth based on the alleles it inherited This individualheterozygosity then informed the survival rate of that individualbased on the empirically estimated relationship between in-dividual heterozygosity and age‐specific survival rates Hetero-zygosity did not change throughout an individualrsquos lifetime butits effect on survival changed as individuals aged because theeffect of individual heterozygosity on survival rate differedamong age classes Survival rates of genetically sampled kittenswere biased high because some kittens were sampled later in lifewhen they had already survived for some time To correct forthis bias we adjusted kitten survival rates predicted by theheterozygosity model proportionally From the population tra-jectories we estimated the probability of quasi‐extinction (de-fined as the probability that the population will fall below 10 or30 individuals or that individuals of only 1 sex will remain)

Cost‐benefit analysismdashExperts estimated financial costs ofintrogression based on their experience with the geneticintrogression executed in 1995 and we adjusted estimates tocurrent costs to account for inflation (R T McBride RancherrsquosSupply Incorporated M Cunningham and D P OnoratoFlorida Fish and Wildlife Conservation Commission personalcommunication) Costs included capturing and transportingfemale pumas from the Texas source population caring for thepumas while they were in quarantine and post‐introgressionmonitoring To compare the financial costs of the differentscenarios we also calculated the average costs incurred over

Table 2 Sample size (n) number of alleles (Na) allelic richness (AR) observedheterozygosity (Ho) and expected heterozygosity (He) at 16 microsatellite loci inradio‐collared Florida panthers sampled from 1981 to 2013 in South FloridaUSA We calculated these values from a subset (n= 137) of the samples used inthe analyses and they include only adult and subadult panthers We excludedkittens because they can result in an overrepresentation of certain alleles

Locus n Na AR Ho He

FCA090 129 5 50 0333 0401FCA133 136 3 30 0618 0608FCA243 136 5 50 0419 0487F124 137 7 69 0708 0729F37 129 4 40 0395 0397FCA075 131 5 50 0534 0598FCA559 133 5 50 0496 0503FCA057 134 6 59 0679 0624FCA081 134 7 69 0209 0206FCA566 130 6 60 0462 0475F42 133 10 100 0737 0766FCA043 136 5 49 0596 0588FCA161 132 6 60 0500 0515FCA293 132 4 40 0614 0624FCA369 131 6 60 0573 0567FCA668 133 7 70 0406 0462Mean 1329 57 57 0517 0535

Figure 4 Proportional membership (q) of radio‐collared Florida panthers (n= 218) and pumas from Texas USA (n= 7) in 2 clusters identified by STRUCTUREEach individual animal is represented by a separate vertical bar Yellow indicates canonical panther ancestry and blue represents admixed ancestry The admixedancestry of the Texas females and F1 generation panthers that were radiocollared are highlighted red and green respectively to demarcate those groups The pre‐introgression period is inclusive of panthers radiocollared from 10 February 1981 to 6 March 1996 The post‐introgression era inclusive of the F1 panthers includesanimals radiocollared from 4 March 1997 to 18 February 2015 in South Florida USA

16 Wildlife Monographs bull 203

100 years assuming that the population would be managedaccording to a particular strategy Our goal with theseexploratory analyses was to evaluate the relative costs andbenefits of alternative management scenarios

RESULTS

Genetic VariationWe assessed genetic variation at 16 microsatellite loci for 137DNA samples collected from adult and subadult panthers(1981ndash2013) that were captured and subsequently radiocollaredThe number of alleles at these loci ranged from 3ndash10 and allelicrichness averaged 57 (Table 2) Average expected hetero-zygosity for this sample was 0535 and average observed het-erozygosity was 0517 (Table 2)We determined individual heterozygosity (1 minusHL) and an-

cestry for the complete sample of panther genotypes (n= 503)collected from 1981 to 2015 for use as covariates in subsequentanalyses Average individual heterozygosity was 0559 andranged from 0ndash0980 Our STRUCTURE analysis providedstrong support for K= 2 clusters based on the probability of thedata (log Pr(D)|K) and ΔK in STRUCTURE HARVESTERBased on this result we categorized the ancestry of each pantheras either canonical (n= 123) or admixed (n= 380) using valuesof proportional membership (q Fig 4) The average individualheterozygosity of canonical panthers was 0386plusmn 0012 (SE) foradmixed panthers it was 0615plusmn 0007

Survival and Cause‐Specific Mortality RatesOf the 395 kittens that were PIT‐tagged and released 55 werelater encountered alive 15 were recovered dead and 85 werepart of failed litters (Table 1) We radio‐tracked 209 subadult oradult panthers for a total of 244195 panther‐days and docu-mented 126 mortalities (Table 1)Survival depended strongly on age and kittens had the lowest

overall rate (032plusmn 009 Fig 5A Table 3) The model‐averagedkitten survival was 047plusmn 009 when we included only geneti-cally sampled individuals in the analysis (Table 3) as statedabove this estimate is biased high because many kittenswere genetically sampled when they were recaptured alive orrecovered dead at older ages We found no evidence for sex‐specific differences in kitten survival but females survived betterthan males in all older age classes For females subadults hadthe highest survival rates followed by prime adults and oldadults For males prime adults had the highest survival ratefollowed by subadults and old adults (Fig 5A and Table 3)For panthers of all ages there was substantial evidence that

ancestry affected survival with F1 admixed panthers of all ageshaving the highest rates and canonical individuals having thelowest (Fig 5B) The combined AIC weights of models thatincluded an ancestry covariate were 0880 for kittens and 0992for older panthers The survival rates of subadult prime‐adultand old‐adult panthers in the period immediately after the in-trogression (1995ndash2004) were similar to those in more recentyears (2005ndash2013 Fig 5C)After genetic introgression individual panther heterozygosity

increased (Fig 6) and varied among ancestry categories individualheterozygosity was the highest for F1 panthers (078plusmn 001)

A

B

C

Figure 5 Overall age‐ and sex‐specific survival rates (plusmnSE A) age‐ and sex‐specific survival rates (plusmnSE) for Florida panthers of canonical F1 admixed andother admixed ancestry (B) and age‐ and sex‐specific survival estimates (plusmnSE)for subadult and adult Florida panthers in the period immediately post‐introgression (1995ndash2004) and more recently (2005ndash2013 C) in SouthFlorida USA

Table 3 Model‐averaged estimates (plusmnSE) of demographic parameters for theFlorida panther obtained using data collected during 1981ndash2013 in SouthFlorida USA

Parameter Sex Age class Estimate

Survival Both Kitten 032plusmn 009a

Female Subadult 097plusmn 002Prime adult 086+ 003Old adult 078plusmn 009

Male Subadult 066plusmn 006Prime adult 077plusmn 005Old adult 065plusmn 010

Probability of reproduction Female Subadult 035plusmn 008Prime adult 050plusmn 005Old adult 025plusmn 006

Kittens produced annually Female Subadult 280plusmn 005Prime adult 267plusmn 001Old adult 228plusmn 013

a Overall kitten survival (based on all kittens in our data set including thosethat were not genetically sampled) Estimate of survival of geneticallysampled kittens was 047plusmn 009 this estimate is biased high because manykittens were genetically sampled when they were recaptured alive or re-covered dead at older ages

van de Kerk et al bull Florida Panther Population and Genetic Viability 17

followed by those for non‐F1 admixed (062plusmn 001) and canonical(039plusmn 001) panthers Individual heterozygosity positively affectedsurvival rates of kittens (slope parameter η= 1455 95CI= 0505ndash2405) Individual heterozygosity also positively af-fected survival of subadult and adult panthers but the evidence forthis effect was weaker because the confidence interval for the ha-zard ratio overlapped 10 (hazard ratio exp(ɩ)= 0311 95CI= 0092ndash1054 Table 4 and Fig 7)The MPC increased after genetic introgression (Fig 2) This

abundance index was also included in top‐ranking modelsindicating that population density affected survival (Table 4)Abundance reduced the survival of kittens (slope parameterη= minus0035 95 CI= minus0049 to minus0021) evidence was weak forthe effect of abundance on survival of subadult and adult panthers(hazard ratio exp(ɩ)= 1002 95 CI= 0995ndash1010 Fig 7) Re-gression coefficients associated with the effect of abundance andindividual heterozygosity on demographic parameters are presentedin Table B1 (available online in Supporting Information)The most frequently observed causes of death for radio‐

collared panthers were vehicle collision and intraspecific ag-gression Cause‐specific mortality analyses revealed thatmortality rates among possible causes were generally similar forthe 2 sexes but that males were more likely to die from in-traspecific aggression than were females (Fig 8A) Male mor-tality from intraspecific aggression was higher for subadult andold adult panthers than for prime adults (Fig 8B) For bothmales and females canonical panthers were more likely to diefrom intraspecific aggression than were admixed panthers(Fig 8C)

Reproductive ParametersOf 101 radio‐collared females 67 reproduced at least once duringour monitoring we used these individuals to assess the annualprobability of reproduction The top‐ranked model for the annual

probability of reproduction included age effect only suggesting thatadding ancestry or abundance did not improve model parsimony(Table 4) Model‐averaged annual probabilities of reproductiondiffered between female subadults (035plusmn 008) prime adults(050plusmn 005) and older adults (025plusmn 006)The most parsimonious model for the number of kittens

produced annually by females that produced at least 1 litter (ieconditional on reproduction) included effects of age ancestryand abundance a model that included only the effect of age wasalmost equally supported (Table 4) The model‐averagednumber of kittens produced annually conditional on re-production was 280plusmn 075 for subadults 267plusmn 043 for primeadults and 228plusmn 083 for old adults Confidence intervals forthe slope parameter (β) overlapped 0 for both abundance(β= 0010 95 CI= minus0001 to 0021) and ancestry (β= minus073795 CI= minus1637 to 0162)

Population Dynamics and PersistenceDeterministic and stochastic demographymdashThe deterministic (λ)

and stochastic (λs) annual population growth rate calculated usingthe matrix population model were 106 (5th and 95th percentiles099ndash114) and 104 (072ndash141) respectively The generation timewas 47 years A female starting in age class one (kitten) wasexpected to live another 58 years and produce 10 kittens onaverage during her lifetime Overall λs was proportionately(elasticity) most sensitive to changes in female prime adultsurvival on the absolute scale (sensitivity) however it was mostsensitive to changes in kitten survival (Fig 9) Populationtrajectories probabilities of quasi‐extinction and times untilquasi‐extinction estimated using the matrix population modeland IBM for comparable scenarios were nearly identical for acritical quasi‐extinction threshold of 10 panthers (Figs 10 and 11)and for a critical quasi‐extinction threshold of 30 panthers(Appendix E available online in Supporting Information)Minimum population count scenariomdashUnder the MPC scenario

(ie density dependence estimated using the minimumpopulation count data) with a critical threshold of 10panthers the population size reached 99 (72ndash104) and 100(77ndash105) at 200 years (Fig 10) as predicted by the IBM and thematrix model respectively The cumulative quasi‐extinctionprobabilities within 100 years based on the MPC scenario were15 (IBM 0ndash48) and 30 (matrix model 0ndash76) Theseprobabilities increased to 25 (IBM 0ndash114) and 38 (matrixmodel 0ndash154) by year 200 (Fig 10) The mean times untilquasi‐extinction were 15 years (IBM 0ndash67) and 15 years (matrixmodel 0ndash72) conditioned on going quasi‐extinct within 100years Expected times until quasi‐extinction conditioned onquasi‐extinction within 200 years were higher 34 years (IBM0ndash137) and 32 years (matrix model 0ndash134 Fig 10)Motor vehicle mortality scenariomdashUnder the MVM scenario

(ie density dependence estimated using estimates of pantherpopulation size from McClintock et al 2015) with a criticalthreshold of 10 panthers the projected population reached 188(142ndash217) and 190 (143ndash219) at 200 years as predicted by theIBM and the matrix model respectively (Fig 11) Thecumulative quasi‐extinction probabilities within 100 years were14 (IBM 0ndash08) and 13 (matrix model 0ndash06) Theseprobabilities increased to 20 (IBM 0ndash17) and 19 (matrix

000

025

050

075

100

1995 2000 2005 2010Year of birth

Indi

vidu

al h

eter

ozyg

osity

Figure 6 Temporal changes in the individual heterozygosity of Floridapanthers born during the period 1995ndash2013 in South Florida USA Pointsare measurements solid line is linear regression shaded area is 95 confidenceinterval of slope Slope= 0006plusmn 0001 R2= 009

18 Wildlife Monographs bull 203

model 0ndash16) within year 200 (Fig 11) The mean times until quasi‐extinction based on the MVM scenario were 5 years (IBM 0ndash61)and 6 years (matrix model 0ndash64) conditioned on going quasi‐extinctwithin 100 years Expected times until quasi‐extinction conditionedon quasi‐extinction within 200 years were 11 years (IBM 0ndash113)and 11 years (matrix model 0ndash112 Fig 11)

Sensitivity of extinction probability to demographic parametersmdashThe partial rank correlation coefficients between quasi‐extinctionprobabilities and demographic parameters were negative anincrease in any of the demographic parameters decreased thequasi‐extinction probability Probabilities of quasi‐extinction within200 years were most sensitive to changes in kitten survival for bothscenarios (Fig 12) followed by survival of subadult females in theMVM scenarios and survival of prime adult females in the MPCscenario The probability of reproduction of prime adult femaleshad the third‐largest partial rank correlation coefficient with quasi‐extinction probability for both scenarios

Benefits and Costs of Genetic ManagementMinimum population count scenariomdashThe MPC scenario with

a critical threshold of 10 panthers predicted that withoutmanagement intervention average individual heterozygositywould decrease reaching 057 (049ndash064) at 100 years and053 (042ndash062) at 200 years (Fig 13) The population woulddecrease slightly reaching an average of 79 (41ndash99) at 100 yearsand 76 (43ndash98) at 200 years (Fig 14) The probabilities of quasi‐extinction under the no‐intervention MPC scenario when weconsidered the effects of genetic erosion were 13 (0ndash99) at 100years and 23 (0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with

the translocation of 5 Texas females average individualheterozygosity under the MPC scenario increased and thenstabilized at 068 (064ndash072) at 100 years and remained atthat level at 200 years (Fig 13) The population reached anaverage of 88 (62ndash104) at 100 years and stayed the same at200 years (Fig 14) The probabilities of quasi‐extinctionunder this introgression strategy were 6 (0ndash53) within 100years and 7 (0ndash82) within 200 years (Fig 15) Results ofanalyses using a critical threshold of 30 panthers revealed asimilar pattern of relative benefits However the prob-abilities of quasi‐extinction were substantially higher (ge45)across all scenarios (Appendix E)

Motor vehicle mortality scenariomdashWithout managementintervention the average individual heterozygosity predictedby the MVM scenario and a critical threshold of 10 panthersdecreased to 060 (046ndash069) at 100 years and to 059 (039ndash070) at 200 years (Fig 13) The population increased slightlyand reached an equilibrium at 154 individuals (46ndash217) at 100years and 157 (57ndash218) at 200 years (Fig 14) The probabilitiesof quasi‐extinction were 17 (0ndash100) at 100 years and 22(0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with 5

female pumas from Texas average individual heterozygosityincreased slightly reaching 067 (060ndash073) at 100 years and068 (059ndash075) at 200 years (Fig 13) Under this strategy thepopulation reached an average of 164 individuals (44ndash226) at

Table 4 Model comparison results testing for the effects of several covariateson survival of all kittens survival of genetically sampled kittens survival ofsubadult and adult Florida panthers probability of reproduction and kittensproduced annually for Florida panthers sampled from 1981ndash2013 in SouthFlorida USA

Model Parameters ΔAICa Weight

Kitten survival (all data)Base+ F1b+ abundancec 10 000 0988Base+ abundance 9 1018 0006Base+ F1 9 1035 0005Base 8 2711 lt0001Base+ sexd 9 2912 lt0001Base+ litter sizee 9 2914 lt0001

Kitten survival (genetically sampled kittens only)Base+ F1+ abundance 10 000 0541Base+ F1CanAdmf+ abundance 11 099 0329Base+Hetg+ abundance 10 310 0115Base+CanAdmh+ abundance 10 791 0010Base+ abundance 9 944 0005Base+ F1 9 1638 lt0001Base+ F1CanAdm 10 1837 lt0001Base+Het 9 2872 lt0001Base 8 3296 lt0001Base+CanAdm 9 3348 lt0001

Subadult and adult survivalBase+ F1CanAdm+ abundance 7 000 0360Base+ F1 5 053 0276Base+ F1CanAdm 6 105 0213Base+ F1+ abundance 6 222 0119Base+CanAdm+ abundance 6 575 0020Base+CanAdm 5 908 0004Base+Het 5 964 0003Base+Het+ abundance 6 984 0003Base 4 1118 0001Base+ abundance 5 1278 0001

Probability of reproductionAge 3 000 0242Age+CanAdm 4 012 0228Age+ abundance 4 173 0102Age+ F1 4 190 0094Age+CanAdm+ abundance 5 216 0082Age+Het 4 219 0081Age+ F1CanAdm 5 232 0076Age+ F1+ abundance 5 372 0038Age+Het+ abundance 5 399 0033Age+ F1CanAdm+ abundance 6 444 0026

Kittens produced annuallyAge+ F1+ abundance 5 000 0194Age 3 060 0144Age+ abundance 4 061 0143Age+ F1 4 097 0120Age+CanAdm 4 139 0097Age+ F1CanAdm+ abundance 6 196 0073Age+ F1CanAdm 5 231 0061Age+CanAdm+ abundance 5 238 0059Age+Het+ abundance 5 250 0056Age+Het 4 259 0053

a Akaikersquos information criterion (AIC) for kitten survival models was adjustedfor small sample size and overdispersion (QAICc)

b F1 divides Florida panthers into 2 groups F1 admixed and all other panthersc Index of Florida panther abundanced Sex of an individual Florida panther (male or female)e Size of the litter in which a Florida panther kitten was bornf F1CanAdm divides panthers into 3 groups F1 admixed canonical andother admixed panthers

g Het is the individual heterozygosityh CanAdm divides panthers into 2 groups canonical and admixed (includingF1 admixed) panthers

van de Kerk et al bull Florida Panther Population and Genetic Viability 19

100 years and 170 (67ndash231) at 200 years based on the MVMscenario (Fig 14) The probabilities of quasi‐extinction underthis introgression strategy were 10 (0ndash99) at 100 years and12 (0ndash99) at 200 years (Fig 15)The projected population size and average individual hetero-

zygosity reached equilibrium levels with periodic fluctuations inthese values when introgression was implemented (Figs 16ndash17)Genetic introgression decreased the time until quasi‐extinctionacross all scenarios (Fig 18) Results of analyses using a criticalthreshold of 30 panthers revealed a similar pattern of relative ben-efits However the probabilities of quasi‐extinction were sub-stantially higher (ge45) across all scenarios (Appendix E)

Addition of pumas from Texas increased both the populationsize and average individual heterozygosity consequentlyintrogression caused periodic increases in average individualheterozygosity and population size at the interval at which theintrogression occurred Whereas average individual hetero-zygosity gradually decreased following introgression densitydependence caused the population size to fluctuate especiallywhen a larger number of pumas from Texas were released Thepositive effect of genetic introgression initiatives on quasi‐extinction probabilities decreased as the time interval betweenthese events increased More frequent genetic introgressions ledto greater increases in average individual heterozygosity and

Old adult

Prime adult

Subadult

Kitten

025 050 075

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Individual heterozygosity

Ann

ual s

urvi

val r

ate

Old adult

Prime adult

Subadult

Kitten

40 60 80 100 120

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Abundance index

Ann

ual s

urvi

val r

ate

Kittens Males Females

Figure 7 The effect of individual heterozygosity (left) and the abundance index (right) on Florida panther age‐ and sex‐specific survival estimates (shaded area isSE) in South Florida USA 1981ndash2013 Abundance index data are from McBride et al (2008) and R T McBride and C McBride (Rancherrsquos Supply Incorporatedunpublished report)

20 Wildlife Monographs bull 203

equilibrium population size while reducing the probability ofquasi‐extinction Comparatively performing genetic introgres-sion less often but with more individuals per release was lesseffective at improving the long‐term prospects for the pantherpopulation (Figs 13ndash15)

Financial cost and persistence benefitsmdashThe total estimated costof 1 genetic introgression was $40000 $80000 or $120000for releasing 5 10 or 15 pumas from Texas respectively(Table 5) The total cost incurred over 100 years for eachstrategy ranged from $50000 to $12 million (Table 6)The most expensive strategy involved the introduction of15 pumas every 10 years this strategy reduced the probability

of quasi‐extinction by 58 and 40 under the MPC and MVMscenarios respectively (Table 6) Introducing 5 pumas every80 years cost the least but improvements in populationpersistence under this strategy were insubstantial (Table 6)Releasing more panthers more frequently caused increasedpopulation fluctuations but did not necessarily reduce the quasi‐extinction probability (Figs 14 and 15 Table 6)

DISCUSSION

The duration (34 yr of data) and sample sizes associated with theFlorida Panther Project allowed us to obtain a comprehensiveunderstanding of genetic variation demographic parameters

A

B

C

Figure 8 Cause‐specific mortality rates (plusmnSE) for male and female Florida panthers (A) subadult prime‐adult and old‐adult Florida panthers of both sexes (B)and canonical and admixed Florida panthers of both sexes (C) in South Florida USA 1981ndash2013 Causes of mortality are vehicle collision intraspecific aggression(ISA) other causes (known causes such as diseases injuries and infections unrelated to the first 2 causes) and unknown cause (mortalities for which we could notassign a cause)

van de Kerk et al bull Florida Panther Population and Genetic Viability 21

probability of population persistence and the duration of thepositive impacts of genetic restoration on this endangered po-pulation The Florida panther population was in dire straits inthe early 1990s and our results clearly demonstrated that the

implementation of the introgression project in 1995 subse-quently accrued a series of genetic and demographic improve-ments that have led to the change from a declining population toone that is growing and expanding its current breeding range

Sensitivity Elasticity

0 1 2 3 00 02 04

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

Para

met

er

Figure 9 Sensitivity and elasticity (proportional sensitivity) of stochastic population growth rate (λs) of the Florida panther to vital demographic parameters in SouthFlorida USA 1981ndash2013 Horizontal lines represent 5th and 95th percentiles

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

70

80

90

100

110

0

5

10

15

0

50

100

Years

StatisticMeanMedian

Figure 10 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the minimum population count(MPC) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

125

150

175

200

00

05

10

15

20

0

30

60

90

Years

StatisticMeanMedian

Figure 11 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the motor vehicle mortality(MVM) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

22 Wildlife Monographs bull 203

MPC MVM

minus08 minus06 minus04 minus02 00 minus08 minus06 minus04 minus02 00

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Male subadult survival

Male prime adult survival

Male old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

PRCC

Para

met

er

Figure 12 Partial rank correlation coefficient (PRCC) between quasi‐extinction probabilities and the demographic parameters of the Florida panther for theminimum population count (MPC) and motor vehicle mortality (MVM) scenarios Error bars indicate standard deviation Based on data collected in South FloridaUSA 1981ndash2013

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 13 Average projected individual heterozygosities in the Florida panther population over 200 years without genetic management intervention and with eachof the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include theintroduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years Average individual heterozygosity peaks when pumas are added to the Floridapanther population Based on data collected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 23

Our research has further validated the success of genetic in-trogression by quantifying the reduction in the probability ofextinction of the panther population because of this manage-ment initiative These findings all bode well for continuedprogress toward recovery That said the Florida panther po-pulation remains small and isolated from other populations ofpuma from western North America thereby denoting that theeventual impact of genetic drift and inbreeding may negativelyaffect the population in the future Realizing this will continueto be an issue that managers will have to contemplate wemodeled the impact of varied genetic management scenarios todetermine the frequency of introgression along with the numberof panthers that should be released during each initiative tominimize cost and maximize benefits to the population Theseresults will prove invaluable to managers attempting to conservethe Florida panther

Genetic Variation Demographic Parametersand Persistence of Heterotic Benefits from GeneticIntrogressionOur measure of individual heterozygosity (1 minusHL) was higheron average for the admixed (0615) panthers than for those ofcanonical ancestry (0386) Levels of individual heterozygosityfor admixed panthers were comparable to levels observed in asample of pumas from west Texas (x plusmn SE= 0695plusmn 0019n= 41) which further demonstrates the improvement in levelsof genetic variation in the Florida population since 1995 The

correlation between HL and several demographic parametershas been noted in wild populations including cheetahs (Terrellet al 2016) and several species of birds such as European shags(Phalacrocorax aristotelis male and female survival probabilityVelando et al 2015) and Egyptian vulture (Neophron percnop-terus age at recruitment Agudo et al 2012) If we focus only onpanthers captured and radiocollared from 2012ndash2015 (FP211ndashFP240) all of which had estimated birth years ge2005 (ge10 yrpost‐introgression) those panthers had an average individualheterozygosity level of 0618plusmn 0029 highlighting the elevatedlevels of genetic variation that continue to persist in cohorts ofpanthers 5 generations after the release of female pumas fromTexasThe proportional membership values (q) from our

STRUCTURE analysis allowed us to delineate 2 clusters ofancestry for Florida panthers (Fig 4) During the pre‐in-trogression era the population was dominated by panthers thatretained a high percentage of the canonical ancestry which wasaffected by inbreeding depression The post‐introgression eraancestry values are comprised of many admixed individuals inessence demonstrating the success of the introgression in termsof increasing genetic variation in the population and mi-micking gene flow that historically occurred with panthers andother populations of pumas (Onorato et al 2010) A somewhatanalogous situation although abetted by natural immigrationwas evident in the ancestral variation in a small population ofpuma intersected by a major freeway in California USA In

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

75

80

85

90

95

100

130

150

170

190

Years

Popu

latio

n si

ze

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 14 Average projected population sizes in the Florida panther population over 200 years without intervention and with each of the genetic introgressionstrategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include the introduction of 5 10 or 15pumas from Texas USA every 10 20 40 or 80 years Peaks occur when pumas are added to the Florida panther population Based on data collected in SouthFlorida USA 1981ndash2013

24 Wildlife Monographs bull 203

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

25

50

75

100

0

25

50

75

100

Introgression interval (yrs)

PQ

E (

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 15 Average probability of quasi‐extinction (PQE) of the Florida panther population within 100 years and within 200 years without genetic managementintervention (N) and with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenariosIntrogression strategies include the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Errorbars indicate 5th and 95th percentiles Based on data collected in South Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

50

60

70

80

90

100

110

50

100

150

200

Years

Popu

latio

n si

ze

Figure 16 Average projected Florida panther population trajectories under a geneticintrogression strategy involving the release of 5 female pumas from Texas USA every20 years for the minimum population count (MPC) and motor vehicle mortality(MVM) scenarios Shaded area indicates 5th and 95th percentiles and isrepresentative of confidence intervals for other scenarios Based on data collected inSouth Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Figure 17 Average projected Florida panther population‐level heterozygositiesunder a genetic introgression strategy involving the release of 5 female pumas fromTexas USA every 20 years for the minimum population count (MPC) and motorvehicle mortality (MVM) scenarios Shaded area bounds the 5th and 95th percentilesand is representative of confidence intervals for other scenarios Based on datacollected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 25

this case the migration of just a single male across the freewayto the south resulted in an increase in the number of in-dividuals with admixed ancestry and substantially improvedgenetic diversity of the subpopulation south of the freeway(Riley et al 2014)Our estimate of overall kitten survival (0321plusmn 0056) was

similar to that reported by Hostetler et al (2010) who used13 years of post‐introgression data (0323plusmn 0071) These valuesare lower than kitten survival estimates derived from severalpuma populations in western North America (044ndash091 Loganand Sweanor 2001 Lambert et al 2006 Laundre et al 2007Robinson et al 2008 Ruth et al 2011) When we included onlydata for individuals that had been genetically sampled our es-timate was 15 higher The proportion of admixed kittens thatwere genetically sampled increased as the population expandedwhich could have resulted in higher estimates of kitten survivalbecause admixed kittens survive better than canonical kittensKitten survival was strongly influenced by genetic ancestry withsurvival rates of F1 kittens being almost twice that of canonicalkittens (Fig 5B) Consistent with the findings of Hostetler et al(2010) increases in heterozygosity coincided with increasedkitten survival whereas population density negatively affectedkitten survival Population density often has a strong negativeimpact on survival of young age classes due to infanticide andcannibalism which has been reported from several carnivore

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

50

100

150

0

50

100

150

Introgression interval (yrs)

TQE

(yrs

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 18 Average times until quasi‐extinction (TQE) for the Florida panther population within 100 years and within 200 years without genetic management interventionand with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategiesinclude the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Error bars indicate 5th and 95thpercentiles Based on data collected in South Florida USA 1981ndash2013

Table 5 Average cost (2017 US dollars) of introgression strategies employedover 100 years that involve the introduction of 5 10 or 15 pumas from TexasUSA into the Florida panther population in South Florida USA at an in-creasingly higher frequency Costs of capture (including transportation) quar-antine and post‐introgression monitoring were estimated by Roy McBrideMark Cunningham and Dave Onorato respectively

Frequency Component 5 pumas 10 pumas 15 pumas

Once Capture 25000 50000 75000

Quarantine 5000 10000 15000

Monitoring 10000 20000 30000

Total 40000 80000 120000

Every 80 years Capture 31250 62500 93750

Quarantine 6250 12500 18750

Monitoring 12500 25000 37500

Total 50000 100000 150000

Every 40 years Capture 62500 125000 187500

Quarantine 12500 25000 37500

Monitoring 25000 50000 75000

Total 100000 200000 300000

Every 20 years Capture 125000 250000 375000

Quarantine 25000 50000 75000

Monitoring 50000 100000 150000

Total 200000 400000 600000

Every 10 years Capture 250000 500000 750000

Quarantine 50000 100000 150000

Monitoring 100000 200000 300000

Total 400000 800000 1200000

26 Wildlife Monographs bull 203

species including polar bears (Ursus maritimus Derocher andWiig 1999) black bears (Ursus americanus Garrison et al 2007)and pumas (Logan and Sweanor 2001)Our estimates of sex‐ and age‐specific survival rates of subadult

and adult panthers (Table 3) and the effects of covariates on theserates were similar to those reported by Benson et al (2011) derivedfrom data collected through 2006 Our survival rates for males(065ndash077) were lower than females (078ndash097) for all 3 ageclasses (subadult prime adult and old adult) which is the typicaltrend observed in most puma populations (eg Ruth et al 19982011) However an unhunted puma population occupying afragmented urban landscape in southern California exhibited lowersurvival (0586 females 0525 males Vickers et al 2015) perhapsas a result of severe habitat fragmentation and the lack of extensiveswaths of public land protected in perpetuity Most populations ofpuma in western North America are typically subject to variedlevels of management via regulated hunting which often is biasedtoward the take of males (Cooley et al 2009 Ruth et al 2011)Consequently annual survival estimates from hunted populationsin northeast Washington (0679ndash0733 females 0341 malesRobinson et al 2008) and southeastern Arizona (0685 females0584 males Cunningham et al 2001) were generally lower thanthose we estimated for Florida panthersCause‐specific mortality analysis showed that male panthers

experienced a substantially greater mortality risk from in-traspecific aggression This differs from the pattern observedduring a long‐term study in Arizona where females were at ahigher risk of intraspecific mortality than males (46 of maleand 53 of female mortality Logan and Sweanor 2001)Mortality associated with intraspecific strife has been docu-mented in hunted and unhunted populations (this study Loganand Sweanor 2001 Stoner et al 2010) and its root cause hasbeen difficult to determine (eg population density and preypopulation trends) That being the case finding ways to reducewhat is often a major source of mortality for puma populationscontinues to pose a challenge to managersCanonical panthers were substantially more likely to die from

intraspecific aggression than were admixed panthers This mayat least partly explain the difference in survival between admixed

and canonical panthers Canonical panthers are known to bemore likely to suffer from varied correlates of inbreeding de-pression than admixed animals including atrial septal defects(Johnson et al 2010) and compromised immune systems(Roelke et al 1993) These factors have the potential to affectfitness and perhaps dominance of individuals when engaged inan intraspecific encounter A somewhat similar scenario wasobserved by Hogg et al (2006) in a population of bighorn sheepthat were part of an introgression project Admixed males in thispopulation had a greater probability of paternity than males thatwere less outbred This included coursing males with higherlevels of admixture being markedly more successful at fightingmales that were defending females in estrous (Hogg et al 2006)Heterosis resulting from genetic introgression is expected to be

the strongest in F1 individuals (Shull 1908 Crow 1948 Burkeand Arnold 2001 Waller 2015) and to progressively decline insubsequent generations (Pickup et al 2013 Frankham 2015)Given that admixed (especially F1) panthers survived better thancanonical panthers and that individual heterozygosity improvedsurvival of both sexes and all age classes our data provide evi-dence for heterotic advantage whereby individuals of mixedancestry had higher fitness (Shull 1908 Crow 1948) Our resultsare consistent with findings of other studies that involved in-tentional genetic introgression for conservation purposes Forexample Hogg et al (2006) detected substantial improvementsin survival and reproduction of introgressed bighorn sheep andMadsen et al (1999 2004) reported a dramatic increase in thepopulation of adders after genetic introgressionKnowledge of the persistence of benefits to a population ac-

crued via genetic introgression is essential for determiningwhen additional introgression may be warranted There is adepauperate amount of information regarding this topic and theidentification of factors that may influence persistence of thebenefits of genetic introgression (Tallmon et al 2004 Hedricket al 2014 Whiteley et al 2015) For example in minks(Neovison vison) Thirstrup et al (2014) found that outcrossingled to larger litters in F2 and F3 but this benefit disappeared insubsequent generations In a population of gray wolves Hedricket al (2014) suggested that benefits of genetic introgression may

Table 6 Costs and benefits accumulated over 100 years for genetic introgression at different intervals and with different numbers of pumas from Texas USAintroduced into the Florida panther population in South Florida USA Costs (2017 US dollars) include capture and transportation quarantine and post‐introgression monitoring Benefits are represented as average probability of quasi‐extinction (PQE and 5th and 95th percentiles) and changes (Δ) therein under thescenario using the minimum population count (McBride et al 2008 MPC) and the scenario using the motor vehicle mortality population estimates (McClintocket al 2015 MVM) The table is sorted by percent change in probability of quasi‐extinction under the MPC scenario

Interval (yr) Pumas Cost MPC PQE () ΔMPC PQE () MVM PQE () ΔMVM PQE ()

ndash 0 0 13 (0ndash99) 0 17 (0ndash100) 080 10 100000 12 (0ndash99) 10 (0ndash58) 16 (0ndash100) 5 (0ndash38)80 15 150000 12 (0ndash99) 13 (0ndash65) 15 (0ndash100) 8 (0ndash56)80 5 50000 12 (0ndash99) 14 (0ndash73) 15 (0ndash99) 9 (0ndash41)40 15 300000 10 (0ndash98) 26 (0ndash104) 14 (0ndash99) 13 (0ndash60)40 10 200000 9 (0ndash94) 35 (0ndash183) 13 (0ndash99) 21 (0ndash95)40 5 100000 8 (0ndash89) 39 (0ndash211) 12 (0ndash99) 26 (0ndash124)20 15 600000 8 (0ndash88) 42 (0ndash257) 13 (0ndash99) 24 (0ndash123)20 10 400000 6 (0ndash67) 54 (0ndash318) 10 (0ndash99) 37 (0ndash217)10 15 1200000 6 (0ndash53) 58 (0ndash372) 10 (0ndash99) 40 (0ndash208)20 5 200000 5 (0ndash50) 59 (0ndash338) 9 (0ndash99) 44 (0ndash352)10 10 800000 4 (0ndash16) 69 (0ndash489) 8 (0ndash92) 55 (0ndash401)10 5 400000 4 (0ndash7) 73 (0ndash502) 6 (0ndash71) 63 (0ndash503)

van de Kerk et al bull Florida Panther Population and Genetic Viability 27

wane after 2 or 3 generations The WrightndashFisher model ofgenetic drift predicts a geometric decline in expected hetero-zygosity at the rate of (1 minus 12Ne) per generation where Ne is theeffective population size (Hartl and Clark 1997 Frankham et al2014) Thus it seems logical to assume that the duration of thebenefits of introgression is limited especially in a species per-sisting in a small population such as Florida panthersWe expected Florida panther survival in the most recent years

(2005ndash2013) post‐introgression to differ from that observed in thedecade following the introgression in 1995 because pantherabundance and heterozygosity factors known to influence panthersurvival (Hostetler et al 2010 Benson et al 2011) have changed(Figs 2 and 6) We found that subadult and adult survival rates inrecent years (2005ndash2013) were similar to those in the period im-mediately following introgression (1995ndash2004 Fig 5C) overallkitten survival was similar to estimates of Hostetler et al (2010) Infact the pattern of covariate effects on Florida panther survivalreported by Benson et al (2011) and Hostetler et al (2010) hasbarely changed The negative effects of population density andpositive effects of heterozygosity may counterbalance survival ratesreducing population fluctuations Our result that benefits of geneticrestoration have remained in the population nearly for 5 genera-tions post‐intervention was surprising but also encouraging Pos-sible explanations for this observation may include 1) genetic ero-sion occurs at slower rate than previously thought 2) introducing 5migrants in 1 genetic intervention may have beneficial effects si-milar to those from 1 migrant per generation over multiple gen-erations or 3) other and as yet unknown factors may reduce therate of genetic erosion and subsequent demographic effects Adensity‐dependent effect on kitten survival could also explain whynon‐F1 admixed kittens did not survive substantially better thandid canonical kittens most kittens sampled from 2005ndash2013 wereadmixed and their survival was lower possibly because of a density‐dependent effect as the Florida panther population continuouslygrew during that period Trinkel et al (2010) also found thatpopulation density and inbreeding coefficient interacted to affectdemographic parameters and growth rate of an African lion po-pulation Likewise population density interacted with winterweather to affect the survival and population growth rates in Soaysheep (Ovis aries Milner et al 1999 Coulson et al 2001)

Population Dynamics and PersistenceThe finite deterministic and stochastic growth rates were gt10reflecting that much of the data used in this study were collectedfollowing genetic introgression in 1995 during which time thepopulation increased substantially Because our demographicparameter estimates did not change markedly in comparison tothose of Hostetler et al (2013) the similarity between thepopulation growth estimates from these studies was expectedBut these estimates reflect population growth during thedata‐collection period and do not predict future growth In factestimates of population size reported by McClintock et al(2015) suggest that population growth may already be slowing(Fig 2) A similar pattern was observed in an introduced po-pulation of African lion which increased steadily from 1995until 2001 then fluctuated around the presumed carrying ca-pacity thereafter (Trinkel et al 2010) Several other African lionpopulations that experienced various degrees of recovery

subsequently became relatively stable or exhibited decliningtrends (Bauer et al 2015)Our estimates of quasi‐extinction probabilities were lower than

those reported by Hostetler et al (2013) using a 2‐sex age‐structured matrix population model Lower probabilities ofquasi‐extinction than those reported by the earlier study forcomparable scenarios were likely a consequence of the fact thatthe estimates of abundance (McBride et al 2008) and thus thedensity‐dependent effects on survival of kittens and old panthers(gt10 yr) have changed in recent years Additionally these earlyestimates of the probability of quasi‐extinction and of time toquasi‐extinction did not consider genetic erosion that will in-evitably occur without management interventionUnder the MVM scenario the equilibrium population size was

twice that attained under the MPC scenario The MVM popula-tion estimates are approximately twice the MPCs (Fig 2) as aresult the simulated population under the MVM scenario achieveda higher equilibrium size Probability of quasi‐extinction under theMVM scenario was generally greater than that under the MPCscenario This result is a consequence of the fact that the estimateddensity dependence on kitten survival based on the MPC scenario isstronger than that for theMVM scenario (regression coefficients forabundance for the MPC scenario= minus0068 vs minus0002 for theMVM scenario Appendix B Table B1) Consequently simulatedpopulations under the MPC scenario have a stronger tendency tomove toward the equilibrium population size which inherentlyreduces the quasi‐extinction probability (eg Royama 1992)As expected for long‐lived mammals (Heppell et al 2000 Oli

and Dobson 2003) the population growth rate was proportio-nately most sensitive to changes in survival of prime adultsfollowed by that in younger age classes Global sensitivity ana-lysis in contrast showed that under all scenarios extinctionparameters were particularly sensitive to changes in survival ofFlorida panther kittens This result is a consequence of the highsensitivity of the population growth rate to kitten survival(Fig 9) and a larger standard error of kitten survival (Table 3)Results of our sensitivity analyses indicate that future researchshould focus on acquiring additional information on early lifestages to improve the accuracy and precision of estimates ofkitten survival Our results suggest that demographic variables(or their environmental drivers) that strongly influence extinc-tion risks are not necessarily those with the largest elasticityvalues A study of island fox (Bakker et al 2009) found thatextinction risk was strongly influenced by survival modifierparameters that had low elasticity values

Genetic Management When and How to InterveneThe use of genetic introgression as a management tool has al-ways been controversial and the probable effectiveness it mighthave in preventing the imminent demise of the panther popu-lation was initially debated (Creel 2006 Maehr et al 2006Pimm et al 2006) These debates notwithstanding the pantherpopulation had suffered from genetic morphological and bio-medical correlates of inbreeding before this management in-tervention (Johnson et al 2010 Onorato et al 2010) whereasthe post‐introgression period has been characterized by a sub-stantial reduction in the biomedical correlates of inbreeding andimprovements in demographic vigor and abundance (McBride

28 Wildlife Monographs bull 203

et al 2008 Hostetler et al 2010 2013 Johnson et al 2010Benson et al 2011 McClintock et al 2015) Nevertheless thepopulation remains small and isolated habitat is still being lostand there is no current possibility of natural gene flow betweenthis and other North American puma populations thus therecurrence of inbreeding‐related problems and subsequent po-pulation declines are inevitable The question therefore is notwhether genetic management of the Florida panther populationis needed but when and how it should be implementedWithout genetic introgression probabilities of quasi‐extinc-

tion were substantially greater than those from models thatincorporated periodic genetic introgression events (Fig 15)highlighting the importance of genetic management in smallisolated populations When 5 female pumas are added to theFlorida panther population every 10 years genetic variationincreased substantially under the MPC and MVM scenariosThe IBM predicted that the equillibrium population size wouldbe substantially larger when 5 pumas were released into thepopulation every 10 years than populations without intervention(ie no introgression) Releasing 15 Texas females did not af-ford substantially greater benefit to the equilibrium populationsize than did releasing only 5 Texas females Instead addingmore individuals caused increasingly large fluctuations in po-pulation size due to the numerical increases and density de-pendence (Fig 14) possibly diminishing the benefits associatedwith increased genetic variationCompared with the no‐intervention scenario (ie no genetic

introgression implemented) the introduction of 5 female pumasevery 10 years reduced quasi‐extinction probabilities from 13to 4 and from 17 to 6 for the MPC and MVM scenariosrespectively The benefit of genetic‐introgression strategies de-creased as the interval between successive management inter-ventions increased and no benefit was evident once the intervalreached 80 years (Fig 15) Introgression reduced the averagetime until quasi‐extinction (Fig 18) This somewhat counter‐intuitive result can be explained by the fact that repeated geneticintrogression makes the population more robust over timewhich will reduce the probability of extinction Consequentlyfew simulated population trajectories that fall below the criticalthreshold do so early reducing the mean time to extinctionAlso density dependence has a stabilizing effect on populations(Royama 1992) populations tend toward equilibrium popula-tion sizes which reduces the probability over time of popula-tions falling below quasi‐extinction threshold However timeuntil quasi‐extinction must be interpreted in conjunction withthe probability of quasi‐extinction which is very low for allscenarios with genetic introgression (Fig 15)Cost is a significant impediment to the implementation of a

genetic introgression program because captures relocations andmonitoring efforts before and after introgression are expensive(Edmands 2007 Frankham 2010b 2015 2016) Costs may beacceptable to society if the management actions result in sub-stantial fitness benefits especially if the species of concern ispopular and charismatic (Frankham 2015) We estimated the100‐year cost of introgression strategies modeled here to rangefrom $50000 to $12 million More expensive strategies gen-erally led to larger decreases in the probability of quasi‐extinc-tion but cheaper strategies also substantially improved

population viability (ie reduced quasi‐extinction probabilityTable 6) One of the cheaper strategies (5 pumas released every20 years) which cost an estimated $200000 over 100 yearsreduced the probability of quasi‐extinction by 59 and 44 forthe MPC and MVM scenarios respectively But these pre-liminary cost estimates are based on expenses incurred duringthe 1995 introgression and include costs of capture quarantinetransportation release and post‐release monitoring of femalesonly Although the cost for future genetic interventions wouldlikely be higher than those reported here the relative differencesin cost between alternative intervention strategies should remainapproximately the sameOur ability to simulate genetics and link it to the viability of

the Florida panther population is based on the empirically es-timated relationship between individual heterozygosity andsurvival Such heterozygosity and fitness correlations have beenstudied for decades (David 1998) but have only recently beenused to predict population viability (Benson et al 2016) noneto our knowledge have incorporated cost into their modelingefforts The mechanisms underlying heterozygosity and fitnesscorrelations remain unclear (David 1998 Szulkin et al 2010)and their significance as a proxy for inbreeding depression hasbeen debated (Balloux et al 2004 Miller and Coltman 2014)Nevertheless evidence continues to mount demonstratingpositive relationships between heterozygosity and survival(Coulson et al 1998ab Hansson et al 2004 Silva et al 2005Acevedo‐Whitehouse et al 2006 Jensen et al 2007 Velandoet al 2015) and between heterozygosity and fecundity (Seddonet al 2004 Charpentier et al 2005 Agudo et al 2012 Velandoet al 2015) Although the reported correlations are generallyweak their consequences can be substantial if population growthand persistence parameters are highly sensitive to the affectedvital rates (Velando et al 2015) Compared with scenarios thatignore genetics incorporating the effects of inbreeding on sur-vival increased quasi‐extinction probabilities within a 100‐yeartimeframe from 15 to 13 and from 14 to 17 for theMPC and MVM scenarios respectively These results high-light the importance of incorporating genetics when analyzingthe viability of small isolated populationsAlthough conservation biologists have recognized the im-

portance of genetics in population viability many still fail toincorporate genetic factors into PVA models (Reed et al 2002)Explicit consideration of genetics in PVAs requires long‐termgenetic and demographic data This permits the assessment ofthe functional relationship between genetic variability and de-mographic parameters Population viability analysis softwarepackages such as VORTEX (Lacy 1993 2000) offer a powerfulframework for individual‐based simulations but they often donot adequately capture a speciesrsquo life history or genetics Basedon a thorough review of the importance of genetic factors inwildlife conservation Frankham et al (2014) concluded thatgenetic factors can strongly influence the dynamics and persis-tence of small andor fragmented populations They furthernoted that ldquoMost PVA‐based risk assessments ignore or in-adequately model genetic factorsrdquo and recommended that ldquoPVAshould routinely include realistic inbreeding depressionhelliprdquo(Frankham et al 201457) Our custom‐built IBM permitted usto develop a model that adequately captured the Florida panther

van de Kerk et al bull Florida Panther Population and Genetic Viability 29

life history and we could parameterize the model with our long‐term genetic and demographic dataThe explicit consideration of the effects of inbreeding on

Florida panther population dynamics and persistence representsan important step forward Nonetheless our model remainsimperfect First we neglected the possible effects of habitat lossdetrimental anthropogenic activities climate change the prob-ability of immigration of pumas from western populationsdisease or catastrophes on demographic rates and populationviability Although immigration from puma populations outsideFlorida is possible its probability is very low given the extensivechallenges the anthropogenically altered landscape would posefor such a migrant to make it successfully to South Florida Weomitted mutations from our model because we have no in-formation on mutation rates though we expect that they are lowbased on values reported for other mammals (Kumar and Sub-ramanian 2002) Finally we only considered possible effects ofinbreeding on age‐specific survival rates because there was noevidence that heterozygosity affected female reproduction Lossof heterozygosity can negatively affect reproduction becauseinbred male panthers can suffer from cryptorchidism which canadversely affect their fecundity However given the polygynousmating system we expect the Florida panther population growthis primarily female‐limitedAlthough it is generally accepted that genetic introgression

only temporarily relieves a population from inbreeding depres-sion we know of no cases in which it has been implementedmore than once to conserve a threatened wildlife populationOur study provides an example of how demographic and mo-lecular data can be used within an IBM framework to evaluatethe efficacy of alternative genetic management strategies via theFlorida panther as a case study Our model can be adjusted forand applied to other species and offers great potential as a toolfor genetic management of small isolated wildlife populations

MANAGEMENT IMPLICATIONS

Our results and those of Hostetler et al (2013) and Johnsonet al (2010) provide persuasive evidence that the genetic in-tervention implemented in 1995 prevented the demise of theFlorida panther and restored demographic vigor to the popu-lation The Florida panther population remains small and iso-lated from other puma populations the closest puma populationis located gt1000 km away in Texas and the intervening land-scape is highly developed and fragmented with many interstate(and other high‐traffic) highways and major urban centersTheory suggests that 1 migrant per generation is sufficient tominimize the loss of genetic diversity (eg Mills and Allendorf1996) However the possibility of successful migration of apuma to South Florida followed by successful mating every ~5years is very low considering the distance between Texas andFlorida (Hedrick 1995) and the many risks and barriers thatdispersing pumas face (Beier 1995 Maehr et al 2002 Thatcheret al 2006) Consequently the pantherrsquos long‐term persistencewill likely depend on subsequent timely genetic introgressionsgiven the near‐impossibility of natural gene flow from otherpuma populations To that end our study offers considerableinsights into benefits of different genetic management strategiesand associated costs

Without genetic management the Florida panther populationfaces a substantial risk of quasi‐extinction within 100 years (10ndash46 depending on quasi‐extinction threshold and scenarios)Twenty‐three years have passed since genetic introgression wasimplemented and the population appears healthy as indicatedby measures of genetic variation increases in the populationsize and improvements in age‐specific survival rates Ourfindings suggest that releasing more pumas more frequently doesnot necessarily improve persistence probability because such astrategy would cause larger fluctuations in population size(Fig 14) which negatively affects persistence probability Thusextinction risks faced by inbred populations of large carnivorescan be substantially reduced by releasing approximately 5 im-migrants every 2 decades We suggest that such a managementstrategy be followed only in conjunction with continued long‐term monitoring of the population Our analyses demonstratethat population growth and persistence are highly sensitive tokitten and female (adult and subadult) survival Continuing tocollect data on these demographic variables along with bio-metric and genetic sampling will remain important to pantherrecovery and vigilance in identifying issues associated with in-breeding depression The collection of these monitoring datashould allow managers to detect any demographic or geneticchanges in a timely fashion and respond with genetic in-trogression or other management actions if warrantedRecovery objectives as defined by the USFWS in its current

recovery plan (USFWS 2008) delineate the need for additionalpopulations in the historical range to downlist or delist theFlorida panther If the only extant population of Floridapanthers continued to grow it could serve as a source ofpanthers to be translocated to suitable habitat to seed addi-tional populations Maintaining current information on thegenetic health of the population and precise estimates of itssize will be important in assessing whether it can withstand theremoval of a subset of animals for translocation Although the2017 documentation of a female panther with kittens north ofthe current breeding range is encouraging in terms of thepotential for population expansionmdashgiven that no female hadbeen recorded north of the Caloosahatchee River since 1972mdashthe re‐establishment of separate new populations will mostlikely require translocations by wildlife managers and a lengthysociopolitical processConservation of threatened or endangered species such as the

Florida panther is inherently challenging For panthers and otherlarge carnivores that require vast extents of natural habitat topersist habitat is typically the most limiting factor that will de-termine whether recovery efforts succeed Although geneticmanagement invariably plays a key role in the long‐term persis-tence of the panther population even a healthy population caneventually succumb to the impacts of habitat loss The humanpopulation of Florida continues to be one of the fastest‐growingin the nation the United States Census Bureau currently ranksFlorida as the third most populous Therefore habitat loss isgoing to continue to be a key issue for panthers Diligence towardpreserving panther habitat in its historical range via conservationeasements or other means and trying to keep such areas inter-connected via corridors is a goal stakeholders such as governmentagencies sportsmen non‐governmental organizations ranchers

30 Wildlife Monographs bull 203

and other private landowners must work on collaboratively toachieve

ACKNOWLEDGMENTS

We thank D Shindle D Jennings T OrsquoMeara D Land andC Belden for valuable advice throughout the project We thankS Bass M Criffield M Cunningham D Giardina D JansenA Johnson D Land M Lotz C McBride Rocky McBrideRowdy McBride Roy McBride L Oberhofer S Schulze staffsat Everglades National Park and Big Cypress National Preserveand others for assistance with fieldwork We also thank JAustin M Conroy J Hines J Laake S Mills J Nichols JHines M Sunquist J Fryxell and T Coulson for advice andassistance regarding data analysis and panther biology TheMuseum of Texas Tech University Natural Science ResearchLaboratory provided samples from pumas from west Texas forcomparative genetic analyses M Ben‐David D Land WMcCown E Hellgren and 4 anonymous reviewers providedmany helpful comments on the manuscript We thank NereaUbierna Lopez for completing the Spanish translation of theabstract We are grateful to the Department of ResearchComputing University of Florida for excellent high‐perfor-mance computing services We would like to thank US Fishand Wildlife Service Florida Fish and Wildlife ConservationCommission (FWC) US National Park Service QuantitativeSpatial Ecology Evolution and Environment (QSE3) In-tegrative Graduate Education and Research Traineeship(IGERT) program funded by the National Science Foundationand the University of Florida for financially supporting thisstudy Funding for panther research conducted by the FWC issupported predominantly via donations accrued with vehicleregistrations for ldquoProtect the Pantherrdquo license plates

LITERATURE CITEDAcevedo‐Whitehouse K T R Spraker E Lyons S R Melin F Gulland RL Delong and W Amos 2006 Contrasting effects of heterozygosity onsurvival and hookworm resistance in California sea lion pups MolecularEcology 151973ndash1982

Adams J R L M Vucetich P W Hedrick R O Peterson and J AVucetich 2011 Genomic sweep and potential genetic rescue during limitingenvironmental conditions in an isolated wolf population Proceedings of theRoyal Society B Biological Sciences 2783336ndash3344

Agresti A 2013 Categorical data analysis Third edition John Wiley amp SonsHoboken New Jersey USA

Agudo R M Carrete M Alcaide C Rico F Hiraldo and J A Donaacutezar2012 Genetic diversity at neutral and adaptive loci determines individualfitness in a long‐lived territorial bird Proceedings of the Royal Society BBiological Sciences 2793241ndash3249

Albeke S E N P Nibbelink and M Ben‐David 2015 Modeling behavior bycoastal river otter (Lontra canadensis) in response to prey availability in PrinceWilliam Sound Alaska a spatially‐explicit individual‐based approach PLOSOne 10e0126208

Alho J S K Vaumllimaumlki and J Merilauml 2010 Rhh an R extension for estimatingmultilocus heterozygosity and heterozygosity‐heterozygosity correlationMolecular Ecology Resources 10720ndash722

Alvarez K 1993 Twilight of the panther Myakka River Publishing SarasotaFlorida USA

Aparicio J M J Ortego and P J Cordero 2006 What should we weigh toestimate heterozygosity alleles or loci Molecular Ecology 154659ndash4665

Bakker V J D F Doak G W Roemer D K Garcelon T J Coonan S AMorrison C Lynch K Ralls and R Shaw 2009 Incorporating ecologicaldrivers and uncertainty into a demographic population viability analysis for theisland fox Ecological Monographs 7977ndash108

Balloux F W Amos and T Coulson 2004 Does heterozygosity estimateinbreeding in real populations Molecular Ecology 133021ndash3031

Barone M A M E Roelke J Howard J L Brown A E Anderson and DE Wildt 1994 Reproductive characteristics of male Florida panthersmdashcomparative studies from Florida Texas Colorado Latin‐America andNorth‐American Zoos Journal of Mammalogy 75150ndash162

Bateson Z W P O Dunn S D Hull A E Henschen J A Johnson and LA Whittingham 2014 Genetic restoration of a threatened population ofgreater prairie‐chickens Biological Conservation 17412ndash19

Bauer H G Chapron K Nowell P Henschel P Funston L T B HunterD W Macdonald and C Packer 2015 Lion (Panthera leo) populations aredeclining rapidly across Africa except in intensively managed areas Pro-ceedings of National Academy of Sciences of the United States of America11214894ndash14899

Beier P 1995 Dispersal of juvenile cougars in fragmented habitat Journal ofWildlife Management 59228ndash237

Benson J F J A Hostetler D P Onorato W E Johnson M E Roelke SJ OrsquoBrien D Jansen and M K Oli 2011 Intentional genetic introgressioninfluences survival of adults and subadults in a small inbred felid populationJournal of Animal Ecology 80958ndash967

Benson J F P J Mahoney J A Sikich L E K Serieys J P Pollinger H BErnest and S P D Riley 2016 Interactions between demography geneticsand landscape connectivity increase extinction probability for a small popu-lation of large carnivores in a major metropolitan area Proceedings of theRoyal Society B Biological Sciences 28320160957

Beverly F 1874 The Florida panther Forest and Stream 3290Boyce M S C V Haridas C T Lee and the NCEAS Stochastic Demo-graphy Working Group 2006 Demography in an increasingly variable worldTrends in Ecology amp Evolution 21141ndash148

Brook B W J J OrsquoGrady A P Chapman M A Burgman H R Akcakayaand R Frankham 2000 Predictive accuracy of population viability analysis inconservation biology Nature 404385ndash387

Brys R and H Jacquemyn 2016 Severe outbreeding and inbreeding de-pression maintain mating system differentiation in Epipactis (Orchidaceae)Journal of Evolutionary Biology 29352ndash359

Burke J M and M L Arnold 2001 Genetics and the fitness of hybridsAnnual Review of Genetics 3531ndash52

Burnham K P 1993 A theory for combined analysis of ring recovery andrecapture data Pages 199ndash213 in J‐D Lebreton and P M North editorsMarked individuals in the study of bird population Springer‐Verlag NewYork New York USA

Burnham K P and D R Anderson 2002 Model selection and multimodelinference a practical information‐theoretic approach Second edition SpringerVerlag New York New York USA

Caswell H 2001 Matrix population models construction analysis and in-terpretation Sinauer Associates Sunderland Massachusetts USA

Charpentier M J M Setchell F Prugnolle L A Knapp E J Wickings PPeignot and M Hossaert‐McKey 2005 Genetic diversity and reproductivesuccess in mandrills (Mandrillus sphinx) Proceedings of the NationalAcademy of Sciences of the United States of America 10216723ndash16728

Cooch E and G C White editors 2018 Program MARK a gentle in-troduction lthttpwwwphidotorgsoftwaremarkdocsbookgt Accessed 7Apr 2016

Cooley H S R B Wielgus G M Koehler H S Robinson and B TMaletzke 2009 Does hunting regulate cougar populations A test of thecompensatory mortality hypothesis Ecology 902913ndash2921

Coulson T E A Catchpole S D Albon B J Morgan J M Pemberton TH Clutton‐Brock M J Crawley and B T Grenfell 2001 Age sex densitywinter weather and population crashes in Soay sheep Science 2921528ndash1531

Coulson T J Pemberton S Albon M Beaumont T Marshall J Slate FGuinness and T Clutton‐Brock 1998a Microsatellites reveal heterosis in reddeer Proceedings of the Royal Society B Biological Sciences 265489ndash495

Coulson T N S D Albon J M Pemberton J Slate F E Guinness and TH Clutton‐Brock 1998b Genotype by environment interactions in wintersurvival in red deer Journal of Animal Ecology 67434ndash445

Cox D 1972 Regression models and life‐tables Journal of the Royal StatisticalSociety Series B Statistical Methodology 34187ndash220

Creel S 2006 Recovery of the Florida panthermdashgenetic rescue demographic rescueor both Response to Pimm et al (2006) Animal Conservation 9125ndash126

Crnokrak P and D A Roff 1999 Inbreeding depression in the wild Heredity83260ndash270

Crow J 1948 Alternative hypotheses of hybrid vigor Genetics 33477ndash487

van de Kerk et al bull Florida Panther Population and Genetic Viability 31

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 13: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

Dynamics and persistence of populations are influenced bymany factors such as demographic and environmental sto-chasticity and density dependence these effects and parametricuncertainty must be considered in population models whenpossible (Caswell 2001 Boyce et al 2006 Bakker et al 2009Hostetler et al 2013) Because by definition IBMs simulate thelife history of individuals in a population they inherently in-corporate the effect of demographic stochasticity Our analysesrevealed that survival of kittens subadults and adults and theprobability of reproduction varied over time so we incorporatedenvironmental stochasticity in the model for these variablesfollowing Hostetler et al (2013) Likewise we found evidenceof density‐dependent effects on kitten survival Because theMPC might be an underestimate of population size (McBrideet al 2008) we performed simulations using both the MPC andMVM abundance indicesTo account for model selection and parametric uncertainty we

used a Monte Carlo simulation approach For each bootstrapsample we selected a model based on the AIC (or QAICc)weights We then sampled the intercept and slope for kittensurvival (on the logit scale) subadult and adult survival (on thelogit scale with log‐hazard effect sizes) and probability of re-production (on the complementary logndashlog scale Appendix B)from multivariate normal distributions with mean vectors equalto the estimated parameters and variance‐covariance matricesequal to the sampling variance‐covariance matrices We thentransformed the results to nominal scale and converted the es-timates to age‐specific survival and reproduction parameters asdescribed in Hostetler et al (2013)We ran 1000 simulations for 1000 parametric bootstraps for

200 years for the MPC and the MVM scenario for both thestochastic 2‐sex matrix‐population model and the IBM Wetracked the population size at each time step and estimatedquasi‐extinction probabilities and times based on the populationtrajectory We considered the population to be quasi‐extinct ifindividuals of only 1 sex remained or if the population size fellbelow critical thresholds set at 10 and 30 individuals We alsoestimated expected time to quasi‐extinction for both thresholdsdefined as the mean and median time until a population be-comes quasi‐extinct conditioned on quasi‐extinction We im-plemented the IBM using the RNetLogo package version 10ndash2(Thiele et al 2012 Thiele 2014) as an interface for the IBMplatform NetLogo (Wilensky 1999)

Sensitivity AnalysisAn important component of demographic analysis is sensitivityanalysis which is the quantification of the sensitivity of amodelrsquos outcome to changes in the input parameters (Caswell2001 Bakker et al 2009 Thiele et al 2014) Using an IBM weperformed global sensitivity analyses for 2 (ie MPC andMVM) density‐dependent scenarios to assess how sensitivequasi‐extinction times and probabilities were to changes (anduncertainties) in the estimated demographic rates We usedLatin hypercube sampling to sample parameters from the entireparameter space (Marino et al 2008 Thiele et al 2014) Wesampled intercept and slope for kitten survival (on logit scale)subadult and adult survival (on logit scale with log‐hazard effectsizes) and probability of reproduction (on complementary

logndashlog scale) from normal distributions We sampled theprobability distribution for the number of kittens producedannually (on the real scale) from a Dirichlet distribution themultivariate generalization of the beta distribution (Kotz et al2000) We sampled 1000 different parameter sets and ran 1000simulations for each scenario We calculated the probability ofquasi‐extinction within 200 years for each density‐dependencescenario (MPC or MVM) using a quasi‐extinction threshold of10 individuals and then calculated the partial rank correlationcoefficient between each of those and each parameter trans-formed to the real scale (Bakker et al 2009 Hostetler et al2013) Partial rank correlation coefficients quantify the sensi-tivity of the probability of quasi‐extinction to input variables(ie survival and reproductive parameters) with higher absolutevalues indicating stronger influence of that variable on quasi‐extinction probabilities

Genetic ManagementOne of our goals was to estimate the benefits (improvements ingenetic variation demographic parameters and populationgrowth and persistence parameters) and costs (financial costs ofcapture transportation quarantine release and monitoring ofintroduced panthers) of genetic management To achieve thisobjective we extended the IBM to include a genetic componentTo simulate individual‐ and population‐level heterozygosity overtime we assigned each panther alleles for 16 microsatellite locibased on the empirical allele frequency in the population (esti-mated from genetic samples collected during 2008ndash2015) whenwe initialized the model (Pierson et al 2015) At each time stepwe simulated Mendelian inheritance for kittens produced suchthat each kitten randomly inherited alleles from its mother andfrom a male panther randomly chosen to have putatively siredthe litter We did not explicitly force inbreeding to occur butthe probability of inbreeding naturally increased as the popula-tion size decreasedIntrogression strategiesmdashWe modeled genetic introgression as a

result of the release of 2‐year‐old female pumas from Texas intothe simulated Florida panther population For the geneticintrogression implemented in 1995 Texas females between 15and 25 years of age were released because females at that agehave lower mortality rates (Benson et al 2011) higherreproductive potential and also because it is much easier todetect with certainty the successful reproduction by females thanby males The ability to more closely monitor reproduction andthe birth of kittens is an important consideration in reducing theprobability of a genomic sweep of Texas alleles into the Floridapanther population Wildlife managers removed the last 2surviving Texas females from the wild population in Florida in2003 to reduce the possibility of genomic sweep To simulatethis strategy in the model we removed any Texas females thatwere still alive 5 years after each introgression event Weassigned Texas females alleles based on the allele frequency ofthe Texas population (based on genotypes from 48 samplescollected in west Texas that was inclusive of the pumas releasedin South Florida in 1995) when they entered the Florida pantherpopulation We could not estimate demographic rates separatelyfor the released Texas females because of small sample size (only8 females were released in 1995) therefore we assumed that

van de Kerk et al bull Florida Panther Population and Genetic Viability 15

survival and reproductive rates and the effects of covariates onthese rates for the released Texas females were identical to thoseof female Florida panthersThe impact of genetic introgression depends on the frequency

of introgression events and the number of individuals introducedat each attempt Thus we defined introgression strategies basedon the number of Texas females to be released (0 5 10 or 15)and the interval between genetic introgressions (10 20 40 and80 yr) for a total of 13 strategies We simulated each manage-ment strategy with density dependence estimated using theMPC and MVM abundance indices

We sampled 1000 parameter sets and for each of these setswe ran all introgression strategies 1000 times for 200 years Weapplied the same parametric uncertainty and environmentalstochasticity across all management scenarios to allow for bettercomparison of introgression strategies At each time step werecorded the projected population size and average individualheterozygosity We calculated the heterozygosity for each in-dividual at birth based on the alleles it inherited This individualheterozygosity then informed the survival rate of that individualbased on the empirically estimated relationship between in-dividual heterozygosity and age‐specific survival rates Hetero-zygosity did not change throughout an individualrsquos lifetime butits effect on survival changed as individuals aged because theeffect of individual heterozygosity on survival rate differedamong age classes Survival rates of genetically sampled kittenswere biased high because some kittens were sampled later in lifewhen they had already survived for some time To correct forthis bias we adjusted kitten survival rates predicted by theheterozygosity model proportionally From the population tra-jectories we estimated the probability of quasi‐extinction (de-fined as the probability that the population will fall below 10 or30 individuals or that individuals of only 1 sex will remain)

Cost‐benefit analysismdashExperts estimated financial costs ofintrogression based on their experience with the geneticintrogression executed in 1995 and we adjusted estimates tocurrent costs to account for inflation (R T McBride RancherrsquosSupply Incorporated M Cunningham and D P OnoratoFlorida Fish and Wildlife Conservation Commission personalcommunication) Costs included capturing and transportingfemale pumas from the Texas source population caring for thepumas while they were in quarantine and post‐introgressionmonitoring To compare the financial costs of the differentscenarios we also calculated the average costs incurred over

Table 2 Sample size (n) number of alleles (Na) allelic richness (AR) observedheterozygosity (Ho) and expected heterozygosity (He) at 16 microsatellite loci inradio‐collared Florida panthers sampled from 1981 to 2013 in South FloridaUSA We calculated these values from a subset (n= 137) of the samples used inthe analyses and they include only adult and subadult panthers We excludedkittens because they can result in an overrepresentation of certain alleles

Locus n Na AR Ho He

FCA090 129 5 50 0333 0401FCA133 136 3 30 0618 0608FCA243 136 5 50 0419 0487F124 137 7 69 0708 0729F37 129 4 40 0395 0397FCA075 131 5 50 0534 0598FCA559 133 5 50 0496 0503FCA057 134 6 59 0679 0624FCA081 134 7 69 0209 0206FCA566 130 6 60 0462 0475F42 133 10 100 0737 0766FCA043 136 5 49 0596 0588FCA161 132 6 60 0500 0515FCA293 132 4 40 0614 0624FCA369 131 6 60 0573 0567FCA668 133 7 70 0406 0462Mean 1329 57 57 0517 0535

Figure 4 Proportional membership (q) of radio‐collared Florida panthers (n= 218) and pumas from Texas USA (n= 7) in 2 clusters identified by STRUCTUREEach individual animal is represented by a separate vertical bar Yellow indicates canonical panther ancestry and blue represents admixed ancestry The admixedancestry of the Texas females and F1 generation panthers that were radiocollared are highlighted red and green respectively to demarcate those groups The pre‐introgression period is inclusive of panthers radiocollared from 10 February 1981 to 6 March 1996 The post‐introgression era inclusive of the F1 panthers includesanimals radiocollared from 4 March 1997 to 18 February 2015 in South Florida USA

16 Wildlife Monographs bull 203

100 years assuming that the population would be managedaccording to a particular strategy Our goal with theseexploratory analyses was to evaluate the relative costs andbenefits of alternative management scenarios

RESULTS

Genetic VariationWe assessed genetic variation at 16 microsatellite loci for 137DNA samples collected from adult and subadult panthers(1981ndash2013) that were captured and subsequently radiocollaredThe number of alleles at these loci ranged from 3ndash10 and allelicrichness averaged 57 (Table 2) Average expected hetero-zygosity for this sample was 0535 and average observed het-erozygosity was 0517 (Table 2)We determined individual heterozygosity (1 minusHL) and an-

cestry for the complete sample of panther genotypes (n= 503)collected from 1981 to 2015 for use as covariates in subsequentanalyses Average individual heterozygosity was 0559 andranged from 0ndash0980 Our STRUCTURE analysis providedstrong support for K= 2 clusters based on the probability of thedata (log Pr(D)|K) and ΔK in STRUCTURE HARVESTERBased on this result we categorized the ancestry of each pantheras either canonical (n= 123) or admixed (n= 380) using valuesof proportional membership (q Fig 4) The average individualheterozygosity of canonical panthers was 0386plusmn 0012 (SE) foradmixed panthers it was 0615plusmn 0007

Survival and Cause‐Specific Mortality RatesOf the 395 kittens that were PIT‐tagged and released 55 werelater encountered alive 15 were recovered dead and 85 werepart of failed litters (Table 1) We radio‐tracked 209 subadult oradult panthers for a total of 244195 panther‐days and docu-mented 126 mortalities (Table 1)Survival depended strongly on age and kittens had the lowest

overall rate (032plusmn 009 Fig 5A Table 3) The model‐averagedkitten survival was 047plusmn 009 when we included only geneti-cally sampled individuals in the analysis (Table 3) as statedabove this estimate is biased high because many kittenswere genetically sampled when they were recaptured alive orrecovered dead at older ages We found no evidence for sex‐specific differences in kitten survival but females survived betterthan males in all older age classes For females subadults hadthe highest survival rates followed by prime adults and oldadults For males prime adults had the highest survival ratefollowed by subadults and old adults (Fig 5A and Table 3)For panthers of all ages there was substantial evidence that

ancestry affected survival with F1 admixed panthers of all ageshaving the highest rates and canonical individuals having thelowest (Fig 5B) The combined AIC weights of models thatincluded an ancestry covariate were 0880 for kittens and 0992for older panthers The survival rates of subadult prime‐adultand old‐adult panthers in the period immediately after the in-trogression (1995ndash2004) were similar to those in more recentyears (2005ndash2013 Fig 5C)After genetic introgression individual panther heterozygosity

increased (Fig 6) and varied among ancestry categories individualheterozygosity was the highest for F1 panthers (078plusmn 001)

A

B

C

Figure 5 Overall age‐ and sex‐specific survival rates (plusmnSE A) age‐ and sex‐specific survival rates (plusmnSE) for Florida panthers of canonical F1 admixed andother admixed ancestry (B) and age‐ and sex‐specific survival estimates (plusmnSE)for subadult and adult Florida panthers in the period immediately post‐introgression (1995ndash2004) and more recently (2005ndash2013 C) in SouthFlorida USA

Table 3 Model‐averaged estimates (plusmnSE) of demographic parameters for theFlorida panther obtained using data collected during 1981ndash2013 in SouthFlorida USA

Parameter Sex Age class Estimate

Survival Both Kitten 032plusmn 009a

Female Subadult 097plusmn 002Prime adult 086+ 003Old adult 078plusmn 009

Male Subadult 066plusmn 006Prime adult 077plusmn 005Old adult 065plusmn 010

Probability of reproduction Female Subadult 035plusmn 008Prime adult 050plusmn 005Old adult 025plusmn 006

Kittens produced annually Female Subadult 280plusmn 005Prime adult 267plusmn 001Old adult 228plusmn 013

a Overall kitten survival (based on all kittens in our data set including thosethat were not genetically sampled) Estimate of survival of geneticallysampled kittens was 047plusmn 009 this estimate is biased high because manykittens were genetically sampled when they were recaptured alive or re-covered dead at older ages

van de Kerk et al bull Florida Panther Population and Genetic Viability 17

followed by those for non‐F1 admixed (062plusmn 001) and canonical(039plusmn 001) panthers Individual heterozygosity positively affectedsurvival rates of kittens (slope parameter η= 1455 95CI= 0505ndash2405) Individual heterozygosity also positively af-fected survival of subadult and adult panthers but the evidence forthis effect was weaker because the confidence interval for the ha-zard ratio overlapped 10 (hazard ratio exp(ɩ)= 0311 95CI= 0092ndash1054 Table 4 and Fig 7)The MPC increased after genetic introgression (Fig 2) This

abundance index was also included in top‐ranking modelsindicating that population density affected survival (Table 4)Abundance reduced the survival of kittens (slope parameterη= minus0035 95 CI= minus0049 to minus0021) evidence was weak forthe effect of abundance on survival of subadult and adult panthers(hazard ratio exp(ɩ)= 1002 95 CI= 0995ndash1010 Fig 7) Re-gression coefficients associated with the effect of abundance andindividual heterozygosity on demographic parameters are presentedin Table B1 (available online in Supporting Information)The most frequently observed causes of death for radio‐

collared panthers were vehicle collision and intraspecific ag-gression Cause‐specific mortality analyses revealed thatmortality rates among possible causes were generally similar forthe 2 sexes but that males were more likely to die from in-traspecific aggression than were females (Fig 8A) Male mor-tality from intraspecific aggression was higher for subadult andold adult panthers than for prime adults (Fig 8B) For bothmales and females canonical panthers were more likely to diefrom intraspecific aggression than were admixed panthers(Fig 8C)

Reproductive ParametersOf 101 radio‐collared females 67 reproduced at least once duringour monitoring we used these individuals to assess the annualprobability of reproduction The top‐ranked model for the annual

probability of reproduction included age effect only suggesting thatadding ancestry or abundance did not improve model parsimony(Table 4) Model‐averaged annual probabilities of reproductiondiffered between female subadults (035plusmn 008) prime adults(050plusmn 005) and older adults (025plusmn 006)The most parsimonious model for the number of kittens

produced annually by females that produced at least 1 litter (ieconditional on reproduction) included effects of age ancestryand abundance a model that included only the effect of age wasalmost equally supported (Table 4) The model‐averagednumber of kittens produced annually conditional on re-production was 280plusmn 075 for subadults 267plusmn 043 for primeadults and 228plusmn 083 for old adults Confidence intervals forthe slope parameter (β) overlapped 0 for both abundance(β= 0010 95 CI= minus0001 to 0021) and ancestry (β= minus073795 CI= minus1637 to 0162)

Population Dynamics and PersistenceDeterministic and stochastic demographymdashThe deterministic (λ)

and stochastic (λs) annual population growth rate calculated usingthe matrix population model were 106 (5th and 95th percentiles099ndash114) and 104 (072ndash141) respectively The generation timewas 47 years A female starting in age class one (kitten) wasexpected to live another 58 years and produce 10 kittens onaverage during her lifetime Overall λs was proportionately(elasticity) most sensitive to changes in female prime adultsurvival on the absolute scale (sensitivity) however it was mostsensitive to changes in kitten survival (Fig 9) Populationtrajectories probabilities of quasi‐extinction and times untilquasi‐extinction estimated using the matrix population modeland IBM for comparable scenarios were nearly identical for acritical quasi‐extinction threshold of 10 panthers (Figs 10 and 11)and for a critical quasi‐extinction threshold of 30 panthers(Appendix E available online in Supporting Information)Minimum population count scenariomdashUnder the MPC scenario

(ie density dependence estimated using the minimumpopulation count data) with a critical threshold of 10panthers the population size reached 99 (72ndash104) and 100(77ndash105) at 200 years (Fig 10) as predicted by the IBM and thematrix model respectively The cumulative quasi‐extinctionprobabilities within 100 years based on the MPC scenario were15 (IBM 0ndash48) and 30 (matrix model 0ndash76) Theseprobabilities increased to 25 (IBM 0ndash114) and 38 (matrixmodel 0ndash154) by year 200 (Fig 10) The mean times untilquasi‐extinction were 15 years (IBM 0ndash67) and 15 years (matrixmodel 0ndash72) conditioned on going quasi‐extinct within 100years Expected times until quasi‐extinction conditioned onquasi‐extinction within 200 years were higher 34 years (IBM0ndash137) and 32 years (matrix model 0ndash134 Fig 10)Motor vehicle mortality scenariomdashUnder the MVM scenario

(ie density dependence estimated using estimates of pantherpopulation size from McClintock et al 2015) with a criticalthreshold of 10 panthers the projected population reached 188(142ndash217) and 190 (143ndash219) at 200 years as predicted by theIBM and the matrix model respectively (Fig 11) Thecumulative quasi‐extinction probabilities within 100 years were14 (IBM 0ndash08) and 13 (matrix model 0ndash06) Theseprobabilities increased to 20 (IBM 0ndash17) and 19 (matrix

000

025

050

075

100

1995 2000 2005 2010Year of birth

Indi

vidu

al h

eter

ozyg

osity

Figure 6 Temporal changes in the individual heterozygosity of Floridapanthers born during the period 1995ndash2013 in South Florida USA Pointsare measurements solid line is linear regression shaded area is 95 confidenceinterval of slope Slope= 0006plusmn 0001 R2= 009

18 Wildlife Monographs bull 203

model 0ndash16) within year 200 (Fig 11) The mean times until quasi‐extinction based on the MVM scenario were 5 years (IBM 0ndash61)and 6 years (matrix model 0ndash64) conditioned on going quasi‐extinctwithin 100 years Expected times until quasi‐extinction conditionedon quasi‐extinction within 200 years were 11 years (IBM 0ndash113)and 11 years (matrix model 0ndash112 Fig 11)

Sensitivity of extinction probability to demographic parametersmdashThe partial rank correlation coefficients between quasi‐extinctionprobabilities and demographic parameters were negative anincrease in any of the demographic parameters decreased thequasi‐extinction probability Probabilities of quasi‐extinction within200 years were most sensitive to changes in kitten survival for bothscenarios (Fig 12) followed by survival of subadult females in theMVM scenarios and survival of prime adult females in the MPCscenario The probability of reproduction of prime adult femaleshad the third‐largest partial rank correlation coefficient with quasi‐extinction probability for both scenarios

Benefits and Costs of Genetic ManagementMinimum population count scenariomdashThe MPC scenario with

a critical threshold of 10 panthers predicted that withoutmanagement intervention average individual heterozygositywould decrease reaching 057 (049ndash064) at 100 years and053 (042ndash062) at 200 years (Fig 13) The population woulddecrease slightly reaching an average of 79 (41ndash99) at 100 yearsand 76 (43ndash98) at 200 years (Fig 14) The probabilities of quasi‐extinction under the no‐intervention MPC scenario when weconsidered the effects of genetic erosion were 13 (0ndash99) at 100years and 23 (0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with

the translocation of 5 Texas females average individualheterozygosity under the MPC scenario increased and thenstabilized at 068 (064ndash072) at 100 years and remained atthat level at 200 years (Fig 13) The population reached anaverage of 88 (62ndash104) at 100 years and stayed the same at200 years (Fig 14) The probabilities of quasi‐extinctionunder this introgression strategy were 6 (0ndash53) within 100years and 7 (0ndash82) within 200 years (Fig 15) Results ofanalyses using a critical threshold of 30 panthers revealed asimilar pattern of relative benefits However the prob-abilities of quasi‐extinction were substantially higher (ge45)across all scenarios (Appendix E)

Motor vehicle mortality scenariomdashWithout managementintervention the average individual heterozygosity predictedby the MVM scenario and a critical threshold of 10 panthersdecreased to 060 (046ndash069) at 100 years and to 059 (039ndash070) at 200 years (Fig 13) The population increased slightlyand reached an equilibrium at 154 individuals (46ndash217) at 100years and 157 (57ndash218) at 200 years (Fig 14) The probabilitiesof quasi‐extinction were 17 (0ndash100) at 100 years and 22(0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with 5

female pumas from Texas average individual heterozygosityincreased slightly reaching 067 (060ndash073) at 100 years and068 (059ndash075) at 200 years (Fig 13) Under this strategy thepopulation reached an average of 164 individuals (44ndash226) at

Table 4 Model comparison results testing for the effects of several covariateson survival of all kittens survival of genetically sampled kittens survival ofsubadult and adult Florida panthers probability of reproduction and kittensproduced annually for Florida panthers sampled from 1981ndash2013 in SouthFlorida USA

Model Parameters ΔAICa Weight

Kitten survival (all data)Base+ F1b+ abundancec 10 000 0988Base+ abundance 9 1018 0006Base+ F1 9 1035 0005Base 8 2711 lt0001Base+ sexd 9 2912 lt0001Base+ litter sizee 9 2914 lt0001

Kitten survival (genetically sampled kittens only)Base+ F1+ abundance 10 000 0541Base+ F1CanAdmf+ abundance 11 099 0329Base+Hetg+ abundance 10 310 0115Base+CanAdmh+ abundance 10 791 0010Base+ abundance 9 944 0005Base+ F1 9 1638 lt0001Base+ F1CanAdm 10 1837 lt0001Base+Het 9 2872 lt0001Base 8 3296 lt0001Base+CanAdm 9 3348 lt0001

Subadult and adult survivalBase+ F1CanAdm+ abundance 7 000 0360Base+ F1 5 053 0276Base+ F1CanAdm 6 105 0213Base+ F1+ abundance 6 222 0119Base+CanAdm+ abundance 6 575 0020Base+CanAdm 5 908 0004Base+Het 5 964 0003Base+Het+ abundance 6 984 0003Base 4 1118 0001Base+ abundance 5 1278 0001

Probability of reproductionAge 3 000 0242Age+CanAdm 4 012 0228Age+ abundance 4 173 0102Age+ F1 4 190 0094Age+CanAdm+ abundance 5 216 0082Age+Het 4 219 0081Age+ F1CanAdm 5 232 0076Age+ F1+ abundance 5 372 0038Age+Het+ abundance 5 399 0033Age+ F1CanAdm+ abundance 6 444 0026

Kittens produced annuallyAge+ F1+ abundance 5 000 0194Age 3 060 0144Age+ abundance 4 061 0143Age+ F1 4 097 0120Age+CanAdm 4 139 0097Age+ F1CanAdm+ abundance 6 196 0073Age+ F1CanAdm 5 231 0061Age+CanAdm+ abundance 5 238 0059Age+Het+ abundance 5 250 0056Age+Het 4 259 0053

a Akaikersquos information criterion (AIC) for kitten survival models was adjustedfor small sample size and overdispersion (QAICc)

b F1 divides Florida panthers into 2 groups F1 admixed and all other panthersc Index of Florida panther abundanced Sex of an individual Florida panther (male or female)e Size of the litter in which a Florida panther kitten was bornf F1CanAdm divides panthers into 3 groups F1 admixed canonical andother admixed panthers

g Het is the individual heterozygosityh CanAdm divides panthers into 2 groups canonical and admixed (includingF1 admixed) panthers

van de Kerk et al bull Florida Panther Population and Genetic Viability 19

100 years and 170 (67ndash231) at 200 years based on the MVMscenario (Fig 14) The probabilities of quasi‐extinction underthis introgression strategy were 10 (0ndash99) at 100 years and12 (0ndash99) at 200 years (Fig 15)The projected population size and average individual hetero-

zygosity reached equilibrium levels with periodic fluctuations inthese values when introgression was implemented (Figs 16ndash17)Genetic introgression decreased the time until quasi‐extinctionacross all scenarios (Fig 18) Results of analyses using a criticalthreshold of 30 panthers revealed a similar pattern of relative ben-efits However the probabilities of quasi‐extinction were sub-stantially higher (ge45) across all scenarios (Appendix E)

Addition of pumas from Texas increased both the populationsize and average individual heterozygosity consequentlyintrogression caused periodic increases in average individualheterozygosity and population size at the interval at which theintrogression occurred Whereas average individual hetero-zygosity gradually decreased following introgression densitydependence caused the population size to fluctuate especiallywhen a larger number of pumas from Texas were released Thepositive effect of genetic introgression initiatives on quasi‐extinction probabilities decreased as the time interval betweenthese events increased More frequent genetic introgressions ledto greater increases in average individual heterozygosity and

Old adult

Prime adult

Subadult

Kitten

025 050 075

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Individual heterozygosity

Ann

ual s

urvi

val r

ate

Old adult

Prime adult

Subadult

Kitten

40 60 80 100 120

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Abundance index

Ann

ual s

urvi

val r

ate

Kittens Males Females

Figure 7 The effect of individual heterozygosity (left) and the abundance index (right) on Florida panther age‐ and sex‐specific survival estimates (shaded area isSE) in South Florida USA 1981ndash2013 Abundance index data are from McBride et al (2008) and R T McBride and C McBride (Rancherrsquos Supply Incorporatedunpublished report)

20 Wildlife Monographs bull 203

equilibrium population size while reducing the probability ofquasi‐extinction Comparatively performing genetic introgres-sion less often but with more individuals per release was lesseffective at improving the long‐term prospects for the pantherpopulation (Figs 13ndash15)

Financial cost and persistence benefitsmdashThe total estimated costof 1 genetic introgression was $40000 $80000 or $120000for releasing 5 10 or 15 pumas from Texas respectively(Table 5) The total cost incurred over 100 years for eachstrategy ranged from $50000 to $12 million (Table 6)The most expensive strategy involved the introduction of15 pumas every 10 years this strategy reduced the probability

of quasi‐extinction by 58 and 40 under the MPC and MVMscenarios respectively (Table 6) Introducing 5 pumas every80 years cost the least but improvements in populationpersistence under this strategy were insubstantial (Table 6)Releasing more panthers more frequently caused increasedpopulation fluctuations but did not necessarily reduce the quasi‐extinction probability (Figs 14 and 15 Table 6)

DISCUSSION

The duration (34 yr of data) and sample sizes associated with theFlorida Panther Project allowed us to obtain a comprehensiveunderstanding of genetic variation demographic parameters

A

B

C

Figure 8 Cause‐specific mortality rates (plusmnSE) for male and female Florida panthers (A) subadult prime‐adult and old‐adult Florida panthers of both sexes (B)and canonical and admixed Florida panthers of both sexes (C) in South Florida USA 1981ndash2013 Causes of mortality are vehicle collision intraspecific aggression(ISA) other causes (known causes such as diseases injuries and infections unrelated to the first 2 causes) and unknown cause (mortalities for which we could notassign a cause)

van de Kerk et al bull Florida Panther Population and Genetic Viability 21

probability of population persistence and the duration of thepositive impacts of genetic restoration on this endangered po-pulation The Florida panther population was in dire straits inthe early 1990s and our results clearly demonstrated that the

implementation of the introgression project in 1995 subse-quently accrued a series of genetic and demographic improve-ments that have led to the change from a declining population toone that is growing and expanding its current breeding range

Sensitivity Elasticity

0 1 2 3 00 02 04

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

Para

met

er

Figure 9 Sensitivity and elasticity (proportional sensitivity) of stochastic population growth rate (λs) of the Florida panther to vital demographic parameters in SouthFlorida USA 1981ndash2013 Horizontal lines represent 5th and 95th percentiles

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

70

80

90

100

110

0

5

10

15

0

50

100

Years

StatisticMeanMedian

Figure 10 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the minimum population count(MPC) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

125

150

175

200

00

05

10

15

20

0

30

60

90

Years

StatisticMeanMedian

Figure 11 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the motor vehicle mortality(MVM) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

22 Wildlife Monographs bull 203

MPC MVM

minus08 minus06 minus04 minus02 00 minus08 minus06 minus04 minus02 00

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Male subadult survival

Male prime adult survival

Male old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

PRCC

Para

met

er

Figure 12 Partial rank correlation coefficient (PRCC) between quasi‐extinction probabilities and the demographic parameters of the Florida panther for theminimum population count (MPC) and motor vehicle mortality (MVM) scenarios Error bars indicate standard deviation Based on data collected in South FloridaUSA 1981ndash2013

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 13 Average projected individual heterozygosities in the Florida panther population over 200 years without genetic management intervention and with eachof the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include theintroduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years Average individual heterozygosity peaks when pumas are added to the Floridapanther population Based on data collected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 23

Our research has further validated the success of genetic in-trogression by quantifying the reduction in the probability ofextinction of the panther population because of this manage-ment initiative These findings all bode well for continuedprogress toward recovery That said the Florida panther po-pulation remains small and isolated from other populations ofpuma from western North America thereby denoting that theeventual impact of genetic drift and inbreeding may negativelyaffect the population in the future Realizing this will continueto be an issue that managers will have to contemplate wemodeled the impact of varied genetic management scenarios todetermine the frequency of introgression along with the numberof panthers that should be released during each initiative tominimize cost and maximize benefits to the population Theseresults will prove invaluable to managers attempting to conservethe Florida panther

Genetic Variation Demographic Parametersand Persistence of Heterotic Benefits from GeneticIntrogressionOur measure of individual heterozygosity (1 minusHL) was higheron average for the admixed (0615) panthers than for those ofcanonical ancestry (0386) Levels of individual heterozygosityfor admixed panthers were comparable to levels observed in asample of pumas from west Texas (x plusmn SE= 0695plusmn 0019n= 41) which further demonstrates the improvement in levelsof genetic variation in the Florida population since 1995 The

correlation between HL and several demographic parametershas been noted in wild populations including cheetahs (Terrellet al 2016) and several species of birds such as European shags(Phalacrocorax aristotelis male and female survival probabilityVelando et al 2015) and Egyptian vulture (Neophron percnop-terus age at recruitment Agudo et al 2012) If we focus only onpanthers captured and radiocollared from 2012ndash2015 (FP211ndashFP240) all of which had estimated birth years ge2005 (ge10 yrpost‐introgression) those panthers had an average individualheterozygosity level of 0618plusmn 0029 highlighting the elevatedlevels of genetic variation that continue to persist in cohorts ofpanthers 5 generations after the release of female pumas fromTexasThe proportional membership values (q) from our

STRUCTURE analysis allowed us to delineate 2 clusters ofancestry for Florida panthers (Fig 4) During the pre‐in-trogression era the population was dominated by panthers thatretained a high percentage of the canonical ancestry which wasaffected by inbreeding depression The post‐introgression eraancestry values are comprised of many admixed individuals inessence demonstrating the success of the introgression in termsof increasing genetic variation in the population and mi-micking gene flow that historically occurred with panthers andother populations of pumas (Onorato et al 2010) A somewhatanalogous situation although abetted by natural immigrationwas evident in the ancestral variation in a small population ofpuma intersected by a major freeway in California USA In

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

75

80

85

90

95

100

130

150

170

190

Years

Popu

latio

n si

ze

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 14 Average projected population sizes in the Florida panther population over 200 years without intervention and with each of the genetic introgressionstrategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include the introduction of 5 10 or 15pumas from Texas USA every 10 20 40 or 80 years Peaks occur when pumas are added to the Florida panther population Based on data collected in SouthFlorida USA 1981ndash2013

24 Wildlife Monographs bull 203

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

25

50

75

100

0

25

50

75

100

Introgression interval (yrs)

PQ

E (

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 15 Average probability of quasi‐extinction (PQE) of the Florida panther population within 100 years and within 200 years without genetic managementintervention (N) and with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenariosIntrogression strategies include the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Errorbars indicate 5th and 95th percentiles Based on data collected in South Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

50

60

70

80

90

100

110

50

100

150

200

Years

Popu

latio

n si

ze

Figure 16 Average projected Florida panther population trajectories under a geneticintrogression strategy involving the release of 5 female pumas from Texas USA every20 years for the minimum population count (MPC) and motor vehicle mortality(MVM) scenarios Shaded area indicates 5th and 95th percentiles and isrepresentative of confidence intervals for other scenarios Based on data collected inSouth Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Figure 17 Average projected Florida panther population‐level heterozygositiesunder a genetic introgression strategy involving the release of 5 female pumas fromTexas USA every 20 years for the minimum population count (MPC) and motorvehicle mortality (MVM) scenarios Shaded area bounds the 5th and 95th percentilesand is representative of confidence intervals for other scenarios Based on datacollected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 25

this case the migration of just a single male across the freewayto the south resulted in an increase in the number of in-dividuals with admixed ancestry and substantially improvedgenetic diversity of the subpopulation south of the freeway(Riley et al 2014)Our estimate of overall kitten survival (0321plusmn 0056) was

similar to that reported by Hostetler et al (2010) who used13 years of post‐introgression data (0323plusmn 0071) These valuesare lower than kitten survival estimates derived from severalpuma populations in western North America (044ndash091 Loganand Sweanor 2001 Lambert et al 2006 Laundre et al 2007Robinson et al 2008 Ruth et al 2011) When we included onlydata for individuals that had been genetically sampled our es-timate was 15 higher The proportion of admixed kittens thatwere genetically sampled increased as the population expandedwhich could have resulted in higher estimates of kitten survivalbecause admixed kittens survive better than canonical kittensKitten survival was strongly influenced by genetic ancestry withsurvival rates of F1 kittens being almost twice that of canonicalkittens (Fig 5B) Consistent with the findings of Hostetler et al(2010) increases in heterozygosity coincided with increasedkitten survival whereas population density negatively affectedkitten survival Population density often has a strong negativeimpact on survival of young age classes due to infanticide andcannibalism which has been reported from several carnivore

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

50

100

150

0

50

100

150

Introgression interval (yrs)

TQE

(yrs

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 18 Average times until quasi‐extinction (TQE) for the Florida panther population within 100 years and within 200 years without genetic management interventionand with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategiesinclude the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Error bars indicate 5th and 95thpercentiles Based on data collected in South Florida USA 1981ndash2013

Table 5 Average cost (2017 US dollars) of introgression strategies employedover 100 years that involve the introduction of 5 10 or 15 pumas from TexasUSA into the Florida panther population in South Florida USA at an in-creasingly higher frequency Costs of capture (including transportation) quar-antine and post‐introgression monitoring were estimated by Roy McBrideMark Cunningham and Dave Onorato respectively

Frequency Component 5 pumas 10 pumas 15 pumas

Once Capture 25000 50000 75000

Quarantine 5000 10000 15000

Monitoring 10000 20000 30000

Total 40000 80000 120000

Every 80 years Capture 31250 62500 93750

Quarantine 6250 12500 18750

Monitoring 12500 25000 37500

Total 50000 100000 150000

Every 40 years Capture 62500 125000 187500

Quarantine 12500 25000 37500

Monitoring 25000 50000 75000

Total 100000 200000 300000

Every 20 years Capture 125000 250000 375000

Quarantine 25000 50000 75000

Monitoring 50000 100000 150000

Total 200000 400000 600000

Every 10 years Capture 250000 500000 750000

Quarantine 50000 100000 150000

Monitoring 100000 200000 300000

Total 400000 800000 1200000

26 Wildlife Monographs bull 203

species including polar bears (Ursus maritimus Derocher andWiig 1999) black bears (Ursus americanus Garrison et al 2007)and pumas (Logan and Sweanor 2001)Our estimates of sex‐ and age‐specific survival rates of subadult

and adult panthers (Table 3) and the effects of covariates on theserates were similar to those reported by Benson et al (2011) derivedfrom data collected through 2006 Our survival rates for males(065ndash077) were lower than females (078ndash097) for all 3 ageclasses (subadult prime adult and old adult) which is the typicaltrend observed in most puma populations (eg Ruth et al 19982011) However an unhunted puma population occupying afragmented urban landscape in southern California exhibited lowersurvival (0586 females 0525 males Vickers et al 2015) perhapsas a result of severe habitat fragmentation and the lack of extensiveswaths of public land protected in perpetuity Most populations ofpuma in western North America are typically subject to variedlevels of management via regulated hunting which often is biasedtoward the take of males (Cooley et al 2009 Ruth et al 2011)Consequently annual survival estimates from hunted populationsin northeast Washington (0679ndash0733 females 0341 malesRobinson et al 2008) and southeastern Arizona (0685 females0584 males Cunningham et al 2001) were generally lower thanthose we estimated for Florida panthersCause‐specific mortality analysis showed that male panthers

experienced a substantially greater mortality risk from in-traspecific aggression This differs from the pattern observedduring a long‐term study in Arizona where females were at ahigher risk of intraspecific mortality than males (46 of maleand 53 of female mortality Logan and Sweanor 2001)Mortality associated with intraspecific strife has been docu-mented in hunted and unhunted populations (this study Loganand Sweanor 2001 Stoner et al 2010) and its root cause hasbeen difficult to determine (eg population density and preypopulation trends) That being the case finding ways to reducewhat is often a major source of mortality for puma populationscontinues to pose a challenge to managersCanonical panthers were substantially more likely to die from

intraspecific aggression than were admixed panthers This mayat least partly explain the difference in survival between admixed

and canonical panthers Canonical panthers are known to bemore likely to suffer from varied correlates of inbreeding de-pression than admixed animals including atrial septal defects(Johnson et al 2010) and compromised immune systems(Roelke et al 1993) These factors have the potential to affectfitness and perhaps dominance of individuals when engaged inan intraspecific encounter A somewhat similar scenario wasobserved by Hogg et al (2006) in a population of bighorn sheepthat were part of an introgression project Admixed males in thispopulation had a greater probability of paternity than males thatwere less outbred This included coursing males with higherlevels of admixture being markedly more successful at fightingmales that were defending females in estrous (Hogg et al 2006)Heterosis resulting from genetic introgression is expected to be

the strongest in F1 individuals (Shull 1908 Crow 1948 Burkeand Arnold 2001 Waller 2015) and to progressively decline insubsequent generations (Pickup et al 2013 Frankham 2015)Given that admixed (especially F1) panthers survived better thancanonical panthers and that individual heterozygosity improvedsurvival of both sexes and all age classes our data provide evi-dence for heterotic advantage whereby individuals of mixedancestry had higher fitness (Shull 1908 Crow 1948) Our resultsare consistent with findings of other studies that involved in-tentional genetic introgression for conservation purposes Forexample Hogg et al (2006) detected substantial improvementsin survival and reproduction of introgressed bighorn sheep andMadsen et al (1999 2004) reported a dramatic increase in thepopulation of adders after genetic introgressionKnowledge of the persistence of benefits to a population ac-

crued via genetic introgression is essential for determiningwhen additional introgression may be warranted There is adepauperate amount of information regarding this topic and theidentification of factors that may influence persistence of thebenefits of genetic introgression (Tallmon et al 2004 Hedricket al 2014 Whiteley et al 2015) For example in minks(Neovison vison) Thirstrup et al (2014) found that outcrossingled to larger litters in F2 and F3 but this benefit disappeared insubsequent generations In a population of gray wolves Hedricket al (2014) suggested that benefits of genetic introgression may

Table 6 Costs and benefits accumulated over 100 years for genetic introgression at different intervals and with different numbers of pumas from Texas USAintroduced into the Florida panther population in South Florida USA Costs (2017 US dollars) include capture and transportation quarantine and post‐introgression monitoring Benefits are represented as average probability of quasi‐extinction (PQE and 5th and 95th percentiles) and changes (Δ) therein under thescenario using the minimum population count (McBride et al 2008 MPC) and the scenario using the motor vehicle mortality population estimates (McClintocket al 2015 MVM) The table is sorted by percent change in probability of quasi‐extinction under the MPC scenario

Interval (yr) Pumas Cost MPC PQE () ΔMPC PQE () MVM PQE () ΔMVM PQE ()

ndash 0 0 13 (0ndash99) 0 17 (0ndash100) 080 10 100000 12 (0ndash99) 10 (0ndash58) 16 (0ndash100) 5 (0ndash38)80 15 150000 12 (0ndash99) 13 (0ndash65) 15 (0ndash100) 8 (0ndash56)80 5 50000 12 (0ndash99) 14 (0ndash73) 15 (0ndash99) 9 (0ndash41)40 15 300000 10 (0ndash98) 26 (0ndash104) 14 (0ndash99) 13 (0ndash60)40 10 200000 9 (0ndash94) 35 (0ndash183) 13 (0ndash99) 21 (0ndash95)40 5 100000 8 (0ndash89) 39 (0ndash211) 12 (0ndash99) 26 (0ndash124)20 15 600000 8 (0ndash88) 42 (0ndash257) 13 (0ndash99) 24 (0ndash123)20 10 400000 6 (0ndash67) 54 (0ndash318) 10 (0ndash99) 37 (0ndash217)10 15 1200000 6 (0ndash53) 58 (0ndash372) 10 (0ndash99) 40 (0ndash208)20 5 200000 5 (0ndash50) 59 (0ndash338) 9 (0ndash99) 44 (0ndash352)10 10 800000 4 (0ndash16) 69 (0ndash489) 8 (0ndash92) 55 (0ndash401)10 5 400000 4 (0ndash7) 73 (0ndash502) 6 (0ndash71) 63 (0ndash503)

van de Kerk et al bull Florida Panther Population and Genetic Viability 27

wane after 2 or 3 generations The WrightndashFisher model ofgenetic drift predicts a geometric decline in expected hetero-zygosity at the rate of (1 minus 12Ne) per generation where Ne is theeffective population size (Hartl and Clark 1997 Frankham et al2014) Thus it seems logical to assume that the duration of thebenefits of introgression is limited especially in a species per-sisting in a small population such as Florida panthersWe expected Florida panther survival in the most recent years

(2005ndash2013) post‐introgression to differ from that observed in thedecade following the introgression in 1995 because pantherabundance and heterozygosity factors known to influence panthersurvival (Hostetler et al 2010 Benson et al 2011) have changed(Figs 2 and 6) We found that subadult and adult survival rates inrecent years (2005ndash2013) were similar to those in the period im-mediately following introgression (1995ndash2004 Fig 5C) overallkitten survival was similar to estimates of Hostetler et al (2010) Infact the pattern of covariate effects on Florida panther survivalreported by Benson et al (2011) and Hostetler et al (2010) hasbarely changed The negative effects of population density andpositive effects of heterozygosity may counterbalance survival ratesreducing population fluctuations Our result that benefits of geneticrestoration have remained in the population nearly for 5 genera-tions post‐intervention was surprising but also encouraging Pos-sible explanations for this observation may include 1) genetic ero-sion occurs at slower rate than previously thought 2) introducing 5migrants in 1 genetic intervention may have beneficial effects si-milar to those from 1 migrant per generation over multiple gen-erations or 3) other and as yet unknown factors may reduce therate of genetic erosion and subsequent demographic effects Adensity‐dependent effect on kitten survival could also explain whynon‐F1 admixed kittens did not survive substantially better thandid canonical kittens most kittens sampled from 2005ndash2013 wereadmixed and their survival was lower possibly because of a density‐dependent effect as the Florida panther population continuouslygrew during that period Trinkel et al (2010) also found thatpopulation density and inbreeding coefficient interacted to affectdemographic parameters and growth rate of an African lion po-pulation Likewise population density interacted with winterweather to affect the survival and population growth rates in Soaysheep (Ovis aries Milner et al 1999 Coulson et al 2001)

Population Dynamics and PersistenceThe finite deterministic and stochastic growth rates were gt10reflecting that much of the data used in this study were collectedfollowing genetic introgression in 1995 during which time thepopulation increased substantially Because our demographicparameter estimates did not change markedly in comparison tothose of Hostetler et al (2013) the similarity between thepopulation growth estimates from these studies was expectedBut these estimates reflect population growth during thedata‐collection period and do not predict future growth In factestimates of population size reported by McClintock et al(2015) suggest that population growth may already be slowing(Fig 2) A similar pattern was observed in an introduced po-pulation of African lion which increased steadily from 1995until 2001 then fluctuated around the presumed carrying ca-pacity thereafter (Trinkel et al 2010) Several other African lionpopulations that experienced various degrees of recovery

subsequently became relatively stable or exhibited decliningtrends (Bauer et al 2015)Our estimates of quasi‐extinction probabilities were lower than

those reported by Hostetler et al (2013) using a 2‐sex age‐structured matrix population model Lower probabilities ofquasi‐extinction than those reported by the earlier study forcomparable scenarios were likely a consequence of the fact thatthe estimates of abundance (McBride et al 2008) and thus thedensity‐dependent effects on survival of kittens and old panthers(gt10 yr) have changed in recent years Additionally these earlyestimates of the probability of quasi‐extinction and of time toquasi‐extinction did not consider genetic erosion that will in-evitably occur without management interventionUnder the MVM scenario the equilibrium population size was

twice that attained under the MPC scenario The MVM popula-tion estimates are approximately twice the MPCs (Fig 2) as aresult the simulated population under the MVM scenario achieveda higher equilibrium size Probability of quasi‐extinction under theMVM scenario was generally greater than that under the MPCscenario This result is a consequence of the fact that the estimateddensity dependence on kitten survival based on the MPC scenario isstronger than that for theMVM scenario (regression coefficients forabundance for the MPC scenario= minus0068 vs minus0002 for theMVM scenario Appendix B Table B1) Consequently simulatedpopulations under the MPC scenario have a stronger tendency tomove toward the equilibrium population size which inherentlyreduces the quasi‐extinction probability (eg Royama 1992)As expected for long‐lived mammals (Heppell et al 2000 Oli

and Dobson 2003) the population growth rate was proportio-nately most sensitive to changes in survival of prime adultsfollowed by that in younger age classes Global sensitivity ana-lysis in contrast showed that under all scenarios extinctionparameters were particularly sensitive to changes in survival ofFlorida panther kittens This result is a consequence of the highsensitivity of the population growth rate to kitten survival(Fig 9) and a larger standard error of kitten survival (Table 3)Results of our sensitivity analyses indicate that future researchshould focus on acquiring additional information on early lifestages to improve the accuracy and precision of estimates ofkitten survival Our results suggest that demographic variables(or their environmental drivers) that strongly influence extinc-tion risks are not necessarily those with the largest elasticityvalues A study of island fox (Bakker et al 2009) found thatextinction risk was strongly influenced by survival modifierparameters that had low elasticity values

Genetic Management When and How to InterveneThe use of genetic introgression as a management tool has al-ways been controversial and the probable effectiveness it mighthave in preventing the imminent demise of the panther popu-lation was initially debated (Creel 2006 Maehr et al 2006Pimm et al 2006) These debates notwithstanding the pantherpopulation had suffered from genetic morphological and bio-medical correlates of inbreeding before this management in-tervention (Johnson et al 2010 Onorato et al 2010) whereasthe post‐introgression period has been characterized by a sub-stantial reduction in the biomedical correlates of inbreeding andimprovements in demographic vigor and abundance (McBride

28 Wildlife Monographs bull 203

et al 2008 Hostetler et al 2010 2013 Johnson et al 2010Benson et al 2011 McClintock et al 2015) Nevertheless thepopulation remains small and isolated habitat is still being lostand there is no current possibility of natural gene flow betweenthis and other North American puma populations thus therecurrence of inbreeding‐related problems and subsequent po-pulation declines are inevitable The question therefore is notwhether genetic management of the Florida panther populationis needed but when and how it should be implementedWithout genetic introgression probabilities of quasi‐extinc-

tion were substantially greater than those from models thatincorporated periodic genetic introgression events (Fig 15)highlighting the importance of genetic management in smallisolated populations When 5 female pumas are added to theFlorida panther population every 10 years genetic variationincreased substantially under the MPC and MVM scenariosThe IBM predicted that the equillibrium population size wouldbe substantially larger when 5 pumas were released into thepopulation every 10 years than populations without intervention(ie no introgression) Releasing 15 Texas females did not af-ford substantially greater benefit to the equilibrium populationsize than did releasing only 5 Texas females Instead addingmore individuals caused increasingly large fluctuations in po-pulation size due to the numerical increases and density de-pendence (Fig 14) possibly diminishing the benefits associatedwith increased genetic variationCompared with the no‐intervention scenario (ie no genetic

introgression implemented) the introduction of 5 female pumasevery 10 years reduced quasi‐extinction probabilities from 13to 4 and from 17 to 6 for the MPC and MVM scenariosrespectively The benefit of genetic‐introgression strategies de-creased as the interval between successive management inter-ventions increased and no benefit was evident once the intervalreached 80 years (Fig 15) Introgression reduced the averagetime until quasi‐extinction (Fig 18) This somewhat counter‐intuitive result can be explained by the fact that repeated geneticintrogression makes the population more robust over timewhich will reduce the probability of extinction Consequentlyfew simulated population trajectories that fall below the criticalthreshold do so early reducing the mean time to extinctionAlso density dependence has a stabilizing effect on populations(Royama 1992) populations tend toward equilibrium popula-tion sizes which reduces the probability over time of popula-tions falling below quasi‐extinction threshold However timeuntil quasi‐extinction must be interpreted in conjunction withthe probability of quasi‐extinction which is very low for allscenarios with genetic introgression (Fig 15)Cost is a significant impediment to the implementation of a

genetic introgression program because captures relocations andmonitoring efforts before and after introgression are expensive(Edmands 2007 Frankham 2010b 2015 2016) Costs may beacceptable to society if the management actions result in sub-stantial fitness benefits especially if the species of concern ispopular and charismatic (Frankham 2015) We estimated the100‐year cost of introgression strategies modeled here to rangefrom $50000 to $12 million More expensive strategies gen-erally led to larger decreases in the probability of quasi‐extinc-tion but cheaper strategies also substantially improved

population viability (ie reduced quasi‐extinction probabilityTable 6) One of the cheaper strategies (5 pumas released every20 years) which cost an estimated $200000 over 100 yearsreduced the probability of quasi‐extinction by 59 and 44 forthe MPC and MVM scenarios respectively But these pre-liminary cost estimates are based on expenses incurred duringthe 1995 introgression and include costs of capture quarantinetransportation release and post‐release monitoring of femalesonly Although the cost for future genetic interventions wouldlikely be higher than those reported here the relative differencesin cost between alternative intervention strategies should remainapproximately the sameOur ability to simulate genetics and link it to the viability of

the Florida panther population is based on the empirically es-timated relationship between individual heterozygosity andsurvival Such heterozygosity and fitness correlations have beenstudied for decades (David 1998) but have only recently beenused to predict population viability (Benson et al 2016) noneto our knowledge have incorporated cost into their modelingefforts The mechanisms underlying heterozygosity and fitnesscorrelations remain unclear (David 1998 Szulkin et al 2010)and their significance as a proxy for inbreeding depression hasbeen debated (Balloux et al 2004 Miller and Coltman 2014)Nevertheless evidence continues to mount demonstratingpositive relationships between heterozygosity and survival(Coulson et al 1998ab Hansson et al 2004 Silva et al 2005Acevedo‐Whitehouse et al 2006 Jensen et al 2007 Velandoet al 2015) and between heterozygosity and fecundity (Seddonet al 2004 Charpentier et al 2005 Agudo et al 2012 Velandoet al 2015) Although the reported correlations are generallyweak their consequences can be substantial if population growthand persistence parameters are highly sensitive to the affectedvital rates (Velando et al 2015) Compared with scenarios thatignore genetics incorporating the effects of inbreeding on sur-vival increased quasi‐extinction probabilities within a 100‐yeartimeframe from 15 to 13 and from 14 to 17 for theMPC and MVM scenarios respectively These results high-light the importance of incorporating genetics when analyzingthe viability of small isolated populationsAlthough conservation biologists have recognized the im-

portance of genetics in population viability many still fail toincorporate genetic factors into PVA models (Reed et al 2002)Explicit consideration of genetics in PVAs requires long‐termgenetic and demographic data This permits the assessment ofthe functional relationship between genetic variability and de-mographic parameters Population viability analysis softwarepackages such as VORTEX (Lacy 1993 2000) offer a powerfulframework for individual‐based simulations but they often donot adequately capture a speciesrsquo life history or genetics Basedon a thorough review of the importance of genetic factors inwildlife conservation Frankham et al (2014) concluded thatgenetic factors can strongly influence the dynamics and persis-tence of small andor fragmented populations They furthernoted that ldquoMost PVA‐based risk assessments ignore or in-adequately model genetic factorsrdquo and recommended that ldquoPVAshould routinely include realistic inbreeding depressionhelliprdquo(Frankham et al 201457) Our custom‐built IBM permitted usto develop a model that adequately captured the Florida panther

van de Kerk et al bull Florida Panther Population and Genetic Viability 29

life history and we could parameterize the model with our long‐term genetic and demographic dataThe explicit consideration of the effects of inbreeding on

Florida panther population dynamics and persistence representsan important step forward Nonetheless our model remainsimperfect First we neglected the possible effects of habitat lossdetrimental anthropogenic activities climate change the prob-ability of immigration of pumas from western populationsdisease or catastrophes on demographic rates and populationviability Although immigration from puma populations outsideFlorida is possible its probability is very low given the extensivechallenges the anthropogenically altered landscape would posefor such a migrant to make it successfully to South Florida Weomitted mutations from our model because we have no in-formation on mutation rates though we expect that they are lowbased on values reported for other mammals (Kumar and Sub-ramanian 2002) Finally we only considered possible effects ofinbreeding on age‐specific survival rates because there was noevidence that heterozygosity affected female reproduction Lossof heterozygosity can negatively affect reproduction becauseinbred male panthers can suffer from cryptorchidism which canadversely affect their fecundity However given the polygynousmating system we expect the Florida panther population growthis primarily female‐limitedAlthough it is generally accepted that genetic introgression

only temporarily relieves a population from inbreeding depres-sion we know of no cases in which it has been implementedmore than once to conserve a threatened wildlife populationOur study provides an example of how demographic and mo-lecular data can be used within an IBM framework to evaluatethe efficacy of alternative genetic management strategies via theFlorida panther as a case study Our model can be adjusted forand applied to other species and offers great potential as a toolfor genetic management of small isolated wildlife populations

MANAGEMENT IMPLICATIONS

Our results and those of Hostetler et al (2013) and Johnsonet al (2010) provide persuasive evidence that the genetic in-tervention implemented in 1995 prevented the demise of theFlorida panther and restored demographic vigor to the popu-lation The Florida panther population remains small and iso-lated from other puma populations the closest puma populationis located gt1000 km away in Texas and the intervening land-scape is highly developed and fragmented with many interstate(and other high‐traffic) highways and major urban centersTheory suggests that 1 migrant per generation is sufficient tominimize the loss of genetic diversity (eg Mills and Allendorf1996) However the possibility of successful migration of apuma to South Florida followed by successful mating every ~5years is very low considering the distance between Texas andFlorida (Hedrick 1995) and the many risks and barriers thatdispersing pumas face (Beier 1995 Maehr et al 2002 Thatcheret al 2006) Consequently the pantherrsquos long‐term persistencewill likely depend on subsequent timely genetic introgressionsgiven the near‐impossibility of natural gene flow from otherpuma populations To that end our study offers considerableinsights into benefits of different genetic management strategiesand associated costs

Without genetic management the Florida panther populationfaces a substantial risk of quasi‐extinction within 100 years (10ndash46 depending on quasi‐extinction threshold and scenarios)Twenty‐three years have passed since genetic introgression wasimplemented and the population appears healthy as indicatedby measures of genetic variation increases in the populationsize and improvements in age‐specific survival rates Ourfindings suggest that releasing more pumas more frequently doesnot necessarily improve persistence probability because such astrategy would cause larger fluctuations in population size(Fig 14) which negatively affects persistence probability Thusextinction risks faced by inbred populations of large carnivorescan be substantially reduced by releasing approximately 5 im-migrants every 2 decades We suggest that such a managementstrategy be followed only in conjunction with continued long‐term monitoring of the population Our analyses demonstratethat population growth and persistence are highly sensitive tokitten and female (adult and subadult) survival Continuing tocollect data on these demographic variables along with bio-metric and genetic sampling will remain important to pantherrecovery and vigilance in identifying issues associated with in-breeding depression The collection of these monitoring datashould allow managers to detect any demographic or geneticchanges in a timely fashion and respond with genetic in-trogression or other management actions if warrantedRecovery objectives as defined by the USFWS in its current

recovery plan (USFWS 2008) delineate the need for additionalpopulations in the historical range to downlist or delist theFlorida panther If the only extant population of Floridapanthers continued to grow it could serve as a source ofpanthers to be translocated to suitable habitat to seed addi-tional populations Maintaining current information on thegenetic health of the population and precise estimates of itssize will be important in assessing whether it can withstand theremoval of a subset of animals for translocation Although the2017 documentation of a female panther with kittens north ofthe current breeding range is encouraging in terms of thepotential for population expansionmdashgiven that no female hadbeen recorded north of the Caloosahatchee River since 1972mdashthe re‐establishment of separate new populations will mostlikely require translocations by wildlife managers and a lengthysociopolitical processConservation of threatened or endangered species such as the

Florida panther is inherently challenging For panthers and otherlarge carnivores that require vast extents of natural habitat topersist habitat is typically the most limiting factor that will de-termine whether recovery efforts succeed Although geneticmanagement invariably plays a key role in the long‐term persis-tence of the panther population even a healthy population caneventually succumb to the impacts of habitat loss The humanpopulation of Florida continues to be one of the fastest‐growingin the nation the United States Census Bureau currently ranksFlorida as the third most populous Therefore habitat loss isgoing to continue to be a key issue for panthers Diligence towardpreserving panther habitat in its historical range via conservationeasements or other means and trying to keep such areas inter-connected via corridors is a goal stakeholders such as governmentagencies sportsmen non‐governmental organizations ranchers

30 Wildlife Monographs bull 203

and other private landowners must work on collaboratively toachieve

ACKNOWLEDGMENTS

We thank D Shindle D Jennings T OrsquoMeara D Land andC Belden for valuable advice throughout the project We thankS Bass M Criffield M Cunningham D Giardina D JansenA Johnson D Land M Lotz C McBride Rocky McBrideRowdy McBride Roy McBride L Oberhofer S Schulze staffsat Everglades National Park and Big Cypress National Preserveand others for assistance with fieldwork We also thank JAustin M Conroy J Hines J Laake S Mills J Nichols JHines M Sunquist J Fryxell and T Coulson for advice andassistance regarding data analysis and panther biology TheMuseum of Texas Tech University Natural Science ResearchLaboratory provided samples from pumas from west Texas forcomparative genetic analyses M Ben‐David D Land WMcCown E Hellgren and 4 anonymous reviewers providedmany helpful comments on the manuscript We thank NereaUbierna Lopez for completing the Spanish translation of theabstract We are grateful to the Department of ResearchComputing University of Florida for excellent high‐perfor-mance computing services We would like to thank US Fishand Wildlife Service Florida Fish and Wildlife ConservationCommission (FWC) US National Park Service QuantitativeSpatial Ecology Evolution and Environment (QSE3) In-tegrative Graduate Education and Research Traineeship(IGERT) program funded by the National Science Foundationand the University of Florida for financially supporting thisstudy Funding for panther research conducted by the FWC issupported predominantly via donations accrued with vehicleregistrations for ldquoProtect the Pantherrdquo license plates

LITERATURE CITEDAcevedo‐Whitehouse K T R Spraker E Lyons S R Melin F Gulland RL Delong and W Amos 2006 Contrasting effects of heterozygosity onsurvival and hookworm resistance in California sea lion pups MolecularEcology 151973ndash1982

Adams J R L M Vucetich P W Hedrick R O Peterson and J AVucetich 2011 Genomic sweep and potential genetic rescue during limitingenvironmental conditions in an isolated wolf population Proceedings of theRoyal Society B Biological Sciences 2783336ndash3344

Agresti A 2013 Categorical data analysis Third edition John Wiley amp SonsHoboken New Jersey USA

Agudo R M Carrete M Alcaide C Rico F Hiraldo and J A Donaacutezar2012 Genetic diversity at neutral and adaptive loci determines individualfitness in a long‐lived territorial bird Proceedings of the Royal Society BBiological Sciences 2793241ndash3249

Albeke S E N P Nibbelink and M Ben‐David 2015 Modeling behavior bycoastal river otter (Lontra canadensis) in response to prey availability in PrinceWilliam Sound Alaska a spatially‐explicit individual‐based approach PLOSOne 10e0126208

Alho J S K Vaumllimaumlki and J Merilauml 2010 Rhh an R extension for estimatingmultilocus heterozygosity and heterozygosity‐heterozygosity correlationMolecular Ecology Resources 10720ndash722

Alvarez K 1993 Twilight of the panther Myakka River Publishing SarasotaFlorida USA

Aparicio J M J Ortego and P J Cordero 2006 What should we weigh toestimate heterozygosity alleles or loci Molecular Ecology 154659ndash4665

Bakker V J D F Doak G W Roemer D K Garcelon T J Coonan S AMorrison C Lynch K Ralls and R Shaw 2009 Incorporating ecologicaldrivers and uncertainty into a demographic population viability analysis for theisland fox Ecological Monographs 7977ndash108

Balloux F W Amos and T Coulson 2004 Does heterozygosity estimateinbreeding in real populations Molecular Ecology 133021ndash3031

Barone M A M E Roelke J Howard J L Brown A E Anderson and DE Wildt 1994 Reproductive characteristics of male Florida panthersmdashcomparative studies from Florida Texas Colorado Latin‐America andNorth‐American Zoos Journal of Mammalogy 75150ndash162

Bateson Z W P O Dunn S D Hull A E Henschen J A Johnson and LA Whittingham 2014 Genetic restoration of a threatened population ofgreater prairie‐chickens Biological Conservation 17412ndash19

Bauer H G Chapron K Nowell P Henschel P Funston L T B HunterD W Macdonald and C Packer 2015 Lion (Panthera leo) populations aredeclining rapidly across Africa except in intensively managed areas Pro-ceedings of National Academy of Sciences of the United States of America11214894ndash14899

Beier P 1995 Dispersal of juvenile cougars in fragmented habitat Journal ofWildlife Management 59228ndash237

Benson J F J A Hostetler D P Onorato W E Johnson M E Roelke SJ OrsquoBrien D Jansen and M K Oli 2011 Intentional genetic introgressioninfluences survival of adults and subadults in a small inbred felid populationJournal of Animal Ecology 80958ndash967

Benson J F P J Mahoney J A Sikich L E K Serieys J P Pollinger H BErnest and S P D Riley 2016 Interactions between demography geneticsand landscape connectivity increase extinction probability for a small popu-lation of large carnivores in a major metropolitan area Proceedings of theRoyal Society B Biological Sciences 28320160957

Beverly F 1874 The Florida panther Forest and Stream 3290Boyce M S C V Haridas C T Lee and the NCEAS Stochastic Demo-graphy Working Group 2006 Demography in an increasingly variable worldTrends in Ecology amp Evolution 21141ndash148

Brook B W J J OrsquoGrady A P Chapman M A Burgman H R Akcakayaand R Frankham 2000 Predictive accuracy of population viability analysis inconservation biology Nature 404385ndash387

Brys R and H Jacquemyn 2016 Severe outbreeding and inbreeding de-pression maintain mating system differentiation in Epipactis (Orchidaceae)Journal of Evolutionary Biology 29352ndash359

Burke J M and M L Arnold 2001 Genetics and the fitness of hybridsAnnual Review of Genetics 3531ndash52

Burnham K P 1993 A theory for combined analysis of ring recovery andrecapture data Pages 199ndash213 in J‐D Lebreton and P M North editorsMarked individuals in the study of bird population Springer‐Verlag NewYork New York USA

Burnham K P and D R Anderson 2002 Model selection and multimodelinference a practical information‐theoretic approach Second edition SpringerVerlag New York New York USA

Caswell H 2001 Matrix population models construction analysis and in-terpretation Sinauer Associates Sunderland Massachusetts USA

Charpentier M J M Setchell F Prugnolle L A Knapp E J Wickings PPeignot and M Hossaert‐McKey 2005 Genetic diversity and reproductivesuccess in mandrills (Mandrillus sphinx) Proceedings of the NationalAcademy of Sciences of the United States of America 10216723ndash16728

Cooch E and G C White editors 2018 Program MARK a gentle in-troduction lthttpwwwphidotorgsoftwaremarkdocsbookgt Accessed 7Apr 2016

Cooley H S R B Wielgus G M Koehler H S Robinson and B TMaletzke 2009 Does hunting regulate cougar populations A test of thecompensatory mortality hypothesis Ecology 902913ndash2921

Coulson T E A Catchpole S D Albon B J Morgan J M Pemberton TH Clutton‐Brock M J Crawley and B T Grenfell 2001 Age sex densitywinter weather and population crashes in Soay sheep Science 2921528ndash1531

Coulson T J Pemberton S Albon M Beaumont T Marshall J Slate FGuinness and T Clutton‐Brock 1998a Microsatellites reveal heterosis in reddeer Proceedings of the Royal Society B Biological Sciences 265489ndash495

Coulson T N S D Albon J M Pemberton J Slate F E Guinness and TH Clutton‐Brock 1998b Genotype by environment interactions in wintersurvival in red deer Journal of Animal Ecology 67434ndash445

Cox D 1972 Regression models and life‐tables Journal of the Royal StatisticalSociety Series B Statistical Methodology 34187ndash220

Creel S 2006 Recovery of the Florida panthermdashgenetic rescue demographic rescueor both Response to Pimm et al (2006) Animal Conservation 9125ndash126

Crnokrak P and D A Roff 1999 Inbreeding depression in the wild Heredity83260ndash270

Crow J 1948 Alternative hypotheses of hybrid vigor Genetics 33477ndash487

van de Kerk et al bull Florida Panther Population and Genetic Viability 31

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 14: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

survival and reproductive rates and the effects of covariates onthese rates for the released Texas females were identical to thoseof female Florida panthersThe impact of genetic introgression depends on the frequency

of introgression events and the number of individuals introducedat each attempt Thus we defined introgression strategies basedon the number of Texas females to be released (0 5 10 or 15)and the interval between genetic introgressions (10 20 40 and80 yr) for a total of 13 strategies We simulated each manage-ment strategy with density dependence estimated using theMPC and MVM abundance indices

We sampled 1000 parameter sets and for each of these setswe ran all introgression strategies 1000 times for 200 years Weapplied the same parametric uncertainty and environmentalstochasticity across all management scenarios to allow for bettercomparison of introgression strategies At each time step werecorded the projected population size and average individualheterozygosity We calculated the heterozygosity for each in-dividual at birth based on the alleles it inherited This individualheterozygosity then informed the survival rate of that individualbased on the empirically estimated relationship between in-dividual heterozygosity and age‐specific survival rates Hetero-zygosity did not change throughout an individualrsquos lifetime butits effect on survival changed as individuals aged because theeffect of individual heterozygosity on survival rate differedamong age classes Survival rates of genetically sampled kittenswere biased high because some kittens were sampled later in lifewhen they had already survived for some time To correct forthis bias we adjusted kitten survival rates predicted by theheterozygosity model proportionally From the population tra-jectories we estimated the probability of quasi‐extinction (de-fined as the probability that the population will fall below 10 or30 individuals or that individuals of only 1 sex will remain)

Cost‐benefit analysismdashExperts estimated financial costs ofintrogression based on their experience with the geneticintrogression executed in 1995 and we adjusted estimates tocurrent costs to account for inflation (R T McBride RancherrsquosSupply Incorporated M Cunningham and D P OnoratoFlorida Fish and Wildlife Conservation Commission personalcommunication) Costs included capturing and transportingfemale pumas from the Texas source population caring for thepumas while they were in quarantine and post‐introgressionmonitoring To compare the financial costs of the differentscenarios we also calculated the average costs incurred over

Table 2 Sample size (n) number of alleles (Na) allelic richness (AR) observedheterozygosity (Ho) and expected heterozygosity (He) at 16 microsatellite loci inradio‐collared Florida panthers sampled from 1981 to 2013 in South FloridaUSA We calculated these values from a subset (n= 137) of the samples used inthe analyses and they include only adult and subadult panthers We excludedkittens because they can result in an overrepresentation of certain alleles

Locus n Na AR Ho He

FCA090 129 5 50 0333 0401FCA133 136 3 30 0618 0608FCA243 136 5 50 0419 0487F124 137 7 69 0708 0729F37 129 4 40 0395 0397FCA075 131 5 50 0534 0598FCA559 133 5 50 0496 0503FCA057 134 6 59 0679 0624FCA081 134 7 69 0209 0206FCA566 130 6 60 0462 0475F42 133 10 100 0737 0766FCA043 136 5 49 0596 0588FCA161 132 6 60 0500 0515FCA293 132 4 40 0614 0624FCA369 131 6 60 0573 0567FCA668 133 7 70 0406 0462Mean 1329 57 57 0517 0535

Figure 4 Proportional membership (q) of radio‐collared Florida panthers (n= 218) and pumas from Texas USA (n= 7) in 2 clusters identified by STRUCTUREEach individual animal is represented by a separate vertical bar Yellow indicates canonical panther ancestry and blue represents admixed ancestry The admixedancestry of the Texas females and F1 generation panthers that were radiocollared are highlighted red and green respectively to demarcate those groups The pre‐introgression period is inclusive of panthers radiocollared from 10 February 1981 to 6 March 1996 The post‐introgression era inclusive of the F1 panthers includesanimals radiocollared from 4 March 1997 to 18 February 2015 in South Florida USA

16 Wildlife Monographs bull 203

100 years assuming that the population would be managedaccording to a particular strategy Our goal with theseexploratory analyses was to evaluate the relative costs andbenefits of alternative management scenarios

RESULTS

Genetic VariationWe assessed genetic variation at 16 microsatellite loci for 137DNA samples collected from adult and subadult panthers(1981ndash2013) that were captured and subsequently radiocollaredThe number of alleles at these loci ranged from 3ndash10 and allelicrichness averaged 57 (Table 2) Average expected hetero-zygosity for this sample was 0535 and average observed het-erozygosity was 0517 (Table 2)We determined individual heterozygosity (1 minusHL) and an-

cestry for the complete sample of panther genotypes (n= 503)collected from 1981 to 2015 for use as covariates in subsequentanalyses Average individual heterozygosity was 0559 andranged from 0ndash0980 Our STRUCTURE analysis providedstrong support for K= 2 clusters based on the probability of thedata (log Pr(D)|K) and ΔK in STRUCTURE HARVESTERBased on this result we categorized the ancestry of each pantheras either canonical (n= 123) or admixed (n= 380) using valuesof proportional membership (q Fig 4) The average individualheterozygosity of canonical panthers was 0386plusmn 0012 (SE) foradmixed panthers it was 0615plusmn 0007

Survival and Cause‐Specific Mortality RatesOf the 395 kittens that were PIT‐tagged and released 55 werelater encountered alive 15 were recovered dead and 85 werepart of failed litters (Table 1) We radio‐tracked 209 subadult oradult panthers for a total of 244195 panther‐days and docu-mented 126 mortalities (Table 1)Survival depended strongly on age and kittens had the lowest

overall rate (032plusmn 009 Fig 5A Table 3) The model‐averagedkitten survival was 047plusmn 009 when we included only geneti-cally sampled individuals in the analysis (Table 3) as statedabove this estimate is biased high because many kittenswere genetically sampled when they were recaptured alive orrecovered dead at older ages We found no evidence for sex‐specific differences in kitten survival but females survived betterthan males in all older age classes For females subadults hadthe highest survival rates followed by prime adults and oldadults For males prime adults had the highest survival ratefollowed by subadults and old adults (Fig 5A and Table 3)For panthers of all ages there was substantial evidence that

ancestry affected survival with F1 admixed panthers of all ageshaving the highest rates and canonical individuals having thelowest (Fig 5B) The combined AIC weights of models thatincluded an ancestry covariate were 0880 for kittens and 0992for older panthers The survival rates of subadult prime‐adultand old‐adult panthers in the period immediately after the in-trogression (1995ndash2004) were similar to those in more recentyears (2005ndash2013 Fig 5C)After genetic introgression individual panther heterozygosity

increased (Fig 6) and varied among ancestry categories individualheterozygosity was the highest for F1 panthers (078plusmn 001)

A

B

C

Figure 5 Overall age‐ and sex‐specific survival rates (plusmnSE A) age‐ and sex‐specific survival rates (plusmnSE) for Florida panthers of canonical F1 admixed andother admixed ancestry (B) and age‐ and sex‐specific survival estimates (plusmnSE)for subadult and adult Florida panthers in the period immediately post‐introgression (1995ndash2004) and more recently (2005ndash2013 C) in SouthFlorida USA

Table 3 Model‐averaged estimates (plusmnSE) of demographic parameters for theFlorida panther obtained using data collected during 1981ndash2013 in SouthFlorida USA

Parameter Sex Age class Estimate

Survival Both Kitten 032plusmn 009a

Female Subadult 097plusmn 002Prime adult 086+ 003Old adult 078plusmn 009

Male Subadult 066plusmn 006Prime adult 077plusmn 005Old adult 065plusmn 010

Probability of reproduction Female Subadult 035plusmn 008Prime adult 050plusmn 005Old adult 025plusmn 006

Kittens produced annually Female Subadult 280plusmn 005Prime adult 267plusmn 001Old adult 228plusmn 013

a Overall kitten survival (based on all kittens in our data set including thosethat were not genetically sampled) Estimate of survival of geneticallysampled kittens was 047plusmn 009 this estimate is biased high because manykittens were genetically sampled when they were recaptured alive or re-covered dead at older ages

van de Kerk et al bull Florida Panther Population and Genetic Viability 17

followed by those for non‐F1 admixed (062plusmn 001) and canonical(039plusmn 001) panthers Individual heterozygosity positively affectedsurvival rates of kittens (slope parameter η= 1455 95CI= 0505ndash2405) Individual heterozygosity also positively af-fected survival of subadult and adult panthers but the evidence forthis effect was weaker because the confidence interval for the ha-zard ratio overlapped 10 (hazard ratio exp(ɩ)= 0311 95CI= 0092ndash1054 Table 4 and Fig 7)The MPC increased after genetic introgression (Fig 2) This

abundance index was also included in top‐ranking modelsindicating that population density affected survival (Table 4)Abundance reduced the survival of kittens (slope parameterη= minus0035 95 CI= minus0049 to minus0021) evidence was weak forthe effect of abundance on survival of subadult and adult panthers(hazard ratio exp(ɩ)= 1002 95 CI= 0995ndash1010 Fig 7) Re-gression coefficients associated with the effect of abundance andindividual heterozygosity on demographic parameters are presentedin Table B1 (available online in Supporting Information)The most frequently observed causes of death for radio‐

collared panthers were vehicle collision and intraspecific ag-gression Cause‐specific mortality analyses revealed thatmortality rates among possible causes were generally similar forthe 2 sexes but that males were more likely to die from in-traspecific aggression than were females (Fig 8A) Male mor-tality from intraspecific aggression was higher for subadult andold adult panthers than for prime adults (Fig 8B) For bothmales and females canonical panthers were more likely to diefrom intraspecific aggression than were admixed panthers(Fig 8C)

Reproductive ParametersOf 101 radio‐collared females 67 reproduced at least once duringour monitoring we used these individuals to assess the annualprobability of reproduction The top‐ranked model for the annual

probability of reproduction included age effect only suggesting thatadding ancestry or abundance did not improve model parsimony(Table 4) Model‐averaged annual probabilities of reproductiondiffered between female subadults (035plusmn 008) prime adults(050plusmn 005) and older adults (025plusmn 006)The most parsimonious model for the number of kittens

produced annually by females that produced at least 1 litter (ieconditional on reproduction) included effects of age ancestryand abundance a model that included only the effect of age wasalmost equally supported (Table 4) The model‐averagednumber of kittens produced annually conditional on re-production was 280plusmn 075 for subadults 267plusmn 043 for primeadults and 228plusmn 083 for old adults Confidence intervals forthe slope parameter (β) overlapped 0 for both abundance(β= 0010 95 CI= minus0001 to 0021) and ancestry (β= minus073795 CI= minus1637 to 0162)

Population Dynamics and PersistenceDeterministic and stochastic demographymdashThe deterministic (λ)

and stochastic (λs) annual population growth rate calculated usingthe matrix population model were 106 (5th and 95th percentiles099ndash114) and 104 (072ndash141) respectively The generation timewas 47 years A female starting in age class one (kitten) wasexpected to live another 58 years and produce 10 kittens onaverage during her lifetime Overall λs was proportionately(elasticity) most sensitive to changes in female prime adultsurvival on the absolute scale (sensitivity) however it was mostsensitive to changes in kitten survival (Fig 9) Populationtrajectories probabilities of quasi‐extinction and times untilquasi‐extinction estimated using the matrix population modeland IBM for comparable scenarios were nearly identical for acritical quasi‐extinction threshold of 10 panthers (Figs 10 and 11)and for a critical quasi‐extinction threshold of 30 panthers(Appendix E available online in Supporting Information)Minimum population count scenariomdashUnder the MPC scenario

(ie density dependence estimated using the minimumpopulation count data) with a critical threshold of 10panthers the population size reached 99 (72ndash104) and 100(77ndash105) at 200 years (Fig 10) as predicted by the IBM and thematrix model respectively The cumulative quasi‐extinctionprobabilities within 100 years based on the MPC scenario were15 (IBM 0ndash48) and 30 (matrix model 0ndash76) Theseprobabilities increased to 25 (IBM 0ndash114) and 38 (matrixmodel 0ndash154) by year 200 (Fig 10) The mean times untilquasi‐extinction were 15 years (IBM 0ndash67) and 15 years (matrixmodel 0ndash72) conditioned on going quasi‐extinct within 100years Expected times until quasi‐extinction conditioned onquasi‐extinction within 200 years were higher 34 years (IBM0ndash137) and 32 years (matrix model 0ndash134 Fig 10)Motor vehicle mortality scenariomdashUnder the MVM scenario

(ie density dependence estimated using estimates of pantherpopulation size from McClintock et al 2015) with a criticalthreshold of 10 panthers the projected population reached 188(142ndash217) and 190 (143ndash219) at 200 years as predicted by theIBM and the matrix model respectively (Fig 11) Thecumulative quasi‐extinction probabilities within 100 years were14 (IBM 0ndash08) and 13 (matrix model 0ndash06) Theseprobabilities increased to 20 (IBM 0ndash17) and 19 (matrix

000

025

050

075

100

1995 2000 2005 2010Year of birth

Indi

vidu

al h

eter

ozyg

osity

Figure 6 Temporal changes in the individual heterozygosity of Floridapanthers born during the period 1995ndash2013 in South Florida USA Pointsare measurements solid line is linear regression shaded area is 95 confidenceinterval of slope Slope= 0006plusmn 0001 R2= 009

18 Wildlife Monographs bull 203

model 0ndash16) within year 200 (Fig 11) The mean times until quasi‐extinction based on the MVM scenario were 5 years (IBM 0ndash61)and 6 years (matrix model 0ndash64) conditioned on going quasi‐extinctwithin 100 years Expected times until quasi‐extinction conditionedon quasi‐extinction within 200 years were 11 years (IBM 0ndash113)and 11 years (matrix model 0ndash112 Fig 11)

Sensitivity of extinction probability to demographic parametersmdashThe partial rank correlation coefficients between quasi‐extinctionprobabilities and demographic parameters were negative anincrease in any of the demographic parameters decreased thequasi‐extinction probability Probabilities of quasi‐extinction within200 years were most sensitive to changes in kitten survival for bothscenarios (Fig 12) followed by survival of subadult females in theMVM scenarios and survival of prime adult females in the MPCscenario The probability of reproduction of prime adult femaleshad the third‐largest partial rank correlation coefficient with quasi‐extinction probability for both scenarios

Benefits and Costs of Genetic ManagementMinimum population count scenariomdashThe MPC scenario with

a critical threshold of 10 panthers predicted that withoutmanagement intervention average individual heterozygositywould decrease reaching 057 (049ndash064) at 100 years and053 (042ndash062) at 200 years (Fig 13) The population woulddecrease slightly reaching an average of 79 (41ndash99) at 100 yearsand 76 (43ndash98) at 200 years (Fig 14) The probabilities of quasi‐extinction under the no‐intervention MPC scenario when weconsidered the effects of genetic erosion were 13 (0ndash99) at 100years and 23 (0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with

the translocation of 5 Texas females average individualheterozygosity under the MPC scenario increased and thenstabilized at 068 (064ndash072) at 100 years and remained atthat level at 200 years (Fig 13) The population reached anaverage of 88 (62ndash104) at 100 years and stayed the same at200 years (Fig 14) The probabilities of quasi‐extinctionunder this introgression strategy were 6 (0ndash53) within 100years and 7 (0ndash82) within 200 years (Fig 15) Results ofanalyses using a critical threshold of 30 panthers revealed asimilar pattern of relative benefits However the prob-abilities of quasi‐extinction were substantially higher (ge45)across all scenarios (Appendix E)

Motor vehicle mortality scenariomdashWithout managementintervention the average individual heterozygosity predictedby the MVM scenario and a critical threshold of 10 panthersdecreased to 060 (046ndash069) at 100 years and to 059 (039ndash070) at 200 years (Fig 13) The population increased slightlyand reached an equilibrium at 154 individuals (46ndash217) at 100years and 157 (57ndash218) at 200 years (Fig 14) The probabilitiesof quasi‐extinction were 17 (0ndash100) at 100 years and 22(0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with 5

female pumas from Texas average individual heterozygosityincreased slightly reaching 067 (060ndash073) at 100 years and068 (059ndash075) at 200 years (Fig 13) Under this strategy thepopulation reached an average of 164 individuals (44ndash226) at

Table 4 Model comparison results testing for the effects of several covariateson survival of all kittens survival of genetically sampled kittens survival ofsubadult and adult Florida panthers probability of reproduction and kittensproduced annually for Florida panthers sampled from 1981ndash2013 in SouthFlorida USA

Model Parameters ΔAICa Weight

Kitten survival (all data)Base+ F1b+ abundancec 10 000 0988Base+ abundance 9 1018 0006Base+ F1 9 1035 0005Base 8 2711 lt0001Base+ sexd 9 2912 lt0001Base+ litter sizee 9 2914 lt0001

Kitten survival (genetically sampled kittens only)Base+ F1+ abundance 10 000 0541Base+ F1CanAdmf+ abundance 11 099 0329Base+Hetg+ abundance 10 310 0115Base+CanAdmh+ abundance 10 791 0010Base+ abundance 9 944 0005Base+ F1 9 1638 lt0001Base+ F1CanAdm 10 1837 lt0001Base+Het 9 2872 lt0001Base 8 3296 lt0001Base+CanAdm 9 3348 lt0001

Subadult and adult survivalBase+ F1CanAdm+ abundance 7 000 0360Base+ F1 5 053 0276Base+ F1CanAdm 6 105 0213Base+ F1+ abundance 6 222 0119Base+CanAdm+ abundance 6 575 0020Base+CanAdm 5 908 0004Base+Het 5 964 0003Base+Het+ abundance 6 984 0003Base 4 1118 0001Base+ abundance 5 1278 0001

Probability of reproductionAge 3 000 0242Age+CanAdm 4 012 0228Age+ abundance 4 173 0102Age+ F1 4 190 0094Age+CanAdm+ abundance 5 216 0082Age+Het 4 219 0081Age+ F1CanAdm 5 232 0076Age+ F1+ abundance 5 372 0038Age+Het+ abundance 5 399 0033Age+ F1CanAdm+ abundance 6 444 0026

Kittens produced annuallyAge+ F1+ abundance 5 000 0194Age 3 060 0144Age+ abundance 4 061 0143Age+ F1 4 097 0120Age+CanAdm 4 139 0097Age+ F1CanAdm+ abundance 6 196 0073Age+ F1CanAdm 5 231 0061Age+CanAdm+ abundance 5 238 0059Age+Het+ abundance 5 250 0056Age+Het 4 259 0053

a Akaikersquos information criterion (AIC) for kitten survival models was adjustedfor small sample size and overdispersion (QAICc)

b F1 divides Florida panthers into 2 groups F1 admixed and all other panthersc Index of Florida panther abundanced Sex of an individual Florida panther (male or female)e Size of the litter in which a Florida panther kitten was bornf F1CanAdm divides panthers into 3 groups F1 admixed canonical andother admixed panthers

g Het is the individual heterozygosityh CanAdm divides panthers into 2 groups canonical and admixed (includingF1 admixed) panthers

van de Kerk et al bull Florida Panther Population and Genetic Viability 19

100 years and 170 (67ndash231) at 200 years based on the MVMscenario (Fig 14) The probabilities of quasi‐extinction underthis introgression strategy were 10 (0ndash99) at 100 years and12 (0ndash99) at 200 years (Fig 15)The projected population size and average individual hetero-

zygosity reached equilibrium levels with periodic fluctuations inthese values when introgression was implemented (Figs 16ndash17)Genetic introgression decreased the time until quasi‐extinctionacross all scenarios (Fig 18) Results of analyses using a criticalthreshold of 30 panthers revealed a similar pattern of relative ben-efits However the probabilities of quasi‐extinction were sub-stantially higher (ge45) across all scenarios (Appendix E)

Addition of pumas from Texas increased both the populationsize and average individual heterozygosity consequentlyintrogression caused periodic increases in average individualheterozygosity and population size at the interval at which theintrogression occurred Whereas average individual hetero-zygosity gradually decreased following introgression densitydependence caused the population size to fluctuate especiallywhen a larger number of pumas from Texas were released Thepositive effect of genetic introgression initiatives on quasi‐extinction probabilities decreased as the time interval betweenthese events increased More frequent genetic introgressions ledto greater increases in average individual heterozygosity and

Old adult

Prime adult

Subadult

Kitten

025 050 075

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Individual heterozygosity

Ann

ual s

urvi

val r

ate

Old adult

Prime adult

Subadult

Kitten

40 60 80 100 120

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Abundance index

Ann

ual s

urvi

val r

ate

Kittens Males Females

Figure 7 The effect of individual heterozygosity (left) and the abundance index (right) on Florida panther age‐ and sex‐specific survival estimates (shaded area isSE) in South Florida USA 1981ndash2013 Abundance index data are from McBride et al (2008) and R T McBride and C McBride (Rancherrsquos Supply Incorporatedunpublished report)

20 Wildlife Monographs bull 203

equilibrium population size while reducing the probability ofquasi‐extinction Comparatively performing genetic introgres-sion less often but with more individuals per release was lesseffective at improving the long‐term prospects for the pantherpopulation (Figs 13ndash15)

Financial cost and persistence benefitsmdashThe total estimated costof 1 genetic introgression was $40000 $80000 or $120000for releasing 5 10 or 15 pumas from Texas respectively(Table 5) The total cost incurred over 100 years for eachstrategy ranged from $50000 to $12 million (Table 6)The most expensive strategy involved the introduction of15 pumas every 10 years this strategy reduced the probability

of quasi‐extinction by 58 and 40 under the MPC and MVMscenarios respectively (Table 6) Introducing 5 pumas every80 years cost the least but improvements in populationpersistence under this strategy were insubstantial (Table 6)Releasing more panthers more frequently caused increasedpopulation fluctuations but did not necessarily reduce the quasi‐extinction probability (Figs 14 and 15 Table 6)

DISCUSSION

The duration (34 yr of data) and sample sizes associated with theFlorida Panther Project allowed us to obtain a comprehensiveunderstanding of genetic variation demographic parameters

A

B

C

Figure 8 Cause‐specific mortality rates (plusmnSE) for male and female Florida panthers (A) subadult prime‐adult and old‐adult Florida panthers of both sexes (B)and canonical and admixed Florida panthers of both sexes (C) in South Florida USA 1981ndash2013 Causes of mortality are vehicle collision intraspecific aggression(ISA) other causes (known causes such as diseases injuries and infections unrelated to the first 2 causes) and unknown cause (mortalities for which we could notassign a cause)

van de Kerk et al bull Florida Panther Population and Genetic Viability 21

probability of population persistence and the duration of thepositive impacts of genetic restoration on this endangered po-pulation The Florida panther population was in dire straits inthe early 1990s and our results clearly demonstrated that the

implementation of the introgression project in 1995 subse-quently accrued a series of genetic and demographic improve-ments that have led to the change from a declining population toone that is growing and expanding its current breeding range

Sensitivity Elasticity

0 1 2 3 00 02 04

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

Para

met

er

Figure 9 Sensitivity and elasticity (proportional sensitivity) of stochastic population growth rate (λs) of the Florida panther to vital demographic parameters in SouthFlorida USA 1981ndash2013 Horizontal lines represent 5th and 95th percentiles

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

70

80

90

100

110

0

5

10

15

0

50

100

Years

StatisticMeanMedian

Figure 10 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the minimum population count(MPC) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

125

150

175

200

00

05

10

15

20

0

30

60

90

Years

StatisticMeanMedian

Figure 11 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the motor vehicle mortality(MVM) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

22 Wildlife Monographs bull 203

MPC MVM

minus08 minus06 minus04 minus02 00 minus08 minus06 minus04 minus02 00

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Male subadult survival

Male prime adult survival

Male old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

PRCC

Para

met

er

Figure 12 Partial rank correlation coefficient (PRCC) between quasi‐extinction probabilities and the demographic parameters of the Florida panther for theminimum population count (MPC) and motor vehicle mortality (MVM) scenarios Error bars indicate standard deviation Based on data collected in South FloridaUSA 1981ndash2013

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 13 Average projected individual heterozygosities in the Florida panther population over 200 years without genetic management intervention and with eachof the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include theintroduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years Average individual heterozygosity peaks when pumas are added to the Floridapanther population Based on data collected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 23

Our research has further validated the success of genetic in-trogression by quantifying the reduction in the probability ofextinction of the panther population because of this manage-ment initiative These findings all bode well for continuedprogress toward recovery That said the Florida panther po-pulation remains small and isolated from other populations ofpuma from western North America thereby denoting that theeventual impact of genetic drift and inbreeding may negativelyaffect the population in the future Realizing this will continueto be an issue that managers will have to contemplate wemodeled the impact of varied genetic management scenarios todetermine the frequency of introgression along with the numberof panthers that should be released during each initiative tominimize cost and maximize benefits to the population Theseresults will prove invaluable to managers attempting to conservethe Florida panther

Genetic Variation Demographic Parametersand Persistence of Heterotic Benefits from GeneticIntrogressionOur measure of individual heterozygosity (1 minusHL) was higheron average for the admixed (0615) panthers than for those ofcanonical ancestry (0386) Levels of individual heterozygosityfor admixed panthers were comparable to levels observed in asample of pumas from west Texas (x plusmn SE= 0695plusmn 0019n= 41) which further demonstrates the improvement in levelsof genetic variation in the Florida population since 1995 The

correlation between HL and several demographic parametershas been noted in wild populations including cheetahs (Terrellet al 2016) and several species of birds such as European shags(Phalacrocorax aristotelis male and female survival probabilityVelando et al 2015) and Egyptian vulture (Neophron percnop-terus age at recruitment Agudo et al 2012) If we focus only onpanthers captured and radiocollared from 2012ndash2015 (FP211ndashFP240) all of which had estimated birth years ge2005 (ge10 yrpost‐introgression) those panthers had an average individualheterozygosity level of 0618plusmn 0029 highlighting the elevatedlevels of genetic variation that continue to persist in cohorts ofpanthers 5 generations after the release of female pumas fromTexasThe proportional membership values (q) from our

STRUCTURE analysis allowed us to delineate 2 clusters ofancestry for Florida panthers (Fig 4) During the pre‐in-trogression era the population was dominated by panthers thatretained a high percentage of the canonical ancestry which wasaffected by inbreeding depression The post‐introgression eraancestry values are comprised of many admixed individuals inessence demonstrating the success of the introgression in termsof increasing genetic variation in the population and mi-micking gene flow that historically occurred with panthers andother populations of pumas (Onorato et al 2010) A somewhatanalogous situation although abetted by natural immigrationwas evident in the ancestral variation in a small population ofpuma intersected by a major freeway in California USA In

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

75

80

85

90

95

100

130

150

170

190

Years

Popu

latio

n si

ze

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 14 Average projected population sizes in the Florida panther population over 200 years without intervention and with each of the genetic introgressionstrategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include the introduction of 5 10 or 15pumas from Texas USA every 10 20 40 or 80 years Peaks occur when pumas are added to the Florida panther population Based on data collected in SouthFlorida USA 1981ndash2013

24 Wildlife Monographs bull 203

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

25

50

75

100

0

25

50

75

100

Introgression interval (yrs)

PQ

E (

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 15 Average probability of quasi‐extinction (PQE) of the Florida panther population within 100 years and within 200 years without genetic managementintervention (N) and with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenariosIntrogression strategies include the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Errorbars indicate 5th and 95th percentiles Based on data collected in South Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

50

60

70

80

90

100

110

50

100

150

200

Years

Popu

latio

n si

ze

Figure 16 Average projected Florida panther population trajectories under a geneticintrogression strategy involving the release of 5 female pumas from Texas USA every20 years for the minimum population count (MPC) and motor vehicle mortality(MVM) scenarios Shaded area indicates 5th and 95th percentiles and isrepresentative of confidence intervals for other scenarios Based on data collected inSouth Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Figure 17 Average projected Florida panther population‐level heterozygositiesunder a genetic introgression strategy involving the release of 5 female pumas fromTexas USA every 20 years for the minimum population count (MPC) and motorvehicle mortality (MVM) scenarios Shaded area bounds the 5th and 95th percentilesand is representative of confidence intervals for other scenarios Based on datacollected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 25

this case the migration of just a single male across the freewayto the south resulted in an increase in the number of in-dividuals with admixed ancestry and substantially improvedgenetic diversity of the subpopulation south of the freeway(Riley et al 2014)Our estimate of overall kitten survival (0321plusmn 0056) was

similar to that reported by Hostetler et al (2010) who used13 years of post‐introgression data (0323plusmn 0071) These valuesare lower than kitten survival estimates derived from severalpuma populations in western North America (044ndash091 Loganand Sweanor 2001 Lambert et al 2006 Laundre et al 2007Robinson et al 2008 Ruth et al 2011) When we included onlydata for individuals that had been genetically sampled our es-timate was 15 higher The proportion of admixed kittens thatwere genetically sampled increased as the population expandedwhich could have resulted in higher estimates of kitten survivalbecause admixed kittens survive better than canonical kittensKitten survival was strongly influenced by genetic ancestry withsurvival rates of F1 kittens being almost twice that of canonicalkittens (Fig 5B) Consistent with the findings of Hostetler et al(2010) increases in heterozygosity coincided with increasedkitten survival whereas population density negatively affectedkitten survival Population density often has a strong negativeimpact on survival of young age classes due to infanticide andcannibalism which has been reported from several carnivore

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

50

100

150

0

50

100

150

Introgression interval (yrs)

TQE

(yrs

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 18 Average times until quasi‐extinction (TQE) for the Florida panther population within 100 years and within 200 years without genetic management interventionand with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategiesinclude the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Error bars indicate 5th and 95thpercentiles Based on data collected in South Florida USA 1981ndash2013

Table 5 Average cost (2017 US dollars) of introgression strategies employedover 100 years that involve the introduction of 5 10 or 15 pumas from TexasUSA into the Florida panther population in South Florida USA at an in-creasingly higher frequency Costs of capture (including transportation) quar-antine and post‐introgression monitoring were estimated by Roy McBrideMark Cunningham and Dave Onorato respectively

Frequency Component 5 pumas 10 pumas 15 pumas

Once Capture 25000 50000 75000

Quarantine 5000 10000 15000

Monitoring 10000 20000 30000

Total 40000 80000 120000

Every 80 years Capture 31250 62500 93750

Quarantine 6250 12500 18750

Monitoring 12500 25000 37500

Total 50000 100000 150000

Every 40 years Capture 62500 125000 187500

Quarantine 12500 25000 37500

Monitoring 25000 50000 75000

Total 100000 200000 300000

Every 20 years Capture 125000 250000 375000

Quarantine 25000 50000 75000

Monitoring 50000 100000 150000

Total 200000 400000 600000

Every 10 years Capture 250000 500000 750000

Quarantine 50000 100000 150000

Monitoring 100000 200000 300000

Total 400000 800000 1200000

26 Wildlife Monographs bull 203

species including polar bears (Ursus maritimus Derocher andWiig 1999) black bears (Ursus americanus Garrison et al 2007)and pumas (Logan and Sweanor 2001)Our estimates of sex‐ and age‐specific survival rates of subadult

and adult panthers (Table 3) and the effects of covariates on theserates were similar to those reported by Benson et al (2011) derivedfrom data collected through 2006 Our survival rates for males(065ndash077) were lower than females (078ndash097) for all 3 ageclasses (subadult prime adult and old adult) which is the typicaltrend observed in most puma populations (eg Ruth et al 19982011) However an unhunted puma population occupying afragmented urban landscape in southern California exhibited lowersurvival (0586 females 0525 males Vickers et al 2015) perhapsas a result of severe habitat fragmentation and the lack of extensiveswaths of public land protected in perpetuity Most populations ofpuma in western North America are typically subject to variedlevels of management via regulated hunting which often is biasedtoward the take of males (Cooley et al 2009 Ruth et al 2011)Consequently annual survival estimates from hunted populationsin northeast Washington (0679ndash0733 females 0341 malesRobinson et al 2008) and southeastern Arizona (0685 females0584 males Cunningham et al 2001) were generally lower thanthose we estimated for Florida panthersCause‐specific mortality analysis showed that male panthers

experienced a substantially greater mortality risk from in-traspecific aggression This differs from the pattern observedduring a long‐term study in Arizona where females were at ahigher risk of intraspecific mortality than males (46 of maleand 53 of female mortality Logan and Sweanor 2001)Mortality associated with intraspecific strife has been docu-mented in hunted and unhunted populations (this study Loganand Sweanor 2001 Stoner et al 2010) and its root cause hasbeen difficult to determine (eg population density and preypopulation trends) That being the case finding ways to reducewhat is often a major source of mortality for puma populationscontinues to pose a challenge to managersCanonical panthers were substantially more likely to die from

intraspecific aggression than were admixed panthers This mayat least partly explain the difference in survival between admixed

and canonical panthers Canonical panthers are known to bemore likely to suffer from varied correlates of inbreeding de-pression than admixed animals including atrial septal defects(Johnson et al 2010) and compromised immune systems(Roelke et al 1993) These factors have the potential to affectfitness and perhaps dominance of individuals when engaged inan intraspecific encounter A somewhat similar scenario wasobserved by Hogg et al (2006) in a population of bighorn sheepthat were part of an introgression project Admixed males in thispopulation had a greater probability of paternity than males thatwere less outbred This included coursing males with higherlevels of admixture being markedly more successful at fightingmales that were defending females in estrous (Hogg et al 2006)Heterosis resulting from genetic introgression is expected to be

the strongest in F1 individuals (Shull 1908 Crow 1948 Burkeand Arnold 2001 Waller 2015) and to progressively decline insubsequent generations (Pickup et al 2013 Frankham 2015)Given that admixed (especially F1) panthers survived better thancanonical panthers and that individual heterozygosity improvedsurvival of both sexes and all age classes our data provide evi-dence for heterotic advantage whereby individuals of mixedancestry had higher fitness (Shull 1908 Crow 1948) Our resultsare consistent with findings of other studies that involved in-tentional genetic introgression for conservation purposes Forexample Hogg et al (2006) detected substantial improvementsin survival and reproduction of introgressed bighorn sheep andMadsen et al (1999 2004) reported a dramatic increase in thepopulation of adders after genetic introgressionKnowledge of the persistence of benefits to a population ac-

crued via genetic introgression is essential for determiningwhen additional introgression may be warranted There is adepauperate amount of information regarding this topic and theidentification of factors that may influence persistence of thebenefits of genetic introgression (Tallmon et al 2004 Hedricket al 2014 Whiteley et al 2015) For example in minks(Neovison vison) Thirstrup et al (2014) found that outcrossingled to larger litters in F2 and F3 but this benefit disappeared insubsequent generations In a population of gray wolves Hedricket al (2014) suggested that benefits of genetic introgression may

Table 6 Costs and benefits accumulated over 100 years for genetic introgression at different intervals and with different numbers of pumas from Texas USAintroduced into the Florida panther population in South Florida USA Costs (2017 US dollars) include capture and transportation quarantine and post‐introgression monitoring Benefits are represented as average probability of quasi‐extinction (PQE and 5th and 95th percentiles) and changes (Δ) therein under thescenario using the minimum population count (McBride et al 2008 MPC) and the scenario using the motor vehicle mortality population estimates (McClintocket al 2015 MVM) The table is sorted by percent change in probability of quasi‐extinction under the MPC scenario

Interval (yr) Pumas Cost MPC PQE () ΔMPC PQE () MVM PQE () ΔMVM PQE ()

ndash 0 0 13 (0ndash99) 0 17 (0ndash100) 080 10 100000 12 (0ndash99) 10 (0ndash58) 16 (0ndash100) 5 (0ndash38)80 15 150000 12 (0ndash99) 13 (0ndash65) 15 (0ndash100) 8 (0ndash56)80 5 50000 12 (0ndash99) 14 (0ndash73) 15 (0ndash99) 9 (0ndash41)40 15 300000 10 (0ndash98) 26 (0ndash104) 14 (0ndash99) 13 (0ndash60)40 10 200000 9 (0ndash94) 35 (0ndash183) 13 (0ndash99) 21 (0ndash95)40 5 100000 8 (0ndash89) 39 (0ndash211) 12 (0ndash99) 26 (0ndash124)20 15 600000 8 (0ndash88) 42 (0ndash257) 13 (0ndash99) 24 (0ndash123)20 10 400000 6 (0ndash67) 54 (0ndash318) 10 (0ndash99) 37 (0ndash217)10 15 1200000 6 (0ndash53) 58 (0ndash372) 10 (0ndash99) 40 (0ndash208)20 5 200000 5 (0ndash50) 59 (0ndash338) 9 (0ndash99) 44 (0ndash352)10 10 800000 4 (0ndash16) 69 (0ndash489) 8 (0ndash92) 55 (0ndash401)10 5 400000 4 (0ndash7) 73 (0ndash502) 6 (0ndash71) 63 (0ndash503)

van de Kerk et al bull Florida Panther Population and Genetic Viability 27

wane after 2 or 3 generations The WrightndashFisher model ofgenetic drift predicts a geometric decline in expected hetero-zygosity at the rate of (1 minus 12Ne) per generation where Ne is theeffective population size (Hartl and Clark 1997 Frankham et al2014) Thus it seems logical to assume that the duration of thebenefits of introgression is limited especially in a species per-sisting in a small population such as Florida panthersWe expected Florida panther survival in the most recent years

(2005ndash2013) post‐introgression to differ from that observed in thedecade following the introgression in 1995 because pantherabundance and heterozygosity factors known to influence panthersurvival (Hostetler et al 2010 Benson et al 2011) have changed(Figs 2 and 6) We found that subadult and adult survival rates inrecent years (2005ndash2013) were similar to those in the period im-mediately following introgression (1995ndash2004 Fig 5C) overallkitten survival was similar to estimates of Hostetler et al (2010) Infact the pattern of covariate effects on Florida panther survivalreported by Benson et al (2011) and Hostetler et al (2010) hasbarely changed The negative effects of population density andpositive effects of heterozygosity may counterbalance survival ratesreducing population fluctuations Our result that benefits of geneticrestoration have remained in the population nearly for 5 genera-tions post‐intervention was surprising but also encouraging Pos-sible explanations for this observation may include 1) genetic ero-sion occurs at slower rate than previously thought 2) introducing 5migrants in 1 genetic intervention may have beneficial effects si-milar to those from 1 migrant per generation over multiple gen-erations or 3) other and as yet unknown factors may reduce therate of genetic erosion and subsequent demographic effects Adensity‐dependent effect on kitten survival could also explain whynon‐F1 admixed kittens did not survive substantially better thandid canonical kittens most kittens sampled from 2005ndash2013 wereadmixed and their survival was lower possibly because of a density‐dependent effect as the Florida panther population continuouslygrew during that period Trinkel et al (2010) also found thatpopulation density and inbreeding coefficient interacted to affectdemographic parameters and growth rate of an African lion po-pulation Likewise population density interacted with winterweather to affect the survival and population growth rates in Soaysheep (Ovis aries Milner et al 1999 Coulson et al 2001)

Population Dynamics and PersistenceThe finite deterministic and stochastic growth rates were gt10reflecting that much of the data used in this study were collectedfollowing genetic introgression in 1995 during which time thepopulation increased substantially Because our demographicparameter estimates did not change markedly in comparison tothose of Hostetler et al (2013) the similarity between thepopulation growth estimates from these studies was expectedBut these estimates reflect population growth during thedata‐collection period and do not predict future growth In factestimates of population size reported by McClintock et al(2015) suggest that population growth may already be slowing(Fig 2) A similar pattern was observed in an introduced po-pulation of African lion which increased steadily from 1995until 2001 then fluctuated around the presumed carrying ca-pacity thereafter (Trinkel et al 2010) Several other African lionpopulations that experienced various degrees of recovery

subsequently became relatively stable or exhibited decliningtrends (Bauer et al 2015)Our estimates of quasi‐extinction probabilities were lower than

those reported by Hostetler et al (2013) using a 2‐sex age‐structured matrix population model Lower probabilities ofquasi‐extinction than those reported by the earlier study forcomparable scenarios were likely a consequence of the fact thatthe estimates of abundance (McBride et al 2008) and thus thedensity‐dependent effects on survival of kittens and old panthers(gt10 yr) have changed in recent years Additionally these earlyestimates of the probability of quasi‐extinction and of time toquasi‐extinction did not consider genetic erosion that will in-evitably occur without management interventionUnder the MVM scenario the equilibrium population size was

twice that attained under the MPC scenario The MVM popula-tion estimates are approximately twice the MPCs (Fig 2) as aresult the simulated population under the MVM scenario achieveda higher equilibrium size Probability of quasi‐extinction under theMVM scenario was generally greater than that under the MPCscenario This result is a consequence of the fact that the estimateddensity dependence on kitten survival based on the MPC scenario isstronger than that for theMVM scenario (regression coefficients forabundance for the MPC scenario= minus0068 vs minus0002 for theMVM scenario Appendix B Table B1) Consequently simulatedpopulations under the MPC scenario have a stronger tendency tomove toward the equilibrium population size which inherentlyreduces the quasi‐extinction probability (eg Royama 1992)As expected for long‐lived mammals (Heppell et al 2000 Oli

and Dobson 2003) the population growth rate was proportio-nately most sensitive to changes in survival of prime adultsfollowed by that in younger age classes Global sensitivity ana-lysis in contrast showed that under all scenarios extinctionparameters were particularly sensitive to changes in survival ofFlorida panther kittens This result is a consequence of the highsensitivity of the population growth rate to kitten survival(Fig 9) and a larger standard error of kitten survival (Table 3)Results of our sensitivity analyses indicate that future researchshould focus on acquiring additional information on early lifestages to improve the accuracy and precision of estimates ofkitten survival Our results suggest that demographic variables(or their environmental drivers) that strongly influence extinc-tion risks are not necessarily those with the largest elasticityvalues A study of island fox (Bakker et al 2009) found thatextinction risk was strongly influenced by survival modifierparameters that had low elasticity values

Genetic Management When and How to InterveneThe use of genetic introgression as a management tool has al-ways been controversial and the probable effectiveness it mighthave in preventing the imminent demise of the panther popu-lation was initially debated (Creel 2006 Maehr et al 2006Pimm et al 2006) These debates notwithstanding the pantherpopulation had suffered from genetic morphological and bio-medical correlates of inbreeding before this management in-tervention (Johnson et al 2010 Onorato et al 2010) whereasthe post‐introgression period has been characterized by a sub-stantial reduction in the biomedical correlates of inbreeding andimprovements in demographic vigor and abundance (McBride

28 Wildlife Monographs bull 203

et al 2008 Hostetler et al 2010 2013 Johnson et al 2010Benson et al 2011 McClintock et al 2015) Nevertheless thepopulation remains small and isolated habitat is still being lostand there is no current possibility of natural gene flow betweenthis and other North American puma populations thus therecurrence of inbreeding‐related problems and subsequent po-pulation declines are inevitable The question therefore is notwhether genetic management of the Florida panther populationis needed but when and how it should be implementedWithout genetic introgression probabilities of quasi‐extinc-

tion were substantially greater than those from models thatincorporated periodic genetic introgression events (Fig 15)highlighting the importance of genetic management in smallisolated populations When 5 female pumas are added to theFlorida panther population every 10 years genetic variationincreased substantially under the MPC and MVM scenariosThe IBM predicted that the equillibrium population size wouldbe substantially larger when 5 pumas were released into thepopulation every 10 years than populations without intervention(ie no introgression) Releasing 15 Texas females did not af-ford substantially greater benefit to the equilibrium populationsize than did releasing only 5 Texas females Instead addingmore individuals caused increasingly large fluctuations in po-pulation size due to the numerical increases and density de-pendence (Fig 14) possibly diminishing the benefits associatedwith increased genetic variationCompared with the no‐intervention scenario (ie no genetic

introgression implemented) the introduction of 5 female pumasevery 10 years reduced quasi‐extinction probabilities from 13to 4 and from 17 to 6 for the MPC and MVM scenariosrespectively The benefit of genetic‐introgression strategies de-creased as the interval between successive management inter-ventions increased and no benefit was evident once the intervalreached 80 years (Fig 15) Introgression reduced the averagetime until quasi‐extinction (Fig 18) This somewhat counter‐intuitive result can be explained by the fact that repeated geneticintrogression makes the population more robust over timewhich will reduce the probability of extinction Consequentlyfew simulated population trajectories that fall below the criticalthreshold do so early reducing the mean time to extinctionAlso density dependence has a stabilizing effect on populations(Royama 1992) populations tend toward equilibrium popula-tion sizes which reduces the probability over time of popula-tions falling below quasi‐extinction threshold However timeuntil quasi‐extinction must be interpreted in conjunction withthe probability of quasi‐extinction which is very low for allscenarios with genetic introgression (Fig 15)Cost is a significant impediment to the implementation of a

genetic introgression program because captures relocations andmonitoring efforts before and after introgression are expensive(Edmands 2007 Frankham 2010b 2015 2016) Costs may beacceptable to society if the management actions result in sub-stantial fitness benefits especially if the species of concern ispopular and charismatic (Frankham 2015) We estimated the100‐year cost of introgression strategies modeled here to rangefrom $50000 to $12 million More expensive strategies gen-erally led to larger decreases in the probability of quasi‐extinc-tion but cheaper strategies also substantially improved

population viability (ie reduced quasi‐extinction probabilityTable 6) One of the cheaper strategies (5 pumas released every20 years) which cost an estimated $200000 over 100 yearsreduced the probability of quasi‐extinction by 59 and 44 forthe MPC and MVM scenarios respectively But these pre-liminary cost estimates are based on expenses incurred duringthe 1995 introgression and include costs of capture quarantinetransportation release and post‐release monitoring of femalesonly Although the cost for future genetic interventions wouldlikely be higher than those reported here the relative differencesin cost between alternative intervention strategies should remainapproximately the sameOur ability to simulate genetics and link it to the viability of

the Florida panther population is based on the empirically es-timated relationship between individual heterozygosity andsurvival Such heterozygosity and fitness correlations have beenstudied for decades (David 1998) but have only recently beenused to predict population viability (Benson et al 2016) noneto our knowledge have incorporated cost into their modelingefforts The mechanisms underlying heterozygosity and fitnesscorrelations remain unclear (David 1998 Szulkin et al 2010)and their significance as a proxy for inbreeding depression hasbeen debated (Balloux et al 2004 Miller and Coltman 2014)Nevertheless evidence continues to mount demonstratingpositive relationships between heterozygosity and survival(Coulson et al 1998ab Hansson et al 2004 Silva et al 2005Acevedo‐Whitehouse et al 2006 Jensen et al 2007 Velandoet al 2015) and between heterozygosity and fecundity (Seddonet al 2004 Charpentier et al 2005 Agudo et al 2012 Velandoet al 2015) Although the reported correlations are generallyweak their consequences can be substantial if population growthand persistence parameters are highly sensitive to the affectedvital rates (Velando et al 2015) Compared with scenarios thatignore genetics incorporating the effects of inbreeding on sur-vival increased quasi‐extinction probabilities within a 100‐yeartimeframe from 15 to 13 and from 14 to 17 for theMPC and MVM scenarios respectively These results high-light the importance of incorporating genetics when analyzingthe viability of small isolated populationsAlthough conservation biologists have recognized the im-

portance of genetics in population viability many still fail toincorporate genetic factors into PVA models (Reed et al 2002)Explicit consideration of genetics in PVAs requires long‐termgenetic and demographic data This permits the assessment ofthe functional relationship between genetic variability and de-mographic parameters Population viability analysis softwarepackages such as VORTEX (Lacy 1993 2000) offer a powerfulframework for individual‐based simulations but they often donot adequately capture a speciesrsquo life history or genetics Basedon a thorough review of the importance of genetic factors inwildlife conservation Frankham et al (2014) concluded thatgenetic factors can strongly influence the dynamics and persis-tence of small andor fragmented populations They furthernoted that ldquoMost PVA‐based risk assessments ignore or in-adequately model genetic factorsrdquo and recommended that ldquoPVAshould routinely include realistic inbreeding depressionhelliprdquo(Frankham et al 201457) Our custom‐built IBM permitted usto develop a model that adequately captured the Florida panther

van de Kerk et al bull Florida Panther Population and Genetic Viability 29

life history and we could parameterize the model with our long‐term genetic and demographic dataThe explicit consideration of the effects of inbreeding on

Florida panther population dynamics and persistence representsan important step forward Nonetheless our model remainsimperfect First we neglected the possible effects of habitat lossdetrimental anthropogenic activities climate change the prob-ability of immigration of pumas from western populationsdisease or catastrophes on demographic rates and populationviability Although immigration from puma populations outsideFlorida is possible its probability is very low given the extensivechallenges the anthropogenically altered landscape would posefor such a migrant to make it successfully to South Florida Weomitted mutations from our model because we have no in-formation on mutation rates though we expect that they are lowbased on values reported for other mammals (Kumar and Sub-ramanian 2002) Finally we only considered possible effects ofinbreeding on age‐specific survival rates because there was noevidence that heterozygosity affected female reproduction Lossof heterozygosity can negatively affect reproduction becauseinbred male panthers can suffer from cryptorchidism which canadversely affect their fecundity However given the polygynousmating system we expect the Florida panther population growthis primarily female‐limitedAlthough it is generally accepted that genetic introgression

only temporarily relieves a population from inbreeding depres-sion we know of no cases in which it has been implementedmore than once to conserve a threatened wildlife populationOur study provides an example of how demographic and mo-lecular data can be used within an IBM framework to evaluatethe efficacy of alternative genetic management strategies via theFlorida panther as a case study Our model can be adjusted forand applied to other species and offers great potential as a toolfor genetic management of small isolated wildlife populations

MANAGEMENT IMPLICATIONS

Our results and those of Hostetler et al (2013) and Johnsonet al (2010) provide persuasive evidence that the genetic in-tervention implemented in 1995 prevented the demise of theFlorida panther and restored demographic vigor to the popu-lation The Florida panther population remains small and iso-lated from other puma populations the closest puma populationis located gt1000 km away in Texas and the intervening land-scape is highly developed and fragmented with many interstate(and other high‐traffic) highways and major urban centersTheory suggests that 1 migrant per generation is sufficient tominimize the loss of genetic diversity (eg Mills and Allendorf1996) However the possibility of successful migration of apuma to South Florida followed by successful mating every ~5years is very low considering the distance between Texas andFlorida (Hedrick 1995) and the many risks and barriers thatdispersing pumas face (Beier 1995 Maehr et al 2002 Thatcheret al 2006) Consequently the pantherrsquos long‐term persistencewill likely depend on subsequent timely genetic introgressionsgiven the near‐impossibility of natural gene flow from otherpuma populations To that end our study offers considerableinsights into benefits of different genetic management strategiesand associated costs

Without genetic management the Florida panther populationfaces a substantial risk of quasi‐extinction within 100 years (10ndash46 depending on quasi‐extinction threshold and scenarios)Twenty‐three years have passed since genetic introgression wasimplemented and the population appears healthy as indicatedby measures of genetic variation increases in the populationsize and improvements in age‐specific survival rates Ourfindings suggest that releasing more pumas more frequently doesnot necessarily improve persistence probability because such astrategy would cause larger fluctuations in population size(Fig 14) which negatively affects persistence probability Thusextinction risks faced by inbred populations of large carnivorescan be substantially reduced by releasing approximately 5 im-migrants every 2 decades We suggest that such a managementstrategy be followed only in conjunction with continued long‐term monitoring of the population Our analyses demonstratethat population growth and persistence are highly sensitive tokitten and female (adult and subadult) survival Continuing tocollect data on these demographic variables along with bio-metric and genetic sampling will remain important to pantherrecovery and vigilance in identifying issues associated with in-breeding depression The collection of these monitoring datashould allow managers to detect any demographic or geneticchanges in a timely fashion and respond with genetic in-trogression or other management actions if warrantedRecovery objectives as defined by the USFWS in its current

recovery plan (USFWS 2008) delineate the need for additionalpopulations in the historical range to downlist or delist theFlorida panther If the only extant population of Floridapanthers continued to grow it could serve as a source ofpanthers to be translocated to suitable habitat to seed addi-tional populations Maintaining current information on thegenetic health of the population and precise estimates of itssize will be important in assessing whether it can withstand theremoval of a subset of animals for translocation Although the2017 documentation of a female panther with kittens north ofthe current breeding range is encouraging in terms of thepotential for population expansionmdashgiven that no female hadbeen recorded north of the Caloosahatchee River since 1972mdashthe re‐establishment of separate new populations will mostlikely require translocations by wildlife managers and a lengthysociopolitical processConservation of threatened or endangered species such as the

Florida panther is inherently challenging For panthers and otherlarge carnivores that require vast extents of natural habitat topersist habitat is typically the most limiting factor that will de-termine whether recovery efforts succeed Although geneticmanagement invariably plays a key role in the long‐term persis-tence of the panther population even a healthy population caneventually succumb to the impacts of habitat loss The humanpopulation of Florida continues to be one of the fastest‐growingin the nation the United States Census Bureau currently ranksFlorida as the third most populous Therefore habitat loss isgoing to continue to be a key issue for panthers Diligence towardpreserving panther habitat in its historical range via conservationeasements or other means and trying to keep such areas inter-connected via corridors is a goal stakeholders such as governmentagencies sportsmen non‐governmental organizations ranchers

30 Wildlife Monographs bull 203

and other private landowners must work on collaboratively toachieve

ACKNOWLEDGMENTS

We thank D Shindle D Jennings T OrsquoMeara D Land andC Belden for valuable advice throughout the project We thankS Bass M Criffield M Cunningham D Giardina D JansenA Johnson D Land M Lotz C McBride Rocky McBrideRowdy McBride Roy McBride L Oberhofer S Schulze staffsat Everglades National Park and Big Cypress National Preserveand others for assistance with fieldwork We also thank JAustin M Conroy J Hines J Laake S Mills J Nichols JHines M Sunquist J Fryxell and T Coulson for advice andassistance regarding data analysis and panther biology TheMuseum of Texas Tech University Natural Science ResearchLaboratory provided samples from pumas from west Texas forcomparative genetic analyses M Ben‐David D Land WMcCown E Hellgren and 4 anonymous reviewers providedmany helpful comments on the manuscript We thank NereaUbierna Lopez for completing the Spanish translation of theabstract We are grateful to the Department of ResearchComputing University of Florida for excellent high‐perfor-mance computing services We would like to thank US Fishand Wildlife Service Florida Fish and Wildlife ConservationCommission (FWC) US National Park Service QuantitativeSpatial Ecology Evolution and Environment (QSE3) In-tegrative Graduate Education and Research Traineeship(IGERT) program funded by the National Science Foundationand the University of Florida for financially supporting thisstudy Funding for panther research conducted by the FWC issupported predominantly via donations accrued with vehicleregistrations for ldquoProtect the Pantherrdquo license plates

LITERATURE CITEDAcevedo‐Whitehouse K T R Spraker E Lyons S R Melin F Gulland RL Delong and W Amos 2006 Contrasting effects of heterozygosity onsurvival and hookworm resistance in California sea lion pups MolecularEcology 151973ndash1982

Adams J R L M Vucetich P W Hedrick R O Peterson and J AVucetich 2011 Genomic sweep and potential genetic rescue during limitingenvironmental conditions in an isolated wolf population Proceedings of theRoyal Society B Biological Sciences 2783336ndash3344

Agresti A 2013 Categorical data analysis Third edition John Wiley amp SonsHoboken New Jersey USA

Agudo R M Carrete M Alcaide C Rico F Hiraldo and J A Donaacutezar2012 Genetic diversity at neutral and adaptive loci determines individualfitness in a long‐lived territorial bird Proceedings of the Royal Society BBiological Sciences 2793241ndash3249

Albeke S E N P Nibbelink and M Ben‐David 2015 Modeling behavior bycoastal river otter (Lontra canadensis) in response to prey availability in PrinceWilliam Sound Alaska a spatially‐explicit individual‐based approach PLOSOne 10e0126208

Alho J S K Vaumllimaumlki and J Merilauml 2010 Rhh an R extension for estimatingmultilocus heterozygosity and heterozygosity‐heterozygosity correlationMolecular Ecology Resources 10720ndash722

Alvarez K 1993 Twilight of the panther Myakka River Publishing SarasotaFlorida USA

Aparicio J M J Ortego and P J Cordero 2006 What should we weigh toestimate heterozygosity alleles or loci Molecular Ecology 154659ndash4665

Bakker V J D F Doak G W Roemer D K Garcelon T J Coonan S AMorrison C Lynch K Ralls and R Shaw 2009 Incorporating ecologicaldrivers and uncertainty into a demographic population viability analysis for theisland fox Ecological Monographs 7977ndash108

Balloux F W Amos and T Coulson 2004 Does heterozygosity estimateinbreeding in real populations Molecular Ecology 133021ndash3031

Barone M A M E Roelke J Howard J L Brown A E Anderson and DE Wildt 1994 Reproductive characteristics of male Florida panthersmdashcomparative studies from Florida Texas Colorado Latin‐America andNorth‐American Zoos Journal of Mammalogy 75150ndash162

Bateson Z W P O Dunn S D Hull A E Henschen J A Johnson and LA Whittingham 2014 Genetic restoration of a threatened population ofgreater prairie‐chickens Biological Conservation 17412ndash19

Bauer H G Chapron K Nowell P Henschel P Funston L T B HunterD W Macdonald and C Packer 2015 Lion (Panthera leo) populations aredeclining rapidly across Africa except in intensively managed areas Pro-ceedings of National Academy of Sciences of the United States of America11214894ndash14899

Beier P 1995 Dispersal of juvenile cougars in fragmented habitat Journal ofWildlife Management 59228ndash237

Benson J F J A Hostetler D P Onorato W E Johnson M E Roelke SJ OrsquoBrien D Jansen and M K Oli 2011 Intentional genetic introgressioninfluences survival of adults and subadults in a small inbred felid populationJournal of Animal Ecology 80958ndash967

Benson J F P J Mahoney J A Sikich L E K Serieys J P Pollinger H BErnest and S P D Riley 2016 Interactions between demography geneticsand landscape connectivity increase extinction probability for a small popu-lation of large carnivores in a major metropolitan area Proceedings of theRoyal Society B Biological Sciences 28320160957

Beverly F 1874 The Florida panther Forest and Stream 3290Boyce M S C V Haridas C T Lee and the NCEAS Stochastic Demo-graphy Working Group 2006 Demography in an increasingly variable worldTrends in Ecology amp Evolution 21141ndash148

Brook B W J J OrsquoGrady A P Chapman M A Burgman H R Akcakayaand R Frankham 2000 Predictive accuracy of population viability analysis inconservation biology Nature 404385ndash387

Brys R and H Jacquemyn 2016 Severe outbreeding and inbreeding de-pression maintain mating system differentiation in Epipactis (Orchidaceae)Journal of Evolutionary Biology 29352ndash359

Burke J M and M L Arnold 2001 Genetics and the fitness of hybridsAnnual Review of Genetics 3531ndash52

Burnham K P 1993 A theory for combined analysis of ring recovery andrecapture data Pages 199ndash213 in J‐D Lebreton and P M North editorsMarked individuals in the study of bird population Springer‐Verlag NewYork New York USA

Burnham K P and D R Anderson 2002 Model selection and multimodelinference a practical information‐theoretic approach Second edition SpringerVerlag New York New York USA

Caswell H 2001 Matrix population models construction analysis and in-terpretation Sinauer Associates Sunderland Massachusetts USA

Charpentier M J M Setchell F Prugnolle L A Knapp E J Wickings PPeignot and M Hossaert‐McKey 2005 Genetic diversity and reproductivesuccess in mandrills (Mandrillus sphinx) Proceedings of the NationalAcademy of Sciences of the United States of America 10216723ndash16728

Cooch E and G C White editors 2018 Program MARK a gentle in-troduction lthttpwwwphidotorgsoftwaremarkdocsbookgt Accessed 7Apr 2016

Cooley H S R B Wielgus G M Koehler H S Robinson and B TMaletzke 2009 Does hunting regulate cougar populations A test of thecompensatory mortality hypothesis Ecology 902913ndash2921

Coulson T E A Catchpole S D Albon B J Morgan J M Pemberton TH Clutton‐Brock M J Crawley and B T Grenfell 2001 Age sex densitywinter weather and population crashes in Soay sheep Science 2921528ndash1531

Coulson T J Pemberton S Albon M Beaumont T Marshall J Slate FGuinness and T Clutton‐Brock 1998a Microsatellites reveal heterosis in reddeer Proceedings of the Royal Society B Biological Sciences 265489ndash495

Coulson T N S D Albon J M Pemberton J Slate F E Guinness and TH Clutton‐Brock 1998b Genotype by environment interactions in wintersurvival in red deer Journal of Animal Ecology 67434ndash445

Cox D 1972 Regression models and life‐tables Journal of the Royal StatisticalSociety Series B Statistical Methodology 34187ndash220

Creel S 2006 Recovery of the Florida panthermdashgenetic rescue demographic rescueor both Response to Pimm et al (2006) Animal Conservation 9125ndash126

Crnokrak P and D A Roff 1999 Inbreeding depression in the wild Heredity83260ndash270

Crow J 1948 Alternative hypotheses of hybrid vigor Genetics 33477ndash487

van de Kerk et al bull Florida Panther Population and Genetic Viability 31

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 15: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

100 years assuming that the population would be managedaccording to a particular strategy Our goal with theseexploratory analyses was to evaluate the relative costs andbenefits of alternative management scenarios

RESULTS

Genetic VariationWe assessed genetic variation at 16 microsatellite loci for 137DNA samples collected from adult and subadult panthers(1981ndash2013) that were captured and subsequently radiocollaredThe number of alleles at these loci ranged from 3ndash10 and allelicrichness averaged 57 (Table 2) Average expected hetero-zygosity for this sample was 0535 and average observed het-erozygosity was 0517 (Table 2)We determined individual heterozygosity (1 minusHL) and an-

cestry for the complete sample of panther genotypes (n= 503)collected from 1981 to 2015 for use as covariates in subsequentanalyses Average individual heterozygosity was 0559 andranged from 0ndash0980 Our STRUCTURE analysis providedstrong support for K= 2 clusters based on the probability of thedata (log Pr(D)|K) and ΔK in STRUCTURE HARVESTERBased on this result we categorized the ancestry of each pantheras either canonical (n= 123) or admixed (n= 380) using valuesof proportional membership (q Fig 4) The average individualheterozygosity of canonical panthers was 0386plusmn 0012 (SE) foradmixed panthers it was 0615plusmn 0007

Survival and Cause‐Specific Mortality RatesOf the 395 kittens that were PIT‐tagged and released 55 werelater encountered alive 15 were recovered dead and 85 werepart of failed litters (Table 1) We radio‐tracked 209 subadult oradult panthers for a total of 244195 panther‐days and docu-mented 126 mortalities (Table 1)Survival depended strongly on age and kittens had the lowest

overall rate (032plusmn 009 Fig 5A Table 3) The model‐averagedkitten survival was 047plusmn 009 when we included only geneti-cally sampled individuals in the analysis (Table 3) as statedabove this estimate is biased high because many kittenswere genetically sampled when they were recaptured alive orrecovered dead at older ages We found no evidence for sex‐specific differences in kitten survival but females survived betterthan males in all older age classes For females subadults hadthe highest survival rates followed by prime adults and oldadults For males prime adults had the highest survival ratefollowed by subadults and old adults (Fig 5A and Table 3)For panthers of all ages there was substantial evidence that

ancestry affected survival with F1 admixed panthers of all ageshaving the highest rates and canonical individuals having thelowest (Fig 5B) The combined AIC weights of models thatincluded an ancestry covariate were 0880 for kittens and 0992for older panthers The survival rates of subadult prime‐adultand old‐adult panthers in the period immediately after the in-trogression (1995ndash2004) were similar to those in more recentyears (2005ndash2013 Fig 5C)After genetic introgression individual panther heterozygosity

increased (Fig 6) and varied among ancestry categories individualheterozygosity was the highest for F1 panthers (078plusmn 001)

A

B

C

Figure 5 Overall age‐ and sex‐specific survival rates (plusmnSE A) age‐ and sex‐specific survival rates (plusmnSE) for Florida panthers of canonical F1 admixed andother admixed ancestry (B) and age‐ and sex‐specific survival estimates (plusmnSE)for subadult and adult Florida panthers in the period immediately post‐introgression (1995ndash2004) and more recently (2005ndash2013 C) in SouthFlorida USA

Table 3 Model‐averaged estimates (plusmnSE) of demographic parameters for theFlorida panther obtained using data collected during 1981ndash2013 in SouthFlorida USA

Parameter Sex Age class Estimate

Survival Both Kitten 032plusmn 009a

Female Subadult 097plusmn 002Prime adult 086+ 003Old adult 078plusmn 009

Male Subadult 066plusmn 006Prime adult 077plusmn 005Old adult 065plusmn 010

Probability of reproduction Female Subadult 035plusmn 008Prime adult 050plusmn 005Old adult 025plusmn 006

Kittens produced annually Female Subadult 280plusmn 005Prime adult 267plusmn 001Old adult 228plusmn 013

a Overall kitten survival (based on all kittens in our data set including thosethat were not genetically sampled) Estimate of survival of geneticallysampled kittens was 047plusmn 009 this estimate is biased high because manykittens were genetically sampled when they were recaptured alive or re-covered dead at older ages

van de Kerk et al bull Florida Panther Population and Genetic Viability 17

followed by those for non‐F1 admixed (062plusmn 001) and canonical(039plusmn 001) panthers Individual heterozygosity positively affectedsurvival rates of kittens (slope parameter η= 1455 95CI= 0505ndash2405) Individual heterozygosity also positively af-fected survival of subadult and adult panthers but the evidence forthis effect was weaker because the confidence interval for the ha-zard ratio overlapped 10 (hazard ratio exp(ɩ)= 0311 95CI= 0092ndash1054 Table 4 and Fig 7)The MPC increased after genetic introgression (Fig 2) This

abundance index was also included in top‐ranking modelsindicating that population density affected survival (Table 4)Abundance reduced the survival of kittens (slope parameterη= minus0035 95 CI= minus0049 to minus0021) evidence was weak forthe effect of abundance on survival of subadult and adult panthers(hazard ratio exp(ɩ)= 1002 95 CI= 0995ndash1010 Fig 7) Re-gression coefficients associated with the effect of abundance andindividual heterozygosity on demographic parameters are presentedin Table B1 (available online in Supporting Information)The most frequently observed causes of death for radio‐

collared panthers were vehicle collision and intraspecific ag-gression Cause‐specific mortality analyses revealed thatmortality rates among possible causes were generally similar forthe 2 sexes but that males were more likely to die from in-traspecific aggression than were females (Fig 8A) Male mor-tality from intraspecific aggression was higher for subadult andold adult panthers than for prime adults (Fig 8B) For bothmales and females canonical panthers were more likely to diefrom intraspecific aggression than were admixed panthers(Fig 8C)

Reproductive ParametersOf 101 radio‐collared females 67 reproduced at least once duringour monitoring we used these individuals to assess the annualprobability of reproduction The top‐ranked model for the annual

probability of reproduction included age effect only suggesting thatadding ancestry or abundance did not improve model parsimony(Table 4) Model‐averaged annual probabilities of reproductiondiffered between female subadults (035plusmn 008) prime adults(050plusmn 005) and older adults (025plusmn 006)The most parsimonious model for the number of kittens

produced annually by females that produced at least 1 litter (ieconditional on reproduction) included effects of age ancestryand abundance a model that included only the effect of age wasalmost equally supported (Table 4) The model‐averagednumber of kittens produced annually conditional on re-production was 280plusmn 075 for subadults 267plusmn 043 for primeadults and 228plusmn 083 for old adults Confidence intervals forthe slope parameter (β) overlapped 0 for both abundance(β= 0010 95 CI= minus0001 to 0021) and ancestry (β= minus073795 CI= minus1637 to 0162)

Population Dynamics and PersistenceDeterministic and stochastic demographymdashThe deterministic (λ)

and stochastic (λs) annual population growth rate calculated usingthe matrix population model were 106 (5th and 95th percentiles099ndash114) and 104 (072ndash141) respectively The generation timewas 47 years A female starting in age class one (kitten) wasexpected to live another 58 years and produce 10 kittens onaverage during her lifetime Overall λs was proportionately(elasticity) most sensitive to changes in female prime adultsurvival on the absolute scale (sensitivity) however it was mostsensitive to changes in kitten survival (Fig 9) Populationtrajectories probabilities of quasi‐extinction and times untilquasi‐extinction estimated using the matrix population modeland IBM for comparable scenarios were nearly identical for acritical quasi‐extinction threshold of 10 panthers (Figs 10 and 11)and for a critical quasi‐extinction threshold of 30 panthers(Appendix E available online in Supporting Information)Minimum population count scenariomdashUnder the MPC scenario

(ie density dependence estimated using the minimumpopulation count data) with a critical threshold of 10panthers the population size reached 99 (72ndash104) and 100(77ndash105) at 200 years (Fig 10) as predicted by the IBM and thematrix model respectively The cumulative quasi‐extinctionprobabilities within 100 years based on the MPC scenario were15 (IBM 0ndash48) and 30 (matrix model 0ndash76) Theseprobabilities increased to 25 (IBM 0ndash114) and 38 (matrixmodel 0ndash154) by year 200 (Fig 10) The mean times untilquasi‐extinction were 15 years (IBM 0ndash67) and 15 years (matrixmodel 0ndash72) conditioned on going quasi‐extinct within 100years Expected times until quasi‐extinction conditioned onquasi‐extinction within 200 years were higher 34 years (IBM0ndash137) and 32 years (matrix model 0ndash134 Fig 10)Motor vehicle mortality scenariomdashUnder the MVM scenario

(ie density dependence estimated using estimates of pantherpopulation size from McClintock et al 2015) with a criticalthreshold of 10 panthers the projected population reached 188(142ndash217) and 190 (143ndash219) at 200 years as predicted by theIBM and the matrix model respectively (Fig 11) Thecumulative quasi‐extinction probabilities within 100 years were14 (IBM 0ndash08) and 13 (matrix model 0ndash06) Theseprobabilities increased to 20 (IBM 0ndash17) and 19 (matrix

000

025

050

075

100

1995 2000 2005 2010Year of birth

Indi

vidu

al h

eter

ozyg

osity

Figure 6 Temporal changes in the individual heterozygosity of Floridapanthers born during the period 1995ndash2013 in South Florida USA Pointsare measurements solid line is linear regression shaded area is 95 confidenceinterval of slope Slope= 0006plusmn 0001 R2= 009

18 Wildlife Monographs bull 203

model 0ndash16) within year 200 (Fig 11) The mean times until quasi‐extinction based on the MVM scenario were 5 years (IBM 0ndash61)and 6 years (matrix model 0ndash64) conditioned on going quasi‐extinctwithin 100 years Expected times until quasi‐extinction conditionedon quasi‐extinction within 200 years were 11 years (IBM 0ndash113)and 11 years (matrix model 0ndash112 Fig 11)

Sensitivity of extinction probability to demographic parametersmdashThe partial rank correlation coefficients between quasi‐extinctionprobabilities and demographic parameters were negative anincrease in any of the demographic parameters decreased thequasi‐extinction probability Probabilities of quasi‐extinction within200 years were most sensitive to changes in kitten survival for bothscenarios (Fig 12) followed by survival of subadult females in theMVM scenarios and survival of prime adult females in the MPCscenario The probability of reproduction of prime adult femaleshad the third‐largest partial rank correlation coefficient with quasi‐extinction probability for both scenarios

Benefits and Costs of Genetic ManagementMinimum population count scenariomdashThe MPC scenario with

a critical threshold of 10 panthers predicted that withoutmanagement intervention average individual heterozygositywould decrease reaching 057 (049ndash064) at 100 years and053 (042ndash062) at 200 years (Fig 13) The population woulddecrease slightly reaching an average of 79 (41ndash99) at 100 yearsand 76 (43ndash98) at 200 years (Fig 14) The probabilities of quasi‐extinction under the no‐intervention MPC scenario when weconsidered the effects of genetic erosion were 13 (0ndash99) at 100years and 23 (0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with

the translocation of 5 Texas females average individualheterozygosity under the MPC scenario increased and thenstabilized at 068 (064ndash072) at 100 years and remained atthat level at 200 years (Fig 13) The population reached anaverage of 88 (62ndash104) at 100 years and stayed the same at200 years (Fig 14) The probabilities of quasi‐extinctionunder this introgression strategy were 6 (0ndash53) within 100years and 7 (0ndash82) within 200 years (Fig 15) Results ofanalyses using a critical threshold of 30 panthers revealed asimilar pattern of relative benefits However the prob-abilities of quasi‐extinction were substantially higher (ge45)across all scenarios (Appendix E)

Motor vehicle mortality scenariomdashWithout managementintervention the average individual heterozygosity predictedby the MVM scenario and a critical threshold of 10 panthersdecreased to 060 (046ndash069) at 100 years and to 059 (039ndash070) at 200 years (Fig 13) The population increased slightlyand reached an equilibrium at 154 individuals (46ndash217) at 100years and 157 (57ndash218) at 200 years (Fig 14) The probabilitiesof quasi‐extinction were 17 (0ndash100) at 100 years and 22(0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with 5

female pumas from Texas average individual heterozygosityincreased slightly reaching 067 (060ndash073) at 100 years and068 (059ndash075) at 200 years (Fig 13) Under this strategy thepopulation reached an average of 164 individuals (44ndash226) at

Table 4 Model comparison results testing for the effects of several covariateson survival of all kittens survival of genetically sampled kittens survival ofsubadult and adult Florida panthers probability of reproduction and kittensproduced annually for Florida panthers sampled from 1981ndash2013 in SouthFlorida USA

Model Parameters ΔAICa Weight

Kitten survival (all data)Base+ F1b+ abundancec 10 000 0988Base+ abundance 9 1018 0006Base+ F1 9 1035 0005Base 8 2711 lt0001Base+ sexd 9 2912 lt0001Base+ litter sizee 9 2914 lt0001

Kitten survival (genetically sampled kittens only)Base+ F1+ abundance 10 000 0541Base+ F1CanAdmf+ abundance 11 099 0329Base+Hetg+ abundance 10 310 0115Base+CanAdmh+ abundance 10 791 0010Base+ abundance 9 944 0005Base+ F1 9 1638 lt0001Base+ F1CanAdm 10 1837 lt0001Base+Het 9 2872 lt0001Base 8 3296 lt0001Base+CanAdm 9 3348 lt0001

Subadult and adult survivalBase+ F1CanAdm+ abundance 7 000 0360Base+ F1 5 053 0276Base+ F1CanAdm 6 105 0213Base+ F1+ abundance 6 222 0119Base+CanAdm+ abundance 6 575 0020Base+CanAdm 5 908 0004Base+Het 5 964 0003Base+Het+ abundance 6 984 0003Base 4 1118 0001Base+ abundance 5 1278 0001

Probability of reproductionAge 3 000 0242Age+CanAdm 4 012 0228Age+ abundance 4 173 0102Age+ F1 4 190 0094Age+CanAdm+ abundance 5 216 0082Age+Het 4 219 0081Age+ F1CanAdm 5 232 0076Age+ F1+ abundance 5 372 0038Age+Het+ abundance 5 399 0033Age+ F1CanAdm+ abundance 6 444 0026

Kittens produced annuallyAge+ F1+ abundance 5 000 0194Age 3 060 0144Age+ abundance 4 061 0143Age+ F1 4 097 0120Age+CanAdm 4 139 0097Age+ F1CanAdm+ abundance 6 196 0073Age+ F1CanAdm 5 231 0061Age+CanAdm+ abundance 5 238 0059Age+Het+ abundance 5 250 0056Age+Het 4 259 0053

a Akaikersquos information criterion (AIC) for kitten survival models was adjustedfor small sample size and overdispersion (QAICc)

b F1 divides Florida panthers into 2 groups F1 admixed and all other panthersc Index of Florida panther abundanced Sex of an individual Florida panther (male or female)e Size of the litter in which a Florida panther kitten was bornf F1CanAdm divides panthers into 3 groups F1 admixed canonical andother admixed panthers

g Het is the individual heterozygosityh CanAdm divides panthers into 2 groups canonical and admixed (includingF1 admixed) panthers

van de Kerk et al bull Florida Panther Population and Genetic Viability 19

100 years and 170 (67ndash231) at 200 years based on the MVMscenario (Fig 14) The probabilities of quasi‐extinction underthis introgression strategy were 10 (0ndash99) at 100 years and12 (0ndash99) at 200 years (Fig 15)The projected population size and average individual hetero-

zygosity reached equilibrium levels with periodic fluctuations inthese values when introgression was implemented (Figs 16ndash17)Genetic introgression decreased the time until quasi‐extinctionacross all scenarios (Fig 18) Results of analyses using a criticalthreshold of 30 panthers revealed a similar pattern of relative ben-efits However the probabilities of quasi‐extinction were sub-stantially higher (ge45) across all scenarios (Appendix E)

Addition of pumas from Texas increased both the populationsize and average individual heterozygosity consequentlyintrogression caused periodic increases in average individualheterozygosity and population size at the interval at which theintrogression occurred Whereas average individual hetero-zygosity gradually decreased following introgression densitydependence caused the population size to fluctuate especiallywhen a larger number of pumas from Texas were released Thepositive effect of genetic introgression initiatives on quasi‐extinction probabilities decreased as the time interval betweenthese events increased More frequent genetic introgressions ledto greater increases in average individual heterozygosity and

Old adult

Prime adult

Subadult

Kitten

025 050 075

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Individual heterozygosity

Ann

ual s

urvi

val r

ate

Old adult

Prime adult

Subadult

Kitten

40 60 80 100 120

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Abundance index

Ann

ual s

urvi

val r

ate

Kittens Males Females

Figure 7 The effect of individual heterozygosity (left) and the abundance index (right) on Florida panther age‐ and sex‐specific survival estimates (shaded area isSE) in South Florida USA 1981ndash2013 Abundance index data are from McBride et al (2008) and R T McBride and C McBride (Rancherrsquos Supply Incorporatedunpublished report)

20 Wildlife Monographs bull 203

equilibrium population size while reducing the probability ofquasi‐extinction Comparatively performing genetic introgres-sion less often but with more individuals per release was lesseffective at improving the long‐term prospects for the pantherpopulation (Figs 13ndash15)

Financial cost and persistence benefitsmdashThe total estimated costof 1 genetic introgression was $40000 $80000 or $120000for releasing 5 10 or 15 pumas from Texas respectively(Table 5) The total cost incurred over 100 years for eachstrategy ranged from $50000 to $12 million (Table 6)The most expensive strategy involved the introduction of15 pumas every 10 years this strategy reduced the probability

of quasi‐extinction by 58 and 40 under the MPC and MVMscenarios respectively (Table 6) Introducing 5 pumas every80 years cost the least but improvements in populationpersistence under this strategy were insubstantial (Table 6)Releasing more panthers more frequently caused increasedpopulation fluctuations but did not necessarily reduce the quasi‐extinction probability (Figs 14 and 15 Table 6)

DISCUSSION

The duration (34 yr of data) and sample sizes associated with theFlorida Panther Project allowed us to obtain a comprehensiveunderstanding of genetic variation demographic parameters

A

B

C

Figure 8 Cause‐specific mortality rates (plusmnSE) for male and female Florida panthers (A) subadult prime‐adult and old‐adult Florida panthers of both sexes (B)and canonical and admixed Florida panthers of both sexes (C) in South Florida USA 1981ndash2013 Causes of mortality are vehicle collision intraspecific aggression(ISA) other causes (known causes such as diseases injuries and infections unrelated to the first 2 causes) and unknown cause (mortalities for which we could notassign a cause)

van de Kerk et al bull Florida Panther Population and Genetic Viability 21

probability of population persistence and the duration of thepositive impacts of genetic restoration on this endangered po-pulation The Florida panther population was in dire straits inthe early 1990s and our results clearly demonstrated that the

implementation of the introgression project in 1995 subse-quently accrued a series of genetic and demographic improve-ments that have led to the change from a declining population toone that is growing and expanding its current breeding range

Sensitivity Elasticity

0 1 2 3 00 02 04

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

Para

met

er

Figure 9 Sensitivity and elasticity (proportional sensitivity) of stochastic population growth rate (λs) of the Florida panther to vital demographic parameters in SouthFlorida USA 1981ndash2013 Horizontal lines represent 5th and 95th percentiles

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

70

80

90

100

110

0

5

10

15

0

50

100

Years

StatisticMeanMedian

Figure 10 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the minimum population count(MPC) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

125

150

175

200

00

05

10

15

20

0

30

60

90

Years

StatisticMeanMedian

Figure 11 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the motor vehicle mortality(MVM) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

22 Wildlife Monographs bull 203

MPC MVM

minus08 minus06 minus04 minus02 00 minus08 minus06 minus04 minus02 00

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Male subadult survival

Male prime adult survival

Male old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

PRCC

Para

met

er

Figure 12 Partial rank correlation coefficient (PRCC) between quasi‐extinction probabilities and the demographic parameters of the Florida panther for theminimum population count (MPC) and motor vehicle mortality (MVM) scenarios Error bars indicate standard deviation Based on data collected in South FloridaUSA 1981ndash2013

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 13 Average projected individual heterozygosities in the Florida panther population over 200 years without genetic management intervention and with eachof the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include theintroduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years Average individual heterozygosity peaks when pumas are added to the Floridapanther population Based on data collected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 23

Our research has further validated the success of genetic in-trogression by quantifying the reduction in the probability ofextinction of the panther population because of this manage-ment initiative These findings all bode well for continuedprogress toward recovery That said the Florida panther po-pulation remains small and isolated from other populations ofpuma from western North America thereby denoting that theeventual impact of genetic drift and inbreeding may negativelyaffect the population in the future Realizing this will continueto be an issue that managers will have to contemplate wemodeled the impact of varied genetic management scenarios todetermine the frequency of introgression along with the numberof panthers that should be released during each initiative tominimize cost and maximize benefits to the population Theseresults will prove invaluable to managers attempting to conservethe Florida panther

Genetic Variation Demographic Parametersand Persistence of Heterotic Benefits from GeneticIntrogressionOur measure of individual heterozygosity (1 minusHL) was higheron average for the admixed (0615) panthers than for those ofcanonical ancestry (0386) Levels of individual heterozygosityfor admixed panthers were comparable to levels observed in asample of pumas from west Texas (x plusmn SE= 0695plusmn 0019n= 41) which further demonstrates the improvement in levelsof genetic variation in the Florida population since 1995 The

correlation between HL and several demographic parametershas been noted in wild populations including cheetahs (Terrellet al 2016) and several species of birds such as European shags(Phalacrocorax aristotelis male and female survival probabilityVelando et al 2015) and Egyptian vulture (Neophron percnop-terus age at recruitment Agudo et al 2012) If we focus only onpanthers captured and radiocollared from 2012ndash2015 (FP211ndashFP240) all of which had estimated birth years ge2005 (ge10 yrpost‐introgression) those panthers had an average individualheterozygosity level of 0618plusmn 0029 highlighting the elevatedlevels of genetic variation that continue to persist in cohorts ofpanthers 5 generations after the release of female pumas fromTexasThe proportional membership values (q) from our

STRUCTURE analysis allowed us to delineate 2 clusters ofancestry for Florida panthers (Fig 4) During the pre‐in-trogression era the population was dominated by panthers thatretained a high percentage of the canonical ancestry which wasaffected by inbreeding depression The post‐introgression eraancestry values are comprised of many admixed individuals inessence demonstrating the success of the introgression in termsof increasing genetic variation in the population and mi-micking gene flow that historically occurred with panthers andother populations of pumas (Onorato et al 2010) A somewhatanalogous situation although abetted by natural immigrationwas evident in the ancestral variation in a small population ofpuma intersected by a major freeway in California USA In

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

75

80

85

90

95

100

130

150

170

190

Years

Popu

latio

n si

ze

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 14 Average projected population sizes in the Florida panther population over 200 years without intervention and with each of the genetic introgressionstrategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include the introduction of 5 10 or 15pumas from Texas USA every 10 20 40 or 80 years Peaks occur when pumas are added to the Florida panther population Based on data collected in SouthFlorida USA 1981ndash2013

24 Wildlife Monographs bull 203

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

25

50

75

100

0

25

50

75

100

Introgression interval (yrs)

PQ

E (

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 15 Average probability of quasi‐extinction (PQE) of the Florida panther population within 100 years and within 200 years without genetic managementintervention (N) and with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenariosIntrogression strategies include the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Errorbars indicate 5th and 95th percentiles Based on data collected in South Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

50

60

70

80

90

100

110

50

100

150

200

Years

Popu

latio

n si

ze

Figure 16 Average projected Florida panther population trajectories under a geneticintrogression strategy involving the release of 5 female pumas from Texas USA every20 years for the minimum population count (MPC) and motor vehicle mortality(MVM) scenarios Shaded area indicates 5th and 95th percentiles and isrepresentative of confidence intervals for other scenarios Based on data collected inSouth Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Figure 17 Average projected Florida panther population‐level heterozygositiesunder a genetic introgression strategy involving the release of 5 female pumas fromTexas USA every 20 years for the minimum population count (MPC) and motorvehicle mortality (MVM) scenarios Shaded area bounds the 5th and 95th percentilesand is representative of confidence intervals for other scenarios Based on datacollected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 25

this case the migration of just a single male across the freewayto the south resulted in an increase in the number of in-dividuals with admixed ancestry and substantially improvedgenetic diversity of the subpopulation south of the freeway(Riley et al 2014)Our estimate of overall kitten survival (0321plusmn 0056) was

similar to that reported by Hostetler et al (2010) who used13 years of post‐introgression data (0323plusmn 0071) These valuesare lower than kitten survival estimates derived from severalpuma populations in western North America (044ndash091 Loganand Sweanor 2001 Lambert et al 2006 Laundre et al 2007Robinson et al 2008 Ruth et al 2011) When we included onlydata for individuals that had been genetically sampled our es-timate was 15 higher The proportion of admixed kittens thatwere genetically sampled increased as the population expandedwhich could have resulted in higher estimates of kitten survivalbecause admixed kittens survive better than canonical kittensKitten survival was strongly influenced by genetic ancestry withsurvival rates of F1 kittens being almost twice that of canonicalkittens (Fig 5B) Consistent with the findings of Hostetler et al(2010) increases in heterozygosity coincided with increasedkitten survival whereas population density negatively affectedkitten survival Population density often has a strong negativeimpact on survival of young age classes due to infanticide andcannibalism which has been reported from several carnivore

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

50

100

150

0

50

100

150

Introgression interval (yrs)

TQE

(yrs

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 18 Average times until quasi‐extinction (TQE) for the Florida panther population within 100 years and within 200 years without genetic management interventionand with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategiesinclude the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Error bars indicate 5th and 95thpercentiles Based on data collected in South Florida USA 1981ndash2013

Table 5 Average cost (2017 US dollars) of introgression strategies employedover 100 years that involve the introduction of 5 10 or 15 pumas from TexasUSA into the Florida panther population in South Florida USA at an in-creasingly higher frequency Costs of capture (including transportation) quar-antine and post‐introgression monitoring were estimated by Roy McBrideMark Cunningham and Dave Onorato respectively

Frequency Component 5 pumas 10 pumas 15 pumas

Once Capture 25000 50000 75000

Quarantine 5000 10000 15000

Monitoring 10000 20000 30000

Total 40000 80000 120000

Every 80 years Capture 31250 62500 93750

Quarantine 6250 12500 18750

Monitoring 12500 25000 37500

Total 50000 100000 150000

Every 40 years Capture 62500 125000 187500

Quarantine 12500 25000 37500

Monitoring 25000 50000 75000

Total 100000 200000 300000

Every 20 years Capture 125000 250000 375000

Quarantine 25000 50000 75000

Monitoring 50000 100000 150000

Total 200000 400000 600000

Every 10 years Capture 250000 500000 750000

Quarantine 50000 100000 150000

Monitoring 100000 200000 300000

Total 400000 800000 1200000

26 Wildlife Monographs bull 203

species including polar bears (Ursus maritimus Derocher andWiig 1999) black bears (Ursus americanus Garrison et al 2007)and pumas (Logan and Sweanor 2001)Our estimates of sex‐ and age‐specific survival rates of subadult

and adult panthers (Table 3) and the effects of covariates on theserates were similar to those reported by Benson et al (2011) derivedfrom data collected through 2006 Our survival rates for males(065ndash077) were lower than females (078ndash097) for all 3 ageclasses (subadult prime adult and old adult) which is the typicaltrend observed in most puma populations (eg Ruth et al 19982011) However an unhunted puma population occupying afragmented urban landscape in southern California exhibited lowersurvival (0586 females 0525 males Vickers et al 2015) perhapsas a result of severe habitat fragmentation and the lack of extensiveswaths of public land protected in perpetuity Most populations ofpuma in western North America are typically subject to variedlevels of management via regulated hunting which often is biasedtoward the take of males (Cooley et al 2009 Ruth et al 2011)Consequently annual survival estimates from hunted populationsin northeast Washington (0679ndash0733 females 0341 malesRobinson et al 2008) and southeastern Arizona (0685 females0584 males Cunningham et al 2001) were generally lower thanthose we estimated for Florida panthersCause‐specific mortality analysis showed that male panthers

experienced a substantially greater mortality risk from in-traspecific aggression This differs from the pattern observedduring a long‐term study in Arizona where females were at ahigher risk of intraspecific mortality than males (46 of maleand 53 of female mortality Logan and Sweanor 2001)Mortality associated with intraspecific strife has been docu-mented in hunted and unhunted populations (this study Loganand Sweanor 2001 Stoner et al 2010) and its root cause hasbeen difficult to determine (eg population density and preypopulation trends) That being the case finding ways to reducewhat is often a major source of mortality for puma populationscontinues to pose a challenge to managersCanonical panthers were substantially more likely to die from

intraspecific aggression than were admixed panthers This mayat least partly explain the difference in survival between admixed

and canonical panthers Canonical panthers are known to bemore likely to suffer from varied correlates of inbreeding de-pression than admixed animals including atrial septal defects(Johnson et al 2010) and compromised immune systems(Roelke et al 1993) These factors have the potential to affectfitness and perhaps dominance of individuals when engaged inan intraspecific encounter A somewhat similar scenario wasobserved by Hogg et al (2006) in a population of bighorn sheepthat were part of an introgression project Admixed males in thispopulation had a greater probability of paternity than males thatwere less outbred This included coursing males with higherlevels of admixture being markedly more successful at fightingmales that were defending females in estrous (Hogg et al 2006)Heterosis resulting from genetic introgression is expected to be

the strongest in F1 individuals (Shull 1908 Crow 1948 Burkeand Arnold 2001 Waller 2015) and to progressively decline insubsequent generations (Pickup et al 2013 Frankham 2015)Given that admixed (especially F1) panthers survived better thancanonical panthers and that individual heterozygosity improvedsurvival of both sexes and all age classes our data provide evi-dence for heterotic advantage whereby individuals of mixedancestry had higher fitness (Shull 1908 Crow 1948) Our resultsare consistent with findings of other studies that involved in-tentional genetic introgression for conservation purposes Forexample Hogg et al (2006) detected substantial improvementsin survival and reproduction of introgressed bighorn sheep andMadsen et al (1999 2004) reported a dramatic increase in thepopulation of adders after genetic introgressionKnowledge of the persistence of benefits to a population ac-

crued via genetic introgression is essential for determiningwhen additional introgression may be warranted There is adepauperate amount of information regarding this topic and theidentification of factors that may influence persistence of thebenefits of genetic introgression (Tallmon et al 2004 Hedricket al 2014 Whiteley et al 2015) For example in minks(Neovison vison) Thirstrup et al (2014) found that outcrossingled to larger litters in F2 and F3 but this benefit disappeared insubsequent generations In a population of gray wolves Hedricket al (2014) suggested that benefits of genetic introgression may

Table 6 Costs and benefits accumulated over 100 years for genetic introgression at different intervals and with different numbers of pumas from Texas USAintroduced into the Florida panther population in South Florida USA Costs (2017 US dollars) include capture and transportation quarantine and post‐introgression monitoring Benefits are represented as average probability of quasi‐extinction (PQE and 5th and 95th percentiles) and changes (Δ) therein under thescenario using the minimum population count (McBride et al 2008 MPC) and the scenario using the motor vehicle mortality population estimates (McClintocket al 2015 MVM) The table is sorted by percent change in probability of quasi‐extinction under the MPC scenario

Interval (yr) Pumas Cost MPC PQE () ΔMPC PQE () MVM PQE () ΔMVM PQE ()

ndash 0 0 13 (0ndash99) 0 17 (0ndash100) 080 10 100000 12 (0ndash99) 10 (0ndash58) 16 (0ndash100) 5 (0ndash38)80 15 150000 12 (0ndash99) 13 (0ndash65) 15 (0ndash100) 8 (0ndash56)80 5 50000 12 (0ndash99) 14 (0ndash73) 15 (0ndash99) 9 (0ndash41)40 15 300000 10 (0ndash98) 26 (0ndash104) 14 (0ndash99) 13 (0ndash60)40 10 200000 9 (0ndash94) 35 (0ndash183) 13 (0ndash99) 21 (0ndash95)40 5 100000 8 (0ndash89) 39 (0ndash211) 12 (0ndash99) 26 (0ndash124)20 15 600000 8 (0ndash88) 42 (0ndash257) 13 (0ndash99) 24 (0ndash123)20 10 400000 6 (0ndash67) 54 (0ndash318) 10 (0ndash99) 37 (0ndash217)10 15 1200000 6 (0ndash53) 58 (0ndash372) 10 (0ndash99) 40 (0ndash208)20 5 200000 5 (0ndash50) 59 (0ndash338) 9 (0ndash99) 44 (0ndash352)10 10 800000 4 (0ndash16) 69 (0ndash489) 8 (0ndash92) 55 (0ndash401)10 5 400000 4 (0ndash7) 73 (0ndash502) 6 (0ndash71) 63 (0ndash503)

van de Kerk et al bull Florida Panther Population and Genetic Viability 27

wane after 2 or 3 generations The WrightndashFisher model ofgenetic drift predicts a geometric decline in expected hetero-zygosity at the rate of (1 minus 12Ne) per generation where Ne is theeffective population size (Hartl and Clark 1997 Frankham et al2014) Thus it seems logical to assume that the duration of thebenefits of introgression is limited especially in a species per-sisting in a small population such as Florida panthersWe expected Florida panther survival in the most recent years

(2005ndash2013) post‐introgression to differ from that observed in thedecade following the introgression in 1995 because pantherabundance and heterozygosity factors known to influence panthersurvival (Hostetler et al 2010 Benson et al 2011) have changed(Figs 2 and 6) We found that subadult and adult survival rates inrecent years (2005ndash2013) were similar to those in the period im-mediately following introgression (1995ndash2004 Fig 5C) overallkitten survival was similar to estimates of Hostetler et al (2010) Infact the pattern of covariate effects on Florida panther survivalreported by Benson et al (2011) and Hostetler et al (2010) hasbarely changed The negative effects of population density andpositive effects of heterozygosity may counterbalance survival ratesreducing population fluctuations Our result that benefits of geneticrestoration have remained in the population nearly for 5 genera-tions post‐intervention was surprising but also encouraging Pos-sible explanations for this observation may include 1) genetic ero-sion occurs at slower rate than previously thought 2) introducing 5migrants in 1 genetic intervention may have beneficial effects si-milar to those from 1 migrant per generation over multiple gen-erations or 3) other and as yet unknown factors may reduce therate of genetic erosion and subsequent demographic effects Adensity‐dependent effect on kitten survival could also explain whynon‐F1 admixed kittens did not survive substantially better thandid canonical kittens most kittens sampled from 2005ndash2013 wereadmixed and their survival was lower possibly because of a density‐dependent effect as the Florida panther population continuouslygrew during that period Trinkel et al (2010) also found thatpopulation density and inbreeding coefficient interacted to affectdemographic parameters and growth rate of an African lion po-pulation Likewise population density interacted with winterweather to affect the survival and population growth rates in Soaysheep (Ovis aries Milner et al 1999 Coulson et al 2001)

Population Dynamics and PersistenceThe finite deterministic and stochastic growth rates were gt10reflecting that much of the data used in this study were collectedfollowing genetic introgression in 1995 during which time thepopulation increased substantially Because our demographicparameter estimates did not change markedly in comparison tothose of Hostetler et al (2013) the similarity between thepopulation growth estimates from these studies was expectedBut these estimates reflect population growth during thedata‐collection period and do not predict future growth In factestimates of population size reported by McClintock et al(2015) suggest that population growth may already be slowing(Fig 2) A similar pattern was observed in an introduced po-pulation of African lion which increased steadily from 1995until 2001 then fluctuated around the presumed carrying ca-pacity thereafter (Trinkel et al 2010) Several other African lionpopulations that experienced various degrees of recovery

subsequently became relatively stable or exhibited decliningtrends (Bauer et al 2015)Our estimates of quasi‐extinction probabilities were lower than

those reported by Hostetler et al (2013) using a 2‐sex age‐structured matrix population model Lower probabilities ofquasi‐extinction than those reported by the earlier study forcomparable scenarios were likely a consequence of the fact thatthe estimates of abundance (McBride et al 2008) and thus thedensity‐dependent effects on survival of kittens and old panthers(gt10 yr) have changed in recent years Additionally these earlyestimates of the probability of quasi‐extinction and of time toquasi‐extinction did not consider genetic erosion that will in-evitably occur without management interventionUnder the MVM scenario the equilibrium population size was

twice that attained under the MPC scenario The MVM popula-tion estimates are approximately twice the MPCs (Fig 2) as aresult the simulated population under the MVM scenario achieveda higher equilibrium size Probability of quasi‐extinction under theMVM scenario was generally greater than that under the MPCscenario This result is a consequence of the fact that the estimateddensity dependence on kitten survival based on the MPC scenario isstronger than that for theMVM scenario (regression coefficients forabundance for the MPC scenario= minus0068 vs minus0002 for theMVM scenario Appendix B Table B1) Consequently simulatedpopulations under the MPC scenario have a stronger tendency tomove toward the equilibrium population size which inherentlyreduces the quasi‐extinction probability (eg Royama 1992)As expected for long‐lived mammals (Heppell et al 2000 Oli

and Dobson 2003) the population growth rate was proportio-nately most sensitive to changes in survival of prime adultsfollowed by that in younger age classes Global sensitivity ana-lysis in contrast showed that under all scenarios extinctionparameters were particularly sensitive to changes in survival ofFlorida panther kittens This result is a consequence of the highsensitivity of the population growth rate to kitten survival(Fig 9) and a larger standard error of kitten survival (Table 3)Results of our sensitivity analyses indicate that future researchshould focus on acquiring additional information on early lifestages to improve the accuracy and precision of estimates ofkitten survival Our results suggest that demographic variables(or their environmental drivers) that strongly influence extinc-tion risks are not necessarily those with the largest elasticityvalues A study of island fox (Bakker et al 2009) found thatextinction risk was strongly influenced by survival modifierparameters that had low elasticity values

Genetic Management When and How to InterveneThe use of genetic introgression as a management tool has al-ways been controversial and the probable effectiveness it mighthave in preventing the imminent demise of the panther popu-lation was initially debated (Creel 2006 Maehr et al 2006Pimm et al 2006) These debates notwithstanding the pantherpopulation had suffered from genetic morphological and bio-medical correlates of inbreeding before this management in-tervention (Johnson et al 2010 Onorato et al 2010) whereasthe post‐introgression period has been characterized by a sub-stantial reduction in the biomedical correlates of inbreeding andimprovements in demographic vigor and abundance (McBride

28 Wildlife Monographs bull 203

et al 2008 Hostetler et al 2010 2013 Johnson et al 2010Benson et al 2011 McClintock et al 2015) Nevertheless thepopulation remains small and isolated habitat is still being lostand there is no current possibility of natural gene flow betweenthis and other North American puma populations thus therecurrence of inbreeding‐related problems and subsequent po-pulation declines are inevitable The question therefore is notwhether genetic management of the Florida panther populationis needed but when and how it should be implementedWithout genetic introgression probabilities of quasi‐extinc-

tion were substantially greater than those from models thatincorporated periodic genetic introgression events (Fig 15)highlighting the importance of genetic management in smallisolated populations When 5 female pumas are added to theFlorida panther population every 10 years genetic variationincreased substantially under the MPC and MVM scenariosThe IBM predicted that the equillibrium population size wouldbe substantially larger when 5 pumas were released into thepopulation every 10 years than populations without intervention(ie no introgression) Releasing 15 Texas females did not af-ford substantially greater benefit to the equilibrium populationsize than did releasing only 5 Texas females Instead addingmore individuals caused increasingly large fluctuations in po-pulation size due to the numerical increases and density de-pendence (Fig 14) possibly diminishing the benefits associatedwith increased genetic variationCompared with the no‐intervention scenario (ie no genetic

introgression implemented) the introduction of 5 female pumasevery 10 years reduced quasi‐extinction probabilities from 13to 4 and from 17 to 6 for the MPC and MVM scenariosrespectively The benefit of genetic‐introgression strategies de-creased as the interval between successive management inter-ventions increased and no benefit was evident once the intervalreached 80 years (Fig 15) Introgression reduced the averagetime until quasi‐extinction (Fig 18) This somewhat counter‐intuitive result can be explained by the fact that repeated geneticintrogression makes the population more robust over timewhich will reduce the probability of extinction Consequentlyfew simulated population trajectories that fall below the criticalthreshold do so early reducing the mean time to extinctionAlso density dependence has a stabilizing effect on populations(Royama 1992) populations tend toward equilibrium popula-tion sizes which reduces the probability over time of popula-tions falling below quasi‐extinction threshold However timeuntil quasi‐extinction must be interpreted in conjunction withthe probability of quasi‐extinction which is very low for allscenarios with genetic introgression (Fig 15)Cost is a significant impediment to the implementation of a

genetic introgression program because captures relocations andmonitoring efforts before and after introgression are expensive(Edmands 2007 Frankham 2010b 2015 2016) Costs may beacceptable to society if the management actions result in sub-stantial fitness benefits especially if the species of concern ispopular and charismatic (Frankham 2015) We estimated the100‐year cost of introgression strategies modeled here to rangefrom $50000 to $12 million More expensive strategies gen-erally led to larger decreases in the probability of quasi‐extinc-tion but cheaper strategies also substantially improved

population viability (ie reduced quasi‐extinction probabilityTable 6) One of the cheaper strategies (5 pumas released every20 years) which cost an estimated $200000 over 100 yearsreduced the probability of quasi‐extinction by 59 and 44 forthe MPC and MVM scenarios respectively But these pre-liminary cost estimates are based on expenses incurred duringthe 1995 introgression and include costs of capture quarantinetransportation release and post‐release monitoring of femalesonly Although the cost for future genetic interventions wouldlikely be higher than those reported here the relative differencesin cost between alternative intervention strategies should remainapproximately the sameOur ability to simulate genetics and link it to the viability of

the Florida panther population is based on the empirically es-timated relationship between individual heterozygosity andsurvival Such heterozygosity and fitness correlations have beenstudied for decades (David 1998) but have only recently beenused to predict population viability (Benson et al 2016) noneto our knowledge have incorporated cost into their modelingefforts The mechanisms underlying heterozygosity and fitnesscorrelations remain unclear (David 1998 Szulkin et al 2010)and their significance as a proxy for inbreeding depression hasbeen debated (Balloux et al 2004 Miller and Coltman 2014)Nevertheless evidence continues to mount demonstratingpositive relationships between heterozygosity and survival(Coulson et al 1998ab Hansson et al 2004 Silva et al 2005Acevedo‐Whitehouse et al 2006 Jensen et al 2007 Velandoet al 2015) and between heterozygosity and fecundity (Seddonet al 2004 Charpentier et al 2005 Agudo et al 2012 Velandoet al 2015) Although the reported correlations are generallyweak their consequences can be substantial if population growthand persistence parameters are highly sensitive to the affectedvital rates (Velando et al 2015) Compared with scenarios thatignore genetics incorporating the effects of inbreeding on sur-vival increased quasi‐extinction probabilities within a 100‐yeartimeframe from 15 to 13 and from 14 to 17 for theMPC and MVM scenarios respectively These results high-light the importance of incorporating genetics when analyzingthe viability of small isolated populationsAlthough conservation biologists have recognized the im-

portance of genetics in population viability many still fail toincorporate genetic factors into PVA models (Reed et al 2002)Explicit consideration of genetics in PVAs requires long‐termgenetic and demographic data This permits the assessment ofthe functional relationship between genetic variability and de-mographic parameters Population viability analysis softwarepackages such as VORTEX (Lacy 1993 2000) offer a powerfulframework for individual‐based simulations but they often donot adequately capture a speciesrsquo life history or genetics Basedon a thorough review of the importance of genetic factors inwildlife conservation Frankham et al (2014) concluded thatgenetic factors can strongly influence the dynamics and persis-tence of small andor fragmented populations They furthernoted that ldquoMost PVA‐based risk assessments ignore or in-adequately model genetic factorsrdquo and recommended that ldquoPVAshould routinely include realistic inbreeding depressionhelliprdquo(Frankham et al 201457) Our custom‐built IBM permitted usto develop a model that adequately captured the Florida panther

van de Kerk et al bull Florida Panther Population and Genetic Viability 29

life history and we could parameterize the model with our long‐term genetic and demographic dataThe explicit consideration of the effects of inbreeding on

Florida panther population dynamics and persistence representsan important step forward Nonetheless our model remainsimperfect First we neglected the possible effects of habitat lossdetrimental anthropogenic activities climate change the prob-ability of immigration of pumas from western populationsdisease or catastrophes on demographic rates and populationviability Although immigration from puma populations outsideFlorida is possible its probability is very low given the extensivechallenges the anthropogenically altered landscape would posefor such a migrant to make it successfully to South Florida Weomitted mutations from our model because we have no in-formation on mutation rates though we expect that they are lowbased on values reported for other mammals (Kumar and Sub-ramanian 2002) Finally we only considered possible effects ofinbreeding on age‐specific survival rates because there was noevidence that heterozygosity affected female reproduction Lossof heterozygosity can negatively affect reproduction becauseinbred male panthers can suffer from cryptorchidism which canadversely affect their fecundity However given the polygynousmating system we expect the Florida panther population growthis primarily female‐limitedAlthough it is generally accepted that genetic introgression

only temporarily relieves a population from inbreeding depres-sion we know of no cases in which it has been implementedmore than once to conserve a threatened wildlife populationOur study provides an example of how demographic and mo-lecular data can be used within an IBM framework to evaluatethe efficacy of alternative genetic management strategies via theFlorida panther as a case study Our model can be adjusted forand applied to other species and offers great potential as a toolfor genetic management of small isolated wildlife populations

MANAGEMENT IMPLICATIONS

Our results and those of Hostetler et al (2013) and Johnsonet al (2010) provide persuasive evidence that the genetic in-tervention implemented in 1995 prevented the demise of theFlorida panther and restored demographic vigor to the popu-lation The Florida panther population remains small and iso-lated from other puma populations the closest puma populationis located gt1000 km away in Texas and the intervening land-scape is highly developed and fragmented with many interstate(and other high‐traffic) highways and major urban centersTheory suggests that 1 migrant per generation is sufficient tominimize the loss of genetic diversity (eg Mills and Allendorf1996) However the possibility of successful migration of apuma to South Florida followed by successful mating every ~5years is very low considering the distance between Texas andFlorida (Hedrick 1995) and the many risks and barriers thatdispersing pumas face (Beier 1995 Maehr et al 2002 Thatcheret al 2006) Consequently the pantherrsquos long‐term persistencewill likely depend on subsequent timely genetic introgressionsgiven the near‐impossibility of natural gene flow from otherpuma populations To that end our study offers considerableinsights into benefits of different genetic management strategiesand associated costs

Without genetic management the Florida panther populationfaces a substantial risk of quasi‐extinction within 100 years (10ndash46 depending on quasi‐extinction threshold and scenarios)Twenty‐three years have passed since genetic introgression wasimplemented and the population appears healthy as indicatedby measures of genetic variation increases in the populationsize and improvements in age‐specific survival rates Ourfindings suggest that releasing more pumas more frequently doesnot necessarily improve persistence probability because such astrategy would cause larger fluctuations in population size(Fig 14) which negatively affects persistence probability Thusextinction risks faced by inbred populations of large carnivorescan be substantially reduced by releasing approximately 5 im-migrants every 2 decades We suggest that such a managementstrategy be followed only in conjunction with continued long‐term monitoring of the population Our analyses demonstratethat population growth and persistence are highly sensitive tokitten and female (adult and subadult) survival Continuing tocollect data on these demographic variables along with bio-metric and genetic sampling will remain important to pantherrecovery and vigilance in identifying issues associated with in-breeding depression The collection of these monitoring datashould allow managers to detect any demographic or geneticchanges in a timely fashion and respond with genetic in-trogression or other management actions if warrantedRecovery objectives as defined by the USFWS in its current

recovery plan (USFWS 2008) delineate the need for additionalpopulations in the historical range to downlist or delist theFlorida panther If the only extant population of Floridapanthers continued to grow it could serve as a source ofpanthers to be translocated to suitable habitat to seed addi-tional populations Maintaining current information on thegenetic health of the population and precise estimates of itssize will be important in assessing whether it can withstand theremoval of a subset of animals for translocation Although the2017 documentation of a female panther with kittens north ofthe current breeding range is encouraging in terms of thepotential for population expansionmdashgiven that no female hadbeen recorded north of the Caloosahatchee River since 1972mdashthe re‐establishment of separate new populations will mostlikely require translocations by wildlife managers and a lengthysociopolitical processConservation of threatened or endangered species such as the

Florida panther is inherently challenging For panthers and otherlarge carnivores that require vast extents of natural habitat topersist habitat is typically the most limiting factor that will de-termine whether recovery efforts succeed Although geneticmanagement invariably plays a key role in the long‐term persis-tence of the panther population even a healthy population caneventually succumb to the impacts of habitat loss The humanpopulation of Florida continues to be one of the fastest‐growingin the nation the United States Census Bureau currently ranksFlorida as the third most populous Therefore habitat loss isgoing to continue to be a key issue for panthers Diligence towardpreserving panther habitat in its historical range via conservationeasements or other means and trying to keep such areas inter-connected via corridors is a goal stakeholders such as governmentagencies sportsmen non‐governmental organizations ranchers

30 Wildlife Monographs bull 203

and other private landowners must work on collaboratively toachieve

ACKNOWLEDGMENTS

We thank D Shindle D Jennings T OrsquoMeara D Land andC Belden for valuable advice throughout the project We thankS Bass M Criffield M Cunningham D Giardina D JansenA Johnson D Land M Lotz C McBride Rocky McBrideRowdy McBride Roy McBride L Oberhofer S Schulze staffsat Everglades National Park and Big Cypress National Preserveand others for assistance with fieldwork We also thank JAustin M Conroy J Hines J Laake S Mills J Nichols JHines M Sunquist J Fryxell and T Coulson for advice andassistance regarding data analysis and panther biology TheMuseum of Texas Tech University Natural Science ResearchLaboratory provided samples from pumas from west Texas forcomparative genetic analyses M Ben‐David D Land WMcCown E Hellgren and 4 anonymous reviewers providedmany helpful comments on the manuscript We thank NereaUbierna Lopez for completing the Spanish translation of theabstract We are grateful to the Department of ResearchComputing University of Florida for excellent high‐perfor-mance computing services We would like to thank US Fishand Wildlife Service Florida Fish and Wildlife ConservationCommission (FWC) US National Park Service QuantitativeSpatial Ecology Evolution and Environment (QSE3) In-tegrative Graduate Education and Research Traineeship(IGERT) program funded by the National Science Foundationand the University of Florida for financially supporting thisstudy Funding for panther research conducted by the FWC issupported predominantly via donations accrued with vehicleregistrations for ldquoProtect the Pantherrdquo license plates

LITERATURE CITEDAcevedo‐Whitehouse K T R Spraker E Lyons S R Melin F Gulland RL Delong and W Amos 2006 Contrasting effects of heterozygosity onsurvival and hookworm resistance in California sea lion pups MolecularEcology 151973ndash1982

Adams J R L M Vucetich P W Hedrick R O Peterson and J AVucetich 2011 Genomic sweep and potential genetic rescue during limitingenvironmental conditions in an isolated wolf population Proceedings of theRoyal Society B Biological Sciences 2783336ndash3344

Agresti A 2013 Categorical data analysis Third edition John Wiley amp SonsHoboken New Jersey USA

Agudo R M Carrete M Alcaide C Rico F Hiraldo and J A Donaacutezar2012 Genetic diversity at neutral and adaptive loci determines individualfitness in a long‐lived territorial bird Proceedings of the Royal Society BBiological Sciences 2793241ndash3249

Albeke S E N P Nibbelink and M Ben‐David 2015 Modeling behavior bycoastal river otter (Lontra canadensis) in response to prey availability in PrinceWilliam Sound Alaska a spatially‐explicit individual‐based approach PLOSOne 10e0126208

Alho J S K Vaumllimaumlki and J Merilauml 2010 Rhh an R extension for estimatingmultilocus heterozygosity and heterozygosity‐heterozygosity correlationMolecular Ecology Resources 10720ndash722

Alvarez K 1993 Twilight of the panther Myakka River Publishing SarasotaFlorida USA

Aparicio J M J Ortego and P J Cordero 2006 What should we weigh toestimate heterozygosity alleles or loci Molecular Ecology 154659ndash4665

Bakker V J D F Doak G W Roemer D K Garcelon T J Coonan S AMorrison C Lynch K Ralls and R Shaw 2009 Incorporating ecologicaldrivers and uncertainty into a demographic population viability analysis for theisland fox Ecological Monographs 7977ndash108

Balloux F W Amos and T Coulson 2004 Does heterozygosity estimateinbreeding in real populations Molecular Ecology 133021ndash3031

Barone M A M E Roelke J Howard J L Brown A E Anderson and DE Wildt 1994 Reproductive characteristics of male Florida panthersmdashcomparative studies from Florida Texas Colorado Latin‐America andNorth‐American Zoos Journal of Mammalogy 75150ndash162

Bateson Z W P O Dunn S D Hull A E Henschen J A Johnson and LA Whittingham 2014 Genetic restoration of a threatened population ofgreater prairie‐chickens Biological Conservation 17412ndash19

Bauer H G Chapron K Nowell P Henschel P Funston L T B HunterD W Macdonald and C Packer 2015 Lion (Panthera leo) populations aredeclining rapidly across Africa except in intensively managed areas Pro-ceedings of National Academy of Sciences of the United States of America11214894ndash14899

Beier P 1995 Dispersal of juvenile cougars in fragmented habitat Journal ofWildlife Management 59228ndash237

Benson J F J A Hostetler D P Onorato W E Johnson M E Roelke SJ OrsquoBrien D Jansen and M K Oli 2011 Intentional genetic introgressioninfluences survival of adults and subadults in a small inbred felid populationJournal of Animal Ecology 80958ndash967

Benson J F P J Mahoney J A Sikich L E K Serieys J P Pollinger H BErnest and S P D Riley 2016 Interactions between demography geneticsand landscape connectivity increase extinction probability for a small popu-lation of large carnivores in a major metropolitan area Proceedings of theRoyal Society B Biological Sciences 28320160957

Beverly F 1874 The Florida panther Forest and Stream 3290Boyce M S C V Haridas C T Lee and the NCEAS Stochastic Demo-graphy Working Group 2006 Demography in an increasingly variable worldTrends in Ecology amp Evolution 21141ndash148

Brook B W J J OrsquoGrady A P Chapman M A Burgman H R Akcakayaand R Frankham 2000 Predictive accuracy of population viability analysis inconservation biology Nature 404385ndash387

Brys R and H Jacquemyn 2016 Severe outbreeding and inbreeding de-pression maintain mating system differentiation in Epipactis (Orchidaceae)Journal of Evolutionary Biology 29352ndash359

Burke J M and M L Arnold 2001 Genetics and the fitness of hybridsAnnual Review of Genetics 3531ndash52

Burnham K P 1993 A theory for combined analysis of ring recovery andrecapture data Pages 199ndash213 in J‐D Lebreton and P M North editorsMarked individuals in the study of bird population Springer‐Verlag NewYork New York USA

Burnham K P and D R Anderson 2002 Model selection and multimodelinference a practical information‐theoretic approach Second edition SpringerVerlag New York New York USA

Caswell H 2001 Matrix population models construction analysis and in-terpretation Sinauer Associates Sunderland Massachusetts USA

Charpentier M J M Setchell F Prugnolle L A Knapp E J Wickings PPeignot and M Hossaert‐McKey 2005 Genetic diversity and reproductivesuccess in mandrills (Mandrillus sphinx) Proceedings of the NationalAcademy of Sciences of the United States of America 10216723ndash16728

Cooch E and G C White editors 2018 Program MARK a gentle in-troduction lthttpwwwphidotorgsoftwaremarkdocsbookgt Accessed 7Apr 2016

Cooley H S R B Wielgus G M Koehler H S Robinson and B TMaletzke 2009 Does hunting regulate cougar populations A test of thecompensatory mortality hypothesis Ecology 902913ndash2921

Coulson T E A Catchpole S D Albon B J Morgan J M Pemberton TH Clutton‐Brock M J Crawley and B T Grenfell 2001 Age sex densitywinter weather and population crashes in Soay sheep Science 2921528ndash1531

Coulson T J Pemberton S Albon M Beaumont T Marshall J Slate FGuinness and T Clutton‐Brock 1998a Microsatellites reveal heterosis in reddeer Proceedings of the Royal Society B Biological Sciences 265489ndash495

Coulson T N S D Albon J M Pemberton J Slate F E Guinness and TH Clutton‐Brock 1998b Genotype by environment interactions in wintersurvival in red deer Journal of Animal Ecology 67434ndash445

Cox D 1972 Regression models and life‐tables Journal of the Royal StatisticalSociety Series B Statistical Methodology 34187ndash220

Creel S 2006 Recovery of the Florida panthermdashgenetic rescue demographic rescueor both Response to Pimm et al (2006) Animal Conservation 9125ndash126

Crnokrak P and D A Roff 1999 Inbreeding depression in the wild Heredity83260ndash270

Crow J 1948 Alternative hypotheses of hybrid vigor Genetics 33477ndash487

van de Kerk et al bull Florida Panther Population and Genetic Viability 31

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 16: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

followed by those for non‐F1 admixed (062plusmn 001) and canonical(039plusmn 001) panthers Individual heterozygosity positively affectedsurvival rates of kittens (slope parameter η= 1455 95CI= 0505ndash2405) Individual heterozygosity also positively af-fected survival of subadult and adult panthers but the evidence forthis effect was weaker because the confidence interval for the ha-zard ratio overlapped 10 (hazard ratio exp(ɩ)= 0311 95CI= 0092ndash1054 Table 4 and Fig 7)The MPC increased after genetic introgression (Fig 2) This

abundance index was also included in top‐ranking modelsindicating that population density affected survival (Table 4)Abundance reduced the survival of kittens (slope parameterη= minus0035 95 CI= minus0049 to minus0021) evidence was weak forthe effect of abundance on survival of subadult and adult panthers(hazard ratio exp(ɩ)= 1002 95 CI= 0995ndash1010 Fig 7) Re-gression coefficients associated with the effect of abundance andindividual heterozygosity on demographic parameters are presentedin Table B1 (available online in Supporting Information)The most frequently observed causes of death for radio‐

collared panthers were vehicle collision and intraspecific ag-gression Cause‐specific mortality analyses revealed thatmortality rates among possible causes were generally similar forthe 2 sexes but that males were more likely to die from in-traspecific aggression than were females (Fig 8A) Male mor-tality from intraspecific aggression was higher for subadult andold adult panthers than for prime adults (Fig 8B) For bothmales and females canonical panthers were more likely to diefrom intraspecific aggression than were admixed panthers(Fig 8C)

Reproductive ParametersOf 101 radio‐collared females 67 reproduced at least once duringour monitoring we used these individuals to assess the annualprobability of reproduction The top‐ranked model for the annual

probability of reproduction included age effect only suggesting thatadding ancestry or abundance did not improve model parsimony(Table 4) Model‐averaged annual probabilities of reproductiondiffered between female subadults (035plusmn 008) prime adults(050plusmn 005) and older adults (025plusmn 006)The most parsimonious model for the number of kittens

produced annually by females that produced at least 1 litter (ieconditional on reproduction) included effects of age ancestryand abundance a model that included only the effect of age wasalmost equally supported (Table 4) The model‐averagednumber of kittens produced annually conditional on re-production was 280plusmn 075 for subadults 267plusmn 043 for primeadults and 228plusmn 083 for old adults Confidence intervals forthe slope parameter (β) overlapped 0 for both abundance(β= 0010 95 CI= minus0001 to 0021) and ancestry (β= minus073795 CI= minus1637 to 0162)

Population Dynamics and PersistenceDeterministic and stochastic demographymdashThe deterministic (λ)

and stochastic (λs) annual population growth rate calculated usingthe matrix population model were 106 (5th and 95th percentiles099ndash114) and 104 (072ndash141) respectively The generation timewas 47 years A female starting in age class one (kitten) wasexpected to live another 58 years and produce 10 kittens onaverage during her lifetime Overall λs was proportionately(elasticity) most sensitive to changes in female prime adultsurvival on the absolute scale (sensitivity) however it was mostsensitive to changes in kitten survival (Fig 9) Populationtrajectories probabilities of quasi‐extinction and times untilquasi‐extinction estimated using the matrix population modeland IBM for comparable scenarios were nearly identical for acritical quasi‐extinction threshold of 10 panthers (Figs 10 and 11)and for a critical quasi‐extinction threshold of 30 panthers(Appendix E available online in Supporting Information)Minimum population count scenariomdashUnder the MPC scenario

(ie density dependence estimated using the minimumpopulation count data) with a critical threshold of 10panthers the population size reached 99 (72ndash104) and 100(77ndash105) at 200 years (Fig 10) as predicted by the IBM and thematrix model respectively The cumulative quasi‐extinctionprobabilities within 100 years based on the MPC scenario were15 (IBM 0ndash48) and 30 (matrix model 0ndash76) Theseprobabilities increased to 25 (IBM 0ndash114) and 38 (matrixmodel 0ndash154) by year 200 (Fig 10) The mean times untilquasi‐extinction were 15 years (IBM 0ndash67) and 15 years (matrixmodel 0ndash72) conditioned on going quasi‐extinct within 100years Expected times until quasi‐extinction conditioned onquasi‐extinction within 200 years were higher 34 years (IBM0ndash137) and 32 years (matrix model 0ndash134 Fig 10)Motor vehicle mortality scenariomdashUnder the MVM scenario

(ie density dependence estimated using estimates of pantherpopulation size from McClintock et al 2015) with a criticalthreshold of 10 panthers the projected population reached 188(142ndash217) and 190 (143ndash219) at 200 years as predicted by theIBM and the matrix model respectively (Fig 11) Thecumulative quasi‐extinction probabilities within 100 years were14 (IBM 0ndash08) and 13 (matrix model 0ndash06) Theseprobabilities increased to 20 (IBM 0ndash17) and 19 (matrix

000

025

050

075

100

1995 2000 2005 2010Year of birth

Indi

vidu

al h

eter

ozyg

osity

Figure 6 Temporal changes in the individual heterozygosity of Floridapanthers born during the period 1995ndash2013 in South Florida USA Pointsare measurements solid line is linear regression shaded area is 95 confidenceinterval of slope Slope= 0006plusmn 0001 R2= 009

18 Wildlife Monographs bull 203

model 0ndash16) within year 200 (Fig 11) The mean times until quasi‐extinction based on the MVM scenario were 5 years (IBM 0ndash61)and 6 years (matrix model 0ndash64) conditioned on going quasi‐extinctwithin 100 years Expected times until quasi‐extinction conditionedon quasi‐extinction within 200 years were 11 years (IBM 0ndash113)and 11 years (matrix model 0ndash112 Fig 11)

Sensitivity of extinction probability to demographic parametersmdashThe partial rank correlation coefficients between quasi‐extinctionprobabilities and demographic parameters were negative anincrease in any of the demographic parameters decreased thequasi‐extinction probability Probabilities of quasi‐extinction within200 years were most sensitive to changes in kitten survival for bothscenarios (Fig 12) followed by survival of subadult females in theMVM scenarios and survival of prime adult females in the MPCscenario The probability of reproduction of prime adult femaleshad the third‐largest partial rank correlation coefficient with quasi‐extinction probability for both scenarios

Benefits and Costs of Genetic ManagementMinimum population count scenariomdashThe MPC scenario with

a critical threshold of 10 panthers predicted that withoutmanagement intervention average individual heterozygositywould decrease reaching 057 (049ndash064) at 100 years and053 (042ndash062) at 200 years (Fig 13) The population woulddecrease slightly reaching an average of 79 (41ndash99) at 100 yearsand 76 (43ndash98) at 200 years (Fig 14) The probabilities of quasi‐extinction under the no‐intervention MPC scenario when weconsidered the effects of genetic erosion were 13 (0ndash99) at 100years and 23 (0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with

the translocation of 5 Texas females average individualheterozygosity under the MPC scenario increased and thenstabilized at 068 (064ndash072) at 100 years and remained atthat level at 200 years (Fig 13) The population reached anaverage of 88 (62ndash104) at 100 years and stayed the same at200 years (Fig 14) The probabilities of quasi‐extinctionunder this introgression strategy were 6 (0ndash53) within 100years and 7 (0ndash82) within 200 years (Fig 15) Results ofanalyses using a critical threshold of 30 panthers revealed asimilar pattern of relative benefits However the prob-abilities of quasi‐extinction were substantially higher (ge45)across all scenarios (Appendix E)

Motor vehicle mortality scenariomdashWithout managementintervention the average individual heterozygosity predictedby the MVM scenario and a critical threshold of 10 panthersdecreased to 060 (046ndash069) at 100 years and to 059 (039ndash070) at 200 years (Fig 13) The population increased slightlyand reached an equilibrium at 154 individuals (46ndash217) at 100years and 157 (57ndash218) at 200 years (Fig 14) The probabilitiesof quasi‐extinction were 17 (0ndash100) at 100 years and 22(0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with 5

female pumas from Texas average individual heterozygosityincreased slightly reaching 067 (060ndash073) at 100 years and068 (059ndash075) at 200 years (Fig 13) Under this strategy thepopulation reached an average of 164 individuals (44ndash226) at

Table 4 Model comparison results testing for the effects of several covariateson survival of all kittens survival of genetically sampled kittens survival ofsubadult and adult Florida panthers probability of reproduction and kittensproduced annually for Florida panthers sampled from 1981ndash2013 in SouthFlorida USA

Model Parameters ΔAICa Weight

Kitten survival (all data)Base+ F1b+ abundancec 10 000 0988Base+ abundance 9 1018 0006Base+ F1 9 1035 0005Base 8 2711 lt0001Base+ sexd 9 2912 lt0001Base+ litter sizee 9 2914 lt0001

Kitten survival (genetically sampled kittens only)Base+ F1+ abundance 10 000 0541Base+ F1CanAdmf+ abundance 11 099 0329Base+Hetg+ abundance 10 310 0115Base+CanAdmh+ abundance 10 791 0010Base+ abundance 9 944 0005Base+ F1 9 1638 lt0001Base+ F1CanAdm 10 1837 lt0001Base+Het 9 2872 lt0001Base 8 3296 lt0001Base+CanAdm 9 3348 lt0001

Subadult and adult survivalBase+ F1CanAdm+ abundance 7 000 0360Base+ F1 5 053 0276Base+ F1CanAdm 6 105 0213Base+ F1+ abundance 6 222 0119Base+CanAdm+ abundance 6 575 0020Base+CanAdm 5 908 0004Base+Het 5 964 0003Base+Het+ abundance 6 984 0003Base 4 1118 0001Base+ abundance 5 1278 0001

Probability of reproductionAge 3 000 0242Age+CanAdm 4 012 0228Age+ abundance 4 173 0102Age+ F1 4 190 0094Age+CanAdm+ abundance 5 216 0082Age+Het 4 219 0081Age+ F1CanAdm 5 232 0076Age+ F1+ abundance 5 372 0038Age+Het+ abundance 5 399 0033Age+ F1CanAdm+ abundance 6 444 0026

Kittens produced annuallyAge+ F1+ abundance 5 000 0194Age 3 060 0144Age+ abundance 4 061 0143Age+ F1 4 097 0120Age+CanAdm 4 139 0097Age+ F1CanAdm+ abundance 6 196 0073Age+ F1CanAdm 5 231 0061Age+CanAdm+ abundance 5 238 0059Age+Het+ abundance 5 250 0056Age+Het 4 259 0053

a Akaikersquos information criterion (AIC) for kitten survival models was adjustedfor small sample size and overdispersion (QAICc)

b F1 divides Florida panthers into 2 groups F1 admixed and all other panthersc Index of Florida panther abundanced Sex of an individual Florida panther (male or female)e Size of the litter in which a Florida panther kitten was bornf F1CanAdm divides panthers into 3 groups F1 admixed canonical andother admixed panthers

g Het is the individual heterozygosityh CanAdm divides panthers into 2 groups canonical and admixed (includingF1 admixed) panthers

van de Kerk et al bull Florida Panther Population and Genetic Viability 19

100 years and 170 (67ndash231) at 200 years based on the MVMscenario (Fig 14) The probabilities of quasi‐extinction underthis introgression strategy were 10 (0ndash99) at 100 years and12 (0ndash99) at 200 years (Fig 15)The projected population size and average individual hetero-

zygosity reached equilibrium levels with periodic fluctuations inthese values when introgression was implemented (Figs 16ndash17)Genetic introgression decreased the time until quasi‐extinctionacross all scenarios (Fig 18) Results of analyses using a criticalthreshold of 30 panthers revealed a similar pattern of relative ben-efits However the probabilities of quasi‐extinction were sub-stantially higher (ge45) across all scenarios (Appendix E)

Addition of pumas from Texas increased both the populationsize and average individual heterozygosity consequentlyintrogression caused periodic increases in average individualheterozygosity and population size at the interval at which theintrogression occurred Whereas average individual hetero-zygosity gradually decreased following introgression densitydependence caused the population size to fluctuate especiallywhen a larger number of pumas from Texas were released Thepositive effect of genetic introgression initiatives on quasi‐extinction probabilities decreased as the time interval betweenthese events increased More frequent genetic introgressions ledto greater increases in average individual heterozygosity and

Old adult

Prime adult

Subadult

Kitten

025 050 075

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Individual heterozygosity

Ann

ual s

urvi

val r

ate

Old adult

Prime adult

Subadult

Kitten

40 60 80 100 120

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Abundance index

Ann

ual s

urvi

val r

ate

Kittens Males Females

Figure 7 The effect of individual heterozygosity (left) and the abundance index (right) on Florida panther age‐ and sex‐specific survival estimates (shaded area isSE) in South Florida USA 1981ndash2013 Abundance index data are from McBride et al (2008) and R T McBride and C McBride (Rancherrsquos Supply Incorporatedunpublished report)

20 Wildlife Monographs bull 203

equilibrium population size while reducing the probability ofquasi‐extinction Comparatively performing genetic introgres-sion less often but with more individuals per release was lesseffective at improving the long‐term prospects for the pantherpopulation (Figs 13ndash15)

Financial cost and persistence benefitsmdashThe total estimated costof 1 genetic introgression was $40000 $80000 or $120000for releasing 5 10 or 15 pumas from Texas respectively(Table 5) The total cost incurred over 100 years for eachstrategy ranged from $50000 to $12 million (Table 6)The most expensive strategy involved the introduction of15 pumas every 10 years this strategy reduced the probability

of quasi‐extinction by 58 and 40 under the MPC and MVMscenarios respectively (Table 6) Introducing 5 pumas every80 years cost the least but improvements in populationpersistence under this strategy were insubstantial (Table 6)Releasing more panthers more frequently caused increasedpopulation fluctuations but did not necessarily reduce the quasi‐extinction probability (Figs 14 and 15 Table 6)

DISCUSSION

The duration (34 yr of data) and sample sizes associated with theFlorida Panther Project allowed us to obtain a comprehensiveunderstanding of genetic variation demographic parameters

A

B

C

Figure 8 Cause‐specific mortality rates (plusmnSE) for male and female Florida panthers (A) subadult prime‐adult and old‐adult Florida panthers of both sexes (B)and canonical and admixed Florida panthers of both sexes (C) in South Florida USA 1981ndash2013 Causes of mortality are vehicle collision intraspecific aggression(ISA) other causes (known causes such as diseases injuries and infections unrelated to the first 2 causes) and unknown cause (mortalities for which we could notassign a cause)

van de Kerk et al bull Florida Panther Population and Genetic Viability 21

probability of population persistence and the duration of thepositive impacts of genetic restoration on this endangered po-pulation The Florida panther population was in dire straits inthe early 1990s and our results clearly demonstrated that the

implementation of the introgression project in 1995 subse-quently accrued a series of genetic and demographic improve-ments that have led to the change from a declining population toone that is growing and expanding its current breeding range

Sensitivity Elasticity

0 1 2 3 00 02 04

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

Para

met

er

Figure 9 Sensitivity and elasticity (proportional sensitivity) of stochastic population growth rate (λs) of the Florida panther to vital demographic parameters in SouthFlorida USA 1981ndash2013 Horizontal lines represent 5th and 95th percentiles

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

70

80

90

100

110

0

5

10

15

0

50

100

Years

StatisticMeanMedian

Figure 10 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the minimum population count(MPC) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

125

150

175

200

00

05

10

15

20

0

30

60

90

Years

StatisticMeanMedian

Figure 11 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the motor vehicle mortality(MVM) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

22 Wildlife Monographs bull 203

MPC MVM

minus08 minus06 minus04 minus02 00 minus08 minus06 minus04 minus02 00

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Male subadult survival

Male prime adult survival

Male old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

PRCC

Para

met

er

Figure 12 Partial rank correlation coefficient (PRCC) between quasi‐extinction probabilities and the demographic parameters of the Florida panther for theminimum population count (MPC) and motor vehicle mortality (MVM) scenarios Error bars indicate standard deviation Based on data collected in South FloridaUSA 1981ndash2013

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 13 Average projected individual heterozygosities in the Florida panther population over 200 years without genetic management intervention and with eachof the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include theintroduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years Average individual heterozygosity peaks when pumas are added to the Floridapanther population Based on data collected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 23

Our research has further validated the success of genetic in-trogression by quantifying the reduction in the probability ofextinction of the panther population because of this manage-ment initiative These findings all bode well for continuedprogress toward recovery That said the Florida panther po-pulation remains small and isolated from other populations ofpuma from western North America thereby denoting that theeventual impact of genetic drift and inbreeding may negativelyaffect the population in the future Realizing this will continueto be an issue that managers will have to contemplate wemodeled the impact of varied genetic management scenarios todetermine the frequency of introgression along with the numberof panthers that should be released during each initiative tominimize cost and maximize benefits to the population Theseresults will prove invaluable to managers attempting to conservethe Florida panther

Genetic Variation Demographic Parametersand Persistence of Heterotic Benefits from GeneticIntrogressionOur measure of individual heterozygosity (1 minusHL) was higheron average for the admixed (0615) panthers than for those ofcanonical ancestry (0386) Levels of individual heterozygosityfor admixed panthers were comparable to levels observed in asample of pumas from west Texas (x plusmn SE= 0695plusmn 0019n= 41) which further demonstrates the improvement in levelsof genetic variation in the Florida population since 1995 The

correlation between HL and several demographic parametershas been noted in wild populations including cheetahs (Terrellet al 2016) and several species of birds such as European shags(Phalacrocorax aristotelis male and female survival probabilityVelando et al 2015) and Egyptian vulture (Neophron percnop-terus age at recruitment Agudo et al 2012) If we focus only onpanthers captured and radiocollared from 2012ndash2015 (FP211ndashFP240) all of which had estimated birth years ge2005 (ge10 yrpost‐introgression) those panthers had an average individualheterozygosity level of 0618plusmn 0029 highlighting the elevatedlevels of genetic variation that continue to persist in cohorts ofpanthers 5 generations after the release of female pumas fromTexasThe proportional membership values (q) from our

STRUCTURE analysis allowed us to delineate 2 clusters ofancestry for Florida panthers (Fig 4) During the pre‐in-trogression era the population was dominated by panthers thatretained a high percentage of the canonical ancestry which wasaffected by inbreeding depression The post‐introgression eraancestry values are comprised of many admixed individuals inessence demonstrating the success of the introgression in termsof increasing genetic variation in the population and mi-micking gene flow that historically occurred with panthers andother populations of pumas (Onorato et al 2010) A somewhatanalogous situation although abetted by natural immigrationwas evident in the ancestral variation in a small population ofpuma intersected by a major freeway in California USA In

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

75

80

85

90

95

100

130

150

170

190

Years

Popu

latio

n si

ze

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 14 Average projected population sizes in the Florida panther population over 200 years without intervention and with each of the genetic introgressionstrategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include the introduction of 5 10 or 15pumas from Texas USA every 10 20 40 or 80 years Peaks occur when pumas are added to the Florida panther population Based on data collected in SouthFlorida USA 1981ndash2013

24 Wildlife Monographs bull 203

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

25

50

75

100

0

25

50

75

100

Introgression interval (yrs)

PQ

E (

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 15 Average probability of quasi‐extinction (PQE) of the Florida panther population within 100 years and within 200 years without genetic managementintervention (N) and with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenariosIntrogression strategies include the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Errorbars indicate 5th and 95th percentiles Based on data collected in South Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

50

60

70

80

90

100

110

50

100

150

200

Years

Popu

latio

n si

ze

Figure 16 Average projected Florida panther population trajectories under a geneticintrogression strategy involving the release of 5 female pumas from Texas USA every20 years for the minimum population count (MPC) and motor vehicle mortality(MVM) scenarios Shaded area indicates 5th and 95th percentiles and isrepresentative of confidence intervals for other scenarios Based on data collected inSouth Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Figure 17 Average projected Florida panther population‐level heterozygositiesunder a genetic introgression strategy involving the release of 5 female pumas fromTexas USA every 20 years for the minimum population count (MPC) and motorvehicle mortality (MVM) scenarios Shaded area bounds the 5th and 95th percentilesand is representative of confidence intervals for other scenarios Based on datacollected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 25

this case the migration of just a single male across the freewayto the south resulted in an increase in the number of in-dividuals with admixed ancestry and substantially improvedgenetic diversity of the subpopulation south of the freeway(Riley et al 2014)Our estimate of overall kitten survival (0321plusmn 0056) was

similar to that reported by Hostetler et al (2010) who used13 years of post‐introgression data (0323plusmn 0071) These valuesare lower than kitten survival estimates derived from severalpuma populations in western North America (044ndash091 Loganand Sweanor 2001 Lambert et al 2006 Laundre et al 2007Robinson et al 2008 Ruth et al 2011) When we included onlydata for individuals that had been genetically sampled our es-timate was 15 higher The proportion of admixed kittens thatwere genetically sampled increased as the population expandedwhich could have resulted in higher estimates of kitten survivalbecause admixed kittens survive better than canonical kittensKitten survival was strongly influenced by genetic ancestry withsurvival rates of F1 kittens being almost twice that of canonicalkittens (Fig 5B) Consistent with the findings of Hostetler et al(2010) increases in heterozygosity coincided with increasedkitten survival whereas population density negatively affectedkitten survival Population density often has a strong negativeimpact on survival of young age classes due to infanticide andcannibalism which has been reported from several carnivore

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

50

100

150

0

50

100

150

Introgression interval (yrs)

TQE

(yrs

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 18 Average times until quasi‐extinction (TQE) for the Florida panther population within 100 years and within 200 years without genetic management interventionand with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategiesinclude the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Error bars indicate 5th and 95thpercentiles Based on data collected in South Florida USA 1981ndash2013

Table 5 Average cost (2017 US dollars) of introgression strategies employedover 100 years that involve the introduction of 5 10 or 15 pumas from TexasUSA into the Florida panther population in South Florida USA at an in-creasingly higher frequency Costs of capture (including transportation) quar-antine and post‐introgression monitoring were estimated by Roy McBrideMark Cunningham and Dave Onorato respectively

Frequency Component 5 pumas 10 pumas 15 pumas

Once Capture 25000 50000 75000

Quarantine 5000 10000 15000

Monitoring 10000 20000 30000

Total 40000 80000 120000

Every 80 years Capture 31250 62500 93750

Quarantine 6250 12500 18750

Monitoring 12500 25000 37500

Total 50000 100000 150000

Every 40 years Capture 62500 125000 187500

Quarantine 12500 25000 37500

Monitoring 25000 50000 75000

Total 100000 200000 300000

Every 20 years Capture 125000 250000 375000

Quarantine 25000 50000 75000

Monitoring 50000 100000 150000

Total 200000 400000 600000

Every 10 years Capture 250000 500000 750000

Quarantine 50000 100000 150000

Monitoring 100000 200000 300000

Total 400000 800000 1200000

26 Wildlife Monographs bull 203

species including polar bears (Ursus maritimus Derocher andWiig 1999) black bears (Ursus americanus Garrison et al 2007)and pumas (Logan and Sweanor 2001)Our estimates of sex‐ and age‐specific survival rates of subadult

and adult panthers (Table 3) and the effects of covariates on theserates were similar to those reported by Benson et al (2011) derivedfrom data collected through 2006 Our survival rates for males(065ndash077) were lower than females (078ndash097) for all 3 ageclasses (subadult prime adult and old adult) which is the typicaltrend observed in most puma populations (eg Ruth et al 19982011) However an unhunted puma population occupying afragmented urban landscape in southern California exhibited lowersurvival (0586 females 0525 males Vickers et al 2015) perhapsas a result of severe habitat fragmentation and the lack of extensiveswaths of public land protected in perpetuity Most populations ofpuma in western North America are typically subject to variedlevels of management via regulated hunting which often is biasedtoward the take of males (Cooley et al 2009 Ruth et al 2011)Consequently annual survival estimates from hunted populationsin northeast Washington (0679ndash0733 females 0341 malesRobinson et al 2008) and southeastern Arizona (0685 females0584 males Cunningham et al 2001) were generally lower thanthose we estimated for Florida panthersCause‐specific mortality analysis showed that male panthers

experienced a substantially greater mortality risk from in-traspecific aggression This differs from the pattern observedduring a long‐term study in Arizona where females were at ahigher risk of intraspecific mortality than males (46 of maleand 53 of female mortality Logan and Sweanor 2001)Mortality associated with intraspecific strife has been docu-mented in hunted and unhunted populations (this study Loganand Sweanor 2001 Stoner et al 2010) and its root cause hasbeen difficult to determine (eg population density and preypopulation trends) That being the case finding ways to reducewhat is often a major source of mortality for puma populationscontinues to pose a challenge to managersCanonical panthers were substantially more likely to die from

intraspecific aggression than were admixed panthers This mayat least partly explain the difference in survival between admixed

and canonical panthers Canonical panthers are known to bemore likely to suffer from varied correlates of inbreeding de-pression than admixed animals including atrial septal defects(Johnson et al 2010) and compromised immune systems(Roelke et al 1993) These factors have the potential to affectfitness and perhaps dominance of individuals when engaged inan intraspecific encounter A somewhat similar scenario wasobserved by Hogg et al (2006) in a population of bighorn sheepthat were part of an introgression project Admixed males in thispopulation had a greater probability of paternity than males thatwere less outbred This included coursing males with higherlevels of admixture being markedly more successful at fightingmales that were defending females in estrous (Hogg et al 2006)Heterosis resulting from genetic introgression is expected to be

the strongest in F1 individuals (Shull 1908 Crow 1948 Burkeand Arnold 2001 Waller 2015) and to progressively decline insubsequent generations (Pickup et al 2013 Frankham 2015)Given that admixed (especially F1) panthers survived better thancanonical panthers and that individual heterozygosity improvedsurvival of both sexes and all age classes our data provide evi-dence for heterotic advantage whereby individuals of mixedancestry had higher fitness (Shull 1908 Crow 1948) Our resultsare consistent with findings of other studies that involved in-tentional genetic introgression for conservation purposes Forexample Hogg et al (2006) detected substantial improvementsin survival and reproduction of introgressed bighorn sheep andMadsen et al (1999 2004) reported a dramatic increase in thepopulation of adders after genetic introgressionKnowledge of the persistence of benefits to a population ac-

crued via genetic introgression is essential for determiningwhen additional introgression may be warranted There is adepauperate amount of information regarding this topic and theidentification of factors that may influence persistence of thebenefits of genetic introgression (Tallmon et al 2004 Hedricket al 2014 Whiteley et al 2015) For example in minks(Neovison vison) Thirstrup et al (2014) found that outcrossingled to larger litters in F2 and F3 but this benefit disappeared insubsequent generations In a population of gray wolves Hedricket al (2014) suggested that benefits of genetic introgression may

Table 6 Costs and benefits accumulated over 100 years for genetic introgression at different intervals and with different numbers of pumas from Texas USAintroduced into the Florida panther population in South Florida USA Costs (2017 US dollars) include capture and transportation quarantine and post‐introgression monitoring Benefits are represented as average probability of quasi‐extinction (PQE and 5th and 95th percentiles) and changes (Δ) therein under thescenario using the minimum population count (McBride et al 2008 MPC) and the scenario using the motor vehicle mortality population estimates (McClintocket al 2015 MVM) The table is sorted by percent change in probability of quasi‐extinction under the MPC scenario

Interval (yr) Pumas Cost MPC PQE () ΔMPC PQE () MVM PQE () ΔMVM PQE ()

ndash 0 0 13 (0ndash99) 0 17 (0ndash100) 080 10 100000 12 (0ndash99) 10 (0ndash58) 16 (0ndash100) 5 (0ndash38)80 15 150000 12 (0ndash99) 13 (0ndash65) 15 (0ndash100) 8 (0ndash56)80 5 50000 12 (0ndash99) 14 (0ndash73) 15 (0ndash99) 9 (0ndash41)40 15 300000 10 (0ndash98) 26 (0ndash104) 14 (0ndash99) 13 (0ndash60)40 10 200000 9 (0ndash94) 35 (0ndash183) 13 (0ndash99) 21 (0ndash95)40 5 100000 8 (0ndash89) 39 (0ndash211) 12 (0ndash99) 26 (0ndash124)20 15 600000 8 (0ndash88) 42 (0ndash257) 13 (0ndash99) 24 (0ndash123)20 10 400000 6 (0ndash67) 54 (0ndash318) 10 (0ndash99) 37 (0ndash217)10 15 1200000 6 (0ndash53) 58 (0ndash372) 10 (0ndash99) 40 (0ndash208)20 5 200000 5 (0ndash50) 59 (0ndash338) 9 (0ndash99) 44 (0ndash352)10 10 800000 4 (0ndash16) 69 (0ndash489) 8 (0ndash92) 55 (0ndash401)10 5 400000 4 (0ndash7) 73 (0ndash502) 6 (0ndash71) 63 (0ndash503)

van de Kerk et al bull Florida Panther Population and Genetic Viability 27

wane after 2 or 3 generations The WrightndashFisher model ofgenetic drift predicts a geometric decline in expected hetero-zygosity at the rate of (1 minus 12Ne) per generation where Ne is theeffective population size (Hartl and Clark 1997 Frankham et al2014) Thus it seems logical to assume that the duration of thebenefits of introgression is limited especially in a species per-sisting in a small population such as Florida panthersWe expected Florida panther survival in the most recent years

(2005ndash2013) post‐introgression to differ from that observed in thedecade following the introgression in 1995 because pantherabundance and heterozygosity factors known to influence panthersurvival (Hostetler et al 2010 Benson et al 2011) have changed(Figs 2 and 6) We found that subadult and adult survival rates inrecent years (2005ndash2013) were similar to those in the period im-mediately following introgression (1995ndash2004 Fig 5C) overallkitten survival was similar to estimates of Hostetler et al (2010) Infact the pattern of covariate effects on Florida panther survivalreported by Benson et al (2011) and Hostetler et al (2010) hasbarely changed The negative effects of population density andpositive effects of heterozygosity may counterbalance survival ratesreducing population fluctuations Our result that benefits of geneticrestoration have remained in the population nearly for 5 genera-tions post‐intervention was surprising but also encouraging Pos-sible explanations for this observation may include 1) genetic ero-sion occurs at slower rate than previously thought 2) introducing 5migrants in 1 genetic intervention may have beneficial effects si-milar to those from 1 migrant per generation over multiple gen-erations or 3) other and as yet unknown factors may reduce therate of genetic erosion and subsequent demographic effects Adensity‐dependent effect on kitten survival could also explain whynon‐F1 admixed kittens did not survive substantially better thandid canonical kittens most kittens sampled from 2005ndash2013 wereadmixed and their survival was lower possibly because of a density‐dependent effect as the Florida panther population continuouslygrew during that period Trinkel et al (2010) also found thatpopulation density and inbreeding coefficient interacted to affectdemographic parameters and growth rate of an African lion po-pulation Likewise population density interacted with winterweather to affect the survival and population growth rates in Soaysheep (Ovis aries Milner et al 1999 Coulson et al 2001)

Population Dynamics and PersistenceThe finite deterministic and stochastic growth rates were gt10reflecting that much of the data used in this study were collectedfollowing genetic introgression in 1995 during which time thepopulation increased substantially Because our demographicparameter estimates did not change markedly in comparison tothose of Hostetler et al (2013) the similarity between thepopulation growth estimates from these studies was expectedBut these estimates reflect population growth during thedata‐collection period and do not predict future growth In factestimates of population size reported by McClintock et al(2015) suggest that population growth may already be slowing(Fig 2) A similar pattern was observed in an introduced po-pulation of African lion which increased steadily from 1995until 2001 then fluctuated around the presumed carrying ca-pacity thereafter (Trinkel et al 2010) Several other African lionpopulations that experienced various degrees of recovery

subsequently became relatively stable or exhibited decliningtrends (Bauer et al 2015)Our estimates of quasi‐extinction probabilities were lower than

those reported by Hostetler et al (2013) using a 2‐sex age‐structured matrix population model Lower probabilities ofquasi‐extinction than those reported by the earlier study forcomparable scenarios were likely a consequence of the fact thatthe estimates of abundance (McBride et al 2008) and thus thedensity‐dependent effects on survival of kittens and old panthers(gt10 yr) have changed in recent years Additionally these earlyestimates of the probability of quasi‐extinction and of time toquasi‐extinction did not consider genetic erosion that will in-evitably occur without management interventionUnder the MVM scenario the equilibrium population size was

twice that attained under the MPC scenario The MVM popula-tion estimates are approximately twice the MPCs (Fig 2) as aresult the simulated population under the MVM scenario achieveda higher equilibrium size Probability of quasi‐extinction under theMVM scenario was generally greater than that under the MPCscenario This result is a consequence of the fact that the estimateddensity dependence on kitten survival based on the MPC scenario isstronger than that for theMVM scenario (regression coefficients forabundance for the MPC scenario= minus0068 vs minus0002 for theMVM scenario Appendix B Table B1) Consequently simulatedpopulations under the MPC scenario have a stronger tendency tomove toward the equilibrium population size which inherentlyreduces the quasi‐extinction probability (eg Royama 1992)As expected for long‐lived mammals (Heppell et al 2000 Oli

and Dobson 2003) the population growth rate was proportio-nately most sensitive to changes in survival of prime adultsfollowed by that in younger age classes Global sensitivity ana-lysis in contrast showed that under all scenarios extinctionparameters were particularly sensitive to changes in survival ofFlorida panther kittens This result is a consequence of the highsensitivity of the population growth rate to kitten survival(Fig 9) and a larger standard error of kitten survival (Table 3)Results of our sensitivity analyses indicate that future researchshould focus on acquiring additional information on early lifestages to improve the accuracy and precision of estimates ofkitten survival Our results suggest that demographic variables(or their environmental drivers) that strongly influence extinc-tion risks are not necessarily those with the largest elasticityvalues A study of island fox (Bakker et al 2009) found thatextinction risk was strongly influenced by survival modifierparameters that had low elasticity values

Genetic Management When and How to InterveneThe use of genetic introgression as a management tool has al-ways been controversial and the probable effectiveness it mighthave in preventing the imminent demise of the panther popu-lation was initially debated (Creel 2006 Maehr et al 2006Pimm et al 2006) These debates notwithstanding the pantherpopulation had suffered from genetic morphological and bio-medical correlates of inbreeding before this management in-tervention (Johnson et al 2010 Onorato et al 2010) whereasthe post‐introgression period has been characterized by a sub-stantial reduction in the biomedical correlates of inbreeding andimprovements in demographic vigor and abundance (McBride

28 Wildlife Monographs bull 203

et al 2008 Hostetler et al 2010 2013 Johnson et al 2010Benson et al 2011 McClintock et al 2015) Nevertheless thepopulation remains small and isolated habitat is still being lostand there is no current possibility of natural gene flow betweenthis and other North American puma populations thus therecurrence of inbreeding‐related problems and subsequent po-pulation declines are inevitable The question therefore is notwhether genetic management of the Florida panther populationis needed but when and how it should be implementedWithout genetic introgression probabilities of quasi‐extinc-

tion were substantially greater than those from models thatincorporated periodic genetic introgression events (Fig 15)highlighting the importance of genetic management in smallisolated populations When 5 female pumas are added to theFlorida panther population every 10 years genetic variationincreased substantially under the MPC and MVM scenariosThe IBM predicted that the equillibrium population size wouldbe substantially larger when 5 pumas were released into thepopulation every 10 years than populations without intervention(ie no introgression) Releasing 15 Texas females did not af-ford substantially greater benefit to the equilibrium populationsize than did releasing only 5 Texas females Instead addingmore individuals caused increasingly large fluctuations in po-pulation size due to the numerical increases and density de-pendence (Fig 14) possibly diminishing the benefits associatedwith increased genetic variationCompared with the no‐intervention scenario (ie no genetic

introgression implemented) the introduction of 5 female pumasevery 10 years reduced quasi‐extinction probabilities from 13to 4 and from 17 to 6 for the MPC and MVM scenariosrespectively The benefit of genetic‐introgression strategies de-creased as the interval between successive management inter-ventions increased and no benefit was evident once the intervalreached 80 years (Fig 15) Introgression reduced the averagetime until quasi‐extinction (Fig 18) This somewhat counter‐intuitive result can be explained by the fact that repeated geneticintrogression makes the population more robust over timewhich will reduce the probability of extinction Consequentlyfew simulated population trajectories that fall below the criticalthreshold do so early reducing the mean time to extinctionAlso density dependence has a stabilizing effect on populations(Royama 1992) populations tend toward equilibrium popula-tion sizes which reduces the probability over time of popula-tions falling below quasi‐extinction threshold However timeuntil quasi‐extinction must be interpreted in conjunction withthe probability of quasi‐extinction which is very low for allscenarios with genetic introgression (Fig 15)Cost is a significant impediment to the implementation of a

genetic introgression program because captures relocations andmonitoring efforts before and after introgression are expensive(Edmands 2007 Frankham 2010b 2015 2016) Costs may beacceptable to society if the management actions result in sub-stantial fitness benefits especially if the species of concern ispopular and charismatic (Frankham 2015) We estimated the100‐year cost of introgression strategies modeled here to rangefrom $50000 to $12 million More expensive strategies gen-erally led to larger decreases in the probability of quasi‐extinc-tion but cheaper strategies also substantially improved

population viability (ie reduced quasi‐extinction probabilityTable 6) One of the cheaper strategies (5 pumas released every20 years) which cost an estimated $200000 over 100 yearsreduced the probability of quasi‐extinction by 59 and 44 forthe MPC and MVM scenarios respectively But these pre-liminary cost estimates are based on expenses incurred duringthe 1995 introgression and include costs of capture quarantinetransportation release and post‐release monitoring of femalesonly Although the cost for future genetic interventions wouldlikely be higher than those reported here the relative differencesin cost between alternative intervention strategies should remainapproximately the sameOur ability to simulate genetics and link it to the viability of

the Florida panther population is based on the empirically es-timated relationship between individual heterozygosity andsurvival Such heterozygosity and fitness correlations have beenstudied for decades (David 1998) but have only recently beenused to predict population viability (Benson et al 2016) noneto our knowledge have incorporated cost into their modelingefforts The mechanisms underlying heterozygosity and fitnesscorrelations remain unclear (David 1998 Szulkin et al 2010)and their significance as a proxy for inbreeding depression hasbeen debated (Balloux et al 2004 Miller and Coltman 2014)Nevertheless evidence continues to mount demonstratingpositive relationships between heterozygosity and survival(Coulson et al 1998ab Hansson et al 2004 Silva et al 2005Acevedo‐Whitehouse et al 2006 Jensen et al 2007 Velandoet al 2015) and between heterozygosity and fecundity (Seddonet al 2004 Charpentier et al 2005 Agudo et al 2012 Velandoet al 2015) Although the reported correlations are generallyweak their consequences can be substantial if population growthand persistence parameters are highly sensitive to the affectedvital rates (Velando et al 2015) Compared with scenarios thatignore genetics incorporating the effects of inbreeding on sur-vival increased quasi‐extinction probabilities within a 100‐yeartimeframe from 15 to 13 and from 14 to 17 for theMPC and MVM scenarios respectively These results high-light the importance of incorporating genetics when analyzingthe viability of small isolated populationsAlthough conservation biologists have recognized the im-

portance of genetics in population viability many still fail toincorporate genetic factors into PVA models (Reed et al 2002)Explicit consideration of genetics in PVAs requires long‐termgenetic and demographic data This permits the assessment ofthe functional relationship between genetic variability and de-mographic parameters Population viability analysis softwarepackages such as VORTEX (Lacy 1993 2000) offer a powerfulframework for individual‐based simulations but they often donot adequately capture a speciesrsquo life history or genetics Basedon a thorough review of the importance of genetic factors inwildlife conservation Frankham et al (2014) concluded thatgenetic factors can strongly influence the dynamics and persis-tence of small andor fragmented populations They furthernoted that ldquoMost PVA‐based risk assessments ignore or in-adequately model genetic factorsrdquo and recommended that ldquoPVAshould routinely include realistic inbreeding depressionhelliprdquo(Frankham et al 201457) Our custom‐built IBM permitted usto develop a model that adequately captured the Florida panther

van de Kerk et al bull Florida Panther Population and Genetic Viability 29

life history and we could parameterize the model with our long‐term genetic and demographic dataThe explicit consideration of the effects of inbreeding on

Florida panther population dynamics and persistence representsan important step forward Nonetheless our model remainsimperfect First we neglected the possible effects of habitat lossdetrimental anthropogenic activities climate change the prob-ability of immigration of pumas from western populationsdisease or catastrophes on demographic rates and populationviability Although immigration from puma populations outsideFlorida is possible its probability is very low given the extensivechallenges the anthropogenically altered landscape would posefor such a migrant to make it successfully to South Florida Weomitted mutations from our model because we have no in-formation on mutation rates though we expect that they are lowbased on values reported for other mammals (Kumar and Sub-ramanian 2002) Finally we only considered possible effects ofinbreeding on age‐specific survival rates because there was noevidence that heterozygosity affected female reproduction Lossof heterozygosity can negatively affect reproduction becauseinbred male panthers can suffer from cryptorchidism which canadversely affect their fecundity However given the polygynousmating system we expect the Florida panther population growthis primarily female‐limitedAlthough it is generally accepted that genetic introgression

only temporarily relieves a population from inbreeding depres-sion we know of no cases in which it has been implementedmore than once to conserve a threatened wildlife populationOur study provides an example of how demographic and mo-lecular data can be used within an IBM framework to evaluatethe efficacy of alternative genetic management strategies via theFlorida panther as a case study Our model can be adjusted forand applied to other species and offers great potential as a toolfor genetic management of small isolated wildlife populations

MANAGEMENT IMPLICATIONS

Our results and those of Hostetler et al (2013) and Johnsonet al (2010) provide persuasive evidence that the genetic in-tervention implemented in 1995 prevented the demise of theFlorida panther and restored demographic vigor to the popu-lation The Florida panther population remains small and iso-lated from other puma populations the closest puma populationis located gt1000 km away in Texas and the intervening land-scape is highly developed and fragmented with many interstate(and other high‐traffic) highways and major urban centersTheory suggests that 1 migrant per generation is sufficient tominimize the loss of genetic diversity (eg Mills and Allendorf1996) However the possibility of successful migration of apuma to South Florida followed by successful mating every ~5years is very low considering the distance between Texas andFlorida (Hedrick 1995) and the many risks and barriers thatdispersing pumas face (Beier 1995 Maehr et al 2002 Thatcheret al 2006) Consequently the pantherrsquos long‐term persistencewill likely depend on subsequent timely genetic introgressionsgiven the near‐impossibility of natural gene flow from otherpuma populations To that end our study offers considerableinsights into benefits of different genetic management strategiesand associated costs

Without genetic management the Florida panther populationfaces a substantial risk of quasi‐extinction within 100 years (10ndash46 depending on quasi‐extinction threshold and scenarios)Twenty‐three years have passed since genetic introgression wasimplemented and the population appears healthy as indicatedby measures of genetic variation increases in the populationsize and improvements in age‐specific survival rates Ourfindings suggest that releasing more pumas more frequently doesnot necessarily improve persistence probability because such astrategy would cause larger fluctuations in population size(Fig 14) which negatively affects persistence probability Thusextinction risks faced by inbred populations of large carnivorescan be substantially reduced by releasing approximately 5 im-migrants every 2 decades We suggest that such a managementstrategy be followed only in conjunction with continued long‐term monitoring of the population Our analyses demonstratethat population growth and persistence are highly sensitive tokitten and female (adult and subadult) survival Continuing tocollect data on these demographic variables along with bio-metric and genetic sampling will remain important to pantherrecovery and vigilance in identifying issues associated with in-breeding depression The collection of these monitoring datashould allow managers to detect any demographic or geneticchanges in a timely fashion and respond with genetic in-trogression or other management actions if warrantedRecovery objectives as defined by the USFWS in its current

recovery plan (USFWS 2008) delineate the need for additionalpopulations in the historical range to downlist or delist theFlorida panther If the only extant population of Floridapanthers continued to grow it could serve as a source ofpanthers to be translocated to suitable habitat to seed addi-tional populations Maintaining current information on thegenetic health of the population and precise estimates of itssize will be important in assessing whether it can withstand theremoval of a subset of animals for translocation Although the2017 documentation of a female panther with kittens north ofthe current breeding range is encouraging in terms of thepotential for population expansionmdashgiven that no female hadbeen recorded north of the Caloosahatchee River since 1972mdashthe re‐establishment of separate new populations will mostlikely require translocations by wildlife managers and a lengthysociopolitical processConservation of threatened or endangered species such as the

Florida panther is inherently challenging For panthers and otherlarge carnivores that require vast extents of natural habitat topersist habitat is typically the most limiting factor that will de-termine whether recovery efforts succeed Although geneticmanagement invariably plays a key role in the long‐term persis-tence of the panther population even a healthy population caneventually succumb to the impacts of habitat loss The humanpopulation of Florida continues to be one of the fastest‐growingin the nation the United States Census Bureau currently ranksFlorida as the third most populous Therefore habitat loss isgoing to continue to be a key issue for panthers Diligence towardpreserving panther habitat in its historical range via conservationeasements or other means and trying to keep such areas inter-connected via corridors is a goal stakeholders such as governmentagencies sportsmen non‐governmental organizations ranchers

30 Wildlife Monographs bull 203

and other private landowners must work on collaboratively toachieve

ACKNOWLEDGMENTS

We thank D Shindle D Jennings T OrsquoMeara D Land andC Belden for valuable advice throughout the project We thankS Bass M Criffield M Cunningham D Giardina D JansenA Johnson D Land M Lotz C McBride Rocky McBrideRowdy McBride Roy McBride L Oberhofer S Schulze staffsat Everglades National Park and Big Cypress National Preserveand others for assistance with fieldwork We also thank JAustin M Conroy J Hines J Laake S Mills J Nichols JHines M Sunquist J Fryxell and T Coulson for advice andassistance regarding data analysis and panther biology TheMuseum of Texas Tech University Natural Science ResearchLaboratory provided samples from pumas from west Texas forcomparative genetic analyses M Ben‐David D Land WMcCown E Hellgren and 4 anonymous reviewers providedmany helpful comments on the manuscript We thank NereaUbierna Lopez for completing the Spanish translation of theabstract We are grateful to the Department of ResearchComputing University of Florida for excellent high‐perfor-mance computing services We would like to thank US Fishand Wildlife Service Florida Fish and Wildlife ConservationCommission (FWC) US National Park Service QuantitativeSpatial Ecology Evolution and Environment (QSE3) In-tegrative Graduate Education and Research Traineeship(IGERT) program funded by the National Science Foundationand the University of Florida for financially supporting thisstudy Funding for panther research conducted by the FWC issupported predominantly via donations accrued with vehicleregistrations for ldquoProtect the Pantherrdquo license plates

LITERATURE CITEDAcevedo‐Whitehouse K T R Spraker E Lyons S R Melin F Gulland RL Delong and W Amos 2006 Contrasting effects of heterozygosity onsurvival and hookworm resistance in California sea lion pups MolecularEcology 151973ndash1982

Adams J R L M Vucetich P W Hedrick R O Peterson and J AVucetich 2011 Genomic sweep and potential genetic rescue during limitingenvironmental conditions in an isolated wolf population Proceedings of theRoyal Society B Biological Sciences 2783336ndash3344

Agresti A 2013 Categorical data analysis Third edition John Wiley amp SonsHoboken New Jersey USA

Agudo R M Carrete M Alcaide C Rico F Hiraldo and J A Donaacutezar2012 Genetic diversity at neutral and adaptive loci determines individualfitness in a long‐lived territorial bird Proceedings of the Royal Society BBiological Sciences 2793241ndash3249

Albeke S E N P Nibbelink and M Ben‐David 2015 Modeling behavior bycoastal river otter (Lontra canadensis) in response to prey availability in PrinceWilliam Sound Alaska a spatially‐explicit individual‐based approach PLOSOne 10e0126208

Alho J S K Vaumllimaumlki and J Merilauml 2010 Rhh an R extension for estimatingmultilocus heterozygosity and heterozygosity‐heterozygosity correlationMolecular Ecology Resources 10720ndash722

Alvarez K 1993 Twilight of the panther Myakka River Publishing SarasotaFlorida USA

Aparicio J M J Ortego and P J Cordero 2006 What should we weigh toestimate heterozygosity alleles or loci Molecular Ecology 154659ndash4665

Bakker V J D F Doak G W Roemer D K Garcelon T J Coonan S AMorrison C Lynch K Ralls and R Shaw 2009 Incorporating ecologicaldrivers and uncertainty into a demographic population viability analysis for theisland fox Ecological Monographs 7977ndash108

Balloux F W Amos and T Coulson 2004 Does heterozygosity estimateinbreeding in real populations Molecular Ecology 133021ndash3031

Barone M A M E Roelke J Howard J L Brown A E Anderson and DE Wildt 1994 Reproductive characteristics of male Florida panthersmdashcomparative studies from Florida Texas Colorado Latin‐America andNorth‐American Zoos Journal of Mammalogy 75150ndash162

Bateson Z W P O Dunn S D Hull A E Henschen J A Johnson and LA Whittingham 2014 Genetic restoration of a threatened population ofgreater prairie‐chickens Biological Conservation 17412ndash19

Bauer H G Chapron K Nowell P Henschel P Funston L T B HunterD W Macdonald and C Packer 2015 Lion (Panthera leo) populations aredeclining rapidly across Africa except in intensively managed areas Pro-ceedings of National Academy of Sciences of the United States of America11214894ndash14899

Beier P 1995 Dispersal of juvenile cougars in fragmented habitat Journal ofWildlife Management 59228ndash237

Benson J F J A Hostetler D P Onorato W E Johnson M E Roelke SJ OrsquoBrien D Jansen and M K Oli 2011 Intentional genetic introgressioninfluences survival of adults and subadults in a small inbred felid populationJournal of Animal Ecology 80958ndash967

Benson J F P J Mahoney J A Sikich L E K Serieys J P Pollinger H BErnest and S P D Riley 2016 Interactions between demography geneticsand landscape connectivity increase extinction probability for a small popu-lation of large carnivores in a major metropolitan area Proceedings of theRoyal Society B Biological Sciences 28320160957

Beverly F 1874 The Florida panther Forest and Stream 3290Boyce M S C V Haridas C T Lee and the NCEAS Stochastic Demo-graphy Working Group 2006 Demography in an increasingly variable worldTrends in Ecology amp Evolution 21141ndash148

Brook B W J J OrsquoGrady A P Chapman M A Burgman H R Akcakayaand R Frankham 2000 Predictive accuracy of population viability analysis inconservation biology Nature 404385ndash387

Brys R and H Jacquemyn 2016 Severe outbreeding and inbreeding de-pression maintain mating system differentiation in Epipactis (Orchidaceae)Journal of Evolutionary Biology 29352ndash359

Burke J M and M L Arnold 2001 Genetics and the fitness of hybridsAnnual Review of Genetics 3531ndash52

Burnham K P 1993 A theory for combined analysis of ring recovery andrecapture data Pages 199ndash213 in J‐D Lebreton and P M North editorsMarked individuals in the study of bird population Springer‐Verlag NewYork New York USA

Burnham K P and D R Anderson 2002 Model selection and multimodelinference a practical information‐theoretic approach Second edition SpringerVerlag New York New York USA

Caswell H 2001 Matrix population models construction analysis and in-terpretation Sinauer Associates Sunderland Massachusetts USA

Charpentier M J M Setchell F Prugnolle L A Knapp E J Wickings PPeignot and M Hossaert‐McKey 2005 Genetic diversity and reproductivesuccess in mandrills (Mandrillus sphinx) Proceedings of the NationalAcademy of Sciences of the United States of America 10216723ndash16728

Cooch E and G C White editors 2018 Program MARK a gentle in-troduction lthttpwwwphidotorgsoftwaremarkdocsbookgt Accessed 7Apr 2016

Cooley H S R B Wielgus G M Koehler H S Robinson and B TMaletzke 2009 Does hunting regulate cougar populations A test of thecompensatory mortality hypothesis Ecology 902913ndash2921

Coulson T E A Catchpole S D Albon B J Morgan J M Pemberton TH Clutton‐Brock M J Crawley and B T Grenfell 2001 Age sex densitywinter weather and population crashes in Soay sheep Science 2921528ndash1531

Coulson T J Pemberton S Albon M Beaumont T Marshall J Slate FGuinness and T Clutton‐Brock 1998a Microsatellites reveal heterosis in reddeer Proceedings of the Royal Society B Biological Sciences 265489ndash495

Coulson T N S D Albon J M Pemberton J Slate F E Guinness and TH Clutton‐Brock 1998b Genotype by environment interactions in wintersurvival in red deer Journal of Animal Ecology 67434ndash445

Cox D 1972 Regression models and life‐tables Journal of the Royal StatisticalSociety Series B Statistical Methodology 34187ndash220

Creel S 2006 Recovery of the Florida panthermdashgenetic rescue demographic rescueor both Response to Pimm et al (2006) Animal Conservation 9125ndash126

Crnokrak P and D A Roff 1999 Inbreeding depression in the wild Heredity83260ndash270

Crow J 1948 Alternative hypotheses of hybrid vigor Genetics 33477ndash487

van de Kerk et al bull Florida Panther Population and Genetic Viability 31

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 17: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

model 0ndash16) within year 200 (Fig 11) The mean times until quasi‐extinction based on the MVM scenario were 5 years (IBM 0ndash61)and 6 years (matrix model 0ndash64) conditioned on going quasi‐extinctwithin 100 years Expected times until quasi‐extinction conditionedon quasi‐extinction within 200 years were 11 years (IBM 0ndash113)and 11 years (matrix model 0ndash112 Fig 11)

Sensitivity of extinction probability to demographic parametersmdashThe partial rank correlation coefficients between quasi‐extinctionprobabilities and demographic parameters were negative anincrease in any of the demographic parameters decreased thequasi‐extinction probability Probabilities of quasi‐extinction within200 years were most sensitive to changes in kitten survival for bothscenarios (Fig 12) followed by survival of subadult females in theMVM scenarios and survival of prime adult females in the MPCscenario The probability of reproduction of prime adult femaleshad the third‐largest partial rank correlation coefficient with quasi‐extinction probability for both scenarios

Benefits and Costs of Genetic ManagementMinimum population count scenariomdashThe MPC scenario with

a critical threshold of 10 panthers predicted that withoutmanagement intervention average individual heterozygositywould decrease reaching 057 (049ndash064) at 100 years and053 (042ndash062) at 200 years (Fig 13) The population woulddecrease slightly reaching an average of 79 (41ndash99) at 100 yearsand 76 (43ndash98) at 200 years (Fig 14) The probabilities of quasi‐extinction under the no‐intervention MPC scenario when weconsidered the effects of genetic erosion were 13 (0ndash99) at 100years and 23 (0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with

the translocation of 5 Texas females average individualheterozygosity under the MPC scenario increased and thenstabilized at 068 (064ndash072) at 100 years and remained atthat level at 200 years (Fig 13) The population reached anaverage of 88 (62ndash104) at 100 years and stayed the same at200 years (Fig 14) The probabilities of quasi‐extinctionunder this introgression strategy were 6 (0ndash53) within 100years and 7 (0ndash82) within 200 years (Fig 15) Results ofanalyses using a critical threshold of 30 panthers revealed asimilar pattern of relative benefits However the prob-abilities of quasi‐extinction were substantially higher (ge45)across all scenarios (Appendix E)

Motor vehicle mortality scenariomdashWithout managementintervention the average individual heterozygosity predictedby the MVM scenario and a critical threshold of 10 panthersdecreased to 060 (046ndash069) at 100 years and to 059 (039ndash070) at 200 years (Fig 13) The population increased slightlyand reached an equilibrium at 154 individuals (46ndash217) at 100years and 157 (57ndash218) at 200 years (Fig 14) The probabilitiesof quasi‐extinction were 17 (0ndash100) at 100 years and 22(0ndash100) at 200 years (Fig 15)When introgression was implemented every 10 years with 5

female pumas from Texas average individual heterozygosityincreased slightly reaching 067 (060ndash073) at 100 years and068 (059ndash075) at 200 years (Fig 13) Under this strategy thepopulation reached an average of 164 individuals (44ndash226) at

Table 4 Model comparison results testing for the effects of several covariateson survival of all kittens survival of genetically sampled kittens survival ofsubadult and adult Florida panthers probability of reproduction and kittensproduced annually for Florida panthers sampled from 1981ndash2013 in SouthFlorida USA

Model Parameters ΔAICa Weight

Kitten survival (all data)Base+ F1b+ abundancec 10 000 0988Base+ abundance 9 1018 0006Base+ F1 9 1035 0005Base 8 2711 lt0001Base+ sexd 9 2912 lt0001Base+ litter sizee 9 2914 lt0001

Kitten survival (genetically sampled kittens only)Base+ F1+ abundance 10 000 0541Base+ F1CanAdmf+ abundance 11 099 0329Base+Hetg+ abundance 10 310 0115Base+CanAdmh+ abundance 10 791 0010Base+ abundance 9 944 0005Base+ F1 9 1638 lt0001Base+ F1CanAdm 10 1837 lt0001Base+Het 9 2872 lt0001Base 8 3296 lt0001Base+CanAdm 9 3348 lt0001

Subadult and adult survivalBase+ F1CanAdm+ abundance 7 000 0360Base+ F1 5 053 0276Base+ F1CanAdm 6 105 0213Base+ F1+ abundance 6 222 0119Base+CanAdm+ abundance 6 575 0020Base+CanAdm 5 908 0004Base+Het 5 964 0003Base+Het+ abundance 6 984 0003Base 4 1118 0001Base+ abundance 5 1278 0001

Probability of reproductionAge 3 000 0242Age+CanAdm 4 012 0228Age+ abundance 4 173 0102Age+ F1 4 190 0094Age+CanAdm+ abundance 5 216 0082Age+Het 4 219 0081Age+ F1CanAdm 5 232 0076Age+ F1+ abundance 5 372 0038Age+Het+ abundance 5 399 0033Age+ F1CanAdm+ abundance 6 444 0026

Kittens produced annuallyAge+ F1+ abundance 5 000 0194Age 3 060 0144Age+ abundance 4 061 0143Age+ F1 4 097 0120Age+CanAdm 4 139 0097Age+ F1CanAdm+ abundance 6 196 0073Age+ F1CanAdm 5 231 0061Age+CanAdm+ abundance 5 238 0059Age+Het+ abundance 5 250 0056Age+Het 4 259 0053

a Akaikersquos information criterion (AIC) for kitten survival models was adjustedfor small sample size and overdispersion (QAICc)

b F1 divides Florida panthers into 2 groups F1 admixed and all other panthersc Index of Florida panther abundanced Sex of an individual Florida panther (male or female)e Size of the litter in which a Florida panther kitten was bornf F1CanAdm divides panthers into 3 groups F1 admixed canonical andother admixed panthers

g Het is the individual heterozygosityh CanAdm divides panthers into 2 groups canonical and admixed (includingF1 admixed) panthers

van de Kerk et al bull Florida Panther Population and Genetic Viability 19

100 years and 170 (67ndash231) at 200 years based on the MVMscenario (Fig 14) The probabilities of quasi‐extinction underthis introgression strategy were 10 (0ndash99) at 100 years and12 (0ndash99) at 200 years (Fig 15)The projected population size and average individual hetero-

zygosity reached equilibrium levels with periodic fluctuations inthese values when introgression was implemented (Figs 16ndash17)Genetic introgression decreased the time until quasi‐extinctionacross all scenarios (Fig 18) Results of analyses using a criticalthreshold of 30 panthers revealed a similar pattern of relative ben-efits However the probabilities of quasi‐extinction were sub-stantially higher (ge45) across all scenarios (Appendix E)

Addition of pumas from Texas increased both the populationsize and average individual heterozygosity consequentlyintrogression caused periodic increases in average individualheterozygosity and population size at the interval at which theintrogression occurred Whereas average individual hetero-zygosity gradually decreased following introgression densitydependence caused the population size to fluctuate especiallywhen a larger number of pumas from Texas were released Thepositive effect of genetic introgression initiatives on quasi‐extinction probabilities decreased as the time interval betweenthese events increased More frequent genetic introgressions ledto greater increases in average individual heterozygosity and

Old adult

Prime adult

Subadult

Kitten

025 050 075

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Individual heterozygosity

Ann

ual s

urvi

val r

ate

Old adult

Prime adult

Subadult

Kitten

40 60 80 100 120

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Abundance index

Ann

ual s

urvi

val r

ate

Kittens Males Females

Figure 7 The effect of individual heterozygosity (left) and the abundance index (right) on Florida panther age‐ and sex‐specific survival estimates (shaded area isSE) in South Florida USA 1981ndash2013 Abundance index data are from McBride et al (2008) and R T McBride and C McBride (Rancherrsquos Supply Incorporatedunpublished report)

20 Wildlife Monographs bull 203

equilibrium population size while reducing the probability ofquasi‐extinction Comparatively performing genetic introgres-sion less often but with more individuals per release was lesseffective at improving the long‐term prospects for the pantherpopulation (Figs 13ndash15)

Financial cost and persistence benefitsmdashThe total estimated costof 1 genetic introgression was $40000 $80000 or $120000for releasing 5 10 or 15 pumas from Texas respectively(Table 5) The total cost incurred over 100 years for eachstrategy ranged from $50000 to $12 million (Table 6)The most expensive strategy involved the introduction of15 pumas every 10 years this strategy reduced the probability

of quasi‐extinction by 58 and 40 under the MPC and MVMscenarios respectively (Table 6) Introducing 5 pumas every80 years cost the least but improvements in populationpersistence under this strategy were insubstantial (Table 6)Releasing more panthers more frequently caused increasedpopulation fluctuations but did not necessarily reduce the quasi‐extinction probability (Figs 14 and 15 Table 6)

DISCUSSION

The duration (34 yr of data) and sample sizes associated with theFlorida Panther Project allowed us to obtain a comprehensiveunderstanding of genetic variation demographic parameters

A

B

C

Figure 8 Cause‐specific mortality rates (plusmnSE) for male and female Florida panthers (A) subadult prime‐adult and old‐adult Florida panthers of both sexes (B)and canonical and admixed Florida panthers of both sexes (C) in South Florida USA 1981ndash2013 Causes of mortality are vehicle collision intraspecific aggression(ISA) other causes (known causes such as diseases injuries and infections unrelated to the first 2 causes) and unknown cause (mortalities for which we could notassign a cause)

van de Kerk et al bull Florida Panther Population and Genetic Viability 21

probability of population persistence and the duration of thepositive impacts of genetic restoration on this endangered po-pulation The Florida panther population was in dire straits inthe early 1990s and our results clearly demonstrated that the

implementation of the introgression project in 1995 subse-quently accrued a series of genetic and demographic improve-ments that have led to the change from a declining population toone that is growing and expanding its current breeding range

Sensitivity Elasticity

0 1 2 3 00 02 04

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

Para

met

er

Figure 9 Sensitivity and elasticity (proportional sensitivity) of stochastic population growth rate (λs) of the Florida panther to vital demographic parameters in SouthFlorida USA 1981ndash2013 Horizontal lines represent 5th and 95th percentiles

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

70

80

90

100

110

0

5

10

15

0

50

100

Years

StatisticMeanMedian

Figure 10 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the minimum population count(MPC) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

125

150

175

200

00

05

10

15

20

0

30

60

90

Years

StatisticMeanMedian

Figure 11 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the motor vehicle mortality(MVM) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

22 Wildlife Monographs bull 203

MPC MVM

minus08 minus06 minus04 minus02 00 minus08 minus06 minus04 minus02 00

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Male subadult survival

Male prime adult survival

Male old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

PRCC

Para

met

er

Figure 12 Partial rank correlation coefficient (PRCC) between quasi‐extinction probabilities and the demographic parameters of the Florida panther for theminimum population count (MPC) and motor vehicle mortality (MVM) scenarios Error bars indicate standard deviation Based on data collected in South FloridaUSA 1981ndash2013

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 13 Average projected individual heterozygosities in the Florida panther population over 200 years without genetic management intervention and with eachof the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include theintroduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years Average individual heterozygosity peaks when pumas are added to the Floridapanther population Based on data collected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 23

Our research has further validated the success of genetic in-trogression by quantifying the reduction in the probability ofextinction of the panther population because of this manage-ment initiative These findings all bode well for continuedprogress toward recovery That said the Florida panther po-pulation remains small and isolated from other populations ofpuma from western North America thereby denoting that theeventual impact of genetic drift and inbreeding may negativelyaffect the population in the future Realizing this will continueto be an issue that managers will have to contemplate wemodeled the impact of varied genetic management scenarios todetermine the frequency of introgression along with the numberof panthers that should be released during each initiative tominimize cost and maximize benefits to the population Theseresults will prove invaluable to managers attempting to conservethe Florida panther

Genetic Variation Demographic Parametersand Persistence of Heterotic Benefits from GeneticIntrogressionOur measure of individual heterozygosity (1 minusHL) was higheron average for the admixed (0615) panthers than for those ofcanonical ancestry (0386) Levels of individual heterozygosityfor admixed panthers were comparable to levels observed in asample of pumas from west Texas (x plusmn SE= 0695plusmn 0019n= 41) which further demonstrates the improvement in levelsof genetic variation in the Florida population since 1995 The

correlation between HL and several demographic parametershas been noted in wild populations including cheetahs (Terrellet al 2016) and several species of birds such as European shags(Phalacrocorax aristotelis male and female survival probabilityVelando et al 2015) and Egyptian vulture (Neophron percnop-terus age at recruitment Agudo et al 2012) If we focus only onpanthers captured and radiocollared from 2012ndash2015 (FP211ndashFP240) all of which had estimated birth years ge2005 (ge10 yrpost‐introgression) those panthers had an average individualheterozygosity level of 0618plusmn 0029 highlighting the elevatedlevels of genetic variation that continue to persist in cohorts ofpanthers 5 generations after the release of female pumas fromTexasThe proportional membership values (q) from our

STRUCTURE analysis allowed us to delineate 2 clusters ofancestry for Florida panthers (Fig 4) During the pre‐in-trogression era the population was dominated by panthers thatretained a high percentage of the canonical ancestry which wasaffected by inbreeding depression The post‐introgression eraancestry values are comprised of many admixed individuals inessence demonstrating the success of the introgression in termsof increasing genetic variation in the population and mi-micking gene flow that historically occurred with panthers andother populations of pumas (Onorato et al 2010) A somewhatanalogous situation although abetted by natural immigrationwas evident in the ancestral variation in a small population ofpuma intersected by a major freeway in California USA In

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

75

80

85

90

95

100

130

150

170

190

Years

Popu

latio

n si

ze

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 14 Average projected population sizes in the Florida panther population over 200 years without intervention and with each of the genetic introgressionstrategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include the introduction of 5 10 or 15pumas from Texas USA every 10 20 40 or 80 years Peaks occur when pumas are added to the Florida panther population Based on data collected in SouthFlorida USA 1981ndash2013

24 Wildlife Monographs bull 203

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

25

50

75

100

0

25

50

75

100

Introgression interval (yrs)

PQ

E (

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 15 Average probability of quasi‐extinction (PQE) of the Florida panther population within 100 years and within 200 years without genetic managementintervention (N) and with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenariosIntrogression strategies include the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Errorbars indicate 5th and 95th percentiles Based on data collected in South Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

50

60

70

80

90

100

110

50

100

150

200

Years

Popu

latio

n si

ze

Figure 16 Average projected Florida panther population trajectories under a geneticintrogression strategy involving the release of 5 female pumas from Texas USA every20 years for the minimum population count (MPC) and motor vehicle mortality(MVM) scenarios Shaded area indicates 5th and 95th percentiles and isrepresentative of confidence intervals for other scenarios Based on data collected inSouth Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Figure 17 Average projected Florida panther population‐level heterozygositiesunder a genetic introgression strategy involving the release of 5 female pumas fromTexas USA every 20 years for the minimum population count (MPC) and motorvehicle mortality (MVM) scenarios Shaded area bounds the 5th and 95th percentilesand is representative of confidence intervals for other scenarios Based on datacollected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 25

this case the migration of just a single male across the freewayto the south resulted in an increase in the number of in-dividuals with admixed ancestry and substantially improvedgenetic diversity of the subpopulation south of the freeway(Riley et al 2014)Our estimate of overall kitten survival (0321plusmn 0056) was

similar to that reported by Hostetler et al (2010) who used13 years of post‐introgression data (0323plusmn 0071) These valuesare lower than kitten survival estimates derived from severalpuma populations in western North America (044ndash091 Loganand Sweanor 2001 Lambert et al 2006 Laundre et al 2007Robinson et al 2008 Ruth et al 2011) When we included onlydata for individuals that had been genetically sampled our es-timate was 15 higher The proportion of admixed kittens thatwere genetically sampled increased as the population expandedwhich could have resulted in higher estimates of kitten survivalbecause admixed kittens survive better than canonical kittensKitten survival was strongly influenced by genetic ancestry withsurvival rates of F1 kittens being almost twice that of canonicalkittens (Fig 5B) Consistent with the findings of Hostetler et al(2010) increases in heterozygosity coincided with increasedkitten survival whereas population density negatively affectedkitten survival Population density often has a strong negativeimpact on survival of young age classes due to infanticide andcannibalism which has been reported from several carnivore

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

50

100

150

0

50

100

150

Introgression interval (yrs)

TQE

(yrs

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 18 Average times until quasi‐extinction (TQE) for the Florida panther population within 100 years and within 200 years without genetic management interventionand with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategiesinclude the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Error bars indicate 5th and 95thpercentiles Based on data collected in South Florida USA 1981ndash2013

Table 5 Average cost (2017 US dollars) of introgression strategies employedover 100 years that involve the introduction of 5 10 or 15 pumas from TexasUSA into the Florida panther population in South Florida USA at an in-creasingly higher frequency Costs of capture (including transportation) quar-antine and post‐introgression monitoring were estimated by Roy McBrideMark Cunningham and Dave Onorato respectively

Frequency Component 5 pumas 10 pumas 15 pumas

Once Capture 25000 50000 75000

Quarantine 5000 10000 15000

Monitoring 10000 20000 30000

Total 40000 80000 120000

Every 80 years Capture 31250 62500 93750

Quarantine 6250 12500 18750

Monitoring 12500 25000 37500

Total 50000 100000 150000

Every 40 years Capture 62500 125000 187500

Quarantine 12500 25000 37500

Monitoring 25000 50000 75000

Total 100000 200000 300000

Every 20 years Capture 125000 250000 375000

Quarantine 25000 50000 75000

Monitoring 50000 100000 150000

Total 200000 400000 600000

Every 10 years Capture 250000 500000 750000

Quarantine 50000 100000 150000

Monitoring 100000 200000 300000

Total 400000 800000 1200000

26 Wildlife Monographs bull 203

species including polar bears (Ursus maritimus Derocher andWiig 1999) black bears (Ursus americanus Garrison et al 2007)and pumas (Logan and Sweanor 2001)Our estimates of sex‐ and age‐specific survival rates of subadult

and adult panthers (Table 3) and the effects of covariates on theserates were similar to those reported by Benson et al (2011) derivedfrom data collected through 2006 Our survival rates for males(065ndash077) were lower than females (078ndash097) for all 3 ageclasses (subadult prime adult and old adult) which is the typicaltrend observed in most puma populations (eg Ruth et al 19982011) However an unhunted puma population occupying afragmented urban landscape in southern California exhibited lowersurvival (0586 females 0525 males Vickers et al 2015) perhapsas a result of severe habitat fragmentation and the lack of extensiveswaths of public land protected in perpetuity Most populations ofpuma in western North America are typically subject to variedlevels of management via regulated hunting which often is biasedtoward the take of males (Cooley et al 2009 Ruth et al 2011)Consequently annual survival estimates from hunted populationsin northeast Washington (0679ndash0733 females 0341 malesRobinson et al 2008) and southeastern Arizona (0685 females0584 males Cunningham et al 2001) were generally lower thanthose we estimated for Florida panthersCause‐specific mortality analysis showed that male panthers

experienced a substantially greater mortality risk from in-traspecific aggression This differs from the pattern observedduring a long‐term study in Arizona where females were at ahigher risk of intraspecific mortality than males (46 of maleand 53 of female mortality Logan and Sweanor 2001)Mortality associated with intraspecific strife has been docu-mented in hunted and unhunted populations (this study Loganand Sweanor 2001 Stoner et al 2010) and its root cause hasbeen difficult to determine (eg population density and preypopulation trends) That being the case finding ways to reducewhat is often a major source of mortality for puma populationscontinues to pose a challenge to managersCanonical panthers were substantially more likely to die from

intraspecific aggression than were admixed panthers This mayat least partly explain the difference in survival between admixed

and canonical panthers Canonical panthers are known to bemore likely to suffer from varied correlates of inbreeding de-pression than admixed animals including atrial septal defects(Johnson et al 2010) and compromised immune systems(Roelke et al 1993) These factors have the potential to affectfitness and perhaps dominance of individuals when engaged inan intraspecific encounter A somewhat similar scenario wasobserved by Hogg et al (2006) in a population of bighorn sheepthat were part of an introgression project Admixed males in thispopulation had a greater probability of paternity than males thatwere less outbred This included coursing males with higherlevels of admixture being markedly more successful at fightingmales that were defending females in estrous (Hogg et al 2006)Heterosis resulting from genetic introgression is expected to be

the strongest in F1 individuals (Shull 1908 Crow 1948 Burkeand Arnold 2001 Waller 2015) and to progressively decline insubsequent generations (Pickup et al 2013 Frankham 2015)Given that admixed (especially F1) panthers survived better thancanonical panthers and that individual heterozygosity improvedsurvival of both sexes and all age classes our data provide evi-dence for heterotic advantage whereby individuals of mixedancestry had higher fitness (Shull 1908 Crow 1948) Our resultsare consistent with findings of other studies that involved in-tentional genetic introgression for conservation purposes Forexample Hogg et al (2006) detected substantial improvementsin survival and reproduction of introgressed bighorn sheep andMadsen et al (1999 2004) reported a dramatic increase in thepopulation of adders after genetic introgressionKnowledge of the persistence of benefits to a population ac-

crued via genetic introgression is essential for determiningwhen additional introgression may be warranted There is adepauperate amount of information regarding this topic and theidentification of factors that may influence persistence of thebenefits of genetic introgression (Tallmon et al 2004 Hedricket al 2014 Whiteley et al 2015) For example in minks(Neovison vison) Thirstrup et al (2014) found that outcrossingled to larger litters in F2 and F3 but this benefit disappeared insubsequent generations In a population of gray wolves Hedricket al (2014) suggested that benefits of genetic introgression may

Table 6 Costs and benefits accumulated over 100 years for genetic introgression at different intervals and with different numbers of pumas from Texas USAintroduced into the Florida panther population in South Florida USA Costs (2017 US dollars) include capture and transportation quarantine and post‐introgression monitoring Benefits are represented as average probability of quasi‐extinction (PQE and 5th and 95th percentiles) and changes (Δ) therein under thescenario using the minimum population count (McBride et al 2008 MPC) and the scenario using the motor vehicle mortality population estimates (McClintocket al 2015 MVM) The table is sorted by percent change in probability of quasi‐extinction under the MPC scenario

Interval (yr) Pumas Cost MPC PQE () ΔMPC PQE () MVM PQE () ΔMVM PQE ()

ndash 0 0 13 (0ndash99) 0 17 (0ndash100) 080 10 100000 12 (0ndash99) 10 (0ndash58) 16 (0ndash100) 5 (0ndash38)80 15 150000 12 (0ndash99) 13 (0ndash65) 15 (0ndash100) 8 (0ndash56)80 5 50000 12 (0ndash99) 14 (0ndash73) 15 (0ndash99) 9 (0ndash41)40 15 300000 10 (0ndash98) 26 (0ndash104) 14 (0ndash99) 13 (0ndash60)40 10 200000 9 (0ndash94) 35 (0ndash183) 13 (0ndash99) 21 (0ndash95)40 5 100000 8 (0ndash89) 39 (0ndash211) 12 (0ndash99) 26 (0ndash124)20 15 600000 8 (0ndash88) 42 (0ndash257) 13 (0ndash99) 24 (0ndash123)20 10 400000 6 (0ndash67) 54 (0ndash318) 10 (0ndash99) 37 (0ndash217)10 15 1200000 6 (0ndash53) 58 (0ndash372) 10 (0ndash99) 40 (0ndash208)20 5 200000 5 (0ndash50) 59 (0ndash338) 9 (0ndash99) 44 (0ndash352)10 10 800000 4 (0ndash16) 69 (0ndash489) 8 (0ndash92) 55 (0ndash401)10 5 400000 4 (0ndash7) 73 (0ndash502) 6 (0ndash71) 63 (0ndash503)

van de Kerk et al bull Florida Panther Population and Genetic Viability 27

wane after 2 or 3 generations The WrightndashFisher model ofgenetic drift predicts a geometric decline in expected hetero-zygosity at the rate of (1 minus 12Ne) per generation where Ne is theeffective population size (Hartl and Clark 1997 Frankham et al2014) Thus it seems logical to assume that the duration of thebenefits of introgression is limited especially in a species per-sisting in a small population such as Florida panthersWe expected Florida panther survival in the most recent years

(2005ndash2013) post‐introgression to differ from that observed in thedecade following the introgression in 1995 because pantherabundance and heterozygosity factors known to influence panthersurvival (Hostetler et al 2010 Benson et al 2011) have changed(Figs 2 and 6) We found that subadult and adult survival rates inrecent years (2005ndash2013) were similar to those in the period im-mediately following introgression (1995ndash2004 Fig 5C) overallkitten survival was similar to estimates of Hostetler et al (2010) Infact the pattern of covariate effects on Florida panther survivalreported by Benson et al (2011) and Hostetler et al (2010) hasbarely changed The negative effects of population density andpositive effects of heterozygosity may counterbalance survival ratesreducing population fluctuations Our result that benefits of geneticrestoration have remained in the population nearly for 5 genera-tions post‐intervention was surprising but also encouraging Pos-sible explanations for this observation may include 1) genetic ero-sion occurs at slower rate than previously thought 2) introducing 5migrants in 1 genetic intervention may have beneficial effects si-milar to those from 1 migrant per generation over multiple gen-erations or 3) other and as yet unknown factors may reduce therate of genetic erosion and subsequent demographic effects Adensity‐dependent effect on kitten survival could also explain whynon‐F1 admixed kittens did not survive substantially better thandid canonical kittens most kittens sampled from 2005ndash2013 wereadmixed and their survival was lower possibly because of a density‐dependent effect as the Florida panther population continuouslygrew during that period Trinkel et al (2010) also found thatpopulation density and inbreeding coefficient interacted to affectdemographic parameters and growth rate of an African lion po-pulation Likewise population density interacted with winterweather to affect the survival and population growth rates in Soaysheep (Ovis aries Milner et al 1999 Coulson et al 2001)

Population Dynamics and PersistenceThe finite deterministic and stochastic growth rates were gt10reflecting that much of the data used in this study were collectedfollowing genetic introgression in 1995 during which time thepopulation increased substantially Because our demographicparameter estimates did not change markedly in comparison tothose of Hostetler et al (2013) the similarity between thepopulation growth estimates from these studies was expectedBut these estimates reflect population growth during thedata‐collection period and do not predict future growth In factestimates of population size reported by McClintock et al(2015) suggest that population growth may already be slowing(Fig 2) A similar pattern was observed in an introduced po-pulation of African lion which increased steadily from 1995until 2001 then fluctuated around the presumed carrying ca-pacity thereafter (Trinkel et al 2010) Several other African lionpopulations that experienced various degrees of recovery

subsequently became relatively stable or exhibited decliningtrends (Bauer et al 2015)Our estimates of quasi‐extinction probabilities were lower than

those reported by Hostetler et al (2013) using a 2‐sex age‐structured matrix population model Lower probabilities ofquasi‐extinction than those reported by the earlier study forcomparable scenarios were likely a consequence of the fact thatthe estimates of abundance (McBride et al 2008) and thus thedensity‐dependent effects on survival of kittens and old panthers(gt10 yr) have changed in recent years Additionally these earlyestimates of the probability of quasi‐extinction and of time toquasi‐extinction did not consider genetic erosion that will in-evitably occur without management interventionUnder the MVM scenario the equilibrium population size was

twice that attained under the MPC scenario The MVM popula-tion estimates are approximately twice the MPCs (Fig 2) as aresult the simulated population under the MVM scenario achieveda higher equilibrium size Probability of quasi‐extinction under theMVM scenario was generally greater than that under the MPCscenario This result is a consequence of the fact that the estimateddensity dependence on kitten survival based on the MPC scenario isstronger than that for theMVM scenario (regression coefficients forabundance for the MPC scenario= minus0068 vs minus0002 for theMVM scenario Appendix B Table B1) Consequently simulatedpopulations under the MPC scenario have a stronger tendency tomove toward the equilibrium population size which inherentlyreduces the quasi‐extinction probability (eg Royama 1992)As expected for long‐lived mammals (Heppell et al 2000 Oli

and Dobson 2003) the population growth rate was proportio-nately most sensitive to changes in survival of prime adultsfollowed by that in younger age classes Global sensitivity ana-lysis in contrast showed that under all scenarios extinctionparameters were particularly sensitive to changes in survival ofFlorida panther kittens This result is a consequence of the highsensitivity of the population growth rate to kitten survival(Fig 9) and a larger standard error of kitten survival (Table 3)Results of our sensitivity analyses indicate that future researchshould focus on acquiring additional information on early lifestages to improve the accuracy and precision of estimates ofkitten survival Our results suggest that demographic variables(or their environmental drivers) that strongly influence extinc-tion risks are not necessarily those with the largest elasticityvalues A study of island fox (Bakker et al 2009) found thatextinction risk was strongly influenced by survival modifierparameters that had low elasticity values

Genetic Management When and How to InterveneThe use of genetic introgression as a management tool has al-ways been controversial and the probable effectiveness it mighthave in preventing the imminent demise of the panther popu-lation was initially debated (Creel 2006 Maehr et al 2006Pimm et al 2006) These debates notwithstanding the pantherpopulation had suffered from genetic morphological and bio-medical correlates of inbreeding before this management in-tervention (Johnson et al 2010 Onorato et al 2010) whereasthe post‐introgression period has been characterized by a sub-stantial reduction in the biomedical correlates of inbreeding andimprovements in demographic vigor and abundance (McBride

28 Wildlife Monographs bull 203

et al 2008 Hostetler et al 2010 2013 Johnson et al 2010Benson et al 2011 McClintock et al 2015) Nevertheless thepopulation remains small and isolated habitat is still being lostand there is no current possibility of natural gene flow betweenthis and other North American puma populations thus therecurrence of inbreeding‐related problems and subsequent po-pulation declines are inevitable The question therefore is notwhether genetic management of the Florida panther populationis needed but when and how it should be implementedWithout genetic introgression probabilities of quasi‐extinc-

tion were substantially greater than those from models thatincorporated periodic genetic introgression events (Fig 15)highlighting the importance of genetic management in smallisolated populations When 5 female pumas are added to theFlorida panther population every 10 years genetic variationincreased substantially under the MPC and MVM scenariosThe IBM predicted that the equillibrium population size wouldbe substantially larger when 5 pumas were released into thepopulation every 10 years than populations without intervention(ie no introgression) Releasing 15 Texas females did not af-ford substantially greater benefit to the equilibrium populationsize than did releasing only 5 Texas females Instead addingmore individuals caused increasingly large fluctuations in po-pulation size due to the numerical increases and density de-pendence (Fig 14) possibly diminishing the benefits associatedwith increased genetic variationCompared with the no‐intervention scenario (ie no genetic

introgression implemented) the introduction of 5 female pumasevery 10 years reduced quasi‐extinction probabilities from 13to 4 and from 17 to 6 for the MPC and MVM scenariosrespectively The benefit of genetic‐introgression strategies de-creased as the interval between successive management inter-ventions increased and no benefit was evident once the intervalreached 80 years (Fig 15) Introgression reduced the averagetime until quasi‐extinction (Fig 18) This somewhat counter‐intuitive result can be explained by the fact that repeated geneticintrogression makes the population more robust over timewhich will reduce the probability of extinction Consequentlyfew simulated population trajectories that fall below the criticalthreshold do so early reducing the mean time to extinctionAlso density dependence has a stabilizing effect on populations(Royama 1992) populations tend toward equilibrium popula-tion sizes which reduces the probability over time of popula-tions falling below quasi‐extinction threshold However timeuntil quasi‐extinction must be interpreted in conjunction withthe probability of quasi‐extinction which is very low for allscenarios with genetic introgression (Fig 15)Cost is a significant impediment to the implementation of a

genetic introgression program because captures relocations andmonitoring efforts before and after introgression are expensive(Edmands 2007 Frankham 2010b 2015 2016) Costs may beacceptable to society if the management actions result in sub-stantial fitness benefits especially if the species of concern ispopular and charismatic (Frankham 2015) We estimated the100‐year cost of introgression strategies modeled here to rangefrom $50000 to $12 million More expensive strategies gen-erally led to larger decreases in the probability of quasi‐extinc-tion but cheaper strategies also substantially improved

population viability (ie reduced quasi‐extinction probabilityTable 6) One of the cheaper strategies (5 pumas released every20 years) which cost an estimated $200000 over 100 yearsreduced the probability of quasi‐extinction by 59 and 44 forthe MPC and MVM scenarios respectively But these pre-liminary cost estimates are based on expenses incurred duringthe 1995 introgression and include costs of capture quarantinetransportation release and post‐release monitoring of femalesonly Although the cost for future genetic interventions wouldlikely be higher than those reported here the relative differencesin cost between alternative intervention strategies should remainapproximately the sameOur ability to simulate genetics and link it to the viability of

the Florida panther population is based on the empirically es-timated relationship between individual heterozygosity andsurvival Such heterozygosity and fitness correlations have beenstudied for decades (David 1998) but have only recently beenused to predict population viability (Benson et al 2016) noneto our knowledge have incorporated cost into their modelingefforts The mechanisms underlying heterozygosity and fitnesscorrelations remain unclear (David 1998 Szulkin et al 2010)and their significance as a proxy for inbreeding depression hasbeen debated (Balloux et al 2004 Miller and Coltman 2014)Nevertheless evidence continues to mount demonstratingpositive relationships between heterozygosity and survival(Coulson et al 1998ab Hansson et al 2004 Silva et al 2005Acevedo‐Whitehouse et al 2006 Jensen et al 2007 Velandoet al 2015) and between heterozygosity and fecundity (Seddonet al 2004 Charpentier et al 2005 Agudo et al 2012 Velandoet al 2015) Although the reported correlations are generallyweak their consequences can be substantial if population growthand persistence parameters are highly sensitive to the affectedvital rates (Velando et al 2015) Compared with scenarios thatignore genetics incorporating the effects of inbreeding on sur-vival increased quasi‐extinction probabilities within a 100‐yeartimeframe from 15 to 13 and from 14 to 17 for theMPC and MVM scenarios respectively These results high-light the importance of incorporating genetics when analyzingthe viability of small isolated populationsAlthough conservation biologists have recognized the im-

portance of genetics in population viability many still fail toincorporate genetic factors into PVA models (Reed et al 2002)Explicit consideration of genetics in PVAs requires long‐termgenetic and demographic data This permits the assessment ofthe functional relationship between genetic variability and de-mographic parameters Population viability analysis softwarepackages such as VORTEX (Lacy 1993 2000) offer a powerfulframework for individual‐based simulations but they often donot adequately capture a speciesrsquo life history or genetics Basedon a thorough review of the importance of genetic factors inwildlife conservation Frankham et al (2014) concluded thatgenetic factors can strongly influence the dynamics and persis-tence of small andor fragmented populations They furthernoted that ldquoMost PVA‐based risk assessments ignore or in-adequately model genetic factorsrdquo and recommended that ldquoPVAshould routinely include realistic inbreeding depressionhelliprdquo(Frankham et al 201457) Our custom‐built IBM permitted usto develop a model that adequately captured the Florida panther

van de Kerk et al bull Florida Panther Population and Genetic Viability 29

life history and we could parameterize the model with our long‐term genetic and demographic dataThe explicit consideration of the effects of inbreeding on

Florida panther population dynamics and persistence representsan important step forward Nonetheless our model remainsimperfect First we neglected the possible effects of habitat lossdetrimental anthropogenic activities climate change the prob-ability of immigration of pumas from western populationsdisease or catastrophes on demographic rates and populationviability Although immigration from puma populations outsideFlorida is possible its probability is very low given the extensivechallenges the anthropogenically altered landscape would posefor such a migrant to make it successfully to South Florida Weomitted mutations from our model because we have no in-formation on mutation rates though we expect that they are lowbased on values reported for other mammals (Kumar and Sub-ramanian 2002) Finally we only considered possible effects ofinbreeding on age‐specific survival rates because there was noevidence that heterozygosity affected female reproduction Lossof heterozygosity can negatively affect reproduction becauseinbred male panthers can suffer from cryptorchidism which canadversely affect their fecundity However given the polygynousmating system we expect the Florida panther population growthis primarily female‐limitedAlthough it is generally accepted that genetic introgression

only temporarily relieves a population from inbreeding depres-sion we know of no cases in which it has been implementedmore than once to conserve a threatened wildlife populationOur study provides an example of how demographic and mo-lecular data can be used within an IBM framework to evaluatethe efficacy of alternative genetic management strategies via theFlorida panther as a case study Our model can be adjusted forand applied to other species and offers great potential as a toolfor genetic management of small isolated wildlife populations

MANAGEMENT IMPLICATIONS

Our results and those of Hostetler et al (2013) and Johnsonet al (2010) provide persuasive evidence that the genetic in-tervention implemented in 1995 prevented the demise of theFlorida panther and restored demographic vigor to the popu-lation The Florida panther population remains small and iso-lated from other puma populations the closest puma populationis located gt1000 km away in Texas and the intervening land-scape is highly developed and fragmented with many interstate(and other high‐traffic) highways and major urban centersTheory suggests that 1 migrant per generation is sufficient tominimize the loss of genetic diversity (eg Mills and Allendorf1996) However the possibility of successful migration of apuma to South Florida followed by successful mating every ~5years is very low considering the distance between Texas andFlorida (Hedrick 1995) and the many risks and barriers thatdispersing pumas face (Beier 1995 Maehr et al 2002 Thatcheret al 2006) Consequently the pantherrsquos long‐term persistencewill likely depend on subsequent timely genetic introgressionsgiven the near‐impossibility of natural gene flow from otherpuma populations To that end our study offers considerableinsights into benefits of different genetic management strategiesand associated costs

Without genetic management the Florida panther populationfaces a substantial risk of quasi‐extinction within 100 years (10ndash46 depending on quasi‐extinction threshold and scenarios)Twenty‐three years have passed since genetic introgression wasimplemented and the population appears healthy as indicatedby measures of genetic variation increases in the populationsize and improvements in age‐specific survival rates Ourfindings suggest that releasing more pumas more frequently doesnot necessarily improve persistence probability because such astrategy would cause larger fluctuations in population size(Fig 14) which negatively affects persistence probability Thusextinction risks faced by inbred populations of large carnivorescan be substantially reduced by releasing approximately 5 im-migrants every 2 decades We suggest that such a managementstrategy be followed only in conjunction with continued long‐term monitoring of the population Our analyses demonstratethat population growth and persistence are highly sensitive tokitten and female (adult and subadult) survival Continuing tocollect data on these demographic variables along with bio-metric and genetic sampling will remain important to pantherrecovery and vigilance in identifying issues associated with in-breeding depression The collection of these monitoring datashould allow managers to detect any demographic or geneticchanges in a timely fashion and respond with genetic in-trogression or other management actions if warrantedRecovery objectives as defined by the USFWS in its current

recovery plan (USFWS 2008) delineate the need for additionalpopulations in the historical range to downlist or delist theFlorida panther If the only extant population of Floridapanthers continued to grow it could serve as a source ofpanthers to be translocated to suitable habitat to seed addi-tional populations Maintaining current information on thegenetic health of the population and precise estimates of itssize will be important in assessing whether it can withstand theremoval of a subset of animals for translocation Although the2017 documentation of a female panther with kittens north ofthe current breeding range is encouraging in terms of thepotential for population expansionmdashgiven that no female hadbeen recorded north of the Caloosahatchee River since 1972mdashthe re‐establishment of separate new populations will mostlikely require translocations by wildlife managers and a lengthysociopolitical processConservation of threatened or endangered species such as the

Florida panther is inherently challenging For panthers and otherlarge carnivores that require vast extents of natural habitat topersist habitat is typically the most limiting factor that will de-termine whether recovery efforts succeed Although geneticmanagement invariably plays a key role in the long‐term persis-tence of the panther population even a healthy population caneventually succumb to the impacts of habitat loss The humanpopulation of Florida continues to be one of the fastest‐growingin the nation the United States Census Bureau currently ranksFlorida as the third most populous Therefore habitat loss isgoing to continue to be a key issue for panthers Diligence towardpreserving panther habitat in its historical range via conservationeasements or other means and trying to keep such areas inter-connected via corridors is a goal stakeholders such as governmentagencies sportsmen non‐governmental organizations ranchers

30 Wildlife Monographs bull 203

and other private landowners must work on collaboratively toachieve

ACKNOWLEDGMENTS

We thank D Shindle D Jennings T OrsquoMeara D Land andC Belden for valuable advice throughout the project We thankS Bass M Criffield M Cunningham D Giardina D JansenA Johnson D Land M Lotz C McBride Rocky McBrideRowdy McBride Roy McBride L Oberhofer S Schulze staffsat Everglades National Park and Big Cypress National Preserveand others for assistance with fieldwork We also thank JAustin M Conroy J Hines J Laake S Mills J Nichols JHines M Sunquist J Fryxell and T Coulson for advice andassistance regarding data analysis and panther biology TheMuseum of Texas Tech University Natural Science ResearchLaboratory provided samples from pumas from west Texas forcomparative genetic analyses M Ben‐David D Land WMcCown E Hellgren and 4 anonymous reviewers providedmany helpful comments on the manuscript We thank NereaUbierna Lopez for completing the Spanish translation of theabstract We are grateful to the Department of ResearchComputing University of Florida for excellent high‐perfor-mance computing services We would like to thank US Fishand Wildlife Service Florida Fish and Wildlife ConservationCommission (FWC) US National Park Service QuantitativeSpatial Ecology Evolution and Environment (QSE3) In-tegrative Graduate Education and Research Traineeship(IGERT) program funded by the National Science Foundationand the University of Florida for financially supporting thisstudy Funding for panther research conducted by the FWC issupported predominantly via donations accrued with vehicleregistrations for ldquoProtect the Pantherrdquo license plates

LITERATURE CITEDAcevedo‐Whitehouse K T R Spraker E Lyons S R Melin F Gulland RL Delong and W Amos 2006 Contrasting effects of heterozygosity onsurvival and hookworm resistance in California sea lion pups MolecularEcology 151973ndash1982

Adams J R L M Vucetich P W Hedrick R O Peterson and J AVucetich 2011 Genomic sweep and potential genetic rescue during limitingenvironmental conditions in an isolated wolf population Proceedings of theRoyal Society B Biological Sciences 2783336ndash3344

Agresti A 2013 Categorical data analysis Third edition John Wiley amp SonsHoboken New Jersey USA

Agudo R M Carrete M Alcaide C Rico F Hiraldo and J A Donaacutezar2012 Genetic diversity at neutral and adaptive loci determines individualfitness in a long‐lived territorial bird Proceedings of the Royal Society BBiological Sciences 2793241ndash3249

Albeke S E N P Nibbelink and M Ben‐David 2015 Modeling behavior bycoastal river otter (Lontra canadensis) in response to prey availability in PrinceWilliam Sound Alaska a spatially‐explicit individual‐based approach PLOSOne 10e0126208

Alho J S K Vaumllimaumlki and J Merilauml 2010 Rhh an R extension for estimatingmultilocus heterozygosity and heterozygosity‐heterozygosity correlationMolecular Ecology Resources 10720ndash722

Alvarez K 1993 Twilight of the panther Myakka River Publishing SarasotaFlorida USA

Aparicio J M J Ortego and P J Cordero 2006 What should we weigh toestimate heterozygosity alleles or loci Molecular Ecology 154659ndash4665

Bakker V J D F Doak G W Roemer D K Garcelon T J Coonan S AMorrison C Lynch K Ralls and R Shaw 2009 Incorporating ecologicaldrivers and uncertainty into a demographic population viability analysis for theisland fox Ecological Monographs 7977ndash108

Balloux F W Amos and T Coulson 2004 Does heterozygosity estimateinbreeding in real populations Molecular Ecology 133021ndash3031

Barone M A M E Roelke J Howard J L Brown A E Anderson and DE Wildt 1994 Reproductive characteristics of male Florida panthersmdashcomparative studies from Florida Texas Colorado Latin‐America andNorth‐American Zoos Journal of Mammalogy 75150ndash162

Bateson Z W P O Dunn S D Hull A E Henschen J A Johnson and LA Whittingham 2014 Genetic restoration of a threatened population ofgreater prairie‐chickens Biological Conservation 17412ndash19

Bauer H G Chapron K Nowell P Henschel P Funston L T B HunterD W Macdonald and C Packer 2015 Lion (Panthera leo) populations aredeclining rapidly across Africa except in intensively managed areas Pro-ceedings of National Academy of Sciences of the United States of America11214894ndash14899

Beier P 1995 Dispersal of juvenile cougars in fragmented habitat Journal ofWildlife Management 59228ndash237

Benson J F J A Hostetler D P Onorato W E Johnson M E Roelke SJ OrsquoBrien D Jansen and M K Oli 2011 Intentional genetic introgressioninfluences survival of adults and subadults in a small inbred felid populationJournal of Animal Ecology 80958ndash967

Benson J F P J Mahoney J A Sikich L E K Serieys J P Pollinger H BErnest and S P D Riley 2016 Interactions between demography geneticsand landscape connectivity increase extinction probability for a small popu-lation of large carnivores in a major metropolitan area Proceedings of theRoyal Society B Biological Sciences 28320160957

Beverly F 1874 The Florida panther Forest and Stream 3290Boyce M S C V Haridas C T Lee and the NCEAS Stochastic Demo-graphy Working Group 2006 Demography in an increasingly variable worldTrends in Ecology amp Evolution 21141ndash148

Brook B W J J OrsquoGrady A P Chapman M A Burgman H R Akcakayaand R Frankham 2000 Predictive accuracy of population viability analysis inconservation biology Nature 404385ndash387

Brys R and H Jacquemyn 2016 Severe outbreeding and inbreeding de-pression maintain mating system differentiation in Epipactis (Orchidaceae)Journal of Evolutionary Biology 29352ndash359

Burke J M and M L Arnold 2001 Genetics and the fitness of hybridsAnnual Review of Genetics 3531ndash52

Burnham K P 1993 A theory for combined analysis of ring recovery andrecapture data Pages 199ndash213 in J‐D Lebreton and P M North editorsMarked individuals in the study of bird population Springer‐Verlag NewYork New York USA

Burnham K P and D R Anderson 2002 Model selection and multimodelinference a practical information‐theoretic approach Second edition SpringerVerlag New York New York USA

Caswell H 2001 Matrix population models construction analysis and in-terpretation Sinauer Associates Sunderland Massachusetts USA

Charpentier M J M Setchell F Prugnolle L A Knapp E J Wickings PPeignot and M Hossaert‐McKey 2005 Genetic diversity and reproductivesuccess in mandrills (Mandrillus sphinx) Proceedings of the NationalAcademy of Sciences of the United States of America 10216723ndash16728

Cooch E and G C White editors 2018 Program MARK a gentle in-troduction lthttpwwwphidotorgsoftwaremarkdocsbookgt Accessed 7Apr 2016

Cooley H S R B Wielgus G M Koehler H S Robinson and B TMaletzke 2009 Does hunting regulate cougar populations A test of thecompensatory mortality hypothesis Ecology 902913ndash2921

Coulson T E A Catchpole S D Albon B J Morgan J M Pemberton TH Clutton‐Brock M J Crawley and B T Grenfell 2001 Age sex densitywinter weather and population crashes in Soay sheep Science 2921528ndash1531

Coulson T J Pemberton S Albon M Beaumont T Marshall J Slate FGuinness and T Clutton‐Brock 1998a Microsatellites reveal heterosis in reddeer Proceedings of the Royal Society B Biological Sciences 265489ndash495

Coulson T N S D Albon J M Pemberton J Slate F E Guinness and TH Clutton‐Brock 1998b Genotype by environment interactions in wintersurvival in red deer Journal of Animal Ecology 67434ndash445

Cox D 1972 Regression models and life‐tables Journal of the Royal StatisticalSociety Series B Statistical Methodology 34187ndash220

Creel S 2006 Recovery of the Florida panthermdashgenetic rescue demographic rescueor both Response to Pimm et al (2006) Animal Conservation 9125ndash126

Crnokrak P and D A Roff 1999 Inbreeding depression in the wild Heredity83260ndash270

Crow J 1948 Alternative hypotheses of hybrid vigor Genetics 33477ndash487

van de Kerk et al bull Florida Panther Population and Genetic Viability 31

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 18: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

100 years and 170 (67ndash231) at 200 years based on the MVMscenario (Fig 14) The probabilities of quasi‐extinction underthis introgression strategy were 10 (0ndash99) at 100 years and12 (0ndash99) at 200 years (Fig 15)The projected population size and average individual hetero-

zygosity reached equilibrium levels with periodic fluctuations inthese values when introgression was implemented (Figs 16ndash17)Genetic introgression decreased the time until quasi‐extinctionacross all scenarios (Fig 18) Results of analyses using a criticalthreshold of 30 panthers revealed a similar pattern of relative ben-efits However the probabilities of quasi‐extinction were sub-stantially higher (ge45) across all scenarios (Appendix E)

Addition of pumas from Texas increased both the populationsize and average individual heterozygosity consequentlyintrogression caused periodic increases in average individualheterozygosity and population size at the interval at which theintrogression occurred Whereas average individual hetero-zygosity gradually decreased following introgression densitydependence caused the population size to fluctuate especiallywhen a larger number of pumas from Texas were released Thepositive effect of genetic introgression initiatives on quasi‐extinction probabilities decreased as the time interval betweenthese events increased More frequent genetic introgressions ledto greater increases in average individual heterozygosity and

Old adult

Prime adult

Subadult

Kitten

025 050 075

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Individual heterozygosity

Ann

ual s

urvi

val r

ate

Old adult

Prime adult

Subadult

Kitten

40 60 80 100 120

000

025

050

075

100

04

06

08

10

04

06

08

10

04

06

08

10

Abundance index

Ann

ual s

urvi

val r

ate

Kittens Males Females

Figure 7 The effect of individual heterozygosity (left) and the abundance index (right) on Florida panther age‐ and sex‐specific survival estimates (shaded area isSE) in South Florida USA 1981ndash2013 Abundance index data are from McBride et al (2008) and R T McBride and C McBride (Rancherrsquos Supply Incorporatedunpublished report)

20 Wildlife Monographs bull 203

equilibrium population size while reducing the probability ofquasi‐extinction Comparatively performing genetic introgres-sion less often but with more individuals per release was lesseffective at improving the long‐term prospects for the pantherpopulation (Figs 13ndash15)

Financial cost and persistence benefitsmdashThe total estimated costof 1 genetic introgression was $40000 $80000 or $120000for releasing 5 10 or 15 pumas from Texas respectively(Table 5) The total cost incurred over 100 years for eachstrategy ranged from $50000 to $12 million (Table 6)The most expensive strategy involved the introduction of15 pumas every 10 years this strategy reduced the probability

of quasi‐extinction by 58 and 40 under the MPC and MVMscenarios respectively (Table 6) Introducing 5 pumas every80 years cost the least but improvements in populationpersistence under this strategy were insubstantial (Table 6)Releasing more panthers more frequently caused increasedpopulation fluctuations but did not necessarily reduce the quasi‐extinction probability (Figs 14 and 15 Table 6)

DISCUSSION

The duration (34 yr of data) and sample sizes associated with theFlorida Panther Project allowed us to obtain a comprehensiveunderstanding of genetic variation demographic parameters

A

B

C

Figure 8 Cause‐specific mortality rates (plusmnSE) for male and female Florida panthers (A) subadult prime‐adult and old‐adult Florida panthers of both sexes (B)and canonical and admixed Florida panthers of both sexes (C) in South Florida USA 1981ndash2013 Causes of mortality are vehicle collision intraspecific aggression(ISA) other causes (known causes such as diseases injuries and infections unrelated to the first 2 causes) and unknown cause (mortalities for which we could notassign a cause)

van de Kerk et al bull Florida Panther Population and Genetic Viability 21

probability of population persistence and the duration of thepositive impacts of genetic restoration on this endangered po-pulation The Florida panther population was in dire straits inthe early 1990s and our results clearly demonstrated that the

implementation of the introgression project in 1995 subse-quently accrued a series of genetic and demographic improve-ments that have led to the change from a declining population toone that is growing and expanding its current breeding range

Sensitivity Elasticity

0 1 2 3 00 02 04

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

Para

met

er

Figure 9 Sensitivity and elasticity (proportional sensitivity) of stochastic population growth rate (λs) of the Florida panther to vital demographic parameters in SouthFlorida USA 1981ndash2013 Horizontal lines represent 5th and 95th percentiles

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

70

80

90

100

110

0

5

10

15

0

50

100

Years

StatisticMeanMedian

Figure 10 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the minimum population count(MPC) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

125

150

175

200

00

05

10

15

20

0

30

60

90

Years

StatisticMeanMedian

Figure 11 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the motor vehicle mortality(MVM) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

22 Wildlife Monographs bull 203

MPC MVM

minus08 minus06 minus04 minus02 00 minus08 minus06 minus04 minus02 00

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Male subadult survival

Male prime adult survival

Male old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

PRCC

Para

met

er

Figure 12 Partial rank correlation coefficient (PRCC) between quasi‐extinction probabilities and the demographic parameters of the Florida panther for theminimum population count (MPC) and motor vehicle mortality (MVM) scenarios Error bars indicate standard deviation Based on data collected in South FloridaUSA 1981ndash2013

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 13 Average projected individual heterozygosities in the Florida panther population over 200 years without genetic management intervention and with eachof the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include theintroduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years Average individual heterozygosity peaks when pumas are added to the Floridapanther population Based on data collected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 23

Our research has further validated the success of genetic in-trogression by quantifying the reduction in the probability ofextinction of the panther population because of this manage-ment initiative These findings all bode well for continuedprogress toward recovery That said the Florida panther po-pulation remains small and isolated from other populations ofpuma from western North America thereby denoting that theeventual impact of genetic drift and inbreeding may negativelyaffect the population in the future Realizing this will continueto be an issue that managers will have to contemplate wemodeled the impact of varied genetic management scenarios todetermine the frequency of introgression along with the numberof panthers that should be released during each initiative tominimize cost and maximize benefits to the population Theseresults will prove invaluable to managers attempting to conservethe Florida panther

Genetic Variation Demographic Parametersand Persistence of Heterotic Benefits from GeneticIntrogressionOur measure of individual heterozygosity (1 minusHL) was higheron average for the admixed (0615) panthers than for those ofcanonical ancestry (0386) Levels of individual heterozygosityfor admixed panthers were comparable to levels observed in asample of pumas from west Texas (x plusmn SE= 0695plusmn 0019n= 41) which further demonstrates the improvement in levelsof genetic variation in the Florida population since 1995 The

correlation between HL and several demographic parametershas been noted in wild populations including cheetahs (Terrellet al 2016) and several species of birds such as European shags(Phalacrocorax aristotelis male and female survival probabilityVelando et al 2015) and Egyptian vulture (Neophron percnop-terus age at recruitment Agudo et al 2012) If we focus only onpanthers captured and radiocollared from 2012ndash2015 (FP211ndashFP240) all of which had estimated birth years ge2005 (ge10 yrpost‐introgression) those panthers had an average individualheterozygosity level of 0618plusmn 0029 highlighting the elevatedlevels of genetic variation that continue to persist in cohorts ofpanthers 5 generations after the release of female pumas fromTexasThe proportional membership values (q) from our

STRUCTURE analysis allowed us to delineate 2 clusters ofancestry for Florida panthers (Fig 4) During the pre‐in-trogression era the population was dominated by panthers thatretained a high percentage of the canonical ancestry which wasaffected by inbreeding depression The post‐introgression eraancestry values are comprised of many admixed individuals inessence demonstrating the success of the introgression in termsof increasing genetic variation in the population and mi-micking gene flow that historically occurred with panthers andother populations of pumas (Onorato et al 2010) A somewhatanalogous situation although abetted by natural immigrationwas evident in the ancestral variation in a small population ofpuma intersected by a major freeway in California USA In

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

75

80

85

90

95

100

130

150

170

190

Years

Popu

latio

n si

ze

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 14 Average projected population sizes in the Florida panther population over 200 years without intervention and with each of the genetic introgressionstrategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include the introduction of 5 10 or 15pumas from Texas USA every 10 20 40 or 80 years Peaks occur when pumas are added to the Florida panther population Based on data collected in SouthFlorida USA 1981ndash2013

24 Wildlife Monographs bull 203

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

25

50

75

100

0

25

50

75

100

Introgression interval (yrs)

PQ

E (

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 15 Average probability of quasi‐extinction (PQE) of the Florida panther population within 100 years and within 200 years without genetic managementintervention (N) and with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenariosIntrogression strategies include the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Errorbars indicate 5th and 95th percentiles Based on data collected in South Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

50

60

70

80

90

100

110

50

100

150

200

Years

Popu

latio

n si

ze

Figure 16 Average projected Florida panther population trajectories under a geneticintrogression strategy involving the release of 5 female pumas from Texas USA every20 years for the minimum population count (MPC) and motor vehicle mortality(MVM) scenarios Shaded area indicates 5th and 95th percentiles and isrepresentative of confidence intervals for other scenarios Based on data collected inSouth Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Figure 17 Average projected Florida panther population‐level heterozygositiesunder a genetic introgression strategy involving the release of 5 female pumas fromTexas USA every 20 years for the minimum population count (MPC) and motorvehicle mortality (MVM) scenarios Shaded area bounds the 5th and 95th percentilesand is representative of confidence intervals for other scenarios Based on datacollected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 25

this case the migration of just a single male across the freewayto the south resulted in an increase in the number of in-dividuals with admixed ancestry and substantially improvedgenetic diversity of the subpopulation south of the freeway(Riley et al 2014)Our estimate of overall kitten survival (0321plusmn 0056) was

similar to that reported by Hostetler et al (2010) who used13 years of post‐introgression data (0323plusmn 0071) These valuesare lower than kitten survival estimates derived from severalpuma populations in western North America (044ndash091 Loganand Sweanor 2001 Lambert et al 2006 Laundre et al 2007Robinson et al 2008 Ruth et al 2011) When we included onlydata for individuals that had been genetically sampled our es-timate was 15 higher The proportion of admixed kittens thatwere genetically sampled increased as the population expandedwhich could have resulted in higher estimates of kitten survivalbecause admixed kittens survive better than canonical kittensKitten survival was strongly influenced by genetic ancestry withsurvival rates of F1 kittens being almost twice that of canonicalkittens (Fig 5B) Consistent with the findings of Hostetler et al(2010) increases in heterozygosity coincided with increasedkitten survival whereas population density negatively affectedkitten survival Population density often has a strong negativeimpact on survival of young age classes due to infanticide andcannibalism which has been reported from several carnivore

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

50

100

150

0

50

100

150

Introgression interval (yrs)

TQE

(yrs

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 18 Average times until quasi‐extinction (TQE) for the Florida panther population within 100 years and within 200 years without genetic management interventionand with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategiesinclude the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Error bars indicate 5th and 95thpercentiles Based on data collected in South Florida USA 1981ndash2013

Table 5 Average cost (2017 US dollars) of introgression strategies employedover 100 years that involve the introduction of 5 10 or 15 pumas from TexasUSA into the Florida panther population in South Florida USA at an in-creasingly higher frequency Costs of capture (including transportation) quar-antine and post‐introgression monitoring were estimated by Roy McBrideMark Cunningham and Dave Onorato respectively

Frequency Component 5 pumas 10 pumas 15 pumas

Once Capture 25000 50000 75000

Quarantine 5000 10000 15000

Monitoring 10000 20000 30000

Total 40000 80000 120000

Every 80 years Capture 31250 62500 93750

Quarantine 6250 12500 18750

Monitoring 12500 25000 37500

Total 50000 100000 150000

Every 40 years Capture 62500 125000 187500

Quarantine 12500 25000 37500

Monitoring 25000 50000 75000

Total 100000 200000 300000

Every 20 years Capture 125000 250000 375000

Quarantine 25000 50000 75000

Monitoring 50000 100000 150000

Total 200000 400000 600000

Every 10 years Capture 250000 500000 750000

Quarantine 50000 100000 150000

Monitoring 100000 200000 300000

Total 400000 800000 1200000

26 Wildlife Monographs bull 203

species including polar bears (Ursus maritimus Derocher andWiig 1999) black bears (Ursus americanus Garrison et al 2007)and pumas (Logan and Sweanor 2001)Our estimates of sex‐ and age‐specific survival rates of subadult

and adult panthers (Table 3) and the effects of covariates on theserates were similar to those reported by Benson et al (2011) derivedfrom data collected through 2006 Our survival rates for males(065ndash077) were lower than females (078ndash097) for all 3 ageclasses (subadult prime adult and old adult) which is the typicaltrend observed in most puma populations (eg Ruth et al 19982011) However an unhunted puma population occupying afragmented urban landscape in southern California exhibited lowersurvival (0586 females 0525 males Vickers et al 2015) perhapsas a result of severe habitat fragmentation and the lack of extensiveswaths of public land protected in perpetuity Most populations ofpuma in western North America are typically subject to variedlevels of management via regulated hunting which often is biasedtoward the take of males (Cooley et al 2009 Ruth et al 2011)Consequently annual survival estimates from hunted populationsin northeast Washington (0679ndash0733 females 0341 malesRobinson et al 2008) and southeastern Arizona (0685 females0584 males Cunningham et al 2001) were generally lower thanthose we estimated for Florida panthersCause‐specific mortality analysis showed that male panthers

experienced a substantially greater mortality risk from in-traspecific aggression This differs from the pattern observedduring a long‐term study in Arizona where females were at ahigher risk of intraspecific mortality than males (46 of maleand 53 of female mortality Logan and Sweanor 2001)Mortality associated with intraspecific strife has been docu-mented in hunted and unhunted populations (this study Loganand Sweanor 2001 Stoner et al 2010) and its root cause hasbeen difficult to determine (eg population density and preypopulation trends) That being the case finding ways to reducewhat is often a major source of mortality for puma populationscontinues to pose a challenge to managersCanonical panthers were substantially more likely to die from

intraspecific aggression than were admixed panthers This mayat least partly explain the difference in survival between admixed

and canonical panthers Canonical panthers are known to bemore likely to suffer from varied correlates of inbreeding de-pression than admixed animals including atrial septal defects(Johnson et al 2010) and compromised immune systems(Roelke et al 1993) These factors have the potential to affectfitness and perhaps dominance of individuals when engaged inan intraspecific encounter A somewhat similar scenario wasobserved by Hogg et al (2006) in a population of bighorn sheepthat were part of an introgression project Admixed males in thispopulation had a greater probability of paternity than males thatwere less outbred This included coursing males with higherlevels of admixture being markedly more successful at fightingmales that were defending females in estrous (Hogg et al 2006)Heterosis resulting from genetic introgression is expected to be

the strongest in F1 individuals (Shull 1908 Crow 1948 Burkeand Arnold 2001 Waller 2015) and to progressively decline insubsequent generations (Pickup et al 2013 Frankham 2015)Given that admixed (especially F1) panthers survived better thancanonical panthers and that individual heterozygosity improvedsurvival of both sexes and all age classes our data provide evi-dence for heterotic advantage whereby individuals of mixedancestry had higher fitness (Shull 1908 Crow 1948) Our resultsare consistent with findings of other studies that involved in-tentional genetic introgression for conservation purposes Forexample Hogg et al (2006) detected substantial improvementsin survival and reproduction of introgressed bighorn sheep andMadsen et al (1999 2004) reported a dramatic increase in thepopulation of adders after genetic introgressionKnowledge of the persistence of benefits to a population ac-

crued via genetic introgression is essential for determiningwhen additional introgression may be warranted There is adepauperate amount of information regarding this topic and theidentification of factors that may influence persistence of thebenefits of genetic introgression (Tallmon et al 2004 Hedricket al 2014 Whiteley et al 2015) For example in minks(Neovison vison) Thirstrup et al (2014) found that outcrossingled to larger litters in F2 and F3 but this benefit disappeared insubsequent generations In a population of gray wolves Hedricket al (2014) suggested that benefits of genetic introgression may

Table 6 Costs and benefits accumulated over 100 years for genetic introgression at different intervals and with different numbers of pumas from Texas USAintroduced into the Florida panther population in South Florida USA Costs (2017 US dollars) include capture and transportation quarantine and post‐introgression monitoring Benefits are represented as average probability of quasi‐extinction (PQE and 5th and 95th percentiles) and changes (Δ) therein under thescenario using the minimum population count (McBride et al 2008 MPC) and the scenario using the motor vehicle mortality population estimates (McClintocket al 2015 MVM) The table is sorted by percent change in probability of quasi‐extinction under the MPC scenario

Interval (yr) Pumas Cost MPC PQE () ΔMPC PQE () MVM PQE () ΔMVM PQE ()

ndash 0 0 13 (0ndash99) 0 17 (0ndash100) 080 10 100000 12 (0ndash99) 10 (0ndash58) 16 (0ndash100) 5 (0ndash38)80 15 150000 12 (0ndash99) 13 (0ndash65) 15 (0ndash100) 8 (0ndash56)80 5 50000 12 (0ndash99) 14 (0ndash73) 15 (0ndash99) 9 (0ndash41)40 15 300000 10 (0ndash98) 26 (0ndash104) 14 (0ndash99) 13 (0ndash60)40 10 200000 9 (0ndash94) 35 (0ndash183) 13 (0ndash99) 21 (0ndash95)40 5 100000 8 (0ndash89) 39 (0ndash211) 12 (0ndash99) 26 (0ndash124)20 15 600000 8 (0ndash88) 42 (0ndash257) 13 (0ndash99) 24 (0ndash123)20 10 400000 6 (0ndash67) 54 (0ndash318) 10 (0ndash99) 37 (0ndash217)10 15 1200000 6 (0ndash53) 58 (0ndash372) 10 (0ndash99) 40 (0ndash208)20 5 200000 5 (0ndash50) 59 (0ndash338) 9 (0ndash99) 44 (0ndash352)10 10 800000 4 (0ndash16) 69 (0ndash489) 8 (0ndash92) 55 (0ndash401)10 5 400000 4 (0ndash7) 73 (0ndash502) 6 (0ndash71) 63 (0ndash503)

van de Kerk et al bull Florida Panther Population and Genetic Viability 27

wane after 2 or 3 generations The WrightndashFisher model ofgenetic drift predicts a geometric decline in expected hetero-zygosity at the rate of (1 minus 12Ne) per generation where Ne is theeffective population size (Hartl and Clark 1997 Frankham et al2014) Thus it seems logical to assume that the duration of thebenefits of introgression is limited especially in a species per-sisting in a small population such as Florida panthersWe expected Florida panther survival in the most recent years

(2005ndash2013) post‐introgression to differ from that observed in thedecade following the introgression in 1995 because pantherabundance and heterozygosity factors known to influence panthersurvival (Hostetler et al 2010 Benson et al 2011) have changed(Figs 2 and 6) We found that subadult and adult survival rates inrecent years (2005ndash2013) were similar to those in the period im-mediately following introgression (1995ndash2004 Fig 5C) overallkitten survival was similar to estimates of Hostetler et al (2010) Infact the pattern of covariate effects on Florida panther survivalreported by Benson et al (2011) and Hostetler et al (2010) hasbarely changed The negative effects of population density andpositive effects of heterozygosity may counterbalance survival ratesreducing population fluctuations Our result that benefits of geneticrestoration have remained in the population nearly for 5 genera-tions post‐intervention was surprising but also encouraging Pos-sible explanations for this observation may include 1) genetic ero-sion occurs at slower rate than previously thought 2) introducing 5migrants in 1 genetic intervention may have beneficial effects si-milar to those from 1 migrant per generation over multiple gen-erations or 3) other and as yet unknown factors may reduce therate of genetic erosion and subsequent demographic effects Adensity‐dependent effect on kitten survival could also explain whynon‐F1 admixed kittens did not survive substantially better thandid canonical kittens most kittens sampled from 2005ndash2013 wereadmixed and their survival was lower possibly because of a density‐dependent effect as the Florida panther population continuouslygrew during that period Trinkel et al (2010) also found thatpopulation density and inbreeding coefficient interacted to affectdemographic parameters and growth rate of an African lion po-pulation Likewise population density interacted with winterweather to affect the survival and population growth rates in Soaysheep (Ovis aries Milner et al 1999 Coulson et al 2001)

Population Dynamics and PersistenceThe finite deterministic and stochastic growth rates were gt10reflecting that much of the data used in this study were collectedfollowing genetic introgression in 1995 during which time thepopulation increased substantially Because our demographicparameter estimates did not change markedly in comparison tothose of Hostetler et al (2013) the similarity between thepopulation growth estimates from these studies was expectedBut these estimates reflect population growth during thedata‐collection period and do not predict future growth In factestimates of population size reported by McClintock et al(2015) suggest that population growth may already be slowing(Fig 2) A similar pattern was observed in an introduced po-pulation of African lion which increased steadily from 1995until 2001 then fluctuated around the presumed carrying ca-pacity thereafter (Trinkel et al 2010) Several other African lionpopulations that experienced various degrees of recovery

subsequently became relatively stable or exhibited decliningtrends (Bauer et al 2015)Our estimates of quasi‐extinction probabilities were lower than

those reported by Hostetler et al (2013) using a 2‐sex age‐structured matrix population model Lower probabilities ofquasi‐extinction than those reported by the earlier study forcomparable scenarios were likely a consequence of the fact thatthe estimates of abundance (McBride et al 2008) and thus thedensity‐dependent effects on survival of kittens and old panthers(gt10 yr) have changed in recent years Additionally these earlyestimates of the probability of quasi‐extinction and of time toquasi‐extinction did not consider genetic erosion that will in-evitably occur without management interventionUnder the MVM scenario the equilibrium population size was

twice that attained under the MPC scenario The MVM popula-tion estimates are approximately twice the MPCs (Fig 2) as aresult the simulated population under the MVM scenario achieveda higher equilibrium size Probability of quasi‐extinction under theMVM scenario was generally greater than that under the MPCscenario This result is a consequence of the fact that the estimateddensity dependence on kitten survival based on the MPC scenario isstronger than that for theMVM scenario (regression coefficients forabundance for the MPC scenario= minus0068 vs minus0002 for theMVM scenario Appendix B Table B1) Consequently simulatedpopulations under the MPC scenario have a stronger tendency tomove toward the equilibrium population size which inherentlyreduces the quasi‐extinction probability (eg Royama 1992)As expected for long‐lived mammals (Heppell et al 2000 Oli

and Dobson 2003) the population growth rate was proportio-nately most sensitive to changes in survival of prime adultsfollowed by that in younger age classes Global sensitivity ana-lysis in contrast showed that under all scenarios extinctionparameters were particularly sensitive to changes in survival ofFlorida panther kittens This result is a consequence of the highsensitivity of the population growth rate to kitten survival(Fig 9) and a larger standard error of kitten survival (Table 3)Results of our sensitivity analyses indicate that future researchshould focus on acquiring additional information on early lifestages to improve the accuracy and precision of estimates ofkitten survival Our results suggest that demographic variables(or their environmental drivers) that strongly influence extinc-tion risks are not necessarily those with the largest elasticityvalues A study of island fox (Bakker et al 2009) found thatextinction risk was strongly influenced by survival modifierparameters that had low elasticity values

Genetic Management When and How to InterveneThe use of genetic introgression as a management tool has al-ways been controversial and the probable effectiveness it mighthave in preventing the imminent demise of the panther popu-lation was initially debated (Creel 2006 Maehr et al 2006Pimm et al 2006) These debates notwithstanding the pantherpopulation had suffered from genetic morphological and bio-medical correlates of inbreeding before this management in-tervention (Johnson et al 2010 Onorato et al 2010) whereasthe post‐introgression period has been characterized by a sub-stantial reduction in the biomedical correlates of inbreeding andimprovements in demographic vigor and abundance (McBride

28 Wildlife Monographs bull 203

et al 2008 Hostetler et al 2010 2013 Johnson et al 2010Benson et al 2011 McClintock et al 2015) Nevertheless thepopulation remains small and isolated habitat is still being lostand there is no current possibility of natural gene flow betweenthis and other North American puma populations thus therecurrence of inbreeding‐related problems and subsequent po-pulation declines are inevitable The question therefore is notwhether genetic management of the Florida panther populationis needed but when and how it should be implementedWithout genetic introgression probabilities of quasi‐extinc-

tion were substantially greater than those from models thatincorporated periodic genetic introgression events (Fig 15)highlighting the importance of genetic management in smallisolated populations When 5 female pumas are added to theFlorida panther population every 10 years genetic variationincreased substantially under the MPC and MVM scenariosThe IBM predicted that the equillibrium population size wouldbe substantially larger when 5 pumas were released into thepopulation every 10 years than populations without intervention(ie no introgression) Releasing 15 Texas females did not af-ford substantially greater benefit to the equilibrium populationsize than did releasing only 5 Texas females Instead addingmore individuals caused increasingly large fluctuations in po-pulation size due to the numerical increases and density de-pendence (Fig 14) possibly diminishing the benefits associatedwith increased genetic variationCompared with the no‐intervention scenario (ie no genetic

introgression implemented) the introduction of 5 female pumasevery 10 years reduced quasi‐extinction probabilities from 13to 4 and from 17 to 6 for the MPC and MVM scenariosrespectively The benefit of genetic‐introgression strategies de-creased as the interval between successive management inter-ventions increased and no benefit was evident once the intervalreached 80 years (Fig 15) Introgression reduced the averagetime until quasi‐extinction (Fig 18) This somewhat counter‐intuitive result can be explained by the fact that repeated geneticintrogression makes the population more robust over timewhich will reduce the probability of extinction Consequentlyfew simulated population trajectories that fall below the criticalthreshold do so early reducing the mean time to extinctionAlso density dependence has a stabilizing effect on populations(Royama 1992) populations tend toward equilibrium popula-tion sizes which reduces the probability over time of popula-tions falling below quasi‐extinction threshold However timeuntil quasi‐extinction must be interpreted in conjunction withthe probability of quasi‐extinction which is very low for allscenarios with genetic introgression (Fig 15)Cost is a significant impediment to the implementation of a

genetic introgression program because captures relocations andmonitoring efforts before and after introgression are expensive(Edmands 2007 Frankham 2010b 2015 2016) Costs may beacceptable to society if the management actions result in sub-stantial fitness benefits especially if the species of concern ispopular and charismatic (Frankham 2015) We estimated the100‐year cost of introgression strategies modeled here to rangefrom $50000 to $12 million More expensive strategies gen-erally led to larger decreases in the probability of quasi‐extinc-tion but cheaper strategies also substantially improved

population viability (ie reduced quasi‐extinction probabilityTable 6) One of the cheaper strategies (5 pumas released every20 years) which cost an estimated $200000 over 100 yearsreduced the probability of quasi‐extinction by 59 and 44 forthe MPC and MVM scenarios respectively But these pre-liminary cost estimates are based on expenses incurred duringthe 1995 introgression and include costs of capture quarantinetransportation release and post‐release monitoring of femalesonly Although the cost for future genetic interventions wouldlikely be higher than those reported here the relative differencesin cost between alternative intervention strategies should remainapproximately the sameOur ability to simulate genetics and link it to the viability of

the Florida panther population is based on the empirically es-timated relationship between individual heterozygosity andsurvival Such heterozygosity and fitness correlations have beenstudied for decades (David 1998) but have only recently beenused to predict population viability (Benson et al 2016) noneto our knowledge have incorporated cost into their modelingefforts The mechanisms underlying heterozygosity and fitnesscorrelations remain unclear (David 1998 Szulkin et al 2010)and their significance as a proxy for inbreeding depression hasbeen debated (Balloux et al 2004 Miller and Coltman 2014)Nevertheless evidence continues to mount demonstratingpositive relationships between heterozygosity and survival(Coulson et al 1998ab Hansson et al 2004 Silva et al 2005Acevedo‐Whitehouse et al 2006 Jensen et al 2007 Velandoet al 2015) and between heterozygosity and fecundity (Seddonet al 2004 Charpentier et al 2005 Agudo et al 2012 Velandoet al 2015) Although the reported correlations are generallyweak their consequences can be substantial if population growthand persistence parameters are highly sensitive to the affectedvital rates (Velando et al 2015) Compared with scenarios thatignore genetics incorporating the effects of inbreeding on sur-vival increased quasi‐extinction probabilities within a 100‐yeartimeframe from 15 to 13 and from 14 to 17 for theMPC and MVM scenarios respectively These results high-light the importance of incorporating genetics when analyzingthe viability of small isolated populationsAlthough conservation biologists have recognized the im-

portance of genetics in population viability many still fail toincorporate genetic factors into PVA models (Reed et al 2002)Explicit consideration of genetics in PVAs requires long‐termgenetic and demographic data This permits the assessment ofthe functional relationship between genetic variability and de-mographic parameters Population viability analysis softwarepackages such as VORTEX (Lacy 1993 2000) offer a powerfulframework for individual‐based simulations but they often donot adequately capture a speciesrsquo life history or genetics Basedon a thorough review of the importance of genetic factors inwildlife conservation Frankham et al (2014) concluded thatgenetic factors can strongly influence the dynamics and persis-tence of small andor fragmented populations They furthernoted that ldquoMost PVA‐based risk assessments ignore or in-adequately model genetic factorsrdquo and recommended that ldquoPVAshould routinely include realistic inbreeding depressionhelliprdquo(Frankham et al 201457) Our custom‐built IBM permitted usto develop a model that adequately captured the Florida panther

van de Kerk et al bull Florida Panther Population and Genetic Viability 29

life history and we could parameterize the model with our long‐term genetic and demographic dataThe explicit consideration of the effects of inbreeding on

Florida panther population dynamics and persistence representsan important step forward Nonetheless our model remainsimperfect First we neglected the possible effects of habitat lossdetrimental anthropogenic activities climate change the prob-ability of immigration of pumas from western populationsdisease or catastrophes on demographic rates and populationviability Although immigration from puma populations outsideFlorida is possible its probability is very low given the extensivechallenges the anthropogenically altered landscape would posefor such a migrant to make it successfully to South Florida Weomitted mutations from our model because we have no in-formation on mutation rates though we expect that they are lowbased on values reported for other mammals (Kumar and Sub-ramanian 2002) Finally we only considered possible effects ofinbreeding on age‐specific survival rates because there was noevidence that heterozygosity affected female reproduction Lossof heterozygosity can negatively affect reproduction becauseinbred male panthers can suffer from cryptorchidism which canadversely affect their fecundity However given the polygynousmating system we expect the Florida panther population growthis primarily female‐limitedAlthough it is generally accepted that genetic introgression

only temporarily relieves a population from inbreeding depres-sion we know of no cases in which it has been implementedmore than once to conserve a threatened wildlife populationOur study provides an example of how demographic and mo-lecular data can be used within an IBM framework to evaluatethe efficacy of alternative genetic management strategies via theFlorida panther as a case study Our model can be adjusted forand applied to other species and offers great potential as a toolfor genetic management of small isolated wildlife populations

MANAGEMENT IMPLICATIONS

Our results and those of Hostetler et al (2013) and Johnsonet al (2010) provide persuasive evidence that the genetic in-tervention implemented in 1995 prevented the demise of theFlorida panther and restored demographic vigor to the popu-lation The Florida panther population remains small and iso-lated from other puma populations the closest puma populationis located gt1000 km away in Texas and the intervening land-scape is highly developed and fragmented with many interstate(and other high‐traffic) highways and major urban centersTheory suggests that 1 migrant per generation is sufficient tominimize the loss of genetic diversity (eg Mills and Allendorf1996) However the possibility of successful migration of apuma to South Florida followed by successful mating every ~5years is very low considering the distance between Texas andFlorida (Hedrick 1995) and the many risks and barriers thatdispersing pumas face (Beier 1995 Maehr et al 2002 Thatcheret al 2006) Consequently the pantherrsquos long‐term persistencewill likely depend on subsequent timely genetic introgressionsgiven the near‐impossibility of natural gene flow from otherpuma populations To that end our study offers considerableinsights into benefits of different genetic management strategiesand associated costs

Without genetic management the Florida panther populationfaces a substantial risk of quasi‐extinction within 100 years (10ndash46 depending on quasi‐extinction threshold and scenarios)Twenty‐three years have passed since genetic introgression wasimplemented and the population appears healthy as indicatedby measures of genetic variation increases in the populationsize and improvements in age‐specific survival rates Ourfindings suggest that releasing more pumas more frequently doesnot necessarily improve persistence probability because such astrategy would cause larger fluctuations in population size(Fig 14) which negatively affects persistence probability Thusextinction risks faced by inbred populations of large carnivorescan be substantially reduced by releasing approximately 5 im-migrants every 2 decades We suggest that such a managementstrategy be followed only in conjunction with continued long‐term monitoring of the population Our analyses demonstratethat population growth and persistence are highly sensitive tokitten and female (adult and subadult) survival Continuing tocollect data on these demographic variables along with bio-metric and genetic sampling will remain important to pantherrecovery and vigilance in identifying issues associated with in-breeding depression The collection of these monitoring datashould allow managers to detect any demographic or geneticchanges in a timely fashion and respond with genetic in-trogression or other management actions if warrantedRecovery objectives as defined by the USFWS in its current

recovery plan (USFWS 2008) delineate the need for additionalpopulations in the historical range to downlist or delist theFlorida panther If the only extant population of Floridapanthers continued to grow it could serve as a source ofpanthers to be translocated to suitable habitat to seed addi-tional populations Maintaining current information on thegenetic health of the population and precise estimates of itssize will be important in assessing whether it can withstand theremoval of a subset of animals for translocation Although the2017 documentation of a female panther with kittens north ofthe current breeding range is encouraging in terms of thepotential for population expansionmdashgiven that no female hadbeen recorded north of the Caloosahatchee River since 1972mdashthe re‐establishment of separate new populations will mostlikely require translocations by wildlife managers and a lengthysociopolitical processConservation of threatened or endangered species such as the

Florida panther is inherently challenging For panthers and otherlarge carnivores that require vast extents of natural habitat topersist habitat is typically the most limiting factor that will de-termine whether recovery efforts succeed Although geneticmanagement invariably plays a key role in the long‐term persis-tence of the panther population even a healthy population caneventually succumb to the impacts of habitat loss The humanpopulation of Florida continues to be one of the fastest‐growingin the nation the United States Census Bureau currently ranksFlorida as the third most populous Therefore habitat loss isgoing to continue to be a key issue for panthers Diligence towardpreserving panther habitat in its historical range via conservationeasements or other means and trying to keep such areas inter-connected via corridors is a goal stakeholders such as governmentagencies sportsmen non‐governmental organizations ranchers

30 Wildlife Monographs bull 203

and other private landowners must work on collaboratively toachieve

ACKNOWLEDGMENTS

We thank D Shindle D Jennings T OrsquoMeara D Land andC Belden for valuable advice throughout the project We thankS Bass M Criffield M Cunningham D Giardina D JansenA Johnson D Land M Lotz C McBride Rocky McBrideRowdy McBride Roy McBride L Oberhofer S Schulze staffsat Everglades National Park and Big Cypress National Preserveand others for assistance with fieldwork We also thank JAustin M Conroy J Hines J Laake S Mills J Nichols JHines M Sunquist J Fryxell and T Coulson for advice andassistance regarding data analysis and panther biology TheMuseum of Texas Tech University Natural Science ResearchLaboratory provided samples from pumas from west Texas forcomparative genetic analyses M Ben‐David D Land WMcCown E Hellgren and 4 anonymous reviewers providedmany helpful comments on the manuscript We thank NereaUbierna Lopez for completing the Spanish translation of theabstract We are grateful to the Department of ResearchComputing University of Florida for excellent high‐perfor-mance computing services We would like to thank US Fishand Wildlife Service Florida Fish and Wildlife ConservationCommission (FWC) US National Park Service QuantitativeSpatial Ecology Evolution and Environment (QSE3) In-tegrative Graduate Education and Research Traineeship(IGERT) program funded by the National Science Foundationand the University of Florida for financially supporting thisstudy Funding for panther research conducted by the FWC issupported predominantly via donations accrued with vehicleregistrations for ldquoProtect the Pantherrdquo license plates

LITERATURE CITEDAcevedo‐Whitehouse K T R Spraker E Lyons S R Melin F Gulland RL Delong and W Amos 2006 Contrasting effects of heterozygosity onsurvival and hookworm resistance in California sea lion pups MolecularEcology 151973ndash1982

Adams J R L M Vucetich P W Hedrick R O Peterson and J AVucetich 2011 Genomic sweep and potential genetic rescue during limitingenvironmental conditions in an isolated wolf population Proceedings of theRoyal Society B Biological Sciences 2783336ndash3344

Agresti A 2013 Categorical data analysis Third edition John Wiley amp SonsHoboken New Jersey USA

Agudo R M Carrete M Alcaide C Rico F Hiraldo and J A Donaacutezar2012 Genetic diversity at neutral and adaptive loci determines individualfitness in a long‐lived territorial bird Proceedings of the Royal Society BBiological Sciences 2793241ndash3249

Albeke S E N P Nibbelink and M Ben‐David 2015 Modeling behavior bycoastal river otter (Lontra canadensis) in response to prey availability in PrinceWilliam Sound Alaska a spatially‐explicit individual‐based approach PLOSOne 10e0126208

Alho J S K Vaumllimaumlki and J Merilauml 2010 Rhh an R extension for estimatingmultilocus heterozygosity and heterozygosity‐heterozygosity correlationMolecular Ecology Resources 10720ndash722

Alvarez K 1993 Twilight of the panther Myakka River Publishing SarasotaFlorida USA

Aparicio J M J Ortego and P J Cordero 2006 What should we weigh toestimate heterozygosity alleles or loci Molecular Ecology 154659ndash4665

Bakker V J D F Doak G W Roemer D K Garcelon T J Coonan S AMorrison C Lynch K Ralls and R Shaw 2009 Incorporating ecologicaldrivers and uncertainty into a demographic population viability analysis for theisland fox Ecological Monographs 7977ndash108

Balloux F W Amos and T Coulson 2004 Does heterozygosity estimateinbreeding in real populations Molecular Ecology 133021ndash3031

Barone M A M E Roelke J Howard J L Brown A E Anderson and DE Wildt 1994 Reproductive characteristics of male Florida panthersmdashcomparative studies from Florida Texas Colorado Latin‐America andNorth‐American Zoos Journal of Mammalogy 75150ndash162

Bateson Z W P O Dunn S D Hull A E Henschen J A Johnson and LA Whittingham 2014 Genetic restoration of a threatened population ofgreater prairie‐chickens Biological Conservation 17412ndash19

Bauer H G Chapron K Nowell P Henschel P Funston L T B HunterD W Macdonald and C Packer 2015 Lion (Panthera leo) populations aredeclining rapidly across Africa except in intensively managed areas Pro-ceedings of National Academy of Sciences of the United States of America11214894ndash14899

Beier P 1995 Dispersal of juvenile cougars in fragmented habitat Journal ofWildlife Management 59228ndash237

Benson J F J A Hostetler D P Onorato W E Johnson M E Roelke SJ OrsquoBrien D Jansen and M K Oli 2011 Intentional genetic introgressioninfluences survival of adults and subadults in a small inbred felid populationJournal of Animal Ecology 80958ndash967

Benson J F P J Mahoney J A Sikich L E K Serieys J P Pollinger H BErnest and S P D Riley 2016 Interactions between demography geneticsand landscape connectivity increase extinction probability for a small popu-lation of large carnivores in a major metropolitan area Proceedings of theRoyal Society B Biological Sciences 28320160957

Beverly F 1874 The Florida panther Forest and Stream 3290Boyce M S C V Haridas C T Lee and the NCEAS Stochastic Demo-graphy Working Group 2006 Demography in an increasingly variable worldTrends in Ecology amp Evolution 21141ndash148

Brook B W J J OrsquoGrady A P Chapman M A Burgman H R Akcakayaand R Frankham 2000 Predictive accuracy of population viability analysis inconservation biology Nature 404385ndash387

Brys R and H Jacquemyn 2016 Severe outbreeding and inbreeding de-pression maintain mating system differentiation in Epipactis (Orchidaceae)Journal of Evolutionary Biology 29352ndash359

Burke J M and M L Arnold 2001 Genetics and the fitness of hybridsAnnual Review of Genetics 3531ndash52

Burnham K P 1993 A theory for combined analysis of ring recovery andrecapture data Pages 199ndash213 in J‐D Lebreton and P M North editorsMarked individuals in the study of bird population Springer‐Verlag NewYork New York USA

Burnham K P and D R Anderson 2002 Model selection and multimodelinference a practical information‐theoretic approach Second edition SpringerVerlag New York New York USA

Caswell H 2001 Matrix population models construction analysis and in-terpretation Sinauer Associates Sunderland Massachusetts USA

Charpentier M J M Setchell F Prugnolle L A Knapp E J Wickings PPeignot and M Hossaert‐McKey 2005 Genetic diversity and reproductivesuccess in mandrills (Mandrillus sphinx) Proceedings of the NationalAcademy of Sciences of the United States of America 10216723ndash16728

Cooch E and G C White editors 2018 Program MARK a gentle in-troduction lthttpwwwphidotorgsoftwaremarkdocsbookgt Accessed 7Apr 2016

Cooley H S R B Wielgus G M Koehler H S Robinson and B TMaletzke 2009 Does hunting regulate cougar populations A test of thecompensatory mortality hypothesis Ecology 902913ndash2921

Coulson T E A Catchpole S D Albon B J Morgan J M Pemberton TH Clutton‐Brock M J Crawley and B T Grenfell 2001 Age sex densitywinter weather and population crashes in Soay sheep Science 2921528ndash1531

Coulson T J Pemberton S Albon M Beaumont T Marshall J Slate FGuinness and T Clutton‐Brock 1998a Microsatellites reveal heterosis in reddeer Proceedings of the Royal Society B Biological Sciences 265489ndash495

Coulson T N S D Albon J M Pemberton J Slate F E Guinness and TH Clutton‐Brock 1998b Genotype by environment interactions in wintersurvival in red deer Journal of Animal Ecology 67434ndash445

Cox D 1972 Regression models and life‐tables Journal of the Royal StatisticalSociety Series B Statistical Methodology 34187ndash220

Creel S 2006 Recovery of the Florida panthermdashgenetic rescue demographic rescueor both Response to Pimm et al (2006) Animal Conservation 9125ndash126

Crnokrak P and D A Roff 1999 Inbreeding depression in the wild Heredity83260ndash270

Crow J 1948 Alternative hypotheses of hybrid vigor Genetics 33477ndash487

van de Kerk et al bull Florida Panther Population and Genetic Viability 31

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 19: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

equilibrium population size while reducing the probability ofquasi‐extinction Comparatively performing genetic introgres-sion less often but with more individuals per release was lesseffective at improving the long‐term prospects for the pantherpopulation (Figs 13ndash15)

Financial cost and persistence benefitsmdashThe total estimated costof 1 genetic introgression was $40000 $80000 or $120000for releasing 5 10 or 15 pumas from Texas respectively(Table 5) The total cost incurred over 100 years for eachstrategy ranged from $50000 to $12 million (Table 6)The most expensive strategy involved the introduction of15 pumas every 10 years this strategy reduced the probability

of quasi‐extinction by 58 and 40 under the MPC and MVMscenarios respectively (Table 6) Introducing 5 pumas every80 years cost the least but improvements in populationpersistence under this strategy were insubstantial (Table 6)Releasing more panthers more frequently caused increasedpopulation fluctuations but did not necessarily reduce the quasi‐extinction probability (Figs 14 and 15 Table 6)

DISCUSSION

The duration (34 yr of data) and sample sizes associated with theFlorida Panther Project allowed us to obtain a comprehensiveunderstanding of genetic variation demographic parameters

A

B

C

Figure 8 Cause‐specific mortality rates (plusmnSE) for male and female Florida panthers (A) subadult prime‐adult and old‐adult Florida panthers of both sexes (B)and canonical and admixed Florida panthers of both sexes (C) in South Florida USA 1981ndash2013 Causes of mortality are vehicle collision intraspecific aggression(ISA) other causes (known causes such as diseases injuries and infections unrelated to the first 2 causes) and unknown cause (mortalities for which we could notassign a cause)

van de Kerk et al bull Florida Panther Population and Genetic Viability 21

probability of population persistence and the duration of thepositive impacts of genetic restoration on this endangered po-pulation The Florida panther population was in dire straits inthe early 1990s and our results clearly demonstrated that the

implementation of the introgression project in 1995 subse-quently accrued a series of genetic and demographic improve-ments that have led to the change from a declining population toone that is growing and expanding its current breeding range

Sensitivity Elasticity

0 1 2 3 00 02 04

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

Para

met

er

Figure 9 Sensitivity and elasticity (proportional sensitivity) of stochastic population growth rate (λs) of the Florida panther to vital demographic parameters in SouthFlorida USA 1981ndash2013 Horizontal lines represent 5th and 95th percentiles

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

70

80

90

100

110

0

5

10

15

0

50

100

Years

StatisticMeanMedian

Figure 10 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the minimum population count(MPC) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

125

150

175

200

00

05

10

15

20

0

30

60

90

Years

StatisticMeanMedian

Figure 11 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the motor vehicle mortality(MVM) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

22 Wildlife Monographs bull 203

MPC MVM

minus08 minus06 minus04 minus02 00 minus08 minus06 minus04 minus02 00

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Male subadult survival

Male prime adult survival

Male old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

PRCC

Para

met

er

Figure 12 Partial rank correlation coefficient (PRCC) between quasi‐extinction probabilities and the demographic parameters of the Florida panther for theminimum population count (MPC) and motor vehicle mortality (MVM) scenarios Error bars indicate standard deviation Based on data collected in South FloridaUSA 1981ndash2013

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 13 Average projected individual heterozygosities in the Florida panther population over 200 years without genetic management intervention and with eachof the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include theintroduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years Average individual heterozygosity peaks when pumas are added to the Floridapanther population Based on data collected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 23

Our research has further validated the success of genetic in-trogression by quantifying the reduction in the probability ofextinction of the panther population because of this manage-ment initiative These findings all bode well for continuedprogress toward recovery That said the Florida panther po-pulation remains small and isolated from other populations ofpuma from western North America thereby denoting that theeventual impact of genetic drift and inbreeding may negativelyaffect the population in the future Realizing this will continueto be an issue that managers will have to contemplate wemodeled the impact of varied genetic management scenarios todetermine the frequency of introgression along with the numberof panthers that should be released during each initiative tominimize cost and maximize benefits to the population Theseresults will prove invaluable to managers attempting to conservethe Florida panther

Genetic Variation Demographic Parametersand Persistence of Heterotic Benefits from GeneticIntrogressionOur measure of individual heterozygosity (1 minusHL) was higheron average for the admixed (0615) panthers than for those ofcanonical ancestry (0386) Levels of individual heterozygosityfor admixed panthers were comparable to levels observed in asample of pumas from west Texas (x plusmn SE= 0695plusmn 0019n= 41) which further demonstrates the improvement in levelsof genetic variation in the Florida population since 1995 The

correlation between HL and several demographic parametershas been noted in wild populations including cheetahs (Terrellet al 2016) and several species of birds such as European shags(Phalacrocorax aristotelis male and female survival probabilityVelando et al 2015) and Egyptian vulture (Neophron percnop-terus age at recruitment Agudo et al 2012) If we focus only onpanthers captured and radiocollared from 2012ndash2015 (FP211ndashFP240) all of which had estimated birth years ge2005 (ge10 yrpost‐introgression) those panthers had an average individualheterozygosity level of 0618plusmn 0029 highlighting the elevatedlevels of genetic variation that continue to persist in cohorts ofpanthers 5 generations after the release of female pumas fromTexasThe proportional membership values (q) from our

STRUCTURE analysis allowed us to delineate 2 clusters ofancestry for Florida panthers (Fig 4) During the pre‐in-trogression era the population was dominated by panthers thatretained a high percentage of the canonical ancestry which wasaffected by inbreeding depression The post‐introgression eraancestry values are comprised of many admixed individuals inessence demonstrating the success of the introgression in termsof increasing genetic variation in the population and mi-micking gene flow that historically occurred with panthers andother populations of pumas (Onorato et al 2010) A somewhatanalogous situation although abetted by natural immigrationwas evident in the ancestral variation in a small population ofpuma intersected by a major freeway in California USA In

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

75

80

85

90

95

100

130

150

170

190

Years

Popu

latio

n si

ze

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 14 Average projected population sizes in the Florida panther population over 200 years without intervention and with each of the genetic introgressionstrategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include the introduction of 5 10 or 15pumas from Texas USA every 10 20 40 or 80 years Peaks occur when pumas are added to the Florida panther population Based on data collected in SouthFlorida USA 1981ndash2013

24 Wildlife Monographs bull 203

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

25

50

75

100

0

25

50

75

100

Introgression interval (yrs)

PQ

E (

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 15 Average probability of quasi‐extinction (PQE) of the Florida panther population within 100 years and within 200 years without genetic managementintervention (N) and with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenariosIntrogression strategies include the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Errorbars indicate 5th and 95th percentiles Based on data collected in South Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

50

60

70

80

90

100

110

50

100

150

200

Years

Popu

latio

n si

ze

Figure 16 Average projected Florida panther population trajectories under a geneticintrogression strategy involving the release of 5 female pumas from Texas USA every20 years for the minimum population count (MPC) and motor vehicle mortality(MVM) scenarios Shaded area indicates 5th and 95th percentiles and isrepresentative of confidence intervals for other scenarios Based on data collected inSouth Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Figure 17 Average projected Florida panther population‐level heterozygositiesunder a genetic introgression strategy involving the release of 5 female pumas fromTexas USA every 20 years for the minimum population count (MPC) and motorvehicle mortality (MVM) scenarios Shaded area bounds the 5th and 95th percentilesand is representative of confidence intervals for other scenarios Based on datacollected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 25

this case the migration of just a single male across the freewayto the south resulted in an increase in the number of in-dividuals with admixed ancestry and substantially improvedgenetic diversity of the subpopulation south of the freeway(Riley et al 2014)Our estimate of overall kitten survival (0321plusmn 0056) was

similar to that reported by Hostetler et al (2010) who used13 years of post‐introgression data (0323plusmn 0071) These valuesare lower than kitten survival estimates derived from severalpuma populations in western North America (044ndash091 Loganand Sweanor 2001 Lambert et al 2006 Laundre et al 2007Robinson et al 2008 Ruth et al 2011) When we included onlydata for individuals that had been genetically sampled our es-timate was 15 higher The proportion of admixed kittens thatwere genetically sampled increased as the population expandedwhich could have resulted in higher estimates of kitten survivalbecause admixed kittens survive better than canonical kittensKitten survival was strongly influenced by genetic ancestry withsurvival rates of F1 kittens being almost twice that of canonicalkittens (Fig 5B) Consistent with the findings of Hostetler et al(2010) increases in heterozygosity coincided with increasedkitten survival whereas population density negatively affectedkitten survival Population density often has a strong negativeimpact on survival of young age classes due to infanticide andcannibalism which has been reported from several carnivore

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

50

100

150

0

50

100

150

Introgression interval (yrs)

TQE

(yrs

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 18 Average times until quasi‐extinction (TQE) for the Florida panther population within 100 years and within 200 years without genetic management interventionand with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategiesinclude the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Error bars indicate 5th and 95thpercentiles Based on data collected in South Florida USA 1981ndash2013

Table 5 Average cost (2017 US dollars) of introgression strategies employedover 100 years that involve the introduction of 5 10 or 15 pumas from TexasUSA into the Florida panther population in South Florida USA at an in-creasingly higher frequency Costs of capture (including transportation) quar-antine and post‐introgression monitoring were estimated by Roy McBrideMark Cunningham and Dave Onorato respectively

Frequency Component 5 pumas 10 pumas 15 pumas

Once Capture 25000 50000 75000

Quarantine 5000 10000 15000

Monitoring 10000 20000 30000

Total 40000 80000 120000

Every 80 years Capture 31250 62500 93750

Quarantine 6250 12500 18750

Monitoring 12500 25000 37500

Total 50000 100000 150000

Every 40 years Capture 62500 125000 187500

Quarantine 12500 25000 37500

Monitoring 25000 50000 75000

Total 100000 200000 300000

Every 20 years Capture 125000 250000 375000

Quarantine 25000 50000 75000

Monitoring 50000 100000 150000

Total 200000 400000 600000

Every 10 years Capture 250000 500000 750000

Quarantine 50000 100000 150000

Monitoring 100000 200000 300000

Total 400000 800000 1200000

26 Wildlife Monographs bull 203

species including polar bears (Ursus maritimus Derocher andWiig 1999) black bears (Ursus americanus Garrison et al 2007)and pumas (Logan and Sweanor 2001)Our estimates of sex‐ and age‐specific survival rates of subadult

and adult panthers (Table 3) and the effects of covariates on theserates were similar to those reported by Benson et al (2011) derivedfrom data collected through 2006 Our survival rates for males(065ndash077) were lower than females (078ndash097) for all 3 ageclasses (subadult prime adult and old adult) which is the typicaltrend observed in most puma populations (eg Ruth et al 19982011) However an unhunted puma population occupying afragmented urban landscape in southern California exhibited lowersurvival (0586 females 0525 males Vickers et al 2015) perhapsas a result of severe habitat fragmentation and the lack of extensiveswaths of public land protected in perpetuity Most populations ofpuma in western North America are typically subject to variedlevels of management via regulated hunting which often is biasedtoward the take of males (Cooley et al 2009 Ruth et al 2011)Consequently annual survival estimates from hunted populationsin northeast Washington (0679ndash0733 females 0341 malesRobinson et al 2008) and southeastern Arizona (0685 females0584 males Cunningham et al 2001) were generally lower thanthose we estimated for Florida panthersCause‐specific mortality analysis showed that male panthers

experienced a substantially greater mortality risk from in-traspecific aggression This differs from the pattern observedduring a long‐term study in Arizona where females were at ahigher risk of intraspecific mortality than males (46 of maleand 53 of female mortality Logan and Sweanor 2001)Mortality associated with intraspecific strife has been docu-mented in hunted and unhunted populations (this study Loganand Sweanor 2001 Stoner et al 2010) and its root cause hasbeen difficult to determine (eg population density and preypopulation trends) That being the case finding ways to reducewhat is often a major source of mortality for puma populationscontinues to pose a challenge to managersCanonical panthers were substantially more likely to die from

intraspecific aggression than were admixed panthers This mayat least partly explain the difference in survival between admixed

and canonical panthers Canonical panthers are known to bemore likely to suffer from varied correlates of inbreeding de-pression than admixed animals including atrial septal defects(Johnson et al 2010) and compromised immune systems(Roelke et al 1993) These factors have the potential to affectfitness and perhaps dominance of individuals when engaged inan intraspecific encounter A somewhat similar scenario wasobserved by Hogg et al (2006) in a population of bighorn sheepthat were part of an introgression project Admixed males in thispopulation had a greater probability of paternity than males thatwere less outbred This included coursing males with higherlevels of admixture being markedly more successful at fightingmales that were defending females in estrous (Hogg et al 2006)Heterosis resulting from genetic introgression is expected to be

the strongest in F1 individuals (Shull 1908 Crow 1948 Burkeand Arnold 2001 Waller 2015) and to progressively decline insubsequent generations (Pickup et al 2013 Frankham 2015)Given that admixed (especially F1) panthers survived better thancanonical panthers and that individual heterozygosity improvedsurvival of both sexes and all age classes our data provide evi-dence for heterotic advantage whereby individuals of mixedancestry had higher fitness (Shull 1908 Crow 1948) Our resultsare consistent with findings of other studies that involved in-tentional genetic introgression for conservation purposes Forexample Hogg et al (2006) detected substantial improvementsin survival and reproduction of introgressed bighorn sheep andMadsen et al (1999 2004) reported a dramatic increase in thepopulation of adders after genetic introgressionKnowledge of the persistence of benefits to a population ac-

crued via genetic introgression is essential for determiningwhen additional introgression may be warranted There is adepauperate amount of information regarding this topic and theidentification of factors that may influence persistence of thebenefits of genetic introgression (Tallmon et al 2004 Hedricket al 2014 Whiteley et al 2015) For example in minks(Neovison vison) Thirstrup et al (2014) found that outcrossingled to larger litters in F2 and F3 but this benefit disappeared insubsequent generations In a population of gray wolves Hedricket al (2014) suggested that benefits of genetic introgression may

Table 6 Costs and benefits accumulated over 100 years for genetic introgression at different intervals and with different numbers of pumas from Texas USAintroduced into the Florida panther population in South Florida USA Costs (2017 US dollars) include capture and transportation quarantine and post‐introgression monitoring Benefits are represented as average probability of quasi‐extinction (PQE and 5th and 95th percentiles) and changes (Δ) therein under thescenario using the minimum population count (McBride et al 2008 MPC) and the scenario using the motor vehicle mortality population estimates (McClintocket al 2015 MVM) The table is sorted by percent change in probability of quasi‐extinction under the MPC scenario

Interval (yr) Pumas Cost MPC PQE () ΔMPC PQE () MVM PQE () ΔMVM PQE ()

ndash 0 0 13 (0ndash99) 0 17 (0ndash100) 080 10 100000 12 (0ndash99) 10 (0ndash58) 16 (0ndash100) 5 (0ndash38)80 15 150000 12 (0ndash99) 13 (0ndash65) 15 (0ndash100) 8 (0ndash56)80 5 50000 12 (0ndash99) 14 (0ndash73) 15 (0ndash99) 9 (0ndash41)40 15 300000 10 (0ndash98) 26 (0ndash104) 14 (0ndash99) 13 (0ndash60)40 10 200000 9 (0ndash94) 35 (0ndash183) 13 (0ndash99) 21 (0ndash95)40 5 100000 8 (0ndash89) 39 (0ndash211) 12 (0ndash99) 26 (0ndash124)20 15 600000 8 (0ndash88) 42 (0ndash257) 13 (0ndash99) 24 (0ndash123)20 10 400000 6 (0ndash67) 54 (0ndash318) 10 (0ndash99) 37 (0ndash217)10 15 1200000 6 (0ndash53) 58 (0ndash372) 10 (0ndash99) 40 (0ndash208)20 5 200000 5 (0ndash50) 59 (0ndash338) 9 (0ndash99) 44 (0ndash352)10 10 800000 4 (0ndash16) 69 (0ndash489) 8 (0ndash92) 55 (0ndash401)10 5 400000 4 (0ndash7) 73 (0ndash502) 6 (0ndash71) 63 (0ndash503)

van de Kerk et al bull Florida Panther Population and Genetic Viability 27

wane after 2 or 3 generations The WrightndashFisher model ofgenetic drift predicts a geometric decline in expected hetero-zygosity at the rate of (1 minus 12Ne) per generation where Ne is theeffective population size (Hartl and Clark 1997 Frankham et al2014) Thus it seems logical to assume that the duration of thebenefits of introgression is limited especially in a species per-sisting in a small population such as Florida panthersWe expected Florida panther survival in the most recent years

(2005ndash2013) post‐introgression to differ from that observed in thedecade following the introgression in 1995 because pantherabundance and heterozygosity factors known to influence panthersurvival (Hostetler et al 2010 Benson et al 2011) have changed(Figs 2 and 6) We found that subadult and adult survival rates inrecent years (2005ndash2013) were similar to those in the period im-mediately following introgression (1995ndash2004 Fig 5C) overallkitten survival was similar to estimates of Hostetler et al (2010) Infact the pattern of covariate effects on Florida panther survivalreported by Benson et al (2011) and Hostetler et al (2010) hasbarely changed The negative effects of population density andpositive effects of heterozygosity may counterbalance survival ratesreducing population fluctuations Our result that benefits of geneticrestoration have remained in the population nearly for 5 genera-tions post‐intervention was surprising but also encouraging Pos-sible explanations for this observation may include 1) genetic ero-sion occurs at slower rate than previously thought 2) introducing 5migrants in 1 genetic intervention may have beneficial effects si-milar to those from 1 migrant per generation over multiple gen-erations or 3) other and as yet unknown factors may reduce therate of genetic erosion and subsequent demographic effects Adensity‐dependent effect on kitten survival could also explain whynon‐F1 admixed kittens did not survive substantially better thandid canonical kittens most kittens sampled from 2005ndash2013 wereadmixed and their survival was lower possibly because of a density‐dependent effect as the Florida panther population continuouslygrew during that period Trinkel et al (2010) also found thatpopulation density and inbreeding coefficient interacted to affectdemographic parameters and growth rate of an African lion po-pulation Likewise population density interacted with winterweather to affect the survival and population growth rates in Soaysheep (Ovis aries Milner et al 1999 Coulson et al 2001)

Population Dynamics and PersistenceThe finite deterministic and stochastic growth rates were gt10reflecting that much of the data used in this study were collectedfollowing genetic introgression in 1995 during which time thepopulation increased substantially Because our demographicparameter estimates did not change markedly in comparison tothose of Hostetler et al (2013) the similarity between thepopulation growth estimates from these studies was expectedBut these estimates reflect population growth during thedata‐collection period and do not predict future growth In factestimates of population size reported by McClintock et al(2015) suggest that population growth may already be slowing(Fig 2) A similar pattern was observed in an introduced po-pulation of African lion which increased steadily from 1995until 2001 then fluctuated around the presumed carrying ca-pacity thereafter (Trinkel et al 2010) Several other African lionpopulations that experienced various degrees of recovery

subsequently became relatively stable or exhibited decliningtrends (Bauer et al 2015)Our estimates of quasi‐extinction probabilities were lower than

those reported by Hostetler et al (2013) using a 2‐sex age‐structured matrix population model Lower probabilities ofquasi‐extinction than those reported by the earlier study forcomparable scenarios were likely a consequence of the fact thatthe estimates of abundance (McBride et al 2008) and thus thedensity‐dependent effects on survival of kittens and old panthers(gt10 yr) have changed in recent years Additionally these earlyestimates of the probability of quasi‐extinction and of time toquasi‐extinction did not consider genetic erosion that will in-evitably occur without management interventionUnder the MVM scenario the equilibrium population size was

twice that attained under the MPC scenario The MVM popula-tion estimates are approximately twice the MPCs (Fig 2) as aresult the simulated population under the MVM scenario achieveda higher equilibrium size Probability of quasi‐extinction under theMVM scenario was generally greater than that under the MPCscenario This result is a consequence of the fact that the estimateddensity dependence on kitten survival based on the MPC scenario isstronger than that for theMVM scenario (regression coefficients forabundance for the MPC scenario= minus0068 vs minus0002 for theMVM scenario Appendix B Table B1) Consequently simulatedpopulations under the MPC scenario have a stronger tendency tomove toward the equilibrium population size which inherentlyreduces the quasi‐extinction probability (eg Royama 1992)As expected for long‐lived mammals (Heppell et al 2000 Oli

and Dobson 2003) the population growth rate was proportio-nately most sensitive to changes in survival of prime adultsfollowed by that in younger age classes Global sensitivity ana-lysis in contrast showed that under all scenarios extinctionparameters were particularly sensitive to changes in survival ofFlorida panther kittens This result is a consequence of the highsensitivity of the population growth rate to kitten survival(Fig 9) and a larger standard error of kitten survival (Table 3)Results of our sensitivity analyses indicate that future researchshould focus on acquiring additional information on early lifestages to improve the accuracy and precision of estimates ofkitten survival Our results suggest that demographic variables(or their environmental drivers) that strongly influence extinc-tion risks are not necessarily those with the largest elasticityvalues A study of island fox (Bakker et al 2009) found thatextinction risk was strongly influenced by survival modifierparameters that had low elasticity values

Genetic Management When and How to InterveneThe use of genetic introgression as a management tool has al-ways been controversial and the probable effectiveness it mighthave in preventing the imminent demise of the panther popu-lation was initially debated (Creel 2006 Maehr et al 2006Pimm et al 2006) These debates notwithstanding the pantherpopulation had suffered from genetic morphological and bio-medical correlates of inbreeding before this management in-tervention (Johnson et al 2010 Onorato et al 2010) whereasthe post‐introgression period has been characterized by a sub-stantial reduction in the biomedical correlates of inbreeding andimprovements in demographic vigor and abundance (McBride

28 Wildlife Monographs bull 203

et al 2008 Hostetler et al 2010 2013 Johnson et al 2010Benson et al 2011 McClintock et al 2015) Nevertheless thepopulation remains small and isolated habitat is still being lostand there is no current possibility of natural gene flow betweenthis and other North American puma populations thus therecurrence of inbreeding‐related problems and subsequent po-pulation declines are inevitable The question therefore is notwhether genetic management of the Florida panther populationis needed but when and how it should be implementedWithout genetic introgression probabilities of quasi‐extinc-

tion were substantially greater than those from models thatincorporated periodic genetic introgression events (Fig 15)highlighting the importance of genetic management in smallisolated populations When 5 female pumas are added to theFlorida panther population every 10 years genetic variationincreased substantially under the MPC and MVM scenariosThe IBM predicted that the equillibrium population size wouldbe substantially larger when 5 pumas were released into thepopulation every 10 years than populations without intervention(ie no introgression) Releasing 15 Texas females did not af-ford substantially greater benefit to the equilibrium populationsize than did releasing only 5 Texas females Instead addingmore individuals caused increasingly large fluctuations in po-pulation size due to the numerical increases and density de-pendence (Fig 14) possibly diminishing the benefits associatedwith increased genetic variationCompared with the no‐intervention scenario (ie no genetic

introgression implemented) the introduction of 5 female pumasevery 10 years reduced quasi‐extinction probabilities from 13to 4 and from 17 to 6 for the MPC and MVM scenariosrespectively The benefit of genetic‐introgression strategies de-creased as the interval between successive management inter-ventions increased and no benefit was evident once the intervalreached 80 years (Fig 15) Introgression reduced the averagetime until quasi‐extinction (Fig 18) This somewhat counter‐intuitive result can be explained by the fact that repeated geneticintrogression makes the population more robust over timewhich will reduce the probability of extinction Consequentlyfew simulated population trajectories that fall below the criticalthreshold do so early reducing the mean time to extinctionAlso density dependence has a stabilizing effect on populations(Royama 1992) populations tend toward equilibrium popula-tion sizes which reduces the probability over time of popula-tions falling below quasi‐extinction threshold However timeuntil quasi‐extinction must be interpreted in conjunction withthe probability of quasi‐extinction which is very low for allscenarios with genetic introgression (Fig 15)Cost is a significant impediment to the implementation of a

genetic introgression program because captures relocations andmonitoring efforts before and after introgression are expensive(Edmands 2007 Frankham 2010b 2015 2016) Costs may beacceptable to society if the management actions result in sub-stantial fitness benefits especially if the species of concern ispopular and charismatic (Frankham 2015) We estimated the100‐year cost of introgression strategies modeled here to rangefrom $50000 to $12 million More expensive strategies gen-erally led to larger decreases in the probability of quasi‐extinc-tion but cheaper strategies also substantially improved

population viability (ie reduced quasi‐extinction probabilityTable 6) One of the cheaper strategies (5 pumas released every20 years) which cost an estimated $200000 over 100 yearsreduced the probability of quasi‐extinction by 59 and 44 forthe MPC and MVM scenarios respectively But these pre-liminary cost estimates are based on expenses incurred duringthe 1995 introgression and include costs of capture quarantinetransportation release and post‐release monitoring of femalesonly Although the cost for future genetic interventions wouldlikely be higher than those reported here the relative differencesin cost between alternative intervention strategies should remainapproximately the sameOur ability to simulate genetics and link it to the viability of

the Florida panther population is based on the empirically es-timated relationship between individual heterozygosity andsurvival Such heterozygosity and fitness correlations have beenstudied for decades (David 1998) but have only recently beenused to predict population viability (Benson et al 2016) noneto our knowledge have incorporated cost into their modelingefforts The mechanisms underlying heterozygosity and fitnesscorrelations remain unclear (David 1998 Szulkin et al 2010)and their significance as a proxy for inbreeding depression hasbeen debated (Balloux et al 2004 Miller and Coltman 2014)Nevertheless evidence continues to mount demonstratingpositive relationships between heterozygosity and survival(Coulson et al 1998ab Hansson et al 2004 Silva et al 2005Acevedo‐Whitehouse et al 2006 Jensen et al 2007 Velandoet al 2015) and between heterozygosity and fecundity (Seddonet al 2004 Charpentier et al 2005 Agudo et al 2012 Velandoet al 2015) Although the reported correlations are generallyweak their consequences can be substantial if population growthand persistence parameters are highly sensitive to the affectedvital rates (Velando et al 2015) Compared with scenarios thatignore genetics incorporating the effects of inbreeding on sur-vival increased quasi‐extinction probabilities within a 100‐yeartimeframe from 15 to 13 and from 14 to 17 for theMPC and MVM scenarios respectively These results high-light the importance of incorporating genetics when analyzingthe viability of small isolated populationsAlthough conservation biologists have recognized the im-

portance of genetics in population viability many still fail toincorporate genetic factors into PVA models (Reed et al 2002)Explicit consideration of genetics in PVAs requires long‐termgenetic and demographic data This permits the assessment ofthe functional relationship between genetic variability and de-mographic parameters Population viability analysis softwarepackages such as VORTEX (Lacy 1993 2000) offer a powerfulframework for individual‐based simulations but they often donot adequately capture a speciesrsquo life history or genetics Basedon a thorough review of the importance of genetic factors inwildlife conservation Frankham et al (2014) concluded thatgenetic factors can strongly influence the dynamics and persis-tence of small andor fragmented populations They furthernoted that ldquoMost PVA‐based risk assessments ignore or in-adequately model genetic factorsrdquo and recommended that ldquoPVAshould routinely include realistic inbreeding depressionhelliprdquo(Frankham et al 201457) Our custom‐built IBM permitted usto develop a model that adequately captured the Florida panther

van de Kerk et al bull Florida Panther Population and Genetic Viability 29

life history and we could parameterize the model with our long‐term genetic and demographic dataThe explicit consideration of the effects of inbreeding on

Florida panther population dynamics and persistence representsan important step forward Nonetheless our model remainsimperfect First we neglected the possible effects of habitat lossdetrimental anthropogenic activities climate change the prob-ability of immigration of pumas from western populationsdisease or catastrophes on demographic rates and populationviability Although immigration from puma populations outsideFlorida is possible its probability is very low given the extensivechallenges the anthropogenically altered landscape would posefor such a migrant to make it successfully to South Florida Weomitted mutations from our model because we have no in-formation on mutation rates though we expect that they are lowbased on values reported for other mammals (Kumar and Sub-ramanian 2002) Finally we only considered possible effects ofinbreeding on age‐specific survival rates because there was noevidence that heterozygosity affected female reproduction Lossof heterozygosity can negatively affect reproduction becauseinbred male panthers can suffer from cryptorchidism which canadversely affect their fecundity However given the polygynousmating system we expect the Florida panther population growthis primarily female‐limitedAlthough it is generally accepted that genetic introgression

only temporarily relieves a population from inbreeding depres-sion we know of no cases in which it has been implementedmore than once to conserve a threatened wildlife populationOur study provides an example of how demographic and mo-lecular data can be used within an IBM framework to evaluatethe efficacy of alternative genetic management strategies via theFlorida panther as a case study Our model can be adjusted forand applied to other species and offers great potential as a toolfor genetic management of small isolated wildlife populations

MANAGEMENT IMPLICATIONS

Our results and those of Hostetler et al (2013) and Johnsonet al (2010) provide persuasive evidence that the genetic in-tervention implemented in 1995 prevented the demise of theFlorida panther and restored demographic vigor to the popu-lation The Florida panther population remains small and iso-lated from other puma populations the closest puma populationis located gt1000 km away in Texas and the intervening land-scape is highly developed and fragmented with many interstate(and other high‐traffic) highways and major urban centersTheory suggests that 1 migrant per generation is sufficient tominimize the loss of genetic diversity (eg Mills and Allendorf1996) However the possibility of successful migration of apuma to South Florida followed by successful mating every ~5years is very low considering the distance between Texas andFlorida (Hedrick 1995) and the many risks and barriers thatdispersing pumas face (Beier 1995 Maehr et al 2002 Thatcheret al 2006) Consequently the pantherrsquos long‐term persistencewill likely depend on subsequent timely genetic introgressionsgiven the near‐impossibility of natural gene flow from otherpuma populations To that end our study offers considerableinsights into benefits of different genetic management strategiesand associated costs

Without genetic management the Florida panther populationfaces a substantial risk of quasi‐extinction within 100 years (10ndash46 depending on quasi‐extinction threshold and scenarios)Twenty‐three years have passed since genetic introgression wasimplemented and the population appears healthy as indicatedby measures of genetic variation increases in the populationsize and improvements in age‐specific survival rates Ourfindings suggest that releasing more pumas more frequently doesnot necessarily improve persistence probability because such astrategy would cause larger fluctuations in population size(Fig 14) which negatively affects persistence probability Thusextinction risks faced by inbred populations of large carnivorescan be substantially reduced by releasing approximately 5 im-migrants every 2 decades We suggest that such a managementstrategy be followed only in conjunction with continued long‐term monitoring of the population Our analyses demonstratethat population growth and persistence are highly sensitive tokitten and female (adult and subadult) survival Continuing tocollect data on these demographic variables along with bio-metric and genetic sampling will remain important to pantherrecovery and vigilance in identifying issues associated with in-breeding depression The collection of these monitoring datashould allow managers to detect any demographic or geneticchanges in a timely fashion and respond with genetic in-trogression or other management actions if warrantedRecovery objectives as defined by the USFWS in its current

recovery plan (USFWS 2008) delineate the need for additionalpopulations in the historical range to downlist or delist theFlorida panther If the only extant population of Floridapanthers continued to grow it could serve as a source ofpanthers to be translocated to suitable habitat to seed addi-tional populations Maintaining current information on thegenetic health of the population and precise estimates of itssize will be important in assessing whether it can withstand theremoval of a subset of animals for translocation Although the2017 documentation of a female panther with kittens north ofthe current breeding range is encouraging in terms of thepotential for population expansionmdashgiven that no female hadbeen recorded north of the Caloosahatchee River since 1972mdashthe re‐establishment of separate new populations will mostlikely require translocations by wildlife managers and a lengthysociopolitical processConservation of threatened or endangered species such as the

Florida panther is inherently challenging For panthers and otherlarge carnivores that require vast extents of natural habitat topersist habitat is typically the most limiting factor that will de-termine whether recovery efforts succeed Although geneticmanagement invariably plays a key role in the long‐term persis-tence of the panther population even a healthy population caneventually succumb to the impacts of habitat loss The humanpopulation of Florida continues to be one of the fastest‐growingin the nation the United States Census Bureau currently ranksFlorida as the third most populous Therefore habitat loss isgoing to continue to be a key issue for panthers Diligence towardpreserving panther habitat in its historical range via conservationeasements or other means and trying to keep such areas inter-connected via corridors is a goal stakeholders such as governmentagencies sportsmen non‐governmental organizations ranchers

30 Wildlife Monographs bull 203

and other private landowners must work on collaboratively toachieve

ACKNOWLEDGMENTS

We thank D Shindle D Jennings T OrsquoMeara D Land andC Belden for valuable advice throughout the project We thankS Bass M Criffield M Cunningham D Giardina D JansenA Johnson D Land M Lotz C McBride Rocky McBrideRowdy McBride Roy McBride L Oberhofer S Schulze staffsat Everglades National Park and Big Cypress National Preserveand others for assistance with fieldwork We also thank JAustin M Conroy J Hines J Laake S Mills J Nichols JHines M Sunquist J Fryxell and T Coulson for advice andassistance regarding data analysis and panther biology TheMuseum of Texas Tech University Natural Science ResearchLaboratory provided samples from pumas from west Texas forcomparative genetic analyses M Ben‐David D Land WMcCown E Hellgren and 4 anonymous reviewers providedmany helpful comments on the manuscript We thank NereaUbierna Lopez for completing the Spanish translation of theabstract We are grateful to the Department of ResearchComputing University of Florida for excellent high‐perfor-mance computing services We would like to thank US Fishand Wildlife Service Florida Fish and Wildlife ConservationCommission (FWC) US National Park Service QuantitativeSpatial Ecology Evolution and Environment (QSE3) In-tegrative Graduate Education and Research Traineeship(IGERT) program funded by the National Science Foundationand the University of Florida for financially supporting thisstudy Funding for panther research conducted by the FWC issupported predominantly via donations accrued with vehicleregistrations for ldquoProtect the Pantherrdquo license plates

LITERATURE CITEDAcevedo‐Whitehouse K T R Spraker E Lyons S R Melin F Gulland RL Delong and W Amos 2006 Contrasting effects of heterozygosity onsurvival and hookworm resistance in California sea lion pups MolecularEcology 151973ndash1982

Adams J R L M Vucetich P W Hedrick R O Peterson and J AVucetich 2011 Genomic sweep and potential genetic rescue during limitingenvironmental conditions in an isolated wolf population Proceedings of theRoyal Society B Biological Sciences 2783336ndash3344

Agresti A 2013 Categorical data analysis Third edition John Wiley amp SonsHoboken New Jersey USA

Agudo R M Carrete M Alcaide C Rico F Hiraldo and J A Donaacutezar2012 Genetic diversity at neutral and adaptive loci determines individualfitness in a long‐lived territorial bird Proceedings of the Royal Society BBiological Sciences 2793241ndash3249

Albeke S E N P Nibbelink and M Ben‐David 2015 Modeling behavior bycoastal river otter (Lontra canadensis) in response to prey availability in PrinceWilliam Sound Alaska a spatially‐explicit individual‐based approach PLOSOne 10e0126208

Alho J S K Vaumllimaumlki and J Merilauml 2010 Rhh an R extension for estimatingmultilocus heterozygosity and heterozygosity‐heterozygosity correlationMolecular Ecology Resources 10720ndash722

Alvarez K 1993 Twilight of the panther Myakka River Publishing SarasotaFlorida USA

Aparicio J M J Ortego and P J Cordero 2006 What should we weigh toestimate heterozygosity alleles or loci Molecular Ecology 154659ndash4665

Bakker V J D F Doak G W Roemer D K Garcelon T J Coonan S AMorrison C Lynch K Ralls and R Shaw 2009 Incorporating ecologicaldrivers and uncertainty into a demographic population viability analysis for theisland fox Ecological Monographs 7977ndash108

Balloux F W Amos and T Coulson 2004 Does heterozygosity estimateinbreeding in real populations Molecular Ecology 133021ndash3031

Barone M A M E Roelke J Howard J L Brown A E Anderson and DE Wildt 1994 Reproductive characteristics of male Florida panthersmdashcomparative studies from Florida Texas Colorado Latin‐America andNorth‐American Zoos Journal of Mammalogy 75150ndash162

Bateson Z W P O Dunn S D Hull A E Henschen J A Johnson and LA Whittingham 2014 Genetic restoration of a threatened population ofgreater prairie‐chickens Biological Conservation 17412ndash19

Bauer H G Chapron K Nowell P Henschel P Funston L T B HunterD W Macdonald and C Packer 2015 Lion (Panthera leo) populations aredeclining rapidly across Africa except in intensively managed areas Pro-ceedings of National Academy of Sciences of the United States of America11214894ndash14899

Beier P 1995 Dispersal of juvenile cougars in fragmented habitat Journal ofWildlife Management 59228ndash237

Benson J F J A Hostetler D P Onorato W E Johnson M E Roelke SJ OrsquoBrien D Jansen and M K Oli 2011 Intentional genetic introgressioninfluences survival of adults and subadults in a small inbred felid populationJournal of Animal Ecology 80958ndash967

Benson J F P J Mahoney J A Sikich L E K Serieys J P Pollinger H BErnest and S P D Riley 2016 Interactions between demography geneticsand landscape connectivity increase extinction probability for a small popu-lation of large carnivores in a major metropolitan area Proceedings of theRoyal Society B Biological Sciences 28320160957

Beverly F 1874 The Florida panther Forest and Stream 3290Boyce M S C V Haridas C T Lee and the NCEAS Stochastic Demo-graphy Working Group 2006 Demography in an increasingly variable worldTrends in Ecology amp Evolution 21141ndash148

Brook B W J J OrsquoGrady A P Chapman M A Burgman H R Akcakayaand R Frankham 2000 Predictive accuracy of population viability analysis inconservation biology Nature 404385ndash387

Brys R and H Jacquemyn 2016 Severe outbreeding and inbreeding de-pression maintain mating system differentiation in Epipactis (Orchidaceae)Journal of Evolutionary Biology 29352ndash359

Burke J M and M L Arnold 2001 Genetics and the fitness of hybridsAnnual Review of Genetics 3531ndash52

Burnham K P 1993 A theory for combined analysis of ring recovery andrecapture data Pages 199ndash213 in J‐D Lebreton and P M North editorsMarked individuals in the study of bird population Springer‐Verlag NewYork New York USA

Burnham K P and D R Anderson 2002 Model selection and multimodelinference a practical information‐theoretic approach Second edition SpringerVerlag New York New York USA

Caswell H 2001 Matrix population models construction analysis and in-terpretation Sinauer Associates Sunderland Massachusetts USA

Charpentier M J M Setchell F Prugnolle L A Knapp E J Wickings PPeignot and M Hossaert‐McKey 2005 Genetic diversity and reproductivesuccess in mandrills (Mandrillus sphinx) Proceedings of the NationalAcademy of Sciences of the United States of America 10216723ndash16728

Cooch E and G C White editors 2018 Program MARK a gentle in-troduction lthttpwwwphidotorgsoftwaremarkdocsbookgt Accessed 7Apr 2016

Cooley H S R B Wielgus G M Koehler H S Robinson and B TMaletzke 2009 Does hunting regulate cougar populations A test of thecompensatory mortality hypothesis Ecology 902913ndash2921

Coulson T E A Catchpole S D Albon B J Morgan J M Pemberton TH Clutton‐Brock M J Crawley and B T Grenfell 2001 Age sex densitywinter weather and population crashes in Soay sheep Science 2921528ndash1531

Coulson T J Pemberton S Albon M Beaumont T Marshall J Slate FGuinness and T Clutton‐Brock 1998a Microsatellites reveal heterosis in reddeer Proceedings of the Royal Society B Biological Sciences 265489ndash495

Coulson T N S D Albon J M Pemberton J Slate F E Guinness and TH Clutton‐Brock 1998b Genotype by environment interactions in wintersurvival in red deer Journal of Animal Ecology 67434ndash445

Cox D 1972 Regression models and life‐tables Journal of the Royal StatisticalSociety Series B Statistical Methodology 34187ndash220

Creel S 2006 Recovery of the Florida panthermdashgenetic rescue demographic rescueor both Response to Pimm et al (2006) Animal Conservation 9125ndash126

Crnokrak P and D A Roff 1999 Inbreeding depression in the wild Heredity83260ndash270

Crow J 1948 Alternative hypotheses of hybrid vigor Genetics 33477ndash487

van de Kerk et al bull Florida Panther Population and Genetic Viability 31

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 20: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

probability of population persistence and the duration of thepositive impacts of genetic restoration on this endangered po-pulation The Florida panther population was in dire straits inthe early 1990s and our results clearly demonstrated that the

implementation of the introgression project in 1995 subse-quently accrued a series of genetic and demographic improve-ments that have led to the change from a declining population toone that is growing and expanding its current breeding range

Sensitivity Elasticity

0 1 2 3 00 02 04

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

Para

met

er

Figure 9 Sensitivity and elasticity (proportional sensitivity) of stochastic population growth rate (λs) of the Florida panther to vital demographic parameters in SouthFlorida USA 1981ndash2013 Horizontal lines represent 5th and 95th percentiles

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

70

80

90

100

110

0

5

10

15

0

50

100

Years

StatisticMeanMedian

Figure 10 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the minimum population count(MPC) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

IBM Matrix

NP

QE

TQE

0 50 100 150 200 0 50 100 150 200

125

150

175

200

00

05

10

15

20

0

30

60

90

Years

StatisticMeanMedian

Figure 11 Mean (solid lines) and median (dashed lines) Florida pantherprobabilities () of quasi‐extinction (PQE) times in years until quasi‐extinction(TQE) and population trajectories (N) under the motor vehicle mortality(MVM) scenario as predicted by the individual‐based model (IBM) and matrixpopulation model The critical threshold was 10 panthers Shaded area indicates5th and 95th percentiles Based on data collected in South Florida USA1981ndash2013

22 Wildlife Monographs bull 203

MPC MVM

minus08 minus06 minus04 minus02 00 minus08 minus06 minus04 minus02 00

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Male subadult survival

Male prime adult survival

Male old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

PRCC

Para

met

er

Figure 12 Partial rank correlation coefficient (PRCC) between quasi‐extinction probabilities and the demographic parameters of the Florida panther for theminimum population count (MPC) and motor vehicle mortality (MVM) scenarios Error bars indicate standard deviation Based on data collected in South FloridaUSA 1981ndash2013

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 13 Average projected individual heterozygosities in the Florida panther population over 200 years without genetic management intervention and with eachof the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include theintroduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years Average individual heterozygosity peaks when pumas are added to the Floridapanther population Based on data collected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 23

Our research has further validated the success of genetic in-trogression by quantifying the reduction in the probability ofextinction of the panther population because of this manage-ment initiative These findings all bode well for continuedprogress toward recovery That said the Florida panther po-pulation remains small and isolated from other populations ofpuma from western North America thereby denoting that theeventual impact of genetic drift and inbreeding may negativelyaffect the population in the future Realizing this will continueto be an issue that managers will have to contemplate wemodeled the impact of varied genetic management scenarios todetermine the frequency of introgression along with the numberof panthers that should be released during each initiative tominimize cost and maximize benefits to the population Theseresults will prove invaluable to managers attempting to conservethe Florida panther

Genetic Variation Demographic Parametersand Persistence of Heterotic Benefits from GeneticIntrogressionOur measure of individual heterozygosity (1 minusHL) was higheron average for the admixed (0615) panthers than for those ofcanonical ancestry (0386) Levels of individual heterozygosityfor admixed panthers were comparable to levels observed in asample of pumas from west Texas (x plusmn SE= 0695plusmn 0019n= 41) which further demonstrates the improvement in levelsof genetic variation in the Florida population since 1995 The

correlation between HL and several demographic parametershas been noted in wild populations including cheetahs (Terrellet al 2016) and several species of birds such as European shags(Phalacrocorax aristotelis male and female survival probabilityVelando et al 2015) and Egyptian vulture (Neophron percnop-terus age at recruitment Agudo et al 2012) If we focus only onpanthers captured and radiocollared from 2012ndash2015 (FP211ndashFP240) all of which had estimated birth years ge2005 (ge10 yrpost‐introgression) those panthers had an average individualheterozygosity level of 0618plusmn 0029 highlighting the elevatedlevels of genetic variation that continue to persist in cohorts ofpanthers 5 generations after the release of female pumas fromTexasThe proportional membership values (q) from our

STRUCTURE analysis allowed us to delineate 2 clusters ofancestry for Florida panthers (Fig 4) During the pre‐in-trogression era the population was dominated by panthers thatretained a high percentage of the canonical ancestry which wasaffected by inbreeding depression The post‐introgression eraancestry values are comprised of many admixed individuals inessence demonstrating the success of the introgression in termsof increasing genetic variation in the population and mi-micking gene flow that historically occurred with panthers andother populations of pumas (Onorato et al 2010) A somewhatanalogous situation although abetted by natural immigrationwas evident in the ancestral variation in a small population ofpuma intersected by a major freeway in California USA In

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

75

80

85

90

95

100

130

150

170

190

Years

Popu

latio

n si

ze

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 14 Average projected population sizes in the Florida panther population over 200 years without intervention and with each of the genetic introgressionstrategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include the introduction of 5 10 or 15pumas from Texas USA every 10 20 40 or 80 years Peaks occur when pumas are added to the Florida panther population Based on data collected in SouthFlorida USA 1981ndash2013

24 Wildlife Monographs bull 203

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

25

50

75

100

0

25

50

75

100

Introgression interval (yrs)

PQ

E (

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 15 Average probability of quasi‐extinction (PQE) of the Florida panther population within 100 years and within 200 years without genetic managementintervention (N) and with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenariosIntrogression strategies include the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Errorbars indicate 5th and 95th percentiles Based on data collected in South Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

50

60

70

80

90

100

110

50

100

150

200

Years

Popu

latio

n si

ze

Figure 16 Average projected Florida panther population trajectories under a geneticintrogression strategy involving the release of 5 female pumas from Texas USA every20 years for the minimum population count (MPC) and motor vehicle mortality(MVM) scenarios Shaded area indicates 5th and 95th percentiles and isrepresentative of confidence intervals for other scenarios Based on data collected inSouth Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Figure 17 Average projected Florida panther population‐level heterozygositiesunder a genetic introgression strategy involving the release of 5 female pumas fromTexas USA every 20 years for the minimum population count (MPC) and motorvehicle mortality (MVM) scenarios Shaded area bounds the 5th and 95th percentilesand is representative of confidence intervals for other scenarios Based on datacollected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 25

this case the migration of just a single male across the freewayto the south resulted in an increase in the number of in-dividuals with admixed ancestry and substantially improvedgenetic diversity of the subpopulation south of the freeway(Riley et al 2014)Our estimate of overall kitten survival (0321plusmn 0056) was

similar to that reported by Hostetler et al (2010) who used13 years of post‐introgression data (0323plusmn 0071) These valuesare lower than kitten survival estimates derived from severalpuma populations in western North America (044ndash091 Loganand Sweanor 2001 Lambert et al 2006 Laundre et al 2007Robinson et al 2008 Ruth et al 2011) When we included onlydata for individuals that had been genetically sampled our es-timate was 15 higher The proportion of admixed kittens thatwere genetically sampled increased as the population expandedwhich could have resulted in higher estimates of kitten survivalbecause admixed kittens survive better than canonical kittensKitten survival was strongly influenced by genetic ancestry withsurvival rates of F1 kittens being almost twice that of canonicalkittens (Fig 5B) Consistent with the findings of Hostetler et al(2010) increases in heterozygosity coincided with increasedkitten survival whereas population density negatively affectedkitten survival Population density often has a strong negativeimpact on survival of young age classes due to infanticide andcannibalism which has been reported from several carnivore

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

50

100

150

0

50

100

150

Introgression interval (yrs)

TQE

(yrs

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 18 Average times until quasi‐extinction (TQE) for the Florida panther population within 100 years and within 200 years without genetic management interventionand with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategiesinclude the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Error bars indicate 5th and 95thpercentiles Based on data collected in South Florida USA 1981ndash2013

Table 5 Average cost (2017 US dollars) of introgression strategies employedover 100 years that involve the introduction of 5 10 or 15 pumas from TexasUSA into the Florida panther population in South Florida USA at an in-creasingly higher frequency Costs of capture (including transportation) quar-antine and post‐introgression monitoring were estimated by Roy McBrideMark Cunningham and Dave Onorato respectively

Frequency Component 5 pumas 10 pumas 15 pumas

Once Capture 25000 50000 75000

Quarantine 5000 10000 15000

Monitoring 10000 20000 30000

Total 40000 80000 120000

Every 80 years Capture 31250 62500 93750

Quarantine 6250 12500 18750

Monitoring 12500 25000 37500

Total 50000 100000 150000

Every 40 years Capture 62500 125000 187500

Quarantine 12500 25000 37500

Monitoring 25000 50000 75000

Total 100000 200000 300000

Every 20 years Capture 125000 250000 375000

Quarantine 25000 50000 75000

Monitoring 50000 100000 150000

Total 200000 400000 600000

Every 10 years Capture 250000 500000 750000

Quarantine 50000 100000 150000

Monitoring 100000 200000 300000

Total 400000 800000 1200000

26 Wildlife Monographs bull 203

species including polar bears (Ursus maritimus Derocher andWiig 1999) black bears (Ursus americanus Garrison et al 2007)and pumas (Logan and Sweanor 2001)Our estimates of sex‐ and age‐specific survival rates of subadult

and adult panthers (Table 3) and the effects of covariates on theserates were similar to those reported by Benson et al (2011) derivedfrom data collected through 2006 Our survival rates for males(065ndash077) were lower than females (078ndash097) for all 3 ageclasses (subadult prime adult and old adult) which is the typicaltrend observed in most puma populations (eg Ruth et al 19982011) However an unhunted puma population occupying afragmented urban landscape in southern California exhibited lowersurvival (0586 females 0525 males Vickers et al 2015) perhapsas a result of severe habitat fragmentation and the lack of extensiveswaths of public land protected in perpetuity Most populations ofpuma in western North America are typically subject to variedlevels of management via regulated hunting which often is biasedtoward the take of males (Cooley et al 2009 Ruth et al 2011)Consequently annual survival estimates from hunted populationsin northeast Washington (0679ndash0733 females 0341 malesRobinson et al 2008) and southeastern Arizona (0685 females0584 males Cunningham et al 2001) were generally lower thanthose we estimated for Florida panthersCause‐specific mortality analysis showed that male panthers

experienced a substantially greater mortality risk from in-traspecific aggression This differs from the pattern observedduring a long‐term study in Arizona where females were at ahigher risk of intraspecific mortality than males (46 of maleand 53 of female mortality Logan and Sweanor 2001)Mortality associated with intraspecific strife has been docu-mented in hunted and unhunted populations (this study Loganand Sweanor 2001 Stoner et al 2010) and its root cause hasbeen difficult to determine (eg population density and preypopulation trends) That being the case finding ways to reducewhat is often a major source of mortality for puma populationscontinues to pose a challenge to managersCanonical panthers were substantially more likely to die from

intraspecific aggression than were admixed panthers This mayat least partly explain the difference in survival between admixed

and canonical panthers Canonical panthers are known to bemore likely to suffer from varied correlates of inbreeding de-pression than admixed animals including atrial septal defects(Johnson et al 2010) and compromised immune systems(Roelke et al 1993) These factors have the potential to affectfitness and perhaps dominance of individuals when engaged inan intraspecific encounter A somewhat similar scenario wasobserved by Hogg et al (2006) in a population of bighorn sheepthat were part of an introgression project Admixed males in thispopulation had a greater probability of paternity than males thatwere less outbred This included coursing males with higherlevels of admixture being markedly more successful at fightingmales that were defending females in estrous (Hogg et al 2006)Heterosis resulting from genetic introgression is expected to be

the strongest in F1 individuals (Shull 1908 Crow 1948 Burkeand Arnold 2001 Waller 2015) and to progressively decline insubsequent generations (Pickup et al 2013 Frankham 2015)Given that admixed (especially F1) panthers survived better thancanonical panthers and that individual heterozygosity improvedsurvival of both sexes and all age classes our data provide evi-dence for heterotic advantage whereby individuals of mixedancestry had higher fitness (Shull 1908 Crow 1948) Our resultsare consistent with findings of other studies that involved in-tentional genetic introgression for conservation purposes Forexample Hogg et al (2006) detected substantial improvementsin survival and reproduction of introgressed bighorn sheep andMadsen et al (1999 2004) reported a dramatic increase in thepopulation of adders after genetic introgressionKnowledge of the persistence of benefits to a population ac-

crued via genetic introgression is essential for determiningwhen additional introgression may be warranted There is adepauperate amount of information regarding this topic and theidentification of factors that may influence persistence of thebenefits of genetic introgression (Tallmon et al 2004 Hedricket al 2014 Whiteley et al 2015) For example in minks(Neovison vison) Thirstrup et al (2014) found that outcrossingled to larger litters in F2 and F3 but this benefit disappeared insubsequent generations In a population of gray wolves Hedricket al (2014) suggested that benefits of genetic introgression may

Table 6 Costs and benefits accumulated over 100 years for genetic introgression at different intervals and with different numbers of pumas from Texas USAintroduced into the Florida panther population in South Florida USA Costs (2017 US dollars) include capture and transportation quarantine and post‐introgression monitoring Benefits are represented as average probability of quasi‐extinction (PQE and 5th and 95th percentiles) and changes (Δ) therein under thescenario using the minimum population count (McBride et al 2008 MPC) and the scenario using the motor vehicle mortality population estimates (McClintocket al 2015 MVM) The table is sorted by percent change in probability of quasi‐extinction under the MPC scenario

Interval (yr) Pumas Cost MPC PQE () ΔMPC PQE () MVM PQE () ΔMVM PQE ()

ndash 0 0 13 (0ndash99) 0 17 (0ndash100) 080 10 100000 12 (0ndash99) 10 (0ndash58) 16 (0ndash100) 5 (0ndash38)80 15 150000 12 (0ndash99) 13 (0ndash65) 15 (0ndash100) 8 (0ndash56)80 5 50000 12 (0ndash99) 14 (0ndash73) 15 (0ndash99) 9 (0ndash41)40 15 300000 10 (0ndash98) 26 (0ndash104) 14 (0ndash99) 13 (0ndash60)40 10 200000 9 (0ndash94) 35 (0ndash183) 13 (0ndash99) 21 (0ndash95)40 5 100000 8 (0ndash89) 39 (0ndash211) 12 (0ndash99) 26 (0ndash124)20 15 600000 8 (0ndash88) 42 (0ndash257) 13 (0ndash99) 24 (0ndash123)20 10 400000 6 (0ndash67) 54 (0ndash318) 10 (0ndash99) 37 (0ndash217)10 15 1200000 6 (0ndash53) 58 (0ndash372) 10 (0ndash99) 40 (0ndash208)20 5 200000 5 (0ndash50) 59 (0ndash338) 9 (0ndash99) 44 (0ndash352)10 10 800000 4 (0ndash16) 69 (0ndash489) 8 (0ndash92) 55 (0ndash401)10 5 400000 4 (0ndash7) 73 (0ndash502) 6 (0ndash71) 63 (0ndash503)

van de Kerk et al bull Florida Panther Population and Genetic Viability 27

wane after 2 or 3 generations The WrightndashFisher model ofgenetic drift predicts a geometric decline in expected hetero-zygosity at the rate of (1 minus 12Ne) per generation where Ne is theeffective population size (Hartl and Clark 1997 Frankham et al2014) Thus it seems logical to assume that the duration of thebenefits of introgression is limited especially in a species per-sisting in a small population such as Florida panthersWe expected Florida panther survival in the most recent years

(2005ndash2013) post‐introgression to differ from that observed in thedecade following the introgression in 1995 because pantherabundance and heterozygosity factors known to influence panthersurvival (Hostetler et al 2010 Benson et al 2011) have changed(Figs 2 and 6) We found that subadult and adult survival rates inrecent years (2005ndash2013) were similar to those in the period im-mediately following introgression (1995ndash2004 Fig 5C) overallkitten survival was similar to estimates of Hostetler et al (2010) Infact the pattern of covariate effects on Florida panther survivalreported by Benson et al (2011) and Hostetler et al (2010) hasbarely changed The negative effects of population density andpositive effects of heterozygosity may counterbalance survival ratesreducing population fluctuations Our result that benefits of geneticrestoration have remained in the population nearly for 5 genera-tions post‐intervention was surprising but also encouraging Pos-sible explanations for this observation may include 1) genetic ero-sion occurs at slower rate than previously thought 2) introducing 5migrants in 1 genetic intervention may have beneficial effects si-milar to those from 1 migrant per generation over multiple gen-erations or 3) other and as yet unknown factors may reduce therate of genetic erosion and subsequent demographic effects Adensity‐dependent effect on kitten survival could also explain whynon‐F1 admixed kittens did not survive substantially better thandid canonical kittens most kittens sampled from 2005ndash2013 wereadmixed and their survival was lower possibly because of a density‐dependent effect as the Florida panther population continuouslygrew during that period Trinkel et al (2010) also found thatpopulation density and inbreeding coefficient interacted to affectdemographic parameters and growth rate of an African lion po-pulation Likewise population density interacted with winterweather to affect the survival and population growth rates in Soaysheep (Ovis aries Milner et al 1999 Coulson et al 2001)

Population Dynamics and PersistenceThe finite deterministic and stochastic growth rates were gt10reflecting that much of the data used in this study were collectedfollowing genetic introgression in 1995 during which time thepopulation increased substantially Because our demographicparameter estimates did not change markedly in comparison tothose of Hostetler et al (2013) the similarity between thepopulation growth estimates from these studies was expectedBut these estimates reflect population growth during thedata‐collection period and do not predict future growth In factestimates of population size reported by McClintock et al(2015) suggest that population growth may already be slowing(Fig 2) A similar pattern was observed in an introduced po-pulation of African lion which increased steadily from 1995until 2001 then fluctuated around the presumed carrying ca-pacity thereafter (Trinkel et al 2010) Several other African lionpopulations that experienced various degrees of recovery

subsequently became relatively stable or exhibited decliningtrends (Bauer et al 2015)Our estimates of quasi‐extinction probabilities were lower than

those reported by Hostetler et al (2013) using a 2‐sex age‐structured matrix population model Lower probabilities ofquasi‐extinction than those reported by the earlier study forcomparable scenarios were likely a consequence of the fact thatthe estimates of abundance (McBride et al 2008) and thus thedensity‐dependent effects on survival of kittens and old panthers(gt10 yr) have changed in recent years Additionally these earlyestimates of the probability of quasi‐extinction and of time toquasi‐extinction did not consider genetic erosion that will in-evitably occur without management interventionUnder the MVM scenario the equilibrium population size was

twice that attained under the MPC scenario The MVM popula-tion estimates are approximately twice the MPCs (Fig 2) as aresult the simulated population under the MVM scenario achieveda higher equilibrium size Probability of quasi‐extinction under theMVM scenario was generally greater than that under the MPCscenario This result is a consequence of the fact that the estimateddensity dependence on kitten survival based on the MPC scenario isstronger than that for theMVM scenario (regression coefficients forabundance for the MPC scenario= minus0068 vs minus0002 for theMVM scenario Appendix B Table B1) Consequently simulatedpopulations under the MPC scenario have a stronger tendency tomove toward the equilibrium population size which inherentlyreduces the quasi‐extinction probability (eg Royama 1992)As expected for long‐lived mammals (Heppell et al 2000 Oli

and Dobson 2003) the population growth rate was proportio-nately most sensitive to changes in survival of prime adultsfollowed by that in younger age classes Global sensitivity ana-lysis in contrast showed that under all scenarios extinctionparameters were particularly sensitive to changes in survival ofFlorida panther kittens This result is a consequence of the highsensitivity of the population growth rate to kitten survival(Fig 9) and a larger standard error of kitten survival (Table 3)Results of our sensitivity analyses indicate that future researchshould focus on acquiring additional information on early lifestages to improve the accuracy and precision of estimates ofkitten survival Our results suggest that demographic variables(or their environmental drivers) that strongly influence extinc-tion risks are not necessarily those with the largest elasticityvalues A study of island fox (Bakker et al 2009) found thatextinction risk was strongly influenced by survival modifierparameters that had low elasticity values

Genetic Management When and How to InterveneThe use of genetic introgression as a management tool has al-ways been controversial and the probable effectiveness it mighthave in preventing the imminent demise of the panther popu-lation was initially debated (Creel 2006 Maehr et al 2006Pimm et al 2006) These debates notwithstanding the pantherpopulation had suffered from genetic morphological and bio-medical correlates of inbreeding before this management in-tervention (Johnson et al 2010 Onorato et al 2010) whereasthe post‐introgression period has been characterized by a sub-stantial reduction in the biomedical correlates of inbreeding andimprovements in demographic vigor and abundance (McBride

28 Wildlife Monographs bull 203

et al 2008 Hostetler et al 2010 2013 Johnson et al 2010Benson et al 2011 McClintock et al 2015) Nevertheless thepopulation remains small and isolated habitat is still being lostand there is no current possibility of natural gene flow betweenthis and other North American puma populations thus therecurrence of inbreeding‐related problems and subsequent po-pulation declines are inevitable The question therefore is notwhether genetic management of the Florida panther populationis needed but when and how it should be implementedWithout genetic introgression probabilities of quasi‐extinc-

tion were substantially greater than those from models thatincorporated periodic genetic introgression events (Fig 15)highlighting the importance of genetic management in smallisolated populations When 5 female pumas are added to theFlorida panther population every 10 years genetic variationincreased substantially under the MPC and MVM scenariosThe IBM predicted that the equillibrium population size wouldbe substantially larger when 5 pumas were released into thepopulation every 10 years than populations without intervention(ie no introgression) Releasing 15 Texas females did not af-ford substantially greater benefit to the equilibrium populationsize than did releasing only 5 Texas females Instead addingmore individuals caused increasingly large fluctuations in po-pulation size due to the numerical increases and density de-pendence (Fig 14) possibly diminishing the benefits associatedwith increased genetic variationCompared with the no‐intervention scenario (ie no genetic

introgression implemented) the introduction of 5 female pumasevery 10 years reduced quasi‐extinction probabilities from 13to 4 and from 17 to 6 for the MPC and MVM scenariosrespectively The benefit of genetic‐introgression strategies de-creased as the interval between successive management inter-ventions increased and no benefit was evident once the intervalreached 80 years (Fig 15) Introgression reduced the averagetime until quasi‐extinction (Fig 18) This somewhat counter‐intuitive result can be explained by the fact that repeated geneticintrogression makes the population more robust over timewhich will reduce the probability of extinction Consequentlyfew simulated population trajectories that fall below the criticalthreshold do so early reducing the mean time to extinctionAlso density dependence has a stabilizing effect on populations(Royama 1992) populations tend toward equilibrium popula-tion sizes which reduces the probability over time of popula-tions falling below quasi‐extinction threshold However timeuntil quasi‐extinction must be interpreted in conjunction withthe probability of quasi‐extinction which is very low for allscenarios with genetic introgression (Fig 15)Cost is a significant impediment to the implementation of a

genetic introgression program because captures relocations andmonitoring efforts before and after introgression are expensive(Edmands 2007 Frankham 2010b 2015 2016) Costs may beacceptable to society if the management actions result in sub-stantial fitness benefits especially if the species of concern ispopular and charismatic (Frankham 2015) We estimated the100‐year cost of introgression strategies modeled here to rangefrom $50000 to $12 million More expensive strategies gen-erally led to larger decreases in the probability of quasi‐extinc-tion but cheaper strategies also substantially improved

population viability (ie reduced quasi‐extinction probabilityTable 6) One of the cheaper strategies (5 pumas released every20 years) which cost an estimated $200000 over 100 yearsreduced the probability of quasi‐extinction by 59 and 44 forthe MPC and MVM scenarios respectively But these pre-liminary cost estimates are based on expenses incurred duringthe 1995 introgression and include costs of capture quarantinetransportation release and post‐release monitoring of femalesonly Although the cost for future genetic interventions wouldlikely be higher than those reported here the relative differencesin cost between alternative intervention strategies should remainapproximately the sameOur ability to simulate genetics and link it to the viability of

the Florida panther population is based on the empirically es-timated relationship between individual heterozygosity andsurvival Such heterozygosity and fitness correlations have beenstudied for decades (David 1998) but have only recently beenused to predict population viability (Benson et al 2016) noneto our knowledge have incorporated cost into their modelingefforts The mechanisms underlying heterozygosity and fitnesscorrelations remain unclear (David 1998 Szulkin et al 2010)and their significance as a proxy for inbreeding depression hasbeen debated (Balloux et al 2004 Miller and Coltman 2014)Nevertheless evidence continues to mount demonstratingpositive relationships between heterozygosity and survival(Coulson et al 1998ab Hansson et al 2004 Silva et al 2005Acevedo‐Whitehouse et al 2006 Jensen et al 2007 Velandoet al 2015) and between heterozygosity and fecundity (Seddonet al 2004 Charpentier et al 2005 Agudo et al 2012 Velandoet al 2015) Although the reported correlations are generallyweak their consequences can be substantial if population growthand persistence parameters are highly sensitive to the affectedvital rates (Velando et al 2015) Compared with scenarios thatignore genetics incorporating the effects of inbreeding on sur-vival increased quasi‐extinction probabilities within a 100‐yeartimeframe from 15 to 13 and from 14 to 17 for theMPC and MVM scenarios respectively These results high-light the importance of incorporating genetics when analyzingthe viability of small isolated populationsAlthough conservation biologists have recognized the im-

portance of genetics in population viability many still fail toincorporate genetic factors into PVA models (Reed et al 2002)Explicit consideration of genetics in PVAs requires long‐termgenetic and demographic data This permits the assessment ofthe functional relationship between genetic variability and de-mographic parameters Population viability analysis softwarepackages such as VORTEX (Lacy 1993 2000) offer a powerfulframework for individual‐based simulations but they often donot adequately capture a speciesrsquo life history or genetics Basedon a thorough review of the importance of genetic factors inwildlife conservation Frankham et al (2014) concluded thatgenetic factors can strongly influence the dynamics and persis-tence of small andor fragmented populations They furthernoted that ldquoMost PVA‐based risk assessments ignore or in-adequately model genetic factorsrdquo and recommended that ldquoPVAshould routinely include realistic inbreeding depressionhelliprdquo(Frankham et al 201457) Our custom‐built IBM permitted usto develop a model that adequately captured the Florida panther

van de Kerk et al bull Florida Panther Population and Genetic Viability 29

life history and we could parameterize the model with our long‐term genetic and demographic dataThe explicit consideration of the effects of inbreeding on

Florida panther population dynamics and persistence representsan important step forward Nonetheless our model remainsimperfect First we neglected the possible effects of habitat lossdetrimental anthropogenic activities climate change the prob-ability of immigration of pumas from western populationsdisease or catastrophes on demographic rates and populationviability Although immigration from puma populations outsideFlorida is possible its probability is very low given the extensivechallenges the anthropogenically altered landscape would posefor such a migrant to make it successfully to South Florida Weomitted mutations from our model because we have no in-formation on mutation rates though we expect that they are lowbased on values reported for other mammals (Kumar and Sub-ramanian 2002) Finally we only considered possible effects ofinbreeding on age‐specific survival rates because there was noevidence that heterozygosity affected female reproduction Lossof heterozygosity can negatively affect reproduction becauseinbred male panthers can suffer from cryptorchidism which canadversely affect their fecundity However given the polygynousmating system we expect the Florida panther population growthis primarily female‐limitedAlthough it is generally accepted that genetic introgression

only temporarily relieves a population from inbreeding depres-sion we know of no cases in which it has been implementedmore than once to conserve a threatened wildlife populationOur study provides an example of how demographic and mo-lecular data can be used within an IBM framework to evaluatethe efficacy of alternative genetic management strategies via theFlorida panther as a case study Our model can be adjusted forand applied to other species and offers great potential as a toolfor genetic management of small isolated wildlife populations

MANAGEMENT IMPLICATIONS

Our results and those of Hostetler et al (2013) and Johnsonet al (2010) provide persuasive evidence that the genetic in-tervention implemented in 1995 prevented the demise of theFlorida panther and restored demographic vigor to the popu-lation The Florida panther population remains small and iso-lated from other puma populations the closest puma populationis located gt1000 km away in Texas and the intervening land-scape is highly developed and fragmented with many interstate(and other high‐traffic) highways and major urban centersTheory suggests that 1 migrant per generation is sufficient tominimize the loss of genetic diversity (eg Mills and Allendorf1996) However the possibility of successful migration of apuma to South Florida followed by successful mating every ~5years is very low considering the distance between Texas andFlorida (Hedrick 1995) and the many risks and barriers thatdispersing pumas face (Beier 1995 Maehr et al 2002 Thatcheret al 2006) Consequently the pantherrsquos long‐term persistencewill likely depend on subsequent timely genetic introgressionsgiven the near‐impossibility of natural gene flow from otherpuma populations To that end our study offers considerableinsights into benefits of different genetic management strategiesand associated costs

Without genetic management the Florida panther populationfaces a substantial risk of quasi‐extinction within 100 years (10ndash46 depending on quasi‐extinction threshold and scenarios)Twenty‐three years have passed since genetic introgression wasimplemented and the population appears healthy as indicatedby measures of genetic variation increases in the populationsize and improvements in age‐specific survival rates Ourfindings suggest that releasing more pumas more frequently doesnot necessarily improve persistence probability because such astrategy would cause larger fluctuations in population size(Fig 14) which negatively affects persistence probability Thusextinction risks faced by inbred populations of large carnivorescan be substantially reduced by releasing approximately 5 im-migrants every 2 decades We suggest that such a managementstrategy be followed only in conjunction with continued long‐term monitoring of the population Our analyses demonstratethat population growth and persistence are highly sensitive tokitten and female (adult and subadult) survival Continuing tocollect data on these demographic variables along with bio-metric and genetic sampling will remain important to pantherrecovery and vigilance in identifying issues associated with in-breeding depression The collection of these monitoring datashould allow managers to detect any demographic or geneticchanges in a timely fashion and respond with genetic in-trogression or other management actions if warrantedRecovery objectives as defined by the USFWS in its current

recovery plan (USFWS 2008) delineate the need for additionalpopulations in the historical range to downlist or delist theFlorida panther If the only extant population of Floridapanthers continued to grow it could serve as a source ofpanthers to be translocated to suitable habitat to seed addi-tional populations Maintaining current information on thegenetic health of the population and precise estimates of itssize will be important in assessing whether it can withstand theremoval of a subset of animals for translocation Although the2017 documentation of a female panther with kittens north ofthe current breeding range is encouraging in terms of thepotential for population expansionmdashgiven that no female hadbeen recorded north of the Caloosahatchee River since 1972mdashthe re‐establishment of separate new populations will mostlikely require translocations by wildlife managers and a lengthysociopolitical processConservation of threatened or endangered species such as the

Florida panther is inherently challenging For panthers and otherlarge carnivores that require vast extents of natural habitat topersist habitat is typically the most limiting factor that will de-termine whether recovery efforts succeed Although geneticmanagement invariably plays a key role in the long‐term persis-tence of the panther population even a healthy population caneventually succumb to the impacts of habitat loss The humanpopulation of Florida continues to be one of the fastest‐growingin the nation the United States Census Bureau currently ranksFlorida as the third most populous Therefore habitat loss isgoing to continue to be a key issue for panthers Diligence towardpreserving panther habitat in its historical range via conservationeasements or other means and trying to keep such areas inter-connected via corridors is a goal stakeholders such as governmentagencies sportsmen non‐governmental organizations ranchers

30 Wildlife Monographs bull 203

and other private landowners must work on collaboratively toachieve

ACKNOWLEDGMENTS

We thank D Shindle D Jennings T OrsquoMeara D Land andC Belden for valuable advice throughout the project We thankS Bass M Criffield M Cunningham D Giardina D JansenA Johnson D Land M Lotz C McBride Rocky McBrideRowdy McBride Roy McBride L Oberhofer S Schulze staffsat Everglades National Park and Big Cypress National Preserveand others for assistance with fieldwork We also thank JAustin M Conroy J Hines J Laake S Mills J Nichols JHines M Sunquist J Fryxell and T Coulson for advice andassistance regarding data analysis and panther biology TheMuseum of Texas Tech University Natural Science ResearchLaboratory provided samples from pumas from west Texas forcomparative genetic analyses M Ben‐David D Land WMcCown E Hellgren and 4 anonymous reviewers providedmany helpful comments on the manuscript We thank NereaUbierna Lopez for completing the Spanish translation of theabstract We are grateful to the Department of ResearchComputing University of Florida for excellent high‐perfor-mance computing services We would like to thank US Fishand Wildlife Service Florida Fish and Wildlife ConservationCommission (FWC) US National Park Service QuantitativeSpatial Ecology Evolution and Environment (QSE3) In-tegrative Graduate Education and Research Traineeship(IGERT) program funded by the National Science Foundationand the University of Florida for financially supporting thisstudy Funding for panther research conducted by the FWC issupported predominantly via donations accrued with vehicleregistrations for ldquoProtect the Pantherrdquo license plates

LITERATURE CITEDAcevedo‐Whitehouse K T R Spraker E Lyons S R Melin F Gulland RL Delong and W Amos 2006 Contrasting effects of heterozygosity onsurvival and hookworm resistance in California sea lion pups MolecularEcology 151973ndash1982

Adams J R L M Vucetich P W Hedrick R O Peterson and J AVucetich 2011 Genomic sweep and potential genetic rescue during limitingenvironmental conditions in an isolated wolf population Proceedings of theRoyal Society B Biological Sciences 2783336ndash3344

Agresti A 2013 Categorical data analysis Third edition John Wiley amp SonsHoboken New Jersey USA

Agudo R M Carrete M Alcaide C Rico F Hiraldo and J A Donaacutezar2012 Genetic diversity at neutral and adaptive loci determines individualfitness in a long‐lived territorial bird Proceedings of the Royal Society BBiological Sciences 2793241ndash3249

Albeke S E N P Nibbelink and M Ben‐David 2015 Modeling behavior bycoastal river otter (Lontra canadensis) in response to prey availability in PrinceWilliam Sound Alaska a spatially‐explicit individual‐based approach PLOSOne 10e0126208

Alho J S K Vaumllimaumlki and J Merilauml 2010 Rhh an R extension for estimatingmultilocus heterozygosity and heterozygosity‐heterozygosity correlationMolecular Ecology Resources 10720ndash722

Alvarez K 1993 Twilight of the panther Myakka River Publishing SarasotaFlorida USA

Aparicio J M J Ortego and P J Cordero 2006 What should we weigh toestimate heterozygosity alleles or loci Molecular Ecology 154659ndash4665

Bakker V J D F Doak G W Roemer D K Garcelon T J Coonan S AMorrison C Lynch K Ralls and R Shaw 2009 Incorporating ecologicaldrivers and uncertainty into a demographic population viability analysis for theisland fox Ecological Monographs 7977ndash108

Balloux F W Amos and T Coulson 2004 Does heterozygosity estimateinbreeding in real populations Molecular Ecology 133021ndash3031

Barone M A M E Roelke J Howard J L Brown A E Anderson and DE Wildt 1994 Reproductive characteristics of male Florida panthersmdashcomparative studies from Florida Texas Colorado Latin‐America andNorth‐American Zoos Journal of Mammalogy 75150ndash162

Bateson Z W P O Dunn S D Hull A E Henschen J A Johnson and LA Whittingham 2014 Genetic restoration of a threatened population ofgreater prairie‐chickens Biological Conservation 17412ndash19

Bauer H G Chapron K Nowell P Henschel P Funston L T B HunterD W Macdonald and C Packer 2015 Lion (Panthera leo) populations aredeclining rapidly across Africa except in intensively managed areas Pro-ceedings of National Academy of Sciences of the United States of America11214894ndash14899

Beier P 1995 Dispersal of juvenile cougars in fragmented habitat Journal ofWildlife Management 59228ndash237

Benson J F J A Hostetler D P Onorato W E Johnson M E Roelke SJ OrsquoBrien D Jansen and M K Oli 2011 Intentional genetic introgressioninfluences survival of adults and subadults in a small inbred felid populationJournal of Animal Ecology 80958ndash967

Benson J F P J Mahoney J A Sikich L E K Serieys J P Pollinger H BErnest and S P D Riley 2016 Interactions between demography geneticsand landscape connectivity increase extinction probability for a small popu-lation of large carnivores in a major metropolitan area Proceedings of theRoyal Society B Biological Sciences 28320160957

Beverly F 1874 The Florida panther Forest and Stream 3290Boyce M S C V Haridas C T Lee and the NCEAS Stochastic Demo-graphy Working Group 2006 Demography in an increasingly variable worldTrends in Ecology amp Evolution 21141ndash148

Brook B W J J OrsquoGrady A P Chapman M A Burgman H R Akcakayaand R Frankham 2000 Predictive accuracy of population viability analysis inconservation biology Nature 404385ndash387

Brys R and H Jacquemyn 2016 Severe outbreeding and inbreeding de-pression maintain mating system differentiation in Epipactis (Orchidaceae)Journal of Evolutionary Biology 29352ndash359

Burke J M and M L Arnold 2001 Genetics and the fitness of hybridsAnnual Review of Genetics 3531ndash52

Burnham K P 1993 A theory for combined analysis of ring recovery andrecapture data Pages 199ndash213 in J‐D Lebreton and P M North editorsMarked individuals in the study of bird population Springer‐Verlag NewYork New York USA

Burnham K P and D R Anderson 2002 Model selection and multimodelinference a practical information‐theoretic approach Second edition SpringerVerlag New York New York USA

Caswell H 2001 Matrix population models construction analysis and in-terpretation Sinauer Associates Sunderland Massachusetts USA

Charpentier M J M Setchell F Prugnolle L A Knapp E J Wickings PPeignot and M Hossaert‐McKey 2005 Genetic diversity and reproductivesuccess in mandrills (Mandrillus sphinx) Proceedings of the NationalAcademy of Sciences of the United States of America 10216723ndash16728

Cooch E and G C White editors 2018 Program MARK a gentle in-troduction lthttpwwwphidotorgsoftwaremarkdocsbookgt Accessed 7Apr 2016

Cooley H S R B Wielgus G M Koehler H S Robinson and B TMaletzke 2009 Does hunting regulate cougar populations A test of thecompensatory mortality hypothesis Ecology 902913ndash2921

Coulson T E A Catchpole S D Albon B J Morgan J M Pemberton TH Clutton‐Brock M J Crawley and B T Grenfell 2001 Age sex densitywinter weather and population crashes in Soay sheep Science 2921528ndash1531

Coulson T J Pemberton S Albon M Beaumont T Marshall J Slate FGuinness and T Clutton‐Brock 1998a Microsatellites reveal heterosis in reddeer Proceedings of the Royal Society B Biological Sciences 265489ndash495

Coulson T N S D Albon J M Pemberton J Slate F E Guinness and TH Clutton‐Brock 1998b Genotype by environment interactions in wintersurvival in red deer Journal of Animal Ecology 67434ndash445

Cox D 1972 Regression models and life‐tables Journal of the Royal StatisticalSociety Series B Statistical Methodology 34187ndash220

Creel S 2006 Recovery of the Florida panthermdashgenetic rescue demographic rescueor both Response to Pimm et al (2006) Animal Conservation 9125ndash126

Crnokrak P and D A Roff 1999 Inbreeding depression in the wild Heredity83260ndash270

Crow J 1948 Alternative hypotheses of hybrid vigor Genetics 33477ndash487

van de Kerk et al bull Florida Panther Population and Genetic Viability 31

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 21: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

MPC MVM

minus08 minus06 minus04 minus02 00 minus08 minus06 minus04 minus02 00

Kitten survival

Female subadult survival

Female prime adult survival

Female old adult survival

Male subadult survival

Male prime adult survival

Male old adult survival

Subadult reproduction probability

Prime adult reproduction probability

Old adult reproduction probability

Subadult annual kitten production

Prime adult annual kitten production

Old adult annual kitten production

PRCC

Para

met

er

Figure 12 Partial rank correlation coefficient (PRCC) between quasi‐extinction probabilities and the demographic parameters of the Florida panther for theminimum population count (MPC) and motor vehicle mortality (MVM) scenarios Error bars indicate standard deviation Based on data collected in South FloridaUSA 1981ndash2013

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 13 Average projected individual heterozygosities in the Florida panther population over 200 years without genetic management intervention and with eachof the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include theintroduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years Average individual heterozygosity peaks when pumas are added to the Floridapanther population Based on data collected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 23

Our research has further validated the success of genetic in-trogression by quantifying the reduction in the probability ofextinction of the panther population because of this manage-ment initiative These findings all bode well for continuedprogress toward recovery That said the Florida panther po-pulation remains small and isolated from other populations ofpuma from western North America thereby denoting that theeventual impact of genetic drift and inbreeding may negativelyaffect the population in the future Realizing this will continueto be an issue that managers will have to contemplate wemodeled the impact of varied genetic management scenarios todetermine the frequency of introgression along with the numberof panthers that should be released during each initiative tominimize cost and maximize benefits to the population Theseresults will prove invaluable to managers attempting to conservethe Florida panther

Genetic Variation Demographic Parametersand Persistence of Heterotic Benefits from GeneticIntrogressionOur measure of individual heterozygosity (1 minusHL) was higheron average for the admixed (0615) panthers than for those ofcanonical ancestry (0386) Levels of individual heterozygosityfor admixed panthers were comparable to levels observed in asample of pumas from west Texas (x plusmn SE= 0695plusmn 0019n= 41) which further demonstrates the improvement in levelsof genetic variation in the Florida population since 1995 The

correlation between HL and several demographic parametershas been noted in wild populations including cheetahs (Terrellet al 2016) and several species of birds such as European shags(Phalacrocorax aristotelis male and female survival probabilityVelando et al 2015) and Egyptian vulture (Neophron percnop-terus age at recruitment Agudo et al 2012) If we focus only onpanthers captured and radiocollared from 2012ndash2015 (FP211ndashFP240) all of which had estimated birth years ge2005 (ge10 yrpost‐introgression) those panthers had an average individualheterozygosity level of 0618plusmn 0029 highlighting the elevatedlevels of genetic variation that continue to persist in cohorts ofpanthers 5 generations after the release of female pumas fromTexasThe proportional membership values (q) from our

STRUCTURE analysis allowed us to delineate 2 clusters ofancestry for Florida panthers (Fig 4) During the pre‐in-trogression era the population was dominated by panthers thatretained a high percentage of the canonical ancestry which wasaffected by inbreeding depression The post‐introgression eraancestry values are comprised of many admixed individuals inessence demonstrating the success of the introgression in termsof increasing genetic variation in the population and mi-micking gene flow that historically occurred with panthers andother populations of pumas (Onorato et al 2010) A somewhatanalogous situation although abetted by natural immigrationwas evident in the ancestral variation in a small population ofpuma intersected by a major freeway in California USA In

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

75

80

85

90

95

100

130

150

170

190

Years

Popu

latio

n si

ze

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 14 Average projected population sizes in the Florida panther population over 200 years without intervention and with each of the genetic introgressionstrategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include the introduction of 5 10 or 15pumas from Texas USA every 10 20 40 or 80 years Peaks occur when pumas are added to the Florida panther population Based on data collected in SouthFlorida USA 1981ndash2013

24 Wildlife Monographs bull 203

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

25

50

75

100

0

25

50

75

100

Introgression interval (yrs)

PQ

E (

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 15 Average probability of quasi‐extinction (PQE) of the Florida panther population within 100 years and within 200 years without genetic managementintervention (N) and with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenariosIntrogression strategies include the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Errorbars indicate 5th and 95th percentiles Based on data collected in South Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

50

60

70

80

90

100

110

50

100

150

200

Years

Popu

latio

n si

ze

Figure 16 Average projected Florida panther population trajectories under a geneticintrogression strategy involving the release of 5 female pumas from Texas USA every20 years for the minimum population count (MPC) and motor vehicle mortality(MVM) scenarios Shaded area indicates 5th and 95th percentiles and isrepresentative of confidence intervals for other scenarios Based on data collected inSouth Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Figure 17 Average projected Florida panther population‐level heterozygositiesunder a genetic introgression strategy involving the release of 5 female pumas fromTexas USA every 20 years for the minimum population count (MPC) and motorvehicle mortality (MVM) scenarios Shaded area bounds the 5th and 95th percentilesand is representative of confidence intervals for other scenarios Based on datacollected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 25

this case the migration of just a single male across the freewayto the south resulted in an increase in the number of in-dividuals with admixed ancestry and substantially improvedgenetic diversity of the subpopulation south of the freeway(Riley et al 2014)Our estimate of overall kitten survival (0321plusmn 0056) was

similar to that reported by Hostetler et al (2010) who used13 years of post‐introgression data (0323plusmn 0071) These valuesare lower than kitten survival estimates derived from severalpuma populations in western North America (044ndash091 Loganand Sweanor 2001 Lambert et al 2006 Laundre et al 2007Robinson et al 2008 Ruth et al 2011) When we included onlydata for individuals that had been genetically sampled our es-timate was 15 higher The proportion of admixed kittens thatwere genetically sampled increased as the population expandedwhich could have resulted in higher estimates of kitten survivalbecause admixed kittens survive better than canonical kittensKitten survival was strongly influenced by genetic ancestry withsurvival rates of F1 kittens being almost twice that of canonicalkittens (Fig 5B) Consistent with the findings of Hostetler et al(2010) increases in heterozygosity coincided with increasedkitten survival whereas population density negatively affectedkitten survival Population density often has a strong negativeimpact on survival of young age classes due to infanticide andcannibalism which has been reported from several carnivore

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

50

100

150

0

50

100

150

Introgression interval (yrs)

TQE

(yrs

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 18 Average times until quasi‐extinction (TQE) for the Florida panther population within 100 years and within 200 years without genetic management interventionand with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategiesinclude the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Error bars indicate 5th and 95thpercentiles Based on data collected in South Florida USA 1981ndash2013

Table 5 Average cost (2017 US dollars) of introgression strategies employedover 100 years that involve the introduction of 5 10 or 15 pumas from TexasUSA into the Florida panther population in South Florida USA at an in-creasingly higher frequency Costs of capture (including transportation) quar-antine and post‐introgression monitoring were estimated by Roy McBrideMark Cunningham and Dave Onorato respectively

Frequency Component 5 pumas 10 pumas 15 pumas

Once Capture 25000 50000 75000

Quarantine 5000 10000 15000

Monitoring 10000 20000 30000

Total 40000 80000 120000

Every 80 years Capture 31250 62500 93750

Quarantine 6250 12500 18750

Monitoring 12500 25000 37500

Total 50000 100000 150000

Every 40 years Capture 62500 125000 187500

Quarantine 12500 25000 37500

Monitoring 25000 50000 75000

Total 100000 200000 300000

Every 20 years Capture 125000 250000 375000

Quarantine 25000 50000 75000

Monitoring 50000 100000 150000

Total 200000 400000 600000

Every 10 years Capture 250000 500000 750000

Quarantine 50000 100000 150000

Monitoring 100000 200000 300000

Total 400000 800000 1200000

26 Wildlife Monographs bull 203

species including polar bears (Ursus maritimus Derocher andWiig 1999) black bears (Ursus americanus Garrison et al 2007)and pumas (Logan and Sweanor 2001)Our estimates of sex‐ and age‐specific survival rates of subadult

and adult panthers (Table 3) and the effects of covariates on theserates were similar to those reported by Benson et al (2011) derivedfrom data collected through 2006 Our survival rates for males(065ndash077) were lower than females (078ndash097) for all 3 ageclasses (subadult prime adult and old adult) which is the typicaltrend observed in most puma populations (eg Ruth et al 19982011) However an unhunted puma population occupying afragmented urban landscape in southern California exhibited lowersurvival (0586 females 0525 males Vickers et al 2015) perhapsas a result of severe habitat fragmentation and the lack of extensiveswaths of public land protected in perpetuity Most populations ofpuma in western North America are typically subject to variedlevels of management via regulated hunting which often is biasedtoward the take of males (Cooley et al 2009 Ruth et al 2011)Consequently annual survival estimates from hunted populationsin northeast Washington (0679ndash0733 females 0341 malesRobinson et al 2008) and southeastern Arizona (0685 females0584 males Cunningham et al 2001) were generally lower thanthose we estimated for Florida panthersCause‐specific mortality analysis showed that male panthers

experienced a substantially greater mortality risk from in-traspecific aggression This differs from the pattern observedduring a long‐term study in Arizona where females were at ahigher risk of intraspecific mortality than males (46 of maleand 53 of female mortality Logan and Sweanor 2001)Mortality associated with intraspecific strife has been docu-mented in hunted and unhunted populations (this study Loganand Sweanor 2001 Stoner et al 2010) and its root cause hasbeen difficult to determine (eg population density and preypopulation trends) That being the case finding ways to reducewhat is often a major source of mortality for puma populationscontinues to pose a challenge to managersCanonical panthers were substantially more likely to die from

intraspecific aggression than were admixed panthers This mayat least partly explain the difference in survival between admixed

and canonical panthers Canonical panthers are known to bemore likely to suffer from varied correlates of inbreeding de-pression than admixed animals including atrial septal defects(Johnson et al 2010) and compromised immune systems(Roelke et al 1993) These factors have the potential to affectfitness and perhaps dominance of individuals when engaged inan intraspecific encounter A somewhat similar scenario wasobserved by Hogg et al (2006) in a population of bighorn sheepthat were part of an introgression project Admixed males in thispopulation had a greater probability of paternity than males thatwere less outbred This included coursing males with higherlevels of admixture being markedly more successful at fightingmales that were defending females in estrous (Hogg et al 2006)Heterosis resulting from genetic introgression is expected to be

the strongest in F1 individuals (Shull 1908 Crow 1948 Burkeand Arnold 2001 Waller 2015) and to progressively decline insubsequent generations (Pickup et al 2013 Frankham 2015)Given that admixed (especially F1) panthers survived better thancanonical panthers and that individual heterozygosity improvedsurvival of both sexes and all age classes our data provide evi-dence for heterotic advantage whereby individuals of mixedancestry had higher fitness (Shull 1908 Crow 1948) Our resultsare consistent with findings of other studies that involved in-tentional genetic introgression for conservation purposes Forexample Hogg et al (2006) detected substantial improvementsin survival and reproduction of introgressed bighorn sheep andMadsen et al (1999 2004) reported a dramatic increase in thepopulation of adders after genetic introgressionKnowledge of the persistence of benefits to a population ac-

crued via genetic introgression is essential for determiningwhen additional introgression may be warranted There is adepauperate amount of information regarding this topic and theidentification of factors that may influence persistence of thebenefits of genetic introgression (Tallmon et al 2004 Hedricket al 2014 Whiteley et al 2015) For example in minks(Neovison vison) Thirstrup et al (2014) found that outcrossingled to larger litters in F2 and F3 but this benefit disappeared insubsequent generations In a population of gray wolves Hedricket al (2014) suggested that benefits of genetic introgression may

Table 6 Costs and benefits accumulated over 100 years for genetic introgression at different intervals and with different numbers of pumas from Texas USAintroduced into the Florida panther population in South Florida USA Costs (2017 US dollars) include capture and transportation quarantine and post‐introgression monitoring Benefits are represented as average probability of quasi‐extinction (PQE and 5th and 95th percentiles) and changes (Δ) therein under thescenario using the minimum population count (McBride et al 2008 MPC) and the scenario using the motor vehicle mortality population estimates (McClintocket al 2015 MVM) The table is sorted by percent change in probability of quasi‐extinction under the MPC scenario

Interval (yr) Pumas Cost MPC PQE () ΔMPC PQE () MVM PQE () ΔMVM PQE ()

ndash 0 0 13 (0ndash99) 0 17 (0ndash100) 080 10 100000 12 (0ndash99) 10 (0ndash58) 16 (0ndash100) 5 (0ndash38)80 15 150000 12 (0ndash99) 13 (0ndash65) 15 (0ndash100) 8 (0ndash56)80 5 50000 12 (0ndash99) 14 (0ndash73) 15 (0ndash99) 9 (0ndash41)40 15 300000 10 (0ndash98) 26 (0ndash104) 14 (0ndash99) 13 (0ndash60)40 10 200000 9 (0ndash94) 35 (0ndash183) 13 (0ndash99) 21 (0ndash95)40 5 100000 8 (0ndash89) 39 (0ndash211) 12 (0ndash99) 26 (0ndash124)20 15 600000 8 (0ndash88) 42 (0ndash257) 13 (0ndash99) 24 (0ndash123)20 10 400000 6 (0ndash67) 54 (0ndash318) 10 (0ndash99) 37 (0ndash217)10 15 1200000 6 (0ndash53) 58 (0ndash372) 10 (0ndash99) 40 (0ndash208)20 5 200000 5 (0ndash50) 59 (0ndash338) 9 (0ndash99) 44 (0ndash352)10 10 800000 4 (0ndash16) 69 (0ndash489) 8 (0ndash92) 55 (0ndash401)10 5 400000 4 (0ndash7) 73 (0ndash502) 6 (0ndash71) 63 (0ndash503)

van de Kerk et al bull Florida Panther Population and Genetic Viability 27

wane after 2 or 3 generations The WrightndashFisher model ofgenetic drift predicts a geometric decline in expected hetero-zygosity at the rate of (1 minus 12Ne) per generation where Ne is theeffective population size (Hartl and Clark 1997 Frankham et al2014) Thus it seems logical to assume that the duration of thebenefits of introgression is limited especially in a species per-sisting in a small population such as Florida panthersWe expected Florida panther survival in the most recent years

(2005ndash2013) post‐introgression to differ from that observed in thedecade following the introgression in 1995 because pantherabundance and heterozygosity factors known to influence panthersurvival (Hostetler et al 2010 Benson et al 2011) have changed(Figs 2 and 6) We found that subadult and adult survival rates inrecent years (2005ndash2013) were similar to those in the period im-mediately following introgression (1995ndash2004 Fig 5C) overallkitten survival was similar to estimates of Hostetler et al (2010) Infact the pattern of covariate effects on Florida panther survivalreported by Benson et al (2011) and Hostetler et al (2010) hasbarely changed The negative effects of population density andpositive effects of heterozygosity may counterbalance survival ratesreducing population fluctuations Our result that benefits of geneticrestoration have remained in the population nearly for 5 genera-tions post‐intervention was surprising but also encouraging Pos-sible explanations for this observation may include 1) genetic ero-sion occurs at slower rate than previously thought 2) introducing 5migrants in 1 genetic intervention may have beneficial effects si-milar to those from 1 migrant per generation over multiple gen-erations or 3) other and as yet unknown factors may reduce therate of genetic erosion and subsequent demographic effects Adensity‐dependent effect on kitten survival could also explain whynon‐F1 admixed kittens did not survive substantially better thandid canonical kittens most kittens sampled from 2005ndash2013 wereadmixed and their survival was lower possibly because of a density‐dependent effect as the Florida panther population continuouslygrew during that period Trinkel et al (2010) also found thatpopulation density and inbreeding coefficient interacted to affectdemographic parameters and growth rate of an African lion po-pulation Likewise population density interacted with winterweather to affect the survival and population growth rates in Soaysheep (Ovis aries Milner et al 1999 Coulson et al 2001)

Population Dynamics and PersistenceThe finite deterministic and stochastic growth rates were gt10reflecting that much of the data used in this study were collectedfollowing genetic introgression in 1995 during which time thepopulation increased substantially Because our demographicparameter estimates did not change markedly in comparison tothose of Hostetler et al (2013) the similarity between thepopulation growth estimates from these studies was expectedBut these estimates reflect population growth during thedata‐collection period and do not predict future growth In factestimates of population size reported by McClintock et al(2015) suggest that population growth may already be slowing(Fig 2) A similar pattern was observed in an introduced po-pulation of African lion which increased steadily from 1995until 2001 then fluctuated around the presumed carrying ca-pacity thereafter (Trinkel et al 2010) Several other African lionpopulations that experienced various degrees of recovery

subsequently became relatively stable or exhibited decliningtrends (Bauer et al 2015)Our estimates of quasi‐extinction probabilities were lower than

those reported by Hostetler et al (2013) using a 2‐sex age‐structured matrix population model Lower probabilities ofquasi‐extinction than those reported by the earlier study forcomparable scenarios were likely a consequence of the fact thatthe estimates of abundance (McBride et al 2008) and thus thedensity‐dependent effects on survival of kittens and old panthers(gt10 yr) have changed in recent years Additionally these earlyestimates of the probability of quasi‐extinction and of time toquasi‐extinction did not consider genetic erosion that will in-evitably occur without management interventionUnder the MVM scenario the equilibrium population size was

twice that attained under the MPC scenario The MVM popula-tion estimates are approximately twice the MPCs (Fig 2) as aresult the simulated population under the MVM scenario achieveda higher equilibrium size Probability of quasi‐extinction under theMVM scenario was generally greater than that under the MPCscenario This result is a consequence of the fact that the estimateddensity dependence on kitten survival based on the MPC scenario isstronger than that for theMVM scenario (regression coefficients forabundance for the MPC scenario= minus0068 vs minus0002 for theMVM scenario Appendix B Table B1) Consequently simulatedpopulations under the MPC scenario have a stronger tendency tomove toward the equilibrium population size which inherentlyreduces the quasi‐extinction probability (eg Royama 1992)As expected for long‐lived mammals (Heppell et al 2000 Oli

and Dobson 2003) the population growth rate was proportio-nately most sensitive to changes in survival of prime adultsfollowed by that in younger age classes Global sensitivity ana-lysis in contrast showed that under all scenarios extinctionparameters were particularly sensitive to changes in survival ofFlorida panther kittens This result is a consequence of the highsensitivity of the population growth rate to kitten survival(Fig 9) and a larger standard error of kitten survival (Table 3)Results of our sensitivity analyses indicate that future researchshould focus on acquiring additional information on early lifestages to improve the accuracy and precision of estimates ofkitten survival Our results suggest that demographic variables(or their environmental drivers) that strongly influence extinc-tion risks are not necessarily those with the largest elasticityvalues A study of island fox (Bakker et al 2009) found thatextinction risk was strongly influenced by survival modifierparameters that had low elasticity values

Genetic Management When and How to InterveneThe use of genetic introgression as a management tool has al-ways been controversial and the probable effectiveness it mighthave in preventing the imminent demise of the panther popu-lation was initially debated (Creel 2006 Maehr et al 2006Pimm et al 2006) These debates notwithstanding the pantherpopulation had suffered from genetic morphological and bio-medical correlates of inbreeding before this management in-tervention (Johnson et al 2010 Onorato et al 2010) whereasthe post‐introgression period has been characterized by a sub-stantial reduction in the biomedical correlates of inbreeding andimprovements in demographic vigor and abundance (McBride

28 Wildlife Monographs bull 203

et al 2008 Hostetler et al 2010 2013 Johnson et al 2010Benson et al 2011 McClintock et al 2015) Nevertheless thepopulation remains small and isolated habitat is still being lostand there is no current possibility of natural gene flow betweenthis and other North American puma populations thus therecurrence of inbreeding‐related problems and subsequent po-pulation declines are inevitable The question therefore is notwhether genetic management of the Florida panther populationis needed but when and how it should be implementedWithout genetic introgression probabilities of quasi‐extinc-

tion were substantially greater than those from models thatincorporated periodic genetic introgression events (Fig 15)highlighting the importance of genetic management in smallisolated populations When 5 female pumas are added to theFlorida panther population every 10 years genetic variationincreased substantially under the MPC and MVM scenariosThe IBM predicted that the equillibrium population size wouldbe substantially larger when 5 pumas were released into thepopulation every 10 years than populations without intervention(ie no introgression) Releasing 15 Texas females did not af-ford substantially greater benefit to the equilibrium populationsize than did releasing only 5 Texas females Instead addingmore individuals caused increasingly large fluctuations in po-pulation size due to the numerical increases and density de-pendence (Fig 14) possibly diminishing the benefits associatedwith increased genetic variationCompared with the no‐intervention scenario (ie no genetic

introgression implemented) the introduction of 5 female pumasevery 10 years reduced quasi‐extinction probabilities from 13to 4 and from 17 to 6 for the MPC and MVM scenariosrespectively The benefit of genetic‐introgression strategies de-creased as the interval between successive management inter-ventions increased and no benefit was evident once the intervalreached 80 years (Fig 15) Introgression reduced the averagetime until quasi‐extinction (Fig 18) This somewhat counter‐intuitive result can be explained by the fact that repeated geneticintrogression makes the population more robust over timewhich will reduce the probability of extinction Consequentlyfew simulated population trajectories that fall below the criticalthreshold do so early reducing the mean time to extinctionAlso density dependence has a stabilizing effect on populations(Royama 1992) populations tend toward equilibrium popula-tion sizes which reduces the probability over time of popula-tions falling below quasi‐extinction threshold However timeuntil quasi‐extinction must be interpreted in conjunction withthe probability of quasi‐extinction which is very low for allscenarios with genetic introgression (Fig 15)Cost is a significant impediment to the implementation of a

genetic introgression program because captures relocations andmonitoring efforts before and after introgression are expensive(Edmands 2007 Frankham 2010b 2015 2016) Costs may beacceptable to society if the management actions result in sub-stantial fitness benefits especially if the species of concern ispopular and charismatic (Frankham 2015) We estimated the100‐year cost of introgression strategies modeled here to rangefrom $50000 to $12 million More expensive strategies gen-erally led to larger decreases in the probability of quasi‐extinc-tion but cheaper strategies also substantially improved

population viability (ie reduced quasi‐extinction probabilityTable 6) One of the cheaper strategies (5 pumas released every20 years) which cost an estimated $200000 over 100 yearsreduced the probability of quasi‐extinction by 59 and 44 forthe MPC and MVM scenarios respectively But these pre-liminary cost estimates are based on expenses incurred duringthe 1995 introgression and include costs of capture quarantinetransportation release and post‐release monitoring of femalesonly Although the cost for future genetic interventions wouldlikely be higher than those reported here the relative differencesin cost between alternative intervention strategies should remainapproximately the sameOur ability to simulate genetics and link it to the viability of

the Florida panther population is based on the empirically es-timated relationship between individual heterozygosity andsurvival Such heterozygosity and fitness correlations have beenstudied for decades (David 1998) but have only recently beenused to predict population viability (Benson et al 2016) noneto our knowledge have incorporated cost into their modelingefforts The mechanisms underlying heterozygosity and fitnesscorrelations remain unclear (David 1998 Szulkin et al 2010)and their significance as a proxy for inbreeding depression hasbeen debated (Balloux et al 2004 Miller and Coltman 2014)Nevertheless evidence continues to mount demonstratingpositive relationships between heterozygosity and survival(Coulson et al 1998ab Hansson et al 2004 Silva et al 2005Acevedo‐Whitehouse et al 2006 Jensen et al 2007 Velandoet al 2015) and between heterozygosity and fecundity (Seddonet al 2004 Charpentier et al 2005 Agudo et al 2012 Velandoet al 2015) Although the reported correlations are generallyweak their consequences can be substantial if population growthand persistence parameters are highly sensitive to the affectedvital rates (Velando et al 2015) Compared with scenarios thatignore genetics incorporating the effects of inbreeding on sur-vival increased quasi‐extinction probabilities within a 100‐yeartimeframe from 15 to 13 and from 14 to 17 for theMPC and MVM scenarios respectively These results high-light the importance of incorporating genetics when analyzingthe viability of small isolated populationsAlthough conservation biologists have recognized the im-

portance of genetics in population viability many still fail toincorporate genetic factors into PVA models (Reed et al 2002)Explicit consideration of genetics in PVAs requires long‐termgenetic and demographic data This permits the assessment ofthe functional relationship between genetic variability and de-mographic parameters Population viability analysis softwarepackages such as VORTEX (Lacy 1993 2000) offer a powerfulframework for individual‐based simulations but they often donot adequately capture a speciesrsquo life history or genetics Basedon a thorough review of the importance of genetic factors inwildlife conservation Frankham et al (2014) concluded thatgenetic factors can strongly influence the dynamics and persis-tence of small andor fragmented populations They furthernoted that ldquoMost PVA‐based risk assessments ignore or in-adequately model genetic factorsrdquo and recommended that ldquoPVAshould routinely include realistic inbreeding depressionhelliprdquo(Frankham et al 201457) Our custom‐built IBM permitted usto develop a model that adequately captured the Florida panther

van de Kerk et al bull Florida Panther Population and Genetic Viability 29

life history and we could parameterize the model with our long‐term genetic and demographic dataThe explicit consideration of the effects of inbreeding on

Florida panther population dynamics and persistence representsan important step forward Nonetheless our model remainsimperfect First we neglected the possible effects of habitat lossdetrimental anthropogenic activities climate change the prob-ability of immigration of pumas from western populationsdisease or catastrophes on demographic rates and populationviability Although immigration from puma populations outsideFlorida is possible its probability is very low given the extensivechallenges the anthropogenically altered landscape would posefor such a migrant to make it successfully to South Florida Weomitted mutations from our model because we have no in-formation on mutation rates though we expect that they are lowbased on values reported for other mammals (Kumar and Sub-ramanian 2002) Finally we only considered possible effects ofinbreeding on age‐specific survival rates because there was noevidence that heterozygosity affected female reproduction Lossof heterozygosity can negatively affect reproduction becauseinbred male panthers can suffer from cryptorchidism which canadversely affect their fecundity However given the polygynousmating system we expect the Florida panther population growthis primarily female‐limitedAlthough it is generally accepted that genetic introgression

only temporarily relieves a population from inbreeding depres-sion we know of no cases in which it has been implementedmore than once to conserve a threatened wildlife populationOur study provides an example of how demographic and mo-lecular data can be used within an IBM framework to evaluatethe efficacy of alternative genetic management strategies via theFlorida panther as a case study Our model can be adjusted forand applied to other species and offers great potential as a toolfor genetic management of small isolated wildlife populations

MANAGEMENT IMPLICATIONS

Our results and those of Hostetler et al (2013) and Johnsonet al (2010) provide persuasive evidence that the genetic in-tervention implemented in 1995 prevented the demise of theFlorida panther and restored demographic vigor to the popu-lation The Florida panther population remains small and iso-lated from other puma populations the closest puma populationis located gt1000 km away in Texas and the intervening land-scape is highly developed and fragmented with many interstate(and other high‐traffic) highways and major urban centersTheory suggests that 1 migrant per generation is sufficient tominimize the loss of genetic diversity (eg Mills and Allendorf1996) However the possibility of successful migration of apuma to South Florida followed by successful mating every ~5years is very low considering the distance between Texas andFlorida (Hedrick 1995) and the many risks and barriers thatdispersing pumas face (Beier 1995 Maehr et al 2002 Thatcheret al 2006) Consequently the pantherrsquos long‐term persistencewill likely depend on subsequent timely genetic introgressionsgiven the near‐impossibility of natural gene flow from otherpuma populations To that end our study offers considerableinsights into benefits of different genetic management strategiesand associated costs

Without genetic management the Florida panther populationfaces a substantial risk of quasi‐extinction within 100 years (10ndash46 depending on quasi‐extinction threshold and scenarios)Twenty‐three years have passed since genetic introgression wasimplemented and the population appears healthy as indicatedby measures of genetic variation increases in the populationsize and improvements in age‐specific survival rates Ourfindings suggest that releasing more pumas more frequently doesnot necessarily improve persistence probability because such astrategy would cause larger fluctuations in population size(Fig 14) which negatively affects persistence probability Thusextinction risks faced by inbred populations of large carnivorescan be substantially reduced by releasing approximately 5 im-migrants every 2 decades We suggest that such a managementstrategy be followed only in conjunction with continued long‐term monitoring of the population Our analyses demonstratethat population growth and persistence are highly sensitive tokitten and female (adult and subadult) survival Continuing tocollect data on these demographic variables along with bio-metric and genetic sampling will remain important to pantherrecovery and vigilance in identifying issues associated with in-breeding depression The collection of these monitoring datashould allow managers to detect any demographic or geneticchanges in a timely fashion and respond with genetic in-trogression or other management actions if warrantedRecovery objectives as defined by the USFWS in its current

recovery plan (USFWS 2008) delineate the need for additionalpopulations in the historical range to downlist or delist theFlorida panther If the only extant population of Floridapanthers continued to grow it could serve as a source ofpanthers to be translocated to suitable habitat to seed addi-tional populations Maintaining current information on thegenetic health of the population and precise estimates of itssize will be important in assessing whether it can withstand theremoval of a subset of animals for translocation Although the2017 documentation of a female panther with kittens north ofthe current breeding range is encouraging in terms of thepotential for population expansionmdashgiven that no female hadbeen recorded north of the Caloosahatchee River since 1972mdashthe re‐establishment of separate new populations will mostlikely require translocations by wildlife managers and a lengthysociopolitical processConservation of threatened or endangered species such as the

Florida panther is inherently challenging For panthers and otherlarge carnivores that require vast extents of natural habitat topersist habitat is typically the most limiting factor that will de-termine whether recovery efforts succeed Although geneticmanagement invariably plays a key role in the long‐term persis-tence of the panther population even a healthy population caneventually succumb to the impacts of habitat loss The humanpopulation of Florida continues to be one of the fastest‐growingin the nation the United States Census Bureau currently ranksFlorida as the third most populous Therefore habitat loss isgoing to continue to be a key issue for panthers Diligence towardpreserving panther habitat in its historical range via conservationeasements or other means and trying to keep such areas inter-connected via corridors is a goal stakeholders such as governmentagencies sportsmen non‐governmental organizations ranchers

30 Wildlife Monographs bull 203

and other private landowners must work on collaboratively toachieve

ACKNOWLEDGMENTS

We thank D Shindle D Jennings T OrsquoMeara D Land andC Belden for valuable advice throughout the project We thankS Bass M Criffield M Cunningham D Giardina D JansenA Johnson D Land M Lotz C McBride Rocky McBrideRowdy McBride Roy McBride L Oberhofer S Schulze staffsat Everglades National Park and Big Cypress National Preserveand others for assistance with fieldwork We also thank JAustin M Conroy J Hines J Laake S Mills J Nichols JHines M Sunquist J Fryxell and T Coulson for advice andassistance regarding data analysis and panther biology TheMuseum of Texas Tech University Natural Science ResearchLaboratory provided samples from pumas from west Texas forcomparative genetic analyses M Ben‐David D Land WMcCown E Hellgren and 4 anonymous reviewers providedmany helpful comments on the manuscript We thank NereaUbierna Lopez for completing the Spanish translation of theabstract We are grateful to the Department of ResearchComputing University of Florida for excellent high‐perfor-mance computing services We would like to thank US Fishand Wildlife Service Florida Fish and Wildlife ConservationCommission (FWC) US National Park Service QuantitativeSpatial Ecology Evolution and Environment (QSE3) In-tegrative Graduate Education and Research Traineeship(IGERT) program funded by the National Science Foundationand the University of Florida for financially supporting thisstudy Funding for panther research conducted by the FWC issupported predominantly via donations accrued with vehicleregistrations for ldquoProtect the Pantherrdquo license plates

LITERATURE CITEDAcevedo‐Whitehouse K T R Spraker E Lyons S R Melin F Gulland RL Delong and W Amos 2006 Contrasting effects of heterozygosity onsurvival and hookworm resistance in California sea lion pups MolecularEcology 151973ndash1982

Adams J R L M Vucetich P W Hedrick R O Peterson and J AVucetich 2011 Genomic sweep and potential genetic rescue during limitingenvironmental conditions in an isolated wolf population Proceedings of theRoyal Society B Biological Sciences 2783336ndash3344

Agresti A 2013 Categorical data analysis Third edition John Wiley amp SonsHoboken New Jersey USA

Agudo R M Carrete M Alcaide C Rico F Hiraldo and J A Donaacutezar2012 Genetic diversity at neutral and adaptive loci determines individualfitness in a long‐lived territorial bird Proceedings of the Royal Society BBiological Sciences 2793241ndash3249

Albeke S E N P Nibbelink and M Ben‐David 2015 Modeling behavior bycoastal river otter (Lontra canadensis) in response to prey availability in PrinceWilliam Sound Alaska a spatially‐explicit individual‐based approach PLOSOne 10e0126208

Alho J S K Vaumllimaumlki and J Merilauml 2010 Rhh an R extension for estimatingmultilocus heterozygosity and heterozygosity‐heterozygosity correlationMolecular Ecology Resources 10720ndash722

Alvarez K 1993 Twilight of the panther Myakka River Publishing SarasotaFlorida USA

Aparicio J M J Ortego and P J Cordero 2006 What should we weigh toestimate heterozygosity alleles or loci Molecular Ecology 154659ndash4665

Bakker V J D F Doak G W Roemer D K Garcelon T J Coonan S AMorrison C Lynch K Ralls and R Shaw 2009 Incorporating ecologicaldrivers and uncertainty into a demographic population viability analysis for theisland fox Ecological Monographs 7977ndash108

Balloux F W Amos and T Coulson 2004 Does heterozygosity estimateinbreeding in real populations Molecular Ecology 133021ndash3031

Barone M A M E Roelke J Howard J L Brown A E Anderson and DE Wildt 1994 Reproductive characteristics of male Florida panthersmdashcomparative studies from Florida Texas Colorado Latin‐America andNorth‐American Zoos Journal of Mammalogy 75150ndash162

Bateson Z W P O Dunn S D Hull A E Henschen J A Johnson and LA Whittingham 2014 Genetic restoration of a threatened population ofgreater prairie‐chickens Biological Conservation 17412ndash19

Bauer H G Chapron K Nowell P Henschel P Funston L T B HunterD W Macdonald and C Packer 2015 Lion (Panthera leo) populations aredeclining rapidly across Africa except in intensively managed areas Pro-ceedings of National Academy of Sciences of the United States of America11214894ndash14899

Beier P 1995 Dispersal of juvenile cougars in fragmented habitat Journal ofWildlife Management 59228ndash237

Benson J F J A Hostetler D P Onorato W E Johnson M E Roelke SJ OrsquoBrien D Jansen and M K Oli 2011 Intentional genetic introgressioninfluences survival of adults and subadults in a small inbred felid populationJournal of Animal Ecology 80958ndash967

Benson J F P J Mahoney J A Sikich L E K Serieys J P Pollinger H BErnest and S P D Riley 2016 Interactions between demography geneticsand landscape connectivity increase extinction probability for a small popu-lation of large carnivores in a major metropolitan area Proceedings of theRoyal Society B Biological Sciences 28320160957

Beverly F 1874 The Florida panther Forest and Stream 3290Boyce M S C V Haridas C T Lee and the NCEAS Stochastic Demo-graphy Working Group 2006 Demography in an increasingly variable worldTrends in Ecology amp Evolution 21141ndash148

Brook B W J J OrsquoGrady A P Chapman M A Burgman H R Akcakayaand R Frankham 2000 Predictive accuracy of population viability analysis inconservation biology Nature 404385ndash387

Brys R and H Jacquemyn 2016 Severe outbreeding and inbreeding de-pression maintain mating system differentiation in Epipactis (Orchidaceae)Journal of Evolutionary Biology 29352ndash359

Burke J M and M L Arnold 2001 Genetics and the fitness of hybridsAnnual Review of Genetics 3531ndash52

Burnham K P 1993 A theory for combined analysis of ring recovery andrecapture data Pages 199ndash213 in J‐D Lebreton and P M North editorsMarked individuals in the study of bird population Springer‐Verlag NewYork New York USA

Burnham K P and D R Anderson 2002 Model selection and multimodelinference a practical information‐theoretic approach Second edition SpringerVerlag New York New York USA

Caswell H 2001 Matrix population models construction analysis and in-terpretation Sinauer Associates Sunderland Massachusetts USA

Charpentier M J M Setchell F Prugnolle L A Knapp E J Wickings PPeignot and M Hossaert‐McKey 2005 Genetic diversity and reproductivesuccess in mandrills (Mandrillus sphinx) Proceedings of the NationalAcademy of Sciences of the United States of America 10216723ndash16728

Cooch E and G C White editors 2018 Program MARK a gentle in-troduction lthttpwwwphidotorgsoftwaremarkdocsbookgt Accessed 7Apr 2016

Cooley H S R B Wielgus G M Koehler H S Robinson and B TMaletzke 2009 Does hunting regulate cougar populations A test of thecompensatory mortality hypothesis Ecology 902913ndash2921

Coulson T E A Catchpole S D Albon B J Morgan J M Pemberton TH Clutton‐Brock M J Crawley and B T Grenfell 2001 Age sex densitywinter weather and population crashes in Soay sheep Science 2921528ndash1531

Coulson T J Pemberton S Albon M Beaumont T Marshall J Slate FGuinness and T Clutton‐Brock 1998a Microsatellites reveal heterosis in reddeer Proceedings of the Royal Society B Biological Sciences 265489ndash495

Coulson T N S D Albon J M Pemberton J Slate F E Guinness and TH Clutton‐Brock 1998b Genotype by environment interactions in wintersurvival in red deer Journal of Animal Ecology 67434ndash445

Cox D 1972 Regression models and life‐tables Journal of the Royal StatisticalSociety Series B Statistical Methodology 34187ndash220

Creel S 2006 Recovery of the Florida panthermdashgenetic rescue demographic rescueor both Response to Pimm et al (2006) Animal Conservation 9125ndash126

Crnokrak P and D A Roff 1999 Inbreeding depression in the wild Heredity83260ndash270

Crow J 1948 Alternative hypotheses of hybrid vigor Genetics 33477ndash487

van de Kerk et al bull Florida Panther Population and Genetic Viability 31

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 22: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

Our research has further validated the success of genetic in-trogression by quantifying the reduction in the probability ofextinction of the panther population because of this manage-ment initiative These findings all bode well for continuedprogress toward recovery That said the Florida panther po-pulation remains small and isolated from other populations ofpuma from western North America thereby denoting that theeventual impact of genetic drift and inbreeding may negativelyaffect the population in the future Realizing this will continueto be an issue that managers will have to contemplate wemodeled the impact of varied genetic management scenarios todetermine the frequency of introgression along with the numberof panthers that should be released during each initiative tominimize cost and maximize benefits to the population Theseresults will prove invaluable to managers attempting to conservethe Florida panther

Genetic Variation Demographic Parametersand Persistence of Heterotic Benefits from GeneticIntrogressionOur measure of individual heterozygosity (1 minusHL) was higheron average for the admixed (0615) panthers than for those ofcanonical ancestry (0386) Levels of individual heterozygosityfor admixed panthers were comparable to levels observed in asample of pumas from west Texas (x plusmn SE= 0695plusmn 0019n= 41) which further demonstrates the improvement in levelsof genetic variation in the Florida population since 1995 The

correlation between HL and several demographic parametershas been noted in wild populations including cheetahs (Terrellet al 2016) and several species of birds such as European shags(Phalacrocorax aristotelis male and female survival probabilityVelando et al 2015) and Egyptian vulture (Neophron percnop-terus age at recruitment Agudo et al 2012) If we focus only onpanthers captured and radiocollared from 2012ndash2015 (FP211ndashFP240) all of which had estimated birth years ge2005 (ge10 yrpost‐introgression) those panthers had an average individualheterozygosity level of 0618plusmn 0029 highlighting the elevatedlevels of genetic variation that continue to persist in cohorts ofpanthers 5 generations after the release of female pumas fromTexasThe proportional membership values (q) from our

STRUCTURE analysis allowed us to delineate 2 clusters ofancestry for Florida panthers (Fig 4) During the pre‐in-trogression era the population was dominated by panthers thatretained a high percentage of the canonical ancestry which wasaffected by inbreeding depression The post‐introgression eraancestry values are comprised of many admixed individuals inessence demonstrating the success of the introgression in termsof increasing genetic variation in the population and mi-micking gene flow that historically occurred with panthers andother populations of pumas (Onorato et al 2010) A somewhatanalogous situation although abetted by natural immigrationwas evident in the ancestral variation in a small population ofpuma intersected by a major freeway in California USA In

no intervention 5 TX females 10 TX females 15 TX females

MP

CM

VM

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

75

80

85

90

95

100

130

150

170

190

Years

Popu

latio

n si

ze

Introgressioninterval (yrs)

10

20

40

80

Never

Figure 14 Average projected population sizes in the Florida panther population over 200 years without intervention and with each of the genetic introgressionstrategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategies include the introduction of 5 10 or 15pumas from Texas USA every 10 20 40 or 80 years Peaks occur when pumas are added to the Florida panther population Based on data collected in SouthFlorida USA 1981ndash2013

24 Wildlife Monographs bull 203

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

25

50

75

100

0

25

50

75

100

Introgression interval (yrs)

PQ

E (

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 15 Average probability of quasi‐extinction (PQE) of the Florida panther population within 100 years and within 200 years without genetic managementintervention (N) and with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenariosIntrogression strategies include the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Errorbars indicate 5th and 95th percentiles Based on data collected in South Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

50

60

70

80

90

100

110

50

100

150

200

Years

Popu

latio

n si

ze

Figure 16 Average projected Florida panther population trajectories under a geneticintrogression strategy involving the release of 5 female pumas from Texas USA every20 years for the minimum population count (MPC) and motor vehicle mortality(MVM) scenarios Shaded area indicates 5th and 95th percentiles and isrepresentative of confidence intervals for other scenarios Based on data collected inSouth Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Figure 17 Average projected Florida panther population‐level heterozygositiesunder a genetic introgression strategy involving the release of 5 female pumas fromTexas USA every 20 years for the minimum population count (MPC) and motorvehicle mortality (MVM) scenarios Shaded area bounds the 5th and 95th percentilesand is representative of confidence intervals for other scenarios Based on datacollected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 25

this case the migration of just a single male across the freewayto the south resulted in an increase in the number of in-dividuals with admixed ancestry and substantially improvedgenetic diversity of the subpopulation south of the freeway(Riley et al 2014)Our estimate of overall kitten survival (0321plusmn 0056) was

similar to that reported by Hostetler et al (2010) who used13 years of post‐introgression data (0323plusmn 0071) These valuesare lower than kitten survival estimates derived from severalpuma populations in western North America (044ndash091 Loganand Sweanor 2001 Lambert et al 2006 Laundre et al 2007Robinson et al 2008 Ruth et al 2011) When we included onlydata for individuals that had been genetically sampled our es-timate was 15 higher The proportion of admixed kittens thatwere genetically sampled increased as the population expandedwhich could have resulted in higher estimates of kitten survivalbecause admixed kittens survive better than canonical kittensKitten survival was strongly influenced by genetic ancestry withsurvival rates of F1 kittens being almost twice that of canonicalkittens (Fig 5B) Consistent with the findings of Hostetler et al(2010) increases in heterozygosity coincided with increasedkitten survival whereas population density negatively affectedkitten survival Population density often has a strong negativeimpact on survival of young age classes due to infanticide andcannibalism which has been reported from several carnivore

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

50

100

150

0

50

100

150

Introgression interval (yrs)

TQE

(yrs

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 18 Average times until quasi‐extinction (TQE) for the Florida panther population within 100 years and within 200 years without genetic management interventionand with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategiesinclude the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Error bars indicate 5th and 95thpercentiles Based on data collected in South Florida USA 1981ndash2013

Table 5 Average cost (2017 US dollars) of introgression strategies employedover 100 years that involve the introduction of 5 10 or 15 pumas from TexasUSA into the Florida panther population in South Florida USA at an in-creasingly higher frequency Costs of capture (including transportation) quar-antine and post‐introgression monitoring were estimated by Roy McBrideMark Cunningham and Dave Onorato respectively

Frequency Component 5 pumas 10 pumas 15 pumas

Once Capture 25000 50000 75000

Quarantine 5000 10000 15000

Monitoring 10000 20000 30000

Total 40000 80000 120000

Every 80 years Capture 31250 62500 93750

Quarantine 6250 12500 18750

Monitoring 12500 25000 37500

Total 50000 100000 150000

Every 40 years Capture 62500 125000 187500

Quarantine 12500 25000 37500

Monitoring 25000 50000 75000

Total 100000 200000 300000

Every 20 years Capture 125000 250000 375000

Quarantine 25000 50000 75000

Monitoring 50000 100000 150000

Total 200000 400000 600000

Every 10 years Capture 250000 500000 750000

Quarantine 50000 100000 150000

Monitoring 100000 200000 300000

Total 400000 800000 1200000

26 Wildlife Monographs bull 203

species including polar bears (Ursus maritimus Derocher andWiig 1999) black bears (Ursus americanus Garrison et al 2007)and pumas (Logan and Sweanor 2001)Our estimates of sex‐ and age‐specific survival rates of subadult

and adult panthers (Table 3) and the effects of covariates on theserates were similar to those reported by Benson et al (2011) derivedfrom data collected through 2006 Our survival rates for males(065ndash077) were lower than females (078ndash097) for all 3 ageclasses (subadult prime adult and old adult) which is the typicaltrend observed in most puma populations (eg Ruth et al 19982011) However an unhunted puma population occupying afragmented urban landscape in southern California exhibited lowersurvival (0586 females 0525 males Vickers et al 2015) perhapsas a result of severe habitat fragmentation and the lack of extensiveswaths of public land protected in perpetuity Most populations ofpuma in western North America are typically subject to variedlevels of management via regulated hunting which often is biasedtoward the take of males (Cooley et al 2009 Ruth et al 2011)Consequently annual survival estimates from hunted populationsin northeast Washington (0679ndash0733 females 0341 malesRobinson et al 2008) and southeastern Arizona (0685 females0584 males Cunningham et al 2001) were generally lower thanthose we estimated for Florida panthersCause‐specific mortality analysis showed that male panthers

experienced a substantially greater mortality risk from in-traspecific aggression This differs from the pattern observedduring a long‐term study in Arizona where females were at ahigher risk of intraspecific mortality than males (46 of maleand 53 of female mortality Logan and Sweanor 2001)Mortality associated with intraspecific strife has been docu-mented in hunted and unhunted populations (this study Loganand Sweanor 2001 Stoner et al 2010) and its root cause hasbeen difficult to determine (eg population density and preypopulation trends) That being the case finding ways to reducewhat is often a major source of mortality for puma populationscontinues to pose a challenge to managersCanonical panthers were substantially more likely to die from

intraspecific aggression than were admixed panthers This mayat least partly explain the difference in survival between admixed

and canonical panthers Canonical panthers are known to bemore likely to suffer from varied correlates of inbreeding de-pression than admixed animals including atrial septal defects(Johnson et al 2010) and compromised immune systems(Roelke et al 1993) These factors have the potential to affectfitness and perhaps dominance of individuals when engaged inan intraspecific encounter A somewhat similar scenario wasobserved by Hogg et al (2006) in a population of bighorn sheepthat were part of an introgression project Admixed males in thispopulation had a greater probability of paternity than males thatwere less outbred This included coursing males with higherlevels of admixture being markedly more successful at fightingmales that were defending females in estrous (Hogg et al 2006)Heterosis resulting from genetic introgression is expected to be

the strongest in F1 individuals (Shull 1908 Crow 1948 Burkeand Arnold 2001 Waller 2015) and to progressively decline insubsequent generations (Pickup et al 2013 Frankham 2015)Given that admixed (especially F1) panthers survived better thancanonical panthers and that individual heterozygosity improvedsurvival of both sexes and all age classes our data provide evi-dence for heterotic advantage whereby individuals of mixedancestry had higher fitness (Shull 1908 Crow 1948) Our resultsare consistent with findings of other studies that involved in-tentional genetic introgression for conservation purposes Forexample Hogg et al (2006) detected substantial improvementsin survival and reproduction of introgressed bighorn sheep andMadsen et al (1999 2004) reported a dramatic increase in thepopulation of adders after genetic introgressionKnowledge of the persistence of benefits to a population ac-

crued via genetic introgression is essential for determiningwhen additional introgression may be warranted There is adepauperate amount of information regarding this topic and theidentification of factors that may influence persistence of thebenefits of genetic introgression (Tallmon et al 2004 Hedricket al 2014 Whiteley et al 2015) For example in minks(Neovison vison) Thirstrup et al (2014) found that outcrossingled to larger litters in F2 and F3 but this benefit disappeared insubsequent generations In a population of gray wolves Hedricket al (2014) suggested that benefits of genetic introgression may

Table 6 Costs and benefits accumulated over 100 years for genetic introgression at different intervals and with different numbers of pumas from Texas USAintroduced into the Florida panther population in South Florida USA Costs (2017 US dollars) include capture and transportation quarantine and post‐introgression monitoring Benefits are represented as average probability of quasi‐extinction (PQE and 5th and 95th percentiles) and changes (Δ) therein under thescenario using the minimum population count (McBride et al 2008 MPC) and the scenario using the motor vehicle mortality population estimates (McClintocket al 2015 MVM) The table is sorted by percent change in probability of quasi‐extinction under the MPC scenario

Interval (yr) Pumas Cost MPC PQE () ΔMPC PQE () MVM PQE () ΔMVM PQE ()

ndash 0 0 13 (0ndash99) 0 17 (0ndash100) 080 10 100000 12 (0ndash99) 10 (0ndash58) 16 (0ndash100) 5 (0ndash38)80 15 150000 12 (0ndash99) 13 (0ndash65) 15 (0ndash100) 8 (0ndash56)80 5 50000 12 (0ndash99) 14 (0ndash73) 15 (0ndash99) 9 (0ndash41)40 15 300000 10 (0ndash98) 26 (0ndash104) 14 (0ndash99) 13 (0ndash60)40 10 200000 9 (0ndash94) 35 (0ndash183) 13 (0ndash99) 21 (0ndash95)40 5 100000 8 (0ndash89) 39 (0ndash211) 12 (0ndash99) 26 (0ndash124)20 15 600000 8 (0ndash88) 42 (0ndash257) 13 (0ndash99) 24 (0ndash123)20 10 400000 6 (0ndash67) 54 (0ndash318) 10 (0ndash99) 37 (0ndash217)10 15 1200000 6 (0ndash53) 58 (0ndash372) 10 (0ndash99) 40 (0ndash208)20 5 200000 5 (0ndash50) 59 (0ndash338) 9 (0ndash99) 44 (0ndash352)10 10 800000 4 (0ndash16) 69 (0ndash489) 8 (0ndash92) 55 (0ndash401)10 5 400000 4 (0ndash7) 73 (0ndash502) 6 (0ndash71) 63 (0ndash503)

van de Kerk et al bull Florida Panther Population and Genetic Viability 27

wane after 2 or 3 generations The WrightndashFisher model ofgenetic drift predicts a geometric decline in expected hetero-zygosity at the rate of (1 minus 12Ne) per generation where Ne is theeffective population size (Hartl and Clark 1997 Frankham et al2014) Thus it seems logical to assume that the duration of thebenefits of introgression is limited especially in a species per-sisting in a small population such as Florida panthersWe expected Florida panther survival in the most recent years

(2005ndash2013) post‐introgression to differ from that observed in thedecade following the introgression in 1995 because pantherabundance and heterozygosity factors known to influence panthersurvival (Hostetler et al 2010 Benson et al 2011) have changed(Figs 2 and 6) We found that subadult and adult survival rates inrecent years (2005ndash2013) were similar to those in the period im-mediately following introgression (1995ndash2004 Fig 5C) overallkitten survival was similar to estimates of Hostetler et al (2010) Infact the pattern of covariate effects on Florida panther survivalreported by Benson et al (2011) and Hostetler et al (2010) hasbarely changed The negative effects of population density andpositive effects of heterozygosity may counterbalance survival ratesreducing population fluctuations Our result that benefits of geneticrestoration have remained in the population nearly for 5 genera-tions post‐intervention was surprising but also encouraging Pos-sible explanations for this observation may include 1) genetic ero-sion occurs at slower rate than previously thought 2) introducing 5migrants in 1 genetic intervention may have beneficial effects si-milar to those from 1 migrant per generation over multiple gen-erations or 3) other and as yet unknown factors may reduce therate of genetic erosion and subsequent demographic effects Adensity‐dependent effect on kitten survival could also explain whynon‐F1 admixed kittens did not survive substantially better thandid canonical kittens most kittens sampled from 2005ndash2013 wereadmixed and their survival was lower possibly because of a density‐dependent effect as the Florida panther population continuouslygrew during that period Trinkel et al (2010) also found thatpopulation density and inbreeding coefficient interacted to affectdemographic parameters and growth rate of an African lion po-pulation Likewise population density interacted with winterweather to affect the survival and population growth rates in Soaysheep (Ovis aries Milner et al 1999 Coulson et al 2001)

Population Dynamics and PersistenceThe finite deterministic and stochastic growth rates were gt10reflecting that much of the data used in this study were collectedfollowing genetic introgression in 1995 during which time thepopulation increased substantially Because our demographicparameter estimates did not change markedly in comparison tothose of Hostetler et al (2013) the similarity between thepopulation growth estimates from these studies was expectedBut these estimates reflect population growth during thedata‐collection period and do not predict future growth In factestimates of population size reported by McClintock et al(2015) suggest that population growth may already be slowing(Fig 2) A similar pattern was observed in an introduced po-pulation of African lion which increased steadily from 1995until 2001 then fluctuated around the presumed carrying ca-pacity thereafter (Trinkel et al 2010) Several other African lionpopulations that experienced various degrees of recovery

subsequently became relatively stable or exhibited decliningtrends (Bauer et al 2015)Our estimates of quasi‐extinction probabilities were lower than

those reported by Hostetler et al (2013) using a 2‐sex age‐structured matrix population model Lower probabilities ofquasi‐extinction than those reported by the earlier study forcomparable scenarios were likely a consequence of the fact thatthe estimates of abundance (McBride et al 2008) and thus thedensity‐dependent effects on survival of kittens and old panthers(gt10 yr) have changed in recent years Additionally these earlyestimates of the probability of quasi‐extinction and of time toquasi‐extinction did not consider genetic erosion that will in-evitably occur without management interventionUnder the MVM scenario the equilibrium population size was

twice that attained under the MPC scenario The MVM popula-tion estimates are approximately twice the MPCs (Fig 2) as aresult the simulated population under the MVM scenario achieveda higher equilibrium size Probability of quasi‐extinction under theMVM scenario was generally greater than that under the MPCscenario This result is a consequence of the fact that the estimateddensity dependence on kitten survival based on the MPC scenario isstronger than that for theMVM scenario (regression coefficients forabundance for the MPC scenario= minus0068 vs minus0002 for theMVM scenario Appendix B Table B1) Consequently simulatedpopulations under the MPC scenario have a stronger tendency tomove toward the equilibrium population size which inherentlyreduces the quasi‐extinction probability (eg Royama 1992)As expected for long‐lived mammals (Heppell et al 2000 Oli

and Dobson 2003) the population growth rate was proportio-nately most sensitive to changes in survival of prime adultsfollowed by that in younger age classes Global sensitivity ana-lysis in contrast showed that under all scenarios extinctionparameters were particularly sensitive to changes in survival ofFlorida panther kittens This result is a consequence of the highsensitivity of the population growth rate to kitten survival(Fig 9) and a larger standard error of kitten survival (Table 3)Results of our sensitivity analyses indicate that future researchshould focus on acquiring additional information on early lifestages to improve the accuracy and precision of estimates ofkitten survival Our results suggest that demographic variables(or their environmental drivers) that strongly influence extinc-tion risks are not necessarily those with the largest elasticityvalues A study of island fox (Bakker et al 2009) found thatextinction risk was strongly influenced by survival modifierparameters that had low elasticity values

Genetic Management When and How to InterveneThe use of genetic introgression as a management tool has al-ways been controversial and the probable effectiveness it mighthave in preventing the imminent demise of the panther popu-lation was initially debated (Creel 2006 Maehr et al 2006Pimm et al 2006) These debates notwithstanding the pantherpopulation had suffered from genetic morphological and bio-medical correlates of inbreeding before this management in-tervention (Johnson et al 2010 Onorato et al 2010) whereasthe post‐introgression period has been characterized by a sub-stantial reduction in the biomedical correlates of inbreeding andimprovements in demographic vigor and abundance (McBride

28 Wildlife Monographs bull 203

et al 2008 Hostetler et al 2010 2013 Johnson et al 2010Benson et al 2011 McClintock et al 2015) Nevertheless thepopulation remains small and isolated habitat is still being lostand there is no current possibility of natural gene flow betweenthis and other North American puma populations thus therecurrence of inbreeding‐related problems and subsequent po-pulation declines are inevitable The question therefore is notwhether genetic management of the Florida panther populationis needed but when and how it should be implementedWithout genetic introgression probabilities of quasi‐extinc-

tion were substantially greater than those from models thatincorporated periodic genetic introgression events (Fig 15)highlighting the importance of genetic management in smallisolated populations When 5 female pumas are added to theFlorida panther population every 10 years genetic variationincreased substantially under the MPC and MVM scenariosThe IBM predicted that the equillibrium population size wouldbe substantially larger when 5 pumas were released into thepopulation every 10 years than populations without intervention(ie no introgression) Releasing 15 Texas females did not af-ford substantially greater benefit to the equilibrium populationsize than did releasing only 5 Texas females Instead addingmore individuals caused increasingly large fluctuations in po-pulation size due to the numerical increases and density de-pendence (Fig 14) possibly diminishing the benefits associatedwith increased genetic variationCompared with the no‐intervention scenario (ie no genetic

introgression implemented) the introduction of 5 female pumasevery 10 years reduced quasi‐extinction probabilities from 13to 4 and from 17 to 6 for the MPC and MVM scenariosrespectively The benefit of genetic‐introgression strategies de-creased as the interval between successive management inter-ventions increased and no benefit was evident once the intervalreached 80 years (Fig 15) Introgression reduced the averagetime until quasi‐extinction (Fig 18) This somewhat counter‐intuitive result can be explained by the fact that repeated geneticintrogression makes the population more robust over timewhich will reduce the probability of extinction Consequentlyfew simulated population trajectories that fall below the criticalthreshold do so early reducing the mean time to extinctionAlso density dependence has a stabilizing effect on populations(Royama 1992) populations tend toward equilibrium popula-tion sizes which reduces the probability over time of popula-tions falling below quasi‐extinction threshold However timeuntil quasi‐extinction must be interpreted in conjunction withthe probability of quasi‐extinction which is very low for allscenarios with genetic introgression (Fig 15)Cost is a significant impediment to the implementation of a

genetic introgression program because captures relocations andmonitoring efforts before and after introgression are expensive(Edmands 2007 Frankham 2010b 2015 2016) Costs may beacceptable to society if the management actions result in sub-stantial fitness benefits especially if the species of concern ispopular and charismatic (Frankham 2015) We estimated the100‐year cost of introgression strategies modeled here to rangefrom $50000 to $12 million More expensive strategies gen-erally led to larger decreases in the probability of quasi‐extinc-tion but cheaper strategies also substantially improved

population viability (ie reduced quasi‐extinction probabilityTable 6) One of the cheaper strategies (5 pumas released every20 years) which cost an estimated $200000 over 100 yearsreduced the probability of quasi‐extinction by 59 and 44 forthe MPC and MVM scenarios respectively But these pre-liminary cost estimates are based on expenses incurred duringthe 1995 introgression and include costs of capture quarantinetransportation release and post‐release monitoring of femalesonly Although the cost for future genetic interventions wouldlikely be higher than those reported here the relative differencesin cost between alternative intervention strategies should remainapproximately the sameOur ability to simulate genetics and link it to the viability of

the Florida panther population is based on the empirically es-timated relationship between individual heterozygosity andsurvival Such heterozygosity and fitness correlations have beenstudied for decades (David 1998) but have only recently beenused to predict population viability (Benson et al 2016) noneto our knowledge have incorporated cost into their modelingefforts The mechanisms underlying heterozygosity and fitnesscorrelations remain unclear (David 1998 Szulkin et al 2010)and their significance as a proxy for inbreeding depression hasbeen debated (Balloux et al 2004 Miller and Coltman 2014)Nevertheless evidence continues to mount demonstratingpositive relationships between heterozygosity and survival(Coulson et al 1998ab Hansson et al 2004 Silva et al 2005Acevedo‐Whitehouse et al 2006 Jensen et al 2007 Velandoet al 2015) and between heterozygosity and fecundity (Seddonet al 2004 Charpentier et al 2005 Agudo et al 2012 Velandoet al 2015) Although the reported correlations are generallyweak their consequences can be substantial if population growthand persistence parameters are highly sensitive to the affectedvital rates (Velando et al 2015) Compared with scenarios thatignore genetics incorporating the effects of inbreeding on sur-vival increased quasi‐extinction probabilities within a 100‐yeartimeframe from 15 to 13 and from 14 to 17 for theMPC and MVM scenarios respectively These results high-light the importance of incorporating genetics when analyzingthe viability of small isolated populationsAlthough conservation biologists have recognized the im-

portance of genetics in population viability many still fail toincorporate genetic factors into PVA models (Reed et al 2002)Explicit consideration of genetics in PVAs requires long‐termgenetic and demographic data This permits the assessment ofthe functional relationship between genetic variability and de-mographic parameters Population viability analysis softwarepackages such as VORTEX (Lacy 1993 2000) offer a powerfulframework for individual‐based simulations but they often donot adequately capture a speciesrsquo life history or genetics Basedon a thorough review of the importance of genetic factors inwildlife conservation Frankham et al (2014) concluded thatgenetic factors can strongly influence the dynamics and persis-tence of small andor fragmented populations They furthernoted that ldquoMost PVA‐based risk assessments ignore or in-adequately model genetic factorsrdquo and recommended that ldquoPVAshould routinely include realistic inbreeding depressionhelliprdquo(Frankham et al 201457) Our custom‐built IBM permitted usto develop a model that adequately captured the Florida panther

van de Kerk et al bull Florida Panther Population and Genetic Viability 29

life history and we could parameterize the model with our long‐term genetic and demographic dataThe explicit consideration of the effects of inbreeding on

Florida panther population dynamics and persistence representsan important step forward Nonetheless our model remainsimperfect First we neglected the possible effects of habitat lossdetrimental anthropogenic activities climate change the prob-ability of immigration of pumas from western populationsdisease or catastrophes on demographic rates and populationviability Although immigration from puma populations outsideFlorida is possible its probability is very low given the extensivechallenges the anthropogenically altered landscape would posefor such a migrant to make it successfully to South Florida Weomitted mutations from our model because we have no in-formation on mutation rates though we expect that they are lowbased on values reported for other mammals (Kumar and Sub-ramanian 2002) Finally we only considered possible effects ofinbreeding on age‐specific survival rates because there was noevidence that heterozygosity affected female reproduction Lossof heterozygosity can negatively affect reproduction becauseinbred male panthers can suffer from cryptorchidism which canadversely affect their fecundity However given the polygynousmating system we expect the Florida panther population growthis primarily female‐limitedAlthough it is generally accepted that genetic introgression

only temporarily relieves a population from inbreeding depres-sion we know of no cases in which it has been implementedmore than once to conserve a threatened wildlife populationOur study provides an example of how demographic and mo-lecular data can be used within an IBM framework to evaluatethe efficacy of alternative genetic management strategies via theFlorida panther as a case study Our model can be adjusted forand applied to other species and offers great potential as a toolfor genetic management of small isolated wildlife populations

MANAGEMENT IMPLICATIONS

Our results and those of Hostetler et al (2013) and Johnsonet al (2010) provide persuasive evidence that the genetic in-tervention implemented in 1995 prevented the demise of theFlorida panther and restored demographic vigor to the popu-lation The Florida panther population remains small and iso-lated from other puma populations the closest puma populationis located gt1000 km away in Texas and the intervening land-scape is highly developed and fragmented with many interstate(and other high‐traffic) highways and major urban centersTheory suggests that 1 migrant per generation is sufficient tominimize the loss of genetic diversity (eg Mills and Allendorf1996) However the possibility of successful migration of apuma to South Florida followed by successful mating every ~5years is very low considering the distance between Texas andFlorida (Hedrick 1995) and the many risks and barriers thatdispersing pumas face (Beier 1995 Maehr et al 2002 Thatcheret al 2006) Consequently the pantherrsquos long‐term persistencewill likely depend on subsequent timely genetic introgressionsgiven the near‐impossibility of natural gene flow from otherpuma populations To that end our study offers considerableinsights into benefits of different genetic management strategiesand associated costs

Without genetic management the Florida panther populationfaces a substantial risk of quasi‐extinction within 100 years (10ndash46 depending on quasi‐extinction threshold and scenarios)Twenty‐three years have passed since genetic introgression wasimplemented and the population appears healthy as indicatedby measures of genetic variation increases in the populationsize and improvements in age‐specific survival rates Ourfindings suggest that releasing more pumas more frequently doesnot necessarily improve persistence probability because such astrategy would cause larger fluctuations in population size(Fig 14) which negatively affects persistence probability Thusextinction risks faced by inbred populations of large carnivorescan be substantially reduced by releasing approximately 5 im-migrants every 2 decades We suggest that such a managementstrategy be followed only in conjunction with continued long‐term monitoring of the population Our analyses demonstratethat population growth and persistence are highly sensitive tokitten and female (adult and subadult) survival Continuing tocollect data on these demographic variables along with bio-metric and genetic sampling will remain important to pantherrecovery and vigilance in identifying issues associated with in-breeding depression The collection of these monitoring datashould allow managers to detect any demographic or geneticchanges in a timely fashion and respond with genetic in-trogression or other management actions if warrantedRecovery objectives as defined by the USFWS in its current

recovery plan (USFWS 2008) delineate the need for additionalpopulations in the historical range to downlist or delist theFlorida panther If the only extant population of Floridapanthers continued to grow it could serve as a source ofpanthers to be translocated to suitable habitat to seed addi-tional populations Maintaining current information on thegenetic health of the population and precise estimates of itssize will be important in assessing whether it can withstand theremoval of a subset of animals for translocation Although the2017 documentation of a female panther with kittens north ofthe current breeding range is encouraging in terms of thepotential for population expansionmdashgiven that no female hadbeen recorded north of the Caloosahatchee River since 1972mdashthe re‐establishment of separate new populations will mostlikely require translocations by wildlife managers and a lengthysociopolitical processConservation of threatened or endangered species such as the

Florida panther is inherently challenging For panthers and otherlarge carnivores that require vast extents of natural habitat topersist habitat is typically the most limiting factor that will de-termine whether recovery efforts succeed Although geneticmanagement invariably plays a key role in the long‐term persis-tence of the panther population even a healthy population caneventually succumb to the impacts of habitat loss The humanpopulation of Florida continues to be one of the fastest‐growingin the nation the United States Census Bureau currently ranksFlorida as the third most populous Therefore habitat loss isgoing to continue to be a key issue for panthers Diligence towardpreserving panther habitat in its historical range via conservationeasements or other means and trying to keep such areas inter-connected via corridors is a goal stakeholders such as governmentagencies sportsmen non‐governmental organizations ranchers

30 Wildlife Monographs bull 203

and other private landowners must work on collaboratively toachieve

ACKNOWLEDGMENTS

We thank D Shindle D Jennings T OrsquoMeara D Land andC Belden for valuable advice throughout the project We thankS Bass M Criffield M Cunningham D Giardina D JansenA Johnson D Land M Lotz C McBride Rocky McBrideRowdy McBride Roy McBride L Oberhofer S Schulze staffsat Everglades National Park and Big Cypress National Preserveand others for assistance with fieldwork We also thank JAustin M Conroy J Hines J Laake S Mills J Nichols JHines M Sunquist J Fryxell and T Coulson for advice andassistance regarding data analysis and panther biology TheMuseum of Texas Tech University Natural Science ResearchLaboratory provided samples from pumas from west Texas forcomparative genetic analyses M Ben‐David D Land WMcCown E Hellgren and 4 anonymous reviewers providedmany helpful comments on the manuscript We thank NereaUbierna Lopez for completing the Spanish translation of theabstract We are grateful to the Department of ResearchComputing University of Florida for excellent high‐perfor-mance computing services We would like to thank US Fishand Wildlife Service Florida Fish and Wildlife ConservationCommission (FWC) US National Park Service QuantitativeSpatial Ecology Evolution and Environment (QSE3) In-tegrative Graduate Education and Research Traineeship(IGERT) program funded by the National Science Foundationand the University of Florida for financially supporting thisstudy Funding for panther research conducted by the FWC issupported predominantly via donations accrued with vehicleregistrations for ldquoProtect the Pantherrdquo license plates

LITERATURE CITEDAcevedo‐Whitehouse K T R Spraker E Lyons S R Melin F Gulland RL Delong and W Amos 2006 Contrasting effects of heterozygosity onsurvival and hookworm resistance in California sea lion pups MolecularEcology 151973ndash1982

Adams J R L M Vucetich P W Hedrick R O Peterson and J AVucetich 2011 Genomic sweep and potential genetic rescue during limitingenvironmental conditions in an isolated wolf population Proceedings of theRoyal Society B Biological Sciences 2783336ndash3344

Agresti A 2013 Categorical data analysis Third edition John Wiley amp SonsHoboken New Jersey USA

Agudo R M Carrete M Alcaide C Rico F Hiraldo and J A Donaacutezar2012 Genetic diversity at neutral and adaptive loci determines individualfitness in a long‐lived territorial bird Proceedings of the Royal Society BBiological Sciences 2793241ndash3249

Albeke S E N P Nibbelink and M Ben‐David 2015 Modeling behavior bycoastal river otter (Lontra canadensis) in response to prey availability in PrinceWilliam Sound Alaska a spatially‐explicit individual‐based approach PLOSOne 10e0126208

Alho J S K Vaumllimaumlki and J Merilauml 2010 Rhh an R extension for estimatingmultilocus heterozygosity and heterozygosity‐heterozygosity correlationMolecular Ecology Resources 10720ndash722

Alvarez K 1993 Twilight of the panther Myakka River Publishing SarasotaFlorida USA

Aparicio J M J Ortego and P J Cordero 2006 What should we weigh toestimate heterozygosity alleles or loci Molecular Ecology 154659ndash4665

Bakker V J D F Doak G W Roemer D K Garcelon T J Coonan S AMorrison C Lynch K Ralls and R Shaw 2009 Incorporating ecologicaldrivers and uncertainty into a demographic population viability analysis for theisland fox Ecological Monographs 7977ndash108

Balloux F W Amos and T Coulson 2004 Does heterozygosity estimateinbreeding in real populations Molecular Ecology 133021ndash3031

Barone M A M E Roelke J Howard J L Brown A E Anderson and DE Wildt 1994 Reproductive characteristics of male Florida panthersmdashcomparative studies from Florida Texas Colorado Latin‐America andNorth‐American Zoos Journal of Mammalogy 75150ndash162

Bateson Z W P O Dunn S D Hull A E Henschen J A Johnson and LA Whittingham 2014 Genetic restoration of a threatened population ofgreater prairie‐chickens Biological Conservation 17412ndash19

Bauer H G Chapron K Nowell P Henschel P Funston L T B HunterD W Macdonald and C Packer 2015 Lion (Panthera leo) populations aredeclining rapidly across Africa except in intensively managed areas Pro-ceedings of National Academy of Sciences of the United States of America11214894ndash14899

Beier P 1995 Dispersal of juvenile cougars in fragmented habitat Journal ofWildlife Management 59228ndash237

Benson J F J A Hostetler D P Onorato W E Johnson M E Roelke SJ OrsquoBrien D Jansen and M K Oli 2011 Intentional genetic introgressioninfluences survival of adults and subadults in a small inbred felid populationJournal of Animal Ecology 80958ndash967

Benson J F P J Mahoney J A Sikich L E K Serieys J P Pollinger H BErnest and S P D Riley 2016 Interactions between demography geneticsand landscape connectivity increase extinction probability for a small popu-lation of large carnivores in a major metropolitan area Proceedings of theRoyal Society B Biological Sciences 28320160957

Beverly F 1874 The Florida panther Forest and Stream 3290Boyce M S C V Haridas C T Lee and the NCEAS Stochastic Demo-graphy Working Group 2006 Demography in an increasingly variable worldTrends in Ecology amp Evolution 21141ndash148

Brook B W J J OrsquoGrady A P Chapman M A Burgman H R Akcakayaand R Frankham 2000 Predictive accuracy of population viability analysis inconservation biology Nature 404385ndash387

Brys R and H Jacquemyn 2016 Severe outbreeding and inbreeding de-pression maintain mating system differentiation in Epipactis (Orchidaceae)Journal of Evolutionary Biology 29352ndash359

Burke J M and M L Arnold 2001 Genetics and the fitness of hybridsAnnual Review of Genetics 3531ndash52

Burnham K P 1993 A theory for combined analysis of ring recovery andrecapture data Pages 199ndash213 in J‐D Lebreton and P M North editorsMarked individuals in the study of bird population Springer‐Verlag NewYork New York USA

Burnham K P and D R Anderson 2002 Model selection and multimodelinference a practical information‐theoretic approach Second edition SpringerVerlag New York New York USA

Caswell H 2001 Matrix population models construction analysis and in-terpretation Sinauer Associates Sunderland Massachusetts USA

Charpentier M J M Setchell F Prugnolle L A Knapp E J Wickings PPeignot and M Hossaert‐McKey 2005 Genetic diversity and reproductivesuccess in mandrills (Mandrillus sphinx) Proceedings of the NationalAcademy of Sciences of the United States of America 10216723ndash16728

Cooch E and G C White editors 2018 Program MARK a gentle in-troduction lthttpwwwphidotorgsoftwaremarkdocsbookgt Accessed 7Apr 2016

Cooley H S R B Wielgus G M Koehler H S Robinson and B TMaletzke 2009 Does hunting regulate cougar populations A test of thecompensatory mortality hypothesis Ecology 902913ndash2921

Coulson T E A Catchpole S D Albon B J Morgan J M Pemberton TH Clutton‐Brock M J Crawley and B T Grenfell 2001 Age sex densitywinter weather and population crashes in Soay sheep Science 2921528ndash1531

Coulson T J Pemberton S Albon M Beaumont T Marshall J Slate FGuinness and T Clutton‐Brock 1998a Microsatellites reveal heterosis in reddeer Proceedings of the Royal Society B Biological Sciences 265489ndash495

Coulson T N S D Albon J M Pemberton J Slate F E Guinness and TH Clutton‐Brock 1998b Genotype by environment interactions in wintersurvival in red deer Journal of Animal Ecology 67434ndash445

Cox D 1972 Regression models and life‐tables Journal of the Royal StatisticalSociety Series B Statistical Methodology 34187ndash220

Creel S 2006 Recovery of the Florida panthermdashgenetic rescue demographic rescueor both Response to Pimm et al (2006) Animal Conservation 9125ndash126

Crnokrak P and D A Roff 1999 Inbreeding depression in the wild Heredity83260ndash270

Crow J 1948 Alternative hypotheses of hybrid vigor Genetics 33477ndash487

van de Kerk et al bull Florida Panther Population and Genetic Viability 31

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 23: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

25

50

75

100

0

25

50

75

100

Introgression interval (yrs)

PQ

E (

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 15 Average probability of quasi‐extinction (PQE) of the Florida panther population within 100 years and within 200 years without genetic managementintervention (N) and with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenariosIntrogression strategies include the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Errorbars indicate 5th and 95th percentiles Based on data collected in South Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

50

60

70

80

90

100

110

50

100

150

200

Years

Popu

latio

n si

ze

Figure 16 Average projected Florida panther population trajectories under a geneticintrogression strategy involving the release of 5 female pumas from Texas USA every20 years for the minimum population count (MPC) and motor vehicle mortality(MVM) scenarios Shaded area indicates 5th and 95th percentiles and isrepresentative of confidence intervals for other scenarios Based on data collected inSouth Florida USA 1981ndash2013

MP

CM

VM

0 50 100 150 200

055

060

065

070

055

060

065

070

Years

Het

eroz

ygos

ity

Figure 17 Average projected Florida panther population‐level heterozygositiesunder a genetic introgression strategy involving the release of 5 female pumas fromTexas USA every 20 years for the minimum population count (MPC) and motorvehicle mortality (MVM) scenarios Shaded area bounds the 5th and 95th percentilesand is representative of confidence intervals for other scenarios Based on datacollected in South Florida USA 1981ndash2013

van de Kerk et al bull Florida Panther Population and Genetic Viability 25

this case the migration of just a single male across the freewayto the south resulted in an increase in the number of in-dividuals with admixed ancestry and substantially improvedgenetic diversity of the subpopulation south of the freeway(Riley et al 2014)Our estimate of overall kitten survival (0321plusmn 0056) was

similar to that reported by Hostetler et al (2010) who used13 years of post‐introgression data (0323plusmn 0071) These valuesare lower than kitten survival estimates derived from severalpuma populations in western North America (044ndash091 Loganand Sweanor 2001 Lambert et al 2006 Laundre et al 2007Robinson et al 2008 Ruth et al 2011) When we included onlydata for individuals that had been genetically sampled our es-timate was 15 higher The proportion of admixed kittens thatwere genetically sampled increased as the population expandedwhich could have resulted in higher estimates of kitten survivalbecause admixed kittens survive better than canonical kittensKitten survival was strongly influenced by genetic ancestry withsurvival rates of F1 kittens being almost twice that of canonicalkittens (Fig 5B) Consistent with the findings of Hostetler et al(2010) increases in heterozygosity coincided with increasedkitten survival whereas population density negatively affectedkitten survival Population density often has a strong negativeimpact on survival of young age classes due to infanticide andcannibalism which has been reported from several carnivore

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

50

100

150

0

50

100

150

Introgression interval (yrs)

TQE

(yrs

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 18 Average times until quasi‐extinction (TQE) for the Florida panther population within 100 years and within 200 years without genetic management interventionand with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategiesinclude the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Error bars indicate 5th and 95thpercentiles Based on data collected in South Florida USA 1981ndash2013

Table 5 Average cost (2017 US dollars) of introgression strategies employedover 100 years that involve the introduction of 5 10 or 15 pumas from TexasUSA into the Florida panther population in South Florida USA at an in-creasingly higher frequency Costs of capture (including transportation) quar-antine and post‐introgression monitoring were estimated by Roy McBrideMark Cunningham and Dave Onorato respectively

Frequency Component 5 pumas 10 pumas 15 pumas

Once Capture 25000 50000 75000

Quarantine 5000 10000 15000

Monitoring 10000 20000 30000

Total 40000 80000 120000

Every 80 years Capture 31250 62500 93750

Quarantine 6250 12500 18750

Monitoring 12500 25000 37500

Total 50000 100000 150000

Every 40 years Capture 62500 125000 187500

Quarantine 12500 25000 37500

Monitoring 25000 50000 75000

Total 100000 200000 300000

Every 20 years Capture 125000 250000 375000

Quarantine 25000 50000 75000

Monitoring 50000 100000 150000

Total 200000 400000 600000

Every 10 years Capture 250000 500000 750000

Quarantine 50000 100000 150000

Monitoring 100000 200000 300000

Total 400000 800000 1200000

26 Wildlife Monographs bull 203

species including polar bears (Ursus maritimus Derocher andWiig 1999) black bears (Ursus americanus Garrison et al 2007)and pumas (Logan and Sweanor 2001)Our estimates of sex‐ and age‐specific survival rates of subadult

and adult panthers (Table 3) and the effects of covariates on theserates were similar to those reported by Benson et al (2011) derivedfrom data collected through 2006 Our survival rates for males(065ndash077) were lower than females (078ndash097) for all 3 ageclasses (subadult prime adult and old adult) which is the typicaltrend observed in most puma populations (eg Ruth et al 19982011) However an unhunted puma population occupying afragmented urban landscape in southern California exhibited lowersurvival (0586 females 0525 males Vickers et al 2015) perhapsas a result of severe habitat fragmentation and the lack of extensiveswaths of public land protected in perpetuity Most populations ofpuma in western North America are typically subject to variedlevels of management via regulated hunting which often is biasedtoward the take of males (Cooley et al 2009 Ruth et al 2011)Consequently annual survival estimates from hunted populationsin northeast Washington (0679ndash0733 females 0341 malesRobinson et al 2008) and southeastern Arizona (0685 females0584 males Cunningham et al 2001) were generally lower thanthose we estimated for Florida panthersCause‐specific mortality analysis showed that male panthers

experienced a substantially greater mortality risk from in-traspecific aggression This differs from the pattern observedduring a long‐term study in Arizona where females were at ahigher risk of intraspecific mortality than males (46 of maleand 53 of female mortality Logan and Sweanor 2001)Mortality associated with intraspecific strife has been docu-mented in hunted and unhunted populations (this study Loganand Sweanor 2001 Stoner et al 2010) and its root cause hasbeen difficult to determine (eg population density and preypopulation trends) That being the case finding ways to reducewhat is often a major source of mortality for puma populationscontinues to pose a challenge to managersCanonical panthers were substantially more likely to die from

intraspecific aggression than were admixed panthers This mayat least partly explain the difference in survival between admixed

and canonical panthers Canonical panthers are known to bemore likely to suffer from varied correlates of inbreeding de-pression than admixed animals including atrial septal defects(Johnson et al 2010) and compromised immune systems(Roelke et al 1993) These factors have the potential to affectfitness and perhaps dominance of individuals when engaged inan intraspecific encounter A somewhat similar scenario wasobserved by Hogg et al (2006) in a population of bighorn sheepthat were part of an introgression project Admixed males in thispopulation had a greater probability of paternity than males thatwere less outbred This included coursing males with higherlevels of admixture being markedly more successful at fightingmales that were defending females in estrous (Hogg et al 2006)Heterosis resulting from genetic introgression is expected to be

the strongest in F1 individuals (Shull 1908 Crow 1948 Burkeand Arnold 2001 Waller 2015) and to progressively decline insubsequent generations (Pickup et al 2013 Frankham 2015)Given that admixed (especially F1) panthers survived better thancanonical panthers and that individual heterozygosity improvedsurvival of both sexes and all age classes our data provide evi-dence for heterotic advantage whereby individuals of mixedancestry had higher fitness (Shull 1908 Crow 1948) Our resultsare consistent with findings of other studies that involved in-tentional genetic introgression for conservation purposes Forexample Hogg et al (2006) detected substantial improvementsin survival and reproduction of introgressed bighorn sheep andMadsen et al (1999 2004) reported a dramatic increase in thepopulation of adders after genetic introgressionKnowledge of the persistence of benefits to a population ac-

crued via genetic introgression is essential for determiningwhen additional introgression may be warranted There is adepauperate amount of information regarding this topic and theidentification of factors that may influence persistence of thebenefits of genetic introgression (Tallmon et al 2004 Hedricket al 2014 Whiteley et al 2015) For example in minks(Neovison vison) Thirstrup et al (2014) found that outcrossingled to larger litters in F2 and F3 but this benefit disappeared insubsequent generations In a population of gray wolves Hedricket al (2014) suggested that benefits of genetic introgression may

Table 6 Costs and benefits accumulated over 100 years for genetic introgression at different intervals and with different numbers of pumas from Texas USAintroduced into the Florida panther population in South Florida USA Costs (2017 US dollars) include capture and transportation quarantine and post‐introgression monitoring Benefits are represented as average probability of quasi‐extinction (PQE and 5th and 95th percentiles) and changes (Δ) therein under thescenario using the minimum population count (McBride et al 2008 MPC) and the scenario using the motor vehicle mortality population estimates (McClintocket al 2015 MVM) The table is sorted by percent change in probability of quasi‐extinction under the MPC scenario

Interval (yr) Pumas Cost MPC PQE () ΔMPC PQE () MVM PQE () ΔMVM PQE ()

ndash 0 0 13 (0ndash99) 0 17 (0ndash100) 080 10 100000 12 (0ndash99) 10 (0ndash58) 16 (0ndash100) 5 (0ndash38)80 15 150000 12 (0ndash99) 13 (0ndash65) 15 (0ndash100) 8 (0ndash56)80 5 50000 12 (0ndash99) 14 (0ndash73) 15 (0ndash99) 9 (0ndash41)40 15 300000 10 (0ndash98) 26 (0ndash104) 14 (0ndash99) 13 (0ndash60)40 10 200000 9 (0ndash94) 35 (0ndash183) 13 (0ndash99) 21 (0ndash95)40 5 100000 8 (0ndash89) 39 (0ndash211) 12 (0ndash99) 26 (0ndash124)20 15 600000 8 (0ndash88) 42 (0ndash257) 13 (0ndash99) 24 (0ndash123)20 10 400000 6 (0ndash67) 54 (0ndash318) 10 (0ndash99) 37 (0ndash217)10 15 1200000 6 (0ndash53) 58 (0ndash372) 10 (0ndash99) 40 (0ndash208)20 5 200000 5 (0ndash50) 59 (0ndash338) 9 (0ndash99) 44 (0ndash352)10 10 800000 4 (0ndash16) 69 (0ndash489) 8 (0ndash92) 55 (0ndash401)10 5 400000 4 (0ndash7) 73 (0ndash502) 6 (0ndash71) 63 (0ndash503)

van de Kerk et al bull Florida Panther Population and Genetic Viability 27

wane after 2 or 3 generations The WrightndashFisher model ofgenetic drift predicts a geometric decline in expected hetero-zygosity at the rate of (1 minus 12Ne) per generation where Ne is theeffective population size (Hartl and Clark 1997 Frankham et al2014) Thus it seems logical to assume that the duration of thebenefits of introgression is limited especially in a species per-sisting in a small population such as Florida panthersWe expected Florida panther survival in the most recent years

(2005ndash2013) post‐introgression to differ from that observed in thedecade following the introgression in 1995 because pantherabundance and heterozygosity factors known to influence panthersurvival (Hostetler et al 2010 Benson et al 2011) have changed(Figs 2 and 6) We found that subadult and adult survival rates inrecent years (2005ndash2013) were similar to those in the period im-mediately following introgression (1995ndash2004 Fig 5C) overallkitten survival was similar to estimates of Hostetler et al (2010) Infact the pattern of covariate effects on Florida panther survivalreported by Benson et al (2011) and Hostetler et al (2010) hasbarely changed The negative effects of population density andpositive effects of heterozygosity may counterbalance survival ratesreducing population fluctuations Our result that benefits of geneticrestoration have remained in the population nearly for 5 genera-tions post‐intervention was surprising but also encouraging Pos-sible explanations for this observation may include 1) genetic ero-sion occurs at slower rate than previously thought 2) introducing 5migrants in 1 genetic intervention may have beneficial effects si-milar to those from 1 migrant per generation over multiple gen-erations or 3) other and as yet unknown factors may reduce therate of genetic erosion and subsequent demographic effects Adensity‐dependent effect on kitten survival could also explain whynon‐F1 admixed kittens did not survive substantially better thandid canonical kittens most kittens sampled from 2005ndash2013 wereadmixed and their survival was lower possibly because of a density‐dependent effect as the Florida panther population continuouslygrew during that period Trinkel et al (2010) also found thatpopulation density and inbreeding coefficient interacted to affectdemographic parameters and growth rate of an African lion po-pulation Likewise population density interacted with winterweather to affect the survival and population growth rates in Soaysheep (Ovis aries Milner et al 1999 Coulson et al 2001)

Population Dynamics and PersistenceThe finite deterministic and stochastic growth rates were gt10reflecting that much of the data used in this study were collectedfollowing genetic introgression in 1995 during which time thepopulation increased substantially Because our demographicparameter estimates did not change markedly in comparison tothose of Hostetler et al (2013) the similarity between thepopulation growth estimates from these studies was expectedBut these estimates reflect population growth during thedata‐collection period and do not predict future growth In factestimates of population size reported by McClintock et al(2015) suggest that population growth may already be slowing(Fig 2) A similar pattern was observed in an introduced po-pulation of African lion which increased steadily from 1995until 2001 then fluctuated around the presumed carrying ca-pacity thereafter (Trinkel et al 2010) Several other African lionpopulations that experienced various degrees of recovery

subsequently became relatively stable or exhibited decliningtrends (Bauer et al 2015)Our estimates of quasi‐extinction probabilities were lower than

those reported by Hostetler et al (2013) using a 2‐sex age‐structured matrix population model Lower probabilities ofquasi‐extinction than those reported by the earlier study forcomparable scenarios were likely a consequence of the fact thatthe estimates of abundance (McBride et al 2008) and thus thedensity‐dependent effects on survival of kittens and old panthers(gt10 yr) have changed in recent years Additionally these earlyestimates of the probability of quasi‐extinction and of time toquasi‐extinction did not consider genetic erosion that will in-evitably occur without management interventionUnder the MVM scenario the equilibrium population size was

twice that attained under the MPC scenario The MVM popula-tion estimates are approximately twice the MPCs (Fig 2) as aresult the simulated population under the MVM scenario achieveda higher equilibrium size Probability of quasi‐extinction under theMVM scenario was generally greater than that under the MPCscenario This result is a consequence of the fact that the estimateddensity dependence on kitten survival based on the MPC scenario isstronger than that for theMVM scenario (regression coefficients forabundance for the MPC scenario= minus0068 vs minus0002 for theMVM scenario Appendix B Table B1) Consequently simulatedpopulations under the MPC scenario have a stronger tendency tomove toward the equilibrium population size which inherentlyreduces the quasi‐extinction probability (eg Royama 1992)As expected for long‐lived mammals (Heppell et al 2000 Oli

and Dobson 2003) the population growth rate was proportio-nately most sensitive to changes in survival of prime adultsfollowed by that in younger age classes Global sensitivity ana-lysis in contrast showed that under all scenarios extinctionparameters were particularly sensitive to changes in survival ofFlorida panther kittens This result is a consequence of the highsensitivity of the population growth rate to kitten survival(Fig 9) and a larger standard error of kitten survival (Table 3)Results of our sensitivity analyses indicate that future researchshould focus on acquiring additional information on early lifestages to improve the accuracy and precision of estimates ofkitten survival Our results suggest that demographic variables(or their environmental drivers) that strongly influence extinc-tion risks are not necessarily those with the largest elasticityvalues A study of island fox (Bakker et al 2009) found thatextinction risk was strongly influenced by survival modifierparameters that had low elasticity values

Genetic Management When and How to InterveneThe use of genetic introgression as a management tool has al-ways been controversial and the probable effectiveness it mighthave in preventing the imminent demise of the panther popu-lation was initially debated (Creel 2006 Maehr et al 2006Pimm et al 2006) These debates notwithstanding the pantherpopulation had suffered from genetic morphological and bio-medical correlates of inbreeding before this management in-tervention (Johnson et al 2010 Onorato et al 2010) whereasthe post‐introgression period has been characterized by a sub-stantial reduction in the biomedical correlates of inbreeding andimprovements in demographic vigor and abundance (McBride

28 Wildlife Monographs bull 203

et al 2008 Hostetler et al 2010 2013 Johnson et al 2010Benson et al 2011 McClintock et al 2015) Nevertheless thepopulation remains small and isolated habitat is still being lostand there is no current possibility of natural gene flow betweenthis and other North American puma populations thus therecurrence of inbreeding‐related problems and subsequent po-pulation declines are inevitable The question therefore is notwhether genetic management of the Florida panther populationis needed but when and how it should be implementedWithout genetic introgression probabilities of quasi‐extinc-

tion were substantially greater than those from models thatincorporated periodic genetic introgression events (Fig 15)highlighting the importance of genetic management in smallisolated populations When 5 female pumas are added to theFlorida panther population every 10 years genetic variationincreased substantially under the MPC and MVM scenariosThe IBM predicted that the equillibrium population size wouldbe substantially larger when 5 pumas were released into thepopulation every 10 years than populations without intervention(ie no introgression) Releasing 15 Texas females did not af-ford substantially greater benefit to the equilibrium populationsize than did releasing only 5 Texas females Instead addingmore individuals caused increasingly large fluctuations in po-pulation size due to the numerical increases and density de-pendence (Fig 14) possibly diminishing the benefits associatedwith increased genetic variationCompared with the no‐intervention scenario (ie no genetic

introgression implemented) the introduction of 5 female pumasevery 10 years reduced quasi‐extinction probabilities from 13to 4 and from 17 to 6 for the MPC and MVM scenariosrespectively The benefit of genetic‐introgression strategies de-creased as the interval between successive management inter-ventions increased and no benefit was evident once the intervalreached 80 years (Fig 15) Introgression reduced the averagetime until quasi‐extinction (Fig 18) This somewhat counter‐intuitive result can be explained by the fact that repeated geneticintrogression makes the population more robust over timewhich will reduce the probability of extinction Consequentlyfew simulated population trajectories that fall below the criticalthreshold do so early reducing the mean time to extinctionAlso density dependence has a stabilizing effect on populations(Royama 1992) populations tend toward equilibrium popula-tion sizes which reduces the probability over time of popula-tions falling below quasi‐extinction threshold However timeuntil quasi‐extinction must be interpreted in conjunction withthe probability of quasi‐extinction which is very low for allscenarios with genetic introgression (Fig 15)Cost is a significant impediment to the implementation of a

genetic introgression program because captures relocations andmonitoring efforts before and after introgression are expensive(Edmands 2007 Frankham 2010b 2015 2016) Costs may beacceptable to society if the management actions result in sub-stantial fitness benefits especially if the species of concern ispopular and charismatic (Frankham 2015) We estimated the100‐year cost of introgression strategies modeled here to rangefrom $50000 to $12 million More expensive strategies gen-erally led to larger decreases in the probability of quasi‐extinc-tion but cheaper strategies also substantially improved

population viability (ie reduced quasi‐extinction probabilityTable 6) One of the cheaper strategies (5 pumas released every20 years) which cost an estimated $200000 over 100 yearsreduced the probability of quasi‐extinction by 59 and 44 forthe MPC and MVM scenarios respectively But these pre-liminary cost estimates are based on expenses incurred duringthe 1995 introgression and include costs of capture quarantinetransportation release and post‐release monitoring of femalesonly Although the cost for future genetic interventions wouldlikely be higher than those reported here the relative differencesin cost between alternative intervention strategies should remainapproximately the sameOur ability to simulate genetics and link it to the viability of

the Florida panther population is based on the empirically es-timated relationship between individual heterozygosity andsurvival Such heterozygosity and fitness correlations have beenstudied for decades (David 1998) but have only recently beenused to predict population viability (Benson et al 2016) noneto our knowledge have incorporated cost into their modelingefforts The mechanisms underlying heterozygosity and fitnesscorrelations remain unclear (David 1998 Szulkin et al 2010)and their significance as a proxy for inbreeding depression hasbeen debated (Balloux et al 2004 Miller and Coltman 2014)Nevertheless evidence continues to mount demonstratingpositive relationships between heterozygosity and survival(Coulson et al 1998ab Hansson et al 2004 Silva et al 2005Acevedo‐Whitehouse et al 2006 Jensen et al 2007 Velandoet al 2015) and between heterozygosity and fecundity (Seddonet al 2004 Charpentier et al 2005 Agudo et al 2012 Velandoet al 2015) Although the reported correlations are generallyweak their consequences can be substantial if population growthand persistence parameters are highly sensitive to the affectedvital rates (Velando et al 2015) Compared with scenarios thatignore genetics incorporating the effects of inbreeding on sur-vival increased quasi‐extinction probabilities within a 100‐yeartimeframe from 15 to 13 and from 14 to 17 for theMPC and MVM scenarios respectively These results high-light the importance of incorporating genetics when analyzingthe viability of small isolated populationsAlthough conservation biologists have recognized the im-

portance of genetics in population viability many still fail toincorporate genetic factors into PVA models (Reed et al 2002)Explicit consideration of genetics in PVAs requires long‐termgenetic and demographic data This permits the assessment ofthe functional relationship between genetic variability and de-mographic parameters Population viability analysis softwarepackages such as VORTEX (Lacy 1993 2000) offer a powerfulframework for individual‐based simulations but they often donot adequately capture a speciesrsquo life history or genetics Basedon a thorough review of the importance of genetic factors inwildlife conservation Frankham et al (2014) concluded thatgenetic factors can strongly influence the dynamics and persis-tence of small andor fragmented populations They furthernoted that ldquoMost PVA‐based risk assessments ignore or in-adequately model genetic factorsrdquo and recommended that ldquoPVAshould routinely include realistic inbreeding depressionhelliprdquo(Frankham et al 201457) Our custom‐built IBM permitted usto develop a model that adequately captured the Florida panther

van de Kerk et al bull Florida Panther Population and Genetic Viability 29

life history and we could parameterize the model with our long‐term genetic and demographic dataThe explicit consideration of the effects of inbreeding on

Florida panther population dynamics and persistence representsan important step forward Nonetheless our model remainsimperfect First we neglected the possible effects of habitat lossdetrimental anthropogenic activities climate change the prob-ability of immigration of pumas from western populationsdisease or catastrophes on demographic rates and populationviability Although immigration from puma populations outsideFlorida is possible its probability is very low given the extensivechallenges the anthropogenically altered landscape would posefor such a migrant to make it successfully to South Florida Weomitted mutations from our model because we have no in-formation on mutation rates though we expect that they are lowbased on values reported for other mammals (Kumar and Sub-ramanian 2002) Finally we only considered possible effects ofinbreeding on age‐specific survival rates because there was noevidence that heterozygosity affected female reproduction Lossof heterozygosity can negatively affect reproduction becauseinbred male panthers can suffer from cryptorchidism which canadversely affect their fecundity However given the polygynousmating system we expect the Florida panther population growthis primarily female‐limitedAlthough it is generally accepted that genetic introgression

only temporarily relieves a population from inbreeding depres-sion we know of no cases in which it has been implementedmore than once to conserve a threatened wildlife populationOur study provides an example of how demographic and mo-lecular data can be used within an IBM framework to evaluatethe efficacy of alternative genetic management strategies via theFlorida panther as a case study Our model can be adjusted forand applied to other species and offers great potential as a toolfor genetic management of small isolated wildlife populations

MANAGEMENT IMPLICATIONS

Our results and those of Hostetler et al (2013) and Johnsonet al (2010) provide persuasive evidence that the genetic in-tervention implemented in 1995 prevented the demise of theFlorida panther and restored demographic vigor to the popu-lation The Florida panther population remains small and iso-lated from other puma populations the closest puma populationis located gt1000 km away in Texas and the intervening land-scape is highly developed and fragmented with many interstate(and other high‐traffic) highways and major urban centersTheory suggests that 1 migrant per generation is sufficient tominimize the loss of genetic diversity (eg Mills and Allendorf1996) However the possibility of successful migration of apuma to South Florida followed by successful mating every ~5years is very low considering the distance between Texas andFlorida (Hedrick 1995) and the many risks and barriers thatdispersing pumas face (Beier 1995 Maehr et al 2002 Thatcheret al 2006) Consequently the pantherrsquos long‐term persistencewill likely depend on subsequent timely genetic introgressionsgiven the near‐impossibility of natural gene flow from otherpuma populations To that end our study offers considerableinsights into benefits of different genetic management strategiesand associated costs

Without genetic management the Florida panther populationfaces a substantial risk of quasi‐extinction within 100 years (10ndash46 depending on quasi‐extinction threshold and scenarios)Twenty‐three years have passed since genetic introgression wasimplemented and the population appears healthy as indicatedby measures of genetic variation increases in the populationsize and improvements in age‐specific survival rates Ourfindings suggest that releasing more pumas more frequently doesnot necessarily improve persistence probability because such astrategy would cause larger fluctuations in population size(Fig 14) which negatively affects persistence probability Thusextinction risks faced by inbred populations of large carnivorescan be substantially reduced by releasing approximately 5 im-migrants every 2 decades We suggest that such a managementstrategy be followed only in conjunction with continued long‐term monitoring of the population Our analyses demonstratethat population growth and persistence are highly sensitive tokitten and female (adult and subadult) survival Continuing tocollect data on these demographic variables along with bio-metric and genetic sampling will remain important to pantherrecovery and vigilance in identifying issues associated with in-breeding depression The collection of these monitoring datashould allow managers to detect any demographic or geneticchanges in a timely fashion and respond with genetic in-trogression or other management actions if warrantedRecovery objectives as defined by the USFWS in its current

recovery plan (USFWS 2008) delineate the need for additionalpopulations in the historical range to downlist or delist theFlorida panther If the only extant population of Floridapanthers continued to grow it could serve as a source ofpanthers to be translocated to suitable habitat to seed addi-tional populations Maintaining current information on thegenetic health of the population and precise estimates of itssize will be important in assessing whether it can withstand theremoval of a subset of animals for translocation Although the2017 documentation of a female panther with kittens north ofthe current breeding range is encouraging in terms of thepotential for population expansionmdashgiven that no female hadbeen recorded north of the Caloosahatchee River since 1972mdashthe re‐establishment of separate new populations will mostlikely require translocations by wildlife managers and a lengthysociopolitical processConservation of threatened or endangered species such as the

Florida panther is inherently challenging For panthers and otherlarge carnivores that require vast extents of natural habitat topersist habitat is typically the most limiting factor that will de-termine whether recovery efforts succeed Although geneticmanagement invariably plays a key role in the long‐term persis-tence of the panther population even a healthy population caneventually succumb to the impacts of habitat loss The humanpopulation of Florida continues to be one of the fastest‐growingin the nation the United States Census Bureau currently ranksFlorida as the third most populous Therefore habitat loss isgoing to continue to be a key issue for panthers Diligence towardpreserving panther habitat in its historical range via conservationeasements or other means and trying to keep such areas inter-connected via corridors is a goal stakeholders such as governmentagencies sportsmen non‐governmental organizations ranchers

30 Wildlife Monographs bull 203

and other private landowners must work on collaboratively toachieve

ACKNOWLEDGMENTS

We thank D Shindle D Jennings T OrsquoMeara D Land andC Belden for valuable advice throughout the project We thankS Bass M Criffield M Cunningham D Giardina D JansenA Johnson D Land M Lotz C McBride Rocky McBrideRowdy McBride Roy McBride L Oberhofer S Schulze staffsat Everglades National Park and Big Cypress National Preserveand others for assistance with fieldwork We also thank JAustin M Conroy J Hines J Laake S Mills J Nichols JHines M Sunquist J Fryxell and T Coulson for advice andassistance regarding data analysis and panther biology TheMuseum of Texas Tech University Natural Science ResearchLaboratory provided samples from pumas from west Texas forcomparative genetic analyses M Ben‐David D Land WMcCown E Hellgren and 4 anonymous reviewers providedmany helpful comments on the manuscript We thank NereaUbierna Lopez for completing the Spanish translation of theabstract We are grateful to the Department of ResearchComputing University of Florida for excellent high‐perfor-mance computing services We would like to thank US Fishand Wildlife Service Florida Fish and Wildlife ConservationCommission (FWC) US National Park Service QuantitativeSpatial Ecology Evolution and Environment (QSE3) In-tegrative Graduate Education and Research Traineeship(IGERT) program funded by the National Science Foundationand the University of Florida for financially supporting thisstudy Funding for panther research conducted by the FWC issupported predominantly via donations accrued with vehicleregistrations for ldquoProtect the Pantherrdquo license plates

LITERATURE CITEDAcevedo‐Whitehouse K T R Spraker E Lyons S R Melin F Gulland RL Delong and W Amos 2006 Contrasting effects of heterozygosity onsurvival and hookworm resistance in California sea lion pups MolecularEcology 151973ndash1982

Adams J R L M Vucetich P W Hedrick R O Peterson and J AVucetich 2011 Genomic sweep and potential genetic rescue during limitingenvironmental conditions in an isolated wolf population Proceedings of theRoyal Society B Biological Sciences 2783336ndash3344

Agresti A 2013 Categorical data analysis Third edition John Wiley amp SonsHoboken New Jersey USA

Agudo R M Carrete M Alcaide C Rico F Hiraldo and J A Donaacutezar2012 Genetic diversity at neutral and adaptive loci determines individualfitness in a long‐lived territorial bird Proceedings of the Royal Society BBiological Sciences 2793241ndash3249

Albeke S E N P Nibbelink and M Ben‐David 2015 Modeling behavior bycoastal river otter (Lontra canadensis) in response to prey availability in PrinceWilliam Sound Alaska a spatially‐explicit individual‐based approach PLOSOne 10e0126208

Alho J S K Vaumllimaumlki and J Merilauml 2010 Rhh an R extension for estimatingmultilocus heterozygosity and heterozygosity‐heterozygosity correlationMolecular Ecology Resources 10720ndash722

Alvarez K 1993 Twilight of the panther Myakka River Publishing SarasotaFlorida USA

Aparicio J M J Ortego and P J Cordero 2006 What should we weigh toestimate heterozygosity alleles or loci Molecular Ecology 154659ndash4665

Bakker V J D F Doak G W Roemer D K Garcelon T J Coonan S AMorrison C Lynch K Ralls and R Shaw 2009 Incorporating ecologicaldrivers and uncertainty into a demographic population viability analysis for theisland fox Ecological Monographs 7977ndash108

Balloux F W Amos and T Coulson 2004 Does heterozygosity estimateinbreeding in real populations Molecular Ecology 133021ndash3031

Barone M A M E Roelke J Howard J L Brown A E Anderson and DE Wildt 1994 Reproductive characteristics of male Florida panthersmdashcomparative studies from Florida Texas Colorado Latin‐America andNorth‐American Zoos Journal of Mammalogy 75150ndash162

Bateson Z W P O Dunn S D Hull A E Henschen J A Johnson and LA Whittingham 2014 Genetic restoration of a threatened population ofgreater prairie‐chickens Biological Conservation 17412ndash19

Bauer H G Chapron K Nowell P Henschel P Funston L T B HunterD W Macdonald and C Packer 2015 Lion (Panthera leo) populations aredeclining rapidly across Africa except in intensively managed areas Pro-ceedings of National Academy of Sciences of the United States of America11214894ndash14899

Beier P 1995 Dispersal of juvenile cougars in fragmented habitat Journal ofWildlife Management 59228ndash237

Benson J F J A Hostetler D P Onorato W E Johnson M E Roelke SJ OrsquoBrien D Jansen and M K Oli 2011 Intentional genetic introgressioninfluences survival of adults and subadults in a small inbred felid populationJournal of Animal Ecology 80958ndash967

Benson J F P J Mahoney J A Sikich L E K Serieys J P Pollinger H BErnest and S P D Riley 2016 Interactions between demography geneticsand landscape connectivity increase extinction probability for a small popu-lation of large carnivores in a major metropolitan area Proceedings of theRoyal Society B Biological Sciences 28320160957

Beverly F 1874 The Florida panther Forest and Stream 3290Boyce M S C V Haridas C T Lee and the NCEAS Stochastic Demo-graphy Working Group 2006 Demography in an increasingly variable worldTrends in Ecology amp Evolution 21141ndash148

Brook B W J J OrsquoGrady A P Chapman M A Burgman H R Akcakayaand R Frankham 2000 Predictive accuracy of population viability analysis inconservation biology Nature 404385ndash387

Brys R and H Jacquemyn 2016 Severe outbreeding and inbreeding de-pression maintain mating system differentiation in Epipactis (Orchidaceae)Journal of Evolutionary Biology 29352ndash359

Burke J M and M L Arnold 2001 Genetics and the fitness of hybridsAnnual Review of Genetics 3531ndash52

Burnham K P 1993 A theory for combined analysis of ring recovery andrecapture data Pages 199ndash213 in J‐D Lebreton and P M North editorsMarked individuals in the study of bird population Springer‐Verlag NewYork New York USA

Burnham K P and D R Anderson 2002 Model selection and multimodelinference a practical information‐theoretic approach Second edition SpringerVerlag New York New York USA

Caswell H 2001 Matrix population models construction analysis and in-terpretation Sinauer Associates Sunderland Massachusetts USA

Charpentier M J M Setchell F Prugnolle L A Knapp E J Wickings PPeignot and M Hossaert‐McKey 2005 Genetic diversity and reproductivesuccess in mandrills (Mandrillus sphinx) Proceedings of the NationalAcademy of Sciences of the United States of America 10216723ndash16728

Cooch E and G C White editors 2018 Program MARK a gentle in-troduction lthttpwwwphidotorgsoftwaremarkdocsbookgt Accessed 7Apr 2016

Cooley H S R B Wielgus G M Koehler H S Robinson and B TMaletzke 2009 Does hunting regulate cougar populations A test of thecompensatory mortality hypothesis Ecology 902913ndash2921

Coulson T E A Catchpole S D Albon B J Morgan J M Pemberton TH Clutton‐Brock M J Crawley and B T Grenfell 2001 Age sex densitywinter weather and population crashes in Soay sheep Science 2921528ndash1531

Coulson T J Pemberton S Albon M Beaumont T Marshall J Slate FGuinness and T Clutton‐Brock 1998a Microsatellites reveal heterosis in reddeer Proceedings of the Royal Society B Biological Sciences 265489ndash495

Coulson T N S D Albon J M Pemberton J Slate F E Guinness and TH Clutton‐Brock 1998b Genotype by environment interactions in wintersurvival in red deer Journal of Animal Ecology 67434ndash445

Cox D 1972 Regression models and life‐tables Journal of the Royal StatisticalSociety Series B Statistical Methodology 34187ndash220

Creel S 2006 Recovery of the Florida panthermdashgenetic rescue demographic rescueor both Response to Pimm et al (2006) Animal Conservation 9125ndash126

Crnokrak P and D A Roff 1999 Inbreeding depression in the wild Heredity83260ndash270

Crow J 1948 Alternative hypotheses of hybrid vigor Genetics 33477ndash487

van de Kerk et al bull Florida Panther Population and Genetic Viability 31

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 24: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

this case the migration of just a single male across the freewayto the south resulted in an increase in the number of in-dividuals with admixed ancestry and substantially improvedgenetic diversity of the subpopulation south of the freeway(Riley et al 2014)Our estimate of overall kitten survival (0321plusmn 0056) was

similar to that reported by Hostetler et al (2010) who used13 years of post‐introgression data (0323plusmn 0071) These valuesare lower than kitten survival estimates derived from severalpuma populations in western North America (044ndash091 Loganand Sweanor 2001 Lambert et al 2006 Laundre et al 2007Robinson et al 2008 Ruth et al 2011) When we included onlydata for individuals that had been genetically sampled our es-timate was 15 higher The proportion of admixed kittens thatwere genetically sampled increased as the population expandedwhich could have resulted in higher estimates of kitten survivalbecause admixed kittens survive better than canonical kittensKitten survival was strongly influenced by genetic ancestry withsurvival rates of F1 kittens being almost twice that of canonicalkittens (Fig 5B) Consistent with the findings of Hostetler et al(2010) increases in heterozygosity coincided with increasedkitten survival whereas population density negatively affectedkitten survival Population density often has a strong negativeimpact on survival of young age classes due to infanticide andcannibalism which has been reported from several carnivore

after 100 years after 200 years

MP

CM

VM

10 20 40 80 N 10 20 40 80 N

0

50

100

150

0

50

100

150

Introgression interval (yrs)

TQE

(yrs

)

Number of TX females added

no intervention

5 TX females

10 TX females

15 TX females

Figure 18 Average times until quasi‐extinction (TQE) for the Florida panther population within 100 years and within 200 years without genetic management interventionand with each of the genetic introgression strategies for the minimum population count (MPC) and motor vehicle mortality (MVM) scenarios Introgression strategiesinclude the introduction of 5 10 or 15 pumas from Texas USA every 10 20 40 or 80 years The critical threshold was 10 panthers Error bars indicate 5th and 95thpercentiles Based on data collected in South Florida USA 1981ndash2013

Table 5 Average cost (2017 US dollars) of introgression strategies employedover 100 years that involve the introduction of 5 10 or 15 pumas from TexasUSA into the Florida panther population in South Florida USA at an in-creasingly higher frequency Costs of capture (including transportation) quar-antine and post‐introgression monitoring were estimated by Roy McBrideMark Cunningham and Dave Onorato respectively

Frequency Component 5 pumas 10 pumas 15 pumas

Once Capture 25000 50000 75000

Quarantine 5000 10000 15000

Monitoring 10000 20000 30000

Total 40000 80000 120000

Every 80 years Capture 31250 62500 93750

Quarantine 6250 12500 18750

Monitoring 12500 25000 37500

Total 50000 100000 150000

Every 40 years Capture 62500 125000 187500

Quarantine 12500 25000 37500

Monitoring 25000 50000 75000

Total 100000 200000 300000

Every 20 years Capture 125000 250000 375000

Quarantine 25000 50000 75000

Monitoring 50000 100000 150000

Total 200000 400000 600000

Every 10 years Capture 250000 500000 750000

Quarantine 50000 100000 150000

Monitoring 100000 200000 300000

Total 400000 800000 1200000

26 Wildlife Monographs bull 203

species including polar bears (Ursus maritimus Derocher andWiig 1999) black bears (Ursus americanus Garrison et al 2007)and pumas (Logan and Sweanor 2001)Our estimates of sex‐ and age‐specific survival rates of subadult

and adult panthers (Table 3) and the effects of covariates on theserates were similar to those reported by Benson et al (2011) derivedfrom data collected through 2006 Our survival rates for males(065ndash077) were lower than females (078ndash097) for all 3 ageclasses (subadult prime adult and old adult) which is the typicaltrend observed in most puma populations (eg Ruth et al 19982011) However an unhunted puma population occupying afragmented urban landscape in southern California exhibited lowersurvival (0586 females 0525 males Vickers et al 2015) perhapsas a result of severe habitat fragmentation and the lack of extensiveswaths of public land protected in perpetuity Most populations ofpuma in western North America are typically subject to variedlevels of management via regulated hunting which often is biasedtoward the take of males (Cooley et al 2009 Ruth et al 2011)Consequently annual survival estimates from hunted populationsin northeast Washington (0679ndash0733 females 0341 malesRobinson et al 2008) and southeastern Arizona (0685 females0584 males Cunningham et al 2001) were generally lower thanthose we estimated for Florida panthersCause‐specific mortality analysis showed that male panthers

experienced a substantially greater mortality risk from in-traspecific aggression This differs from the pattern observedduring a long‐term study in Arizona where females were at ahigher risk of intraspecific mortality than males (46 of maleand 53 of female mortality Logan and Sweanor 2001)Mortality associated with intraspecific strife has been docu-mented in hunted and unhunted populations (this study Loganand Sweanor 2001 Stoner et al 2010) and its root cause hasbeen difficult to determine (eg population density and preypopulation trends) That being the case finding ways to reducewhat is often a major source of mortality for puma populationscontinues to pose a challenge to managersCanonical panthers were substantially more likely to die from

intraspecific aggression than were admixed panthers This mayat least partly explain the difference in survival between admixed

and canonical panthers Canonical panthers are known to bemore likely to suffer from varied correlates of inbreeding de-pression than admixed animals including atrial septal defects(Johnson et al 2010) and compromised immune systems(Roelke et al 1993) These factors have the potential to affectfitness and perhaps dominance of individuals when engaged inan intraspecific encounter A somewhat similar scenario wasobserved by Hogg et al (2006) in a population of bighorn sheepthat were part of an introgression project Admixed males in thispopulation had a greater probability of paternity than males thatwere less outbred This included coursing males with higherlevels of admixture being markedly more successful at fightingmales that were defending females in estrous (Hogg et al 2006)Heterosis resulting from genetic introgression is expected to be

the strongest in F1 individuals (Shull 1908 Crow 1948 Burkeand Arnold 2001 Waller 2015) and to progressively decline insubsequent generations (Pickup et al 2013 Frankham 2015)Given that admixed (especially F1) panthers survived better thancanonical panthers and that individual heterozygosity improvedsurvival of both sexes and all age classes our data provide evi-dence for heterotic advantage whereby individuals of mixedancestry had higher fitness (Shull 1908 Crow 1948) Our resultsare consistent with findings of other studies that involved in-tentional genetic introgression for conservation purposes Forexample Hogg et al (2006) detected substantial improvementsin survival and reproduction of introgressed bighorn sheep andMadsen et al (1999 2004) reported a dramatic increase in thepopulation of adders after genetic introgressionKnowledge of the persistence of benefits to a population ac-

crued via genetic introgression is essential for determiningwhen additional introgression may be warranted There is adepauperate amount of information regarding this topic and theidentification of factors that may influence persistence of thebenefits of genetic introgression (Tallmon et al 2004 Hedricket al 2014 Whiteley et al 2015) For example in minks(Neovison vison) Thirstrup et al (2014) found that outcrossingled to larger litters in F2 and F3 but this benefit disappeared insubsequent generations In a population of gray wolves Hedricket al (2014) suggested that benefits of genetic introgression may

Table 6 Costs and benefits accumulated over 100 years for genetic introgression at different intervals and with different numbers of pumas from Texas USAintroduced into the Florida panther population in South Florida USA Costs (2017 US dollars) include capture and transportation quarantine and post‐introgression monitoring Benefits are represented as average probability of quasi‐extinction (PQE and 5th and 95th percentiles) and changes (Δ) therein under thescenario using the minimum population count (McBride et al 2008 MPC) and the scenario using the motor vehicle mortality population estimates (McClintocket al 2015 MVM) The table is sorted by percent change in probability of quasi‐extinction under the MPC scenario

Interval (yr) Pumas Cost MPC PQE () ΔMPC PQE () MVM PQE () ΔMVM PQE ()

ndash 0 0 13 (0ndash99) 0 17 (0ndash100) 080 10 100000 12 (0ndash99) 10 (0ndash58) 16 (0ndash100) 5 (0ndash38)80 15 150000 12 (0ndash99) 13 (0ndash65) 15 (0ndash100) 8 (0ndash56)80 5 50000 12 (0ndash99) 14 (0ndash73) 15 (0ndash99) 9 (0ndash41)40 15 300000 10 (0ndash98) 26 (0ndash104) 14 (0ndash99) 13 (0ndash60)40 10 200000 9 (0ndash94) 35 (0ndash183) 13 (0ndash99) 21 (0ndash95)40 5 100000 8 (0ndash89) 39 (0ndash211) 12 (0ndash99) 26 (0ndash124)20 15 600000 8 (0ndash88) 42 (0ndash257) 13 (0ndash99) 24 (0ndash123)20 10 400000 6 (0ndash67) 54 (0ndash318) 10 (0ndash99) 37 (0ndash217)10 15 1200000 6 (0ndash53) 58 (0ndash372) 10 (0ndash99) 40 (0ndash208)20 5 200000 5 (0ndash50) 59 (0ndash338) 9 (0ndash99) 44 (0ndash352)10 10 800000 4 (0ndash16) 69 (0ndash489) 8 (0ndash92) 55 (0ndash401)10 5 400000 4 (0ndash7) 73 (0ndash502) 6 (0ndash71) 63 (0ndash503)

van de Kerk et al bull Florida Panther Population and Genetic Viability 27

wane after 2 or 3 generations The WrightndashFisher model ofgenetic drift predicts a geometric decline in expected hetero-zygosity at the rate of (1 minus 12Ne) per generation where Ne is theeffective population size (Hartl and Clark 1997 Frankham et al2014) Thus it seems logical to assume that the duration of thebenefits of introgression is limited especially in a species per-sisting in a small population such as Florida panthersWe expected Florida panther survival in the most recent years

(2005ndash2013) post‐introgression to differ from that observed in thedecade following the introgression in 1995 because pantherabundance and heterozygosity factors known to influence panthersurvival (Hostetler et al 2010 Benson et al 2011) have changed(Figs 2 and 6) We found that subadult and adult survival rates inrecent years (2005ndash2013) were similar to those in the period im-mediately following introgression (1995ndash2004 Fig 5C) overallkitten survival was similar to estimates of Hostetler et al (2010) Infact the pattern of covariate effects on Florida panther survivalreported by Benson et al (2011) and Hostetler et al (2010) hasbarely changed The negative effects of population density andpositive effects of heterozygosity may counterbalance survival ratesreducing population fluctuations Our result that benefits of geneticrestoration have remained in the population nearly for 5 genera-tions post‐intervention was surprising but also encouraging Pos-sible explanations for this observation may include 1) genetic ero-sion occurs at slower rate than previously thought 2) introducing 5migrants in 1 genetic intervention may have beneficial effects si-milar to those from 1 migrant per generation over multiple gen-erations or 3) other and as yet unknown factors may reduce therate of genetic erosion and subsequent demographic effects Adensity‐dependent effect on kitten survival could also explain whynon‐F1 admixed kittens did not survive substantially better thandid canonical kittens most kittens sampled from 2005ndash2013 wereadmixed and their survival was lower possibly because of a density‐dependent effect as the Florida panther population continuouslygrew during that period Trinkel et al (2010) also found thatpopulation density and inbreeding coefficient interacted to affectdemographic parameters and growth rate of an African lion po-pulation Likewise population density interacted with winterweather to affect the survival and population growth rates in Soaysheep (Ovis aries Milner et al 1999 Coulson et al 2001)

Population Dynamics and PersistenceThe finite deterministic and stochastic growth rates were gt10reflecting that much of the data used in this study were collectedfollowing genetic introgression in 1995 during which time thepopulation increased substantially Because our demographicparameter estimates did not change markedly in comparison tothose of Hostetler et al (2013) the similarity between thepopulation growth estimates from these studies was expectedBut these estimates reflect population growth during thedata‐collection period and do not predict future growth In factestimates of population size reported by McClintock et al(2015) suggest that population growth may already be slowing(Fig 2) A similar pattern was observed in an introduced po-pulation of African lion which increased steadily from 1995until 2001 then fluctuated around the presumed carrying ca-pacity thereafter (Trinkel et al 2010) Several other African lionpopulations that experienced various degrees of recovery

subsequently became relatively stable or exhibited decliningtrends (Bauer et al 2015)Our estimates of quasi‐extinction probabilities were lower than

those reported by Hostetler et al (2013) using a 2‐sex age‐structured matrix population model Lower probabilities ofquasi‐extinction than those reported by the earlier study forcomparable scenarios were likely a consequence of the fact thatthe estimates of abundance (McBride et al 2008) and thus thedensity‐dependent effects on survival of kittens and old panthers(gt10 yr) have changed in recent years Additionally these earlyestimates of the probability of quasi‐extinction and of time toquasi‐extinction did not consider genetic erosion that will in-evitably occur without management interventionUnder the MVM scenario the equilibrium population size was

twice that attained under the MPC scenario The MVM popula-tion estimates are approximately twice the MPCs (Fig 2) as aresult the simulated population under the MVM scenario achieveda higher equilibrium size Probability of quasi‐extinction under theMVM scenario was generally greater than that under the MPCscenario This result is a consequence of the fact that the estimateddensity dependence on kitten survival based on the MPC scenario isstronger than that for theMVM scenario (regression coefficients forabundance for the MPC scenario= minus0068 vs minus0002 for theMVM scenario Appendix B Table B1) Consequently simulatedpopulations under the MPC scenario have a stronger tendency tomove toward the equilibrium population size which inherentlyreduces the quasi‐extinction probability (eg Royama 1992)As expected for long‐lived mammals (Heppell et al 2000 Oli

and Dobson 2003) the population growth rate was proportio-nately most sensitive to changes in survival of prime adultsfollowed by that in younger age classes Global sensitivity ana-lysis in contrast showed that under all scenarios extinctionparameters were particularly sensitive to changes in survival ofFlorida panther kittens This result is a consequence of the highsensitivity of the population growth rate to kitten survival(Fig 9) and a larger standard error of kitten survival (Table 3)Results of our sensitivity analyses indicate that future researchshould focus on acquiring additional information on early lifestages to improve the accuracy and precision of estimates ofkitten survival Our results suggest that demographic variables(or their environmental drivers) that strongly influence extinc-tion risks are not necessarily those with the largest elasticityvalues A study of island fox (Bakker et al 2009) found thatextinction risk was strongly influenced by survival modifierparameters that had low elasticity values

Genetic Management When and How to InterveneThe use of genetic introgression as a management tool has al-ways been controversial and the probable effectiveness it mighthave in preventing the imminent demise of the panther popu-lation was initially debated (Creel 2006 Maehr et al 2006Pimm et al 2006) These debates notwithstanding the pantherpopulation had suffered from genetic morphological and bio-medical correlates of inbreeding before this management in-tervention (Johnson et al 2010 Onorato et al 2010) whereasthe post‐introgression period has been characterized by a sub-stantial reduction in the biomedical correlates of inbreeding andimprovements in demographic vigor and abundance (McBride

28 Wildlife Monographs bull 203

et al 2008 Hostetler et al 2010 2013 Johnson et al 2010Benson et al 2011 McClintock et al 2015) Nevertheless thepopulation remains small and isolated habitat is still being lostand there is no current possibility of natural gene flow betweenthis and other North American puma populations thus therecurrence of inbreeding‐related problems and subsequent po-pulation declines are inevitable The question therefore is notwhether genetic management of the Florida panther populationis needed but when and how it should be implementedWithout genetic introgression probabilities of quasi‐extinc-

tion were substantially greater than those from models thatincorporated periodic genetic introgression events (Fig 15)highlighting the importance of genetic management in smallisolated populations When 5 female pumas are added to theFlorida panther population every 10 years genetic variationincreased substantially under the MPC and MVM scenariosThe IBM predicted that the equillibrium population size wouldbe substantially larger when 5 pumas were released into thepopulation every 10 years than populations without intervention(ie no introgression) Releasing 15 Texas females did not af-ford substantially greater benefit to the equilibrium populationsize than did releasing only 5 Texas females Instead addingmore individuals caused increasingly large fluctuations in po-pulation size due to the numerical increases and density de-pendence (Fig 14) possibly diminishing the benefits associatedwith increased genetic variationCompared with the no‐intervention scenario (ie no genetic

introgression implemented) the introduction of 5 female pumasevery 10 years reduced quasi‐extinction probabilities from 13to 4 and from 17 to 6 for the MPC and MVM scenariosrespectively The benefit of genetic‐introgression strategies de-creased as the interval between successive management inter-ventions increased and no benefit was evident once the intervalreached 80 years (Fig 15) Introgression reduced the averagetime until quasi‐extinction (Fig 18) This somewhat counter‐intuitive result can be explained by the fact that repeated geneticintrogression makes the population more robust over timewhich will reduce the probability of extinction Consequentlyfew simulated population trajectories that fall below the criticalthreshold do so early reducing the mean time to extinctionAlso density dependence has a stabilizing effect on populations(Royama 1992) populations tend toward equilibrium popula-tion sizes which reduces the probability over time of popula-tions falling below quasi‐extinction threshold However timeuntil quasi‐extinction must be interpreted in conjunction withthe probability of quasi‐extinction which is very low for allscenarios with genetic introgression (Fig 15)Cost is a significant impediment to the implementation of a

genetic introgression program because captures relocations andmonitoring efforts before and after introgression are expensive(Edmands 2007 Frankham 2010b 2015 2016) Costs may beacceptable to society if the management actions result in sub-stantial fitness benefits especially if the species of concern ispopular and charismatic (Frankham 2015) We estimated the100‐year cost of introgression strategies modeled here to rangefrom $50000 to $12 million More expensive strategies gen-erally led to larger decreases in the probability of quasi‐extinc-tion but cheaper strategies also substantially improved

population viability (ie reduced quasi‐extinction probabilityTable 6) One of the cheaper strategies (5 pumas released every20 years) which cost an estimated $200000 over 100 yearsreduced the probability of quasi‐extinction by 59 and 44 forthe MPC and MVM scenarios respectively But these pre-liminary cost estimates are based on expenses incurred duringthe 1995 introgression and include costs of capture quarantinetransportation release and post‐release monitoring of femalesonly Although the cost for future genetic interventions wouldlikely be higher than those reported here the relative differencesin cost between alternative intervention strategies should remainapproximately the sameOur ability to simulate genetics and link it to the viability of

the Florida panther population is based on the empirically es-timated relationship between individual heterozygosity andsurvival Such heterozygosity and fitness correlations have beenstudied for decades (David 1998) but have only recently beenused to predict population viability (Benson et al 2016) noneto our knowledge have incorporated cost into their modelingefforts The mechanisms underlying heterozygosity and fitnesscorrelations remain unclear (David 1998 Szulkin et al 2010)and their significance as a proxy for inbreeding depression hasbeen debated (Balloux et al 2004 Miller and Coltman 2014)Nevertheless evidence continues to mount demonstratingpositive relationships between heterozygosity and survival(Coulson et al 1998ab Hansson et al 2004 Silva et al 2005Acevedo‐Whitehouse et al 2006 Jensen et al 2007 Velandoet al 2015) and between heterozygosity and fecundity (Seddonet al 2004 Charpentier et al 2005 Agudo et al 2012 Velandoet al 2015) Although the reported correlations are generallyweak their consequences can be substantial if population growthand persistence parameters are highly sensitive to the affectedvital rates (Velando et al 2015) Compared with scenarios thatignore genetics incorporating the effects of inbreeding on sur-vival increased quasi‐extinction probabilities within a 100‐yeartimeframe from 15 to 13 and from 14 to 17 for theMPC and MVM scenarios respectively These results high-light the importance of incorporating genetics when analyzingthe viability of small isolated populationsAlthough conservation biologists have recognized the im-

portance of genetics in population viability many still fail toincorporate genetic factors into PVA models (Reed et al 2002)Explicit consideration of genetics in PVAs requires long‐termgenetic and demographic data This permits the assessment ofthe functional relationship between genetic variability and de-mographic parameters Population viability analysis softwarepackages such as VORTEX (Lacy 1993 2000) offer a powerfulframework for individual‐based simulations but they often donot adequately capture a speciesrsquo life history or genetics Basedon a thorough review of the importance of genetic factors inwildlife conservation Frankham et al (2014) concluded thatgenetic factors can strongly influence the dynamics and persis-tence of small andor fragmented populations They furthernoted that ldquoMost PVA‐based risk assessments ignore or in-adequately model genetic factorsrdquo and recommended that ldquoPVAshould routinely include realistic inbreeding depressionhelliprdquo(Frankham et al 201457) Our custom‐built IBM permitted usto develop a model that adequately captured the Florida panther

van de Kerk et al bull Florida Panther Population and Genetic Viability 29

life history and we could parameterize the model with our long‐term genetic and demographic dataThe explicit consideration of the effects of inbreeding on

Florida panther population dynamics and persistence representsan important step forward Nonetheless our model remainsimperfect First we neglected the possible effects of habitat lossdetrimental anthropogenic activities climate change the prob-ability of immigration of pumas from western populationsdisease or catastrophes on demographic rates and populationviability Although immigration from puma populations outsideFlorida is possible its probability is very low given the extensivechallenges the anthropogenically altered landscape would posefor such a migrant to make it successfully to South Florida Weomitted mutations from our model because we have no in-formation on mutation rates though we expect that they are lowbased on values reported for other mammals (Kumar and Sub-ramanian 2002) Finally we only considered possible effects ofinbreeding on age‐specific survival rates because there was noevidence that heterozygosity affected female reproduction Lossof heterozygosity can negatively affect reproduction becauseinbred male panthers can suffer from cryptorchidism which canadversely affect their fecundity However given the polygynousmating system we expect the Florida panther population growthis primarily female‐limitedAlthough it is generally accepted that genetic introgression

only temporarily relieves a population from inbreeding depres-sion we know of no cases in which it has been implementedmore than once to conserve a threatened wildlife populationOur study provides an example of how demographic and mo-lecular data can be used within an IBM framework to evaluatethe efficacy of alternative genetic management strategies via theFlorida panther as a case study Our model can be adjusted forand applied to other species and offers great potential as a toolfor genetic management of small isolated wildlife populations

MANAGEMENT IMPLICATIONS

Our results and those of Hostetler et al (2013) and Johnsonet al (2010) provide persuasive evidence that the genetic in-tervention implemented in 1995 prevented the demise of theFlorida panther and restored demographic vigor to the popu-lation The Florida panther population remains small and iso-lated from other puma populations the closest puma populationis located gt1000 km away in Texas and the intervening land-scape is highly developed and fragmented with many interstate(and other high‐traffic) highways and major urban centersTheory suggests that 1 migrant per generation is sufficient tominimize the loss of genetic diversity (eg Mills and Allendorf1996) However the possibility of successful migration of apuma to South Florida followed by successful mating every ~5years is very low considering the distance between Texas andFlorida (Hedrick 1995) and the many risks and barriers thatdispersing pumas face (Beier 1995 Maehr et al 2002 Thatcheret al 2006) Consequently the pantherrsquos long‐term persistencewill likely depend on subsequent timely genetic introgressionsgiven the near‐impossibility of natural gene flow from otherpuma populations To that end our study offers considerableinsights into benefits of different genetic management strategiesand associated costs

Without genetic management the Florida panther populationfaces a substantial risk of quasi‐extinction within 100 years (10ndash46 depending on quasi‐extinction threshold and scenarios)Twenty‐three years have passed since genetic introgression wasimplemented and the population appears healthy as indicatedby measures of genetic variation increases in the populationsize and improvements in age‐specific survival rates Ourfindings suggest that releasing more pumas more frequently doesnot necessarily improve persistence probability because such astrategy would cause larger fluctuations in population size(Fig 14) which negatively affects persistence probability Thusextinction risks faced by inbred populations of large carnivorescan be substantially reduced by releasing approximately 5 im-migrants every 2 decades We suggest that such a managementstrategy be followed only in conjunction with continued long‐term monitoring of the population Our analyses demonstratethat population growth and persistence are highly sensitive tokitten and female (adult and subadult) survival Continuing tocollect data on these demographic variables along with bio-metric and genetic sampling will remain important to pantherrecovery and vigilance in identifying issues associated with in-breeding depression The collection of these monitoring datashould allow managers to detect any demographic or geneticchanges in a timely fashion and respond with genetic in-trogression or other management actions if warrantedRecovery objectives as defined by the USFWS in its current

recovery plan (USFWS 2008) delineate the need for additionalpopulations in the historical range to downlist or delist theFlorida panther If the only extant population of Floridapanthers continued to grow it could serve as a source ofpanthers to be translocated to suitable habitat to seed addi-tional populations Maintaining current information on thegenetic health of the population and precise estimates of itssize will be important in assessing whether it can withstand theremoval of a subset of animals for translocation Although the2017 documentation of a female panther with kittens north ofthe current breeding range is encouraging in terms of thepotential for population expansionmdashgiven that no female hadbeen recorded north of the Caloosahatchee River since 1972mdashthe re‐establishment of separate new populations will mostlikely require translocations by wildlife managers and a lengthysociopolitical processConservation of threatened or endangered species such as the

Florida panther is inherently challenging For panthers and otherlarge carnivores that require vast extents of natural habitat topersist habitat is typically the most limiting factor that will de-termine whether recovery efforts succeed Although geneticmanagement invariably plays a key role in the long‐term persis-tence of the panther population even a healthy population caneventually succumb to the impacts of habitat loss The humanpopulation of Florida continues to be one of the fastest‐growingin the nation the United States Census Bureau currently ranksFlorida as the third most populous Therefore habitat loss isgoing to continue to be a key issue for panthers Diligence towardpreserving panther habitat in its historical range via conservationeasements or other means and trying to keep such areas inter-connected via corridors is a goal stakeholders such as governmentagencies sportsmen non‐governmental organizations ranchers

30 Wildlife Monographs bull 203

and other private landowners must work on collaboratively toachieve

ACKNOWLEDGMENTS

We thank D Shindle D Jennings T OrsquoMeara D Land andC Belden for valuable advice throughout the project We thankS Bass M Criffield M Cunningham D Giardina D JansenA Johnson D Land M Lotz C McBride Rocky McBrideRowdy McBride Roy McBride L Oberhofer S Schulze staffsat Everglades National Park and Big Cypress National Preserveand others for assistance with fieldwork We also thank JAustin M Conroy J Hines J Laake S Mills J Nichols JHines M Sunquist J Fryxell and T Coulson for advice andassistance regarding data analysis and panther biology TheMuseum of Texas Tech University Natural Science ResearchLaboratory provided samples from pumas from west Texas forcomparative genetic analyses M Ben‐David D Land WMcCown E Hellgren and 4 anonymous reviewers providedmany helpful comments on the manuscript We thank NereaUbierna Lopez for completing the Spanish translation of theabstract We are grateful to the Department of ResearchComputing University of Florida for excellent high‐perfor-mance computing services We would like to thank US Fishand Wildlife Service Florida Fish and Wildlife ConservationCommission (FWC) US National Park Service QuantitativeSpatial Ecology Evolution and Environment (QSE3) In-tegrative Graduate Education and Research Traineeship(IGERT) program funded by the National Science Foundationand the University of Florida for financially supporting thisstudy Funding for panther research conducted by the FWC issupported predominantly via donations accrued with vehicleregistrations for ldquoProtect the Pantherrdquo license plates

LITERATURE CITEDAcevedo‐Whitehouse K T R Spraker E Lyons S R Melin F Gulland RL Delong and W Amos 2006 Contrasting effects of heterozygosity onsurvival and hookworm resistance in California sea lion pups MolecularEcology 151973ndash1982

Adams J R L M Vucetich P W Hedrick R O Peterson and J AVucetich 2011 Genomic sweep and potential genetic rescue during limitingenvironmental conditions in an isolated wolf population Proceedings of theRoyal Society B Biological Sciences 2783336ndash3344

Agresti A 2013 Categorical data analysis Third edition John Wiley amp SonsHoboken New Jersey USA

Agudo R M Carrete M Alcaide C Rico F Hiraldo and J A Donaacutezar2012 Genetic diversity at neutral and adaptive loci determines individualfitness in a long‐lived territorial bird Proceedings of the Royal Society BBiological Sciences 2793241ndash3249

Albeke S E N P Nibbelink and M Ben‐David 2015 Modeling behavior bycoastal river otter (Lontra canadensis) in response to prey availability in PrinceWilliam Sound Alaska a spatially‐explicit individual‐based approach PLOSOne 10e0126208

Alho J S K Vaumllimaumlki and J Merilauml 2010 Rhh an R extension for estimatingmultilocus heterozygosity and heterozygosity‐heterozygosity correlationMolecular Ecology Resources 10720ndash722

Alvarez K 1993 Twilight of the panther Myakka River Publishing SarasotaFlorida USA

Aparicio J M J Ortego and P J Cordero 2006 What should we weigh toestimate heterozygosity alleles or loci Molecular Ecology 154659ndash4665

Bakker V J D F Doak G W Roemer D K Garcelon T J Coonan S AMorrison C Lynch K Ralls and R Shaw 2009 Incorporating ecologicaldrivers and uncertainty into a demographic population viability analysis for theisland fox Ecological Monographs 7977ndash108

Balloux F W Amos and T Coulson 2004 Does heterozygosity estimateinbreeding in real populations Molecular Ecology 133021ndash3031

Barone M A M E Roelke J Howard J L Brown A E Anderson and DE Wildt 1994 Reproductive characteristics of male Florida panthersmdashcomparative studies from Florida Texas Colorado Latin‐America andNorth‐American Zoos Journal of Mammalogy 75150ndash162

Bateson Z W P O Dunn S D Hull A E Henschen J A Johnson and LA Whittingham 2014 Genetic restoration of a threatened population ofgreater prairie‐chickens Biological Conservation 17412ndash19

Bauer H G Chapron K Nowell P Henschel P Funston L T B HunterD W Macdonald and C Packer 2015 Lion (Panthera leo) populations aredeclining rapidly across Africa except in intensively managed areas Pro-ceedings of National Academy of Sciences of the United States of America11214894ndash14899

Beier P 1995 Dispersal of juvenile cougars in fragmented habitat Journal ofWildlife Management 59228ndash237

Benson J F J A Hostetler D P Onorato W E Johnson M E Roelke SJ OrsquoBrien D Jansen and M K Oli 2011 Intentional genetic introgressioninfluences survival of adults and subadults in a small inbred felid populationJournal of Animal Ecology 80958ndash967

Benson J F P J Mahoney J A Sikich L E K Serieys J P Pollinger H BErnest and S P D Riley 2016 Interactions between demography geneticsand landscape connectivity increase extinction probability for a small popu-lation of large carnivores in a major metropolitan area Proceedings of theRoyal Society B Biological Sciences 28320160957

Beverly F 1874 The Florida panther Forest and Stream 3290Boyce M S C V Haridas C T Lee and the NCEAS Stochastic Demo-graphy Working Group 2006 Demography in an increasingly variable worldTrends in Ecology amp Evolution 21141ndash148

Brook B W J J OrsquoGrady A P Chapman M A Burgman H R Akcakayaand R Frankham 2000 Predictive accuracy of population viability analysis inconservation biology Nature 404385ndash387

Brys R and H Jacquemyn 2016 Severe outbreeding and inbreeding de-pression maintain mating system differentiation in Epipactis (Orchidaceae)Journal of Evolutionary Biology 29352ndash359

Burke J M and M L Arnold 2001 Genetics and the fitness of hybridsAnnual Review of Genetics 3531ndash52

Burnham K P 1993 A theory for combined analysis of ring recovery andrecapture data Pages 199ndash213 in J‐D Lebreton and P M North editorsMarked individuals in the study of bird population Springer‐Verlag NewYork New York USA

Burnham K P and D R Anderson 2002 Model selection and multimodelinference a practical information‐theoretic approach Second edition SpringerVerlag New York New York USA

Caswell H 2001 Matrix population models construction analysis and in-terpretation Sinauer Associates Sunderland Massachusetts USA

Charpentier M J M Setchell F Prugnolle L A Knapp E J Wickings PPeignot and M Hossaert‐McKey 2005 Genetic diversity and reproductivesuccess in mandrills (Mandrillus sphinx) Proceedings of the NationalAcademy of Sciences of the United States of America 10216723ndash16728

Cooch E and G C White editors 2018 Program MARK a gentle in-troduction lthttpwwwphidotorgsoftwaremarkdocsbookgt Accessed 7Apr 2016

Cooley H S R B Wielgus G M Koehler H S Robinson and B TMaletzke 2009 Does hunting regulate cougar populations A test of thecompensatory mortality hypothesis Ecology 902913ndash2921

Coulson T E A Catchpole S D Albon B J Morgan J M Pemberton TH Clutton‐Brock M J Crawley and B T Grenfell 2001 Age sex densitywinter weather and population crashes in Soay sheep Science 2921528ndash1531

Coulson T J Pemberton S Albon M Beaumont T Marshall J Slate FGuinness and T Clutton‐Brock 1998a Microsatellites reveal heterosis in reddeer Proceedings of the Royal Society B Biological Sciences 265489ndash495

Coulson T N S D Albon J M Pemberton J Slate F E Guinness and TH Clutton‐Brock 1998b Genotype by environment interactions in wintersurvival in red deer Journal of Animal Ecology 67434ndash445

Cox D 1972 Regression models and life‐tables Journal of the Royal StatisticalSociety Series B Statistical Methodology 34187ndash220

Creel S 2006 Recovery of the Florida panthermdashgenetic rescue demographic rescueor both Response to Pimm et al (2006) Animal Conservation 9125ndash126

Crnokrak P and D A Roff 1999 Inbreeding depression in the wild Heredity83260ndash270

Crow J 1948 Alternative hypotheses of hybrid vigor Genetics 33477ndash487

van de Kerk et al bull Florida Panther Population and Genetic Viability 31

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 25: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

species including polar bears (Ursus maritimus Derocher andWiig 1999) black bears (Ursus americanus Garrison et al 2007)and pumas (Logan and Sweanor 2001)Our estimates of sex‐ and age‐specific survival rates of subadult

and adult panthers (Table 3) and the effects of covariates on theserates were similar to those reported by Benson et al (2011) derivedfrom data collected through 2006 Our survival rates for males(065ndash077) were lower than females (078ndash097) for all 3 ageclasses (subadult prime adult and old adult) which is the typicaltrend observed in most puma populations (eg Ruth et al 19982011) However an unhunted puma population occupying afragmented urban landscape in southern California exhibited lowersurvival (0586 females 0525 males Vickers et al 2015) perhapsas a result of severe habitat fragmentation and the lack of extensiveswaths of public land protected in perpetuity Most populations ofpuma in western North America are typically subject to variedlevels of management via regulated hunting which often is biasedtoward the take of males (Cooley et al 2009 Ruth et al 2011)Consequently annual survival estimates from hunted populationsin northeast Washington (0679ndash0733 females 0341 malesRobinson et al 2008) and southeastern Arizona (0685 females0584 males Cunningham et al 2001) were generally lower thanthose we estimated for Florida panthersCause‐specific mortality analysis showed that male panthers

experienced a substantially greater mortality risk from in-traspecific aggression This differs from the pattern observedduring a long‐term study in Arizona where females were at ahigher risk of intraspecific mortality than males (46 of maleand 53 of female mortality Logan and Sweanor 2001)Mortality associated with intraspecific strife has been docu-mented in hunted and unhunted populations (this study Loganand Sweanor 2001 Stoner et al 2010) and its root cause hasbeen difficult to determine (eg population density and preypopulation trends) That being the case finding ways to reducewhat is often a major source of mortality for puma populationscontinues to pose a challenge to managersCanonical panthers were substantially more likely to die from

intraspecific aggression than were admixed panthers This mayat least partly explain the difference in survival between admixed

and canonical panthers Canonical panthers are known to bemore likely to suffer from varied correlates of inbreeding de-pression than admixed animals including atrial septal defects(Johnson et al 2010) and compromised immune systems(Roelke et al 1993) These factors have the potential to affectfitness and perhaps dominance of individuals when engaged inan intraspecific encounter A somewhat similar scenario wasobserved by Hogg et al (2006) in a population of bighorn sheepthat were part of an introgression project Admixed males in thispopulation had a greater probability of paternity than males thatwere less outbred This included coursing males with higherlevels of admixture being markedly more successful at fightingmales that were defending females in estrous (Hogg et al 2006)Heterosis resulting from genetic introgression is expected to be

the strongest in F1 individuals (Shull 1908 Crow 1948 Burkeand Arnold 2001 Waller 2015) and to progressively decline insubsequent generations (Pickup et al 2013 Frankham 2015)Given that admixed (especially F1) panthers survived better thancanonical panthers and that individual heterozygosity improvedsurvival of both sexes and all age classes our data provide evi-dence for heterotic advantage whereby individuals of mixedancestry had higher fitness (Shull 1908 Crow 1948) Our resultsare consistent with findings of other studies that involved in-tentional genetic introgression for conservation purposes Forexample Hogg et al (2006) detected substantial improvementsin survival and reproduction of introgressed bighorn sheep andMadsen et al (1999 2004) reported a dramatic increase in thepopulation of adders after genetic introgressionKnowledge of the persistence of benefits to a population ac-

crued via genetic introgression is essential for determiningwhen additional introgression may be warranted There is adepauperate amount of information regarding this topic and theidentification of factors that may influence persistence of thebenefits of genetic introgression (Tallmon et al 2004 Hedricket al 2014 Whiteley et al 2015) For example in minks(Neovison vison) Thirstrup et al (2014) found that outcrossingled to larger litters in F2 and F3 but this benefit disappeared insubsequent generations In a population of gray wolves Hedricket al (2014) suggested that benefits of genetic introgression may

Table 6 Costs and benefits accumulated over 100 years for genetic introgression at different intervals and with different numbers of pumas from Texas USAintroduced into the Florida panther population in South Florida USA Costs (2017 US dollars) include capture and transportation quarantine and post‐introgression monitoring Benefits are represented as average probability of quasi‐extinction (PQE and 5th and 95th percentiles) and changes (Δ) therein under thescenario using the minimum population count (McBride et al 2008 MPC) and the scenario using the motor vehicle mortality population estimates (McClintocket al 2015 MVM) The table is sorted by percent change in probability of quasi‐extinction under the MPC scenario

Interval (yr) Pumas Cost MPC PQE () ΔMPC PQE () MVM PQE () ΔMVM PQE ()

ndash 0 0 13 (0ndash99) 0 17 (0ndash100) 080 10 100000 12 (0ndash99) 10 (0ndash58) 16 (0ndash100) 5 (0ndash38)80 15 150000 12 (0ndash99) 13 (0ndash65) 15 (0ndash100) 8 (0ndash56)80 5 50000 12 (0ndash99) 14 (0ndash73) 15 (0ndash99) 9 (0ndash41)40 15 300000 10 (0ndash98) 26 (0ndash104) 14 (0ndash99) 13 (0ndash60)40 10 200000 9 (0ndash94) 35 (0ndash183) 13 (0ndash99) 21 (0ndash95)40 5 100000 8 (0ndash89) 39 (0ndash211) 12 (0ndash99) 26 (0ndash124)20 15 600000 8 (0ndash88) 42 (0ndash257) 13 (0ndash99) 24 (0ndash123)20 10 400000 6 (0ndash67) 54 (0ndash318) 10 (0ndash99) 37 (0ndash217)10 15 1200000 6 (0ndash53) 58 (0ndash372) 10 (0ndash99) 40 (0ndash208)20 5 200000 5 (0ndash50) 59 (0ndash338) 9 (0ndash99) 44 (0ndash352)10 10 800000 4 (0ndash16) 69 (0ndash489) 8 (0ndash92) 55 (0ndash401)10 5 400000 4 (0ndash7) 73 (0ndash502) 6 (0ndash71) 63 (0ndash503)

van de Kerk et al bull Florida Panther Population and Genetic Viability 27

wane after 2 or 3 generations The WrightndashFisher model ofgenetic drift predicts a geometric decline in expected hetero-zygosity at the rate of (1 minus 12Ne) per generation where Ne is theeffective population size (Hartl and Clark 1997 Frankham et al2014) Thus it seems logical to assume that the duration of thebenefits of introgression is limited especially in a species per-sisting in a small population such as Florida panthersWe expected Florida panther survival in the most recent years

(2005ndash2013) post‐introgression to differ from that observed in thedecade following the introgression in 1995 because pantherabundance and heterozygosity factors known to influence panthersurvival (Hostetler et al 2010 Benson et al 2011) have changed(Figs 2 and 6) We found that subadult and adult survival rates inrecent years (2005ndash2013) were similar to those in the period im-mediately following introgression (1995ndash2004 Fig 5C) overallkitten survival was similar to estimates of Hostetler et al (2010) Infact the pattern of covariate effects on Florida panther survivalreported by Benson et al (2011) and Hostetler et al (2010) hasbarely changed The negative effects of population density andpositive effects of heterozygosity may counterbalance survival ratesreducing population fluctuations Our result that benefits of geneticrestoration have remained in the population nearly for 5 genera-tions post‐intervention was surprising but also encouraging Pos-sible explanations for this observation may include 1) genetic ero-sion occurs at slower rate than previously thought 2) introducing 5migrants in 1 genetic intervention may have beneficial effects si-milar to those from 1 migrant per generation over multiple gen-erations or 3) other and as yet unknown factors may reduce therate of genetic erosion and subsequent demographic effects Adensity‐dependent effect on kitten survival could also explain whynon‐F1 admixed kittens did not survive substantially better thandid canonical kittens most kittens sampled from 2005ndash2013 wereadmixed and their survival was lower possibly because of a density‐dependent effect as the Florida panther population continuouslygrew during that period Trinkel et al (2010) also found thatpopulation density and inbreeding coefficient interacted to affectdemographic parameters and growth rate of an African lion po-pulation Likewise population density interacted with winterweather to affect the survival and population growth rates in Soaysheep (Ovis aries Milner et al 1999 Coulson et al 2001)

Population Dynamics and PersistenceThe finite deterministic and stochastic growth rates were gt10reflecting that much of the data used in this study were collectedfollowing genetic introgression in 1995 during which time thepopulation increased substantially Because our demographicparameter estimates did not change markedly in comparison tothose of Hostetler et al (2013) the similarity between thepopulation growth estimates from these studies was expectedBut these estimates reflect population growth during thedata‐collection period and do not predict future growth In factestimates of population size reported by McClintock et al(2015) suggest that population growth may already be slowing(Fig 2) A similar pattern was observed in an introduced po-pulation of African lion which increased steadily from 1995until 2001 then fluctuated around the presumed carrying ca-pacity thereafter (Trinkel et al 2010) Several other African lionpopulations that experienced various degrees of recovery

subsequently became relatively stable or exhibited decliningtrends (Bauer et al 2015)Our estimates of quasi‐extinction probabilities were lower than

those reported by Hostetler et al (2013) using a 2‐sex age‐structured matrix population model Lower probabilities ofquasi‐extinction than those reported by the earlier study forcomparable scenarios were likely a consequence of the fact thatthe estimates of abundance (McBride et al 2008) and thus thedensity‐dependent effects on survival of kittens and old panthers(gt10 yr) have changed in recent years Additionally these earlyestimates of the probability of quasi‐extinction and of time toquasi‐extinction did not consider genetic erosion that will in-evitably occur without management interventionUnder the MVM scenario the equilibrium population size was

twice that attained under the MPC scenario The MVM popula-tion estimates are approximately twice the MPCs (Fig 2) as aresult the simulated population under the MVM scenario achieveda higher equilibrium size Probability of quasi‐extinction under theMVM scenario was generally greater than that under the MPCscenario This result is a consequence of the fact that the estimateddensity dependence on kitten survival based on the MPC scenario isstronger than that for theMVM scenario (regression coefficients forabundance for the MPC scenario= minus0068 vs minus0002 for theMVM scenario Appendix B Table B1) Consequently simulatedpopulations under the MPC scenario have a stronger tendency tomove toward the equilibrium population size which inherentlyreduces the quasi‐extinction probability (eg Royama 1992)As expected for long‐lived mammals (Heppell et al 2000 Oli

and Dobson 2003) the population growth rate was proportio-nately most sensitive to changes in survival of prime adultsfollowed by that in younger age classes Global sensitivity ana-lysis in contrast showed that under all scenarios extinctionparameters were particularly sensitive to changes in survival ofFlorida panther kittens This result is a consequence of the highsensitivity of the population growth rate to kitten survival(Fig 9) and a larger standard error of kitten survival (Table 3)Results of our sensitivity analyses indicate that future researchshould focus on acquiring additional information on early lifestages to improve the accuracy and precision of estimates ofkitten survival Our results suggest that demographic variables(or their environmental drivers) that strongly influence extinc-tion risks are not necessarily those with the largest elasticityvalues A study of island fox (Bakker et al 2009) found thatextinction risk was strongly influenced by survival modifierparameters that had low elasticity values

Genetic Management When and How to InterveneThe use of genetic introgression as a management tool has al-ways been controversial and the probable effectiveness it mighthave in preventing the imminent demise of the panther popu-lation was initially debated (Creel 2006 Maehr et al 2006Pimm et al 2006) These debates notwithstanding the pantherpopulation had suffered from genetic morphological and bio-medical correlates of inbreeding before this management in-tervention (Johnson et al 2010 Onorato et al 2010) whereasthe post‐introgression period has been characterized by a sub-stantial reduction in the biomedical correlates of inbreeding andimprovements in demographic vigor and abundance (McBride

28 Wildlife Monographs bull 203

et al 2008 Hostetler et al 2010 2013 Johnson et al 2010Benson et al 2011 McClintock et al 2015) Nevertheless thepopulation remains small and isolated habitat is still being lostand there is no current possibility of natural gene flow betweenthis and other North American puma populations thus therecurrence of inbreeding‐related problems and subsequent po-pulation declines are inevitable The question therefore is notwhether genetic management of the Florida panther populationis needed but when and how it should be implementedWithout genetic introgression probabilities of quasi‐extinc-

tion were substantially greater than those from models thatincorporated periodic genetic introgression events (Fig 15)highlighting the importance of genetic management in smallisolated populations When 5 female pumas are added to theFlorida panther population every 10 years genetic variationincreased substantially under the MPC and MVM scenariosThe IBM predicted that the equillibrium population size wouldbe substantially larger when 5 pumas were released into thepopulation every 10 years than populations without intervention(ie no introgression) Releasing 15 Texas females did not af-ford substantially greater benefit to the equilibrium populationsize than did releasing only 5 Texas females Instead addingmore individuals caused increasingly large fluctuations in po-pulation size due to the numerical increases and density de-pendence (Fig 14) possibly diminishing the benefits associatedwith increased genetic variationCompared with the no‐intervention scenario (ie no genetic

introgression implemented) the introduction of 5 female pumasevery 10 years reduced quasi‐extinction probabilities from 13to 4 and from 17 to 6 for the MPC and MVM scenariosrespectively The benefit of genetic‐introgression strategies de-creased as the interval between successive management inter-ventions increased and no benefit was evident once the intervalreached 80 years (Fig 15) Introgression reduced the averagetime until quasi‐extinction (Fig 18) This somewhat counter‐intuitive result can be explained by the fact that repeated geneticintrogression makes the population more robust over timewhich will reduce the probability of extinction Consequentlyfew simulated population trajectories that fall below the criticalthreshold do so early reducing the mean time to extinctionAlso density dependence has a stabilizing effect on populations(Royama 1992) populations tend toward equilibrium popula-tion sizes which reduces the probability over time of popula-tions falling below quasi‐extinction threshold However timeuntil quasi‐extinction must be interpreted in conjunction withthe probability of quasi‐extinction which is very low for allscenarios with genetic introgression (Fig 15)Cost is a significant impediment to the implementation of a

genetic introgression program because captures relocations andmonitoring efforts before and after introgression are expensive(Edmands 2007 Frankham 2010b 2015 2016) Costs may beacceptable to society if the management actions result in sub-stantial fitness benefits especially if the species of concern ispopular and charismatic (Frankham 2015) We estimated the100‐year cost of introgression strategies modeled here to rangefrom $50000 to $12 million More expensive strategies gen-erally led to larger decreases in the probability of quasi‐extinc-tion but cheaper strategies also substantially improved

population viability (ie reduced quasi‐extinction probabilityTable 6) One of the cheaper strategies (5 pumas released every20 years) which cost an estimated $200000 over 100 yearsreduced the probability of quasi‐extinction by 59 and 44 forthe MPC and MVM scenarios respectively But these pre-liminary cost estimates are based on expenses incurred duringthe 1995 introgression and include costs of capture quarantinetransportation release and post‐release monitoring of femalesonly Although the cost for future genetic interventions wouldlikely be higher than those reported here the relative differencesin cost between alternative intervention strategies should remainapproximately the sameOur ability to simulate genetics and link it to the viability of

the Florida panther population is based on the empirically es-timated relationship between individual heterozygosity andsurvival Such heterozygosity and fitness correlations have beenstudied for decades (David 1998) but have only recently beenused to predict population viability (Benson et al 2016) noneto our knowledge have incorporated cost into their modelingefforts The mechanisms underlying heterozygosity and fitnesscorrelations remain unclear (David 1998 Szulkin et al 2010)and their significance as a proxy for inbreeding depression hasbeen debated (Balloux et al 2004 Miller and Coltman 2014)Nevertheless evidence continues to mount demonstratingpositive relationships between heterozygosity and survival(Coulson et al 1998ab Hansson et al 2004 Silva et al 2005Acevedo‐Whitehouse et al 2006 Jensen et al 2007 Velandoet al 2015) and between heterozygosity and fecundity (Seddonet al 2004 Charpentier et al 2005 Agudo et al 2012 Velandoet al 2015) Although the reported correlations are generallyweak their consequences can be substantial if population growthand persistence parameters are highly sensitive to the affectedvital rates (Velando et al 2015) Compared with scenarios thatignore genetics incorporating the effects of inbreeding on sur-vival increased quasi‐extinction probabilities within a 100‐yeartimeframe from 15 to 13 and from 14 to 17 for theMPC and MVM scenarios respectively These results high-light the importance of incorporating genetics when analyzingthe viability of small isolated populationsAlthough conservation biologists have recognized the im-

portance of genetics in population viability many still fail toincorporate genetic factors into PVA models (Reed et al 2002)Explicit consideration of genetics in PVAs requires long‐termgenetic and demographic data This permits the assessment ofthe functional relationship between genetic variability and de-mographic parameters Population viability analysis softwarepackages such as VORTEX (Lacy 1993 2000) offer a powerfulframework for individual‐based simulations but they often donot adequately capture a speciesrsquo life history or genetics Basedon a thorough review of the importance of genetic factors inwildlife conservation Frankham et al (2014) concluded thatgenetic factors can strongly influence the dynamics and persis-tence of small andor fragmented populations They furthernoted that ldquoMost PVA‐based risk assessments ignore or in-adequately model genetic factorsrdquo and recommended that ldquoPVAshould routinely include realistic inbreeding depressionhelliprdquo(Frankham et al 201457) Our custom‐built IBM permitted usto develop a model that adequately captured the Florida panther

van de Kerk et al bull Florida Panther Population and Genetic Viability 29

life history and we could parameterize the model with our long‐term genetic and demographic dataThe explicit consideration of the effects of inbreeding on

Florida panther population dynamics and persistence representsan important step forward Nonetheless our model remainsimperfect First we neglected the possible effects of habitat lossdetrimental anthropogenic activities climate change the prob-ability of immigration of pumas from western populationsdisease or catastrophes on demographic rates and populationviability Although immigration from puma populations outsideFlorida is possible its probability is very low given the extensivechallenges the anthropogenically altered landscape would posefor such a migrant to make it successfully to South Florida Weomitted mutations from our model because we have no in-formation on mutation rates though we expect that they are lowbased on values reported for other mammals (Kumar and Sub-ramanian 2002) Finally we only considered possible effects ofinbreeding on age‐specific survival rates because there was noevidence that heterozygosity affected female reproduction Lossof heterozygosity can negatively affect reproduction becauseinbred male panthers can suffer from cryptorchidism which canadversely affect their fecundity However given the polygynousmating system we expect the Florida panther population growthis primarily female‐limitedAlthough it is generally accepted that genetic introgression

only temporarily relieves a population from inbreeding depres-sion we know of no cases in which it has been implementedmore than once to conserve a threatened wildlife populationOur study provides an example of how demographic and mo-lecular data can be used within an IBM framework to evaluatethe efficacy of alternative genetic management strategies via theFlorida panther as a case study Our model can be adjusted forand applied to other species and offers great potential as a toolfor genetic management of small isolated wildlife populations

MANAGEMENT IMPLICATIONS

Our results and those of Hostetler et al (2013) and Johnsonet al (2010) provide persuasive evidence that the genetic in-tervention implemented in 1995 prevented the demise of theFlorida panther and restored demographic vigor to the popu-lation The Florida panther population remains small and iso-lated from other puma populations the closest puma populationis located gt1000 km away in Texas and the intervening land-scape is highly developed and fragmented with many interstate(and other high‐traffic) highways and major urban centersTheory suggests that 1 migrant per generation is sufficient tominimize the loss of genetic diversity (eg Mills and Allendorf1996) However the possibility of successful migration of apuma to South Florida followed by successful mating every ~5years is very low considering the distance between Texas andFlorida (Hedrick 1995) and the many risks and barriers thatdispersing pumas face (Beier 1995 Maehr et al 2002 Thatcheret al 2006) Consequently the pantherrsquos long‐term persistencewill likely depend on subsequent timely genetic introgressionsgiven the near‐impossibility of natural gene flow from otherpuma populations To that end our study offers considerableinsights into benefits of different genetic management strategiesand associated costs

Without genetic management the Florida panther populationfaces a substantial risk of quasi‐extinction within 100 years (10ndash46 depending on quasi‐extinction threshold and scenarios)Twenty‐three years have passed since genetic introgression wasimplemented and the population appears healthy as indicatedby measures of genetic variation increases in the populationsize and improvements in age‐specific survival rates Ourfindings suggest that releasing more pumas more frequently doesnot necessarily improve persistence probability because such astrategy would cause larger fluctuations in population size(Fig 14) which negatively affects persistence probability Thusextinction risks faced by inbred populations of large carnivorescan be substantially reduced by releasing approximately 5 im-migrants every 2 decades We suggest that such a managementstrategy be followed only in conjunction with continued long‐term monitoring of the population Our analyses demonstratethat population growth and persistence are highly sensitive tokitten and female (adult and subadult) survival Continuing tocollect data on these demographic variables along with bio-metric and genetic sampling will remain important to pantherrecovery and vigilance in identifying issues associated with in-breeding depression The collection of these monitoring datashould allow managers to detect any demographic or geneticchanges in a timely fashion and respond with genetic in-trogression or other management actions if warrantedRecovery objectives as defined by the USFWS in its current

recovery plan (USFWS 2008) delineate the need for additionalpopulations in the historical range to downlist or delist theFlorida panther If the only extant population of Floridapanthers continued to grow it could serve as a source ofpanthers to be translocated to suitable habitat to seed addi-tional populations Maintaining current information on thegenetic health of the population and precise estimates of itssize will be important in assessing whether it can withstand theremoval of a subset of animals for translocation Although the2017 documentation of a female panther with kittens north ofthe current breeding range is encouraging in terms of thepotential for population expansionmdashgiven that no female hadbeen recorded north of the Caloosahatchee River since 1972mdashthe re‐establishment of separate new populations will mostlikely require translocations by wildlife managers and a lengthysociopolitical processConservation of threatened or endangered species such as the

Florida panther is inherently challenging For panthers and otherlarge carnivores that require vast extents of natural habitat topersist habitat is typically the most limiting factor that will de-termine whether recovery efforts succeed Although geneticmanagement invariably plays a key role in the long‐term persis-tence of the panther population even a healthy population caneventually succumb to the impacts of habitat loss The humanpopulation of Florida continues to be one of the fastest‐growingin the nation the United States Census Bureau currently ranksFlorida as the third most populous Therefore habitat loss isgoing to continue to be a key issue for panthers Diligence towardpreserving panther habitat in its historical range via conservationeasements or other means and trying to keep such areas inter-connected via corridors is a goal stakeholders such as governmentagencies sportsmen non‐governmental organizations ranchers

30 Wildlife Monographs bull 203

and other private landowners must work on collaboratively toachieve

ACKNOWLEDGMENTS

We thank D Shindle D Jennings T OrsquoMeara D Land andC Belden for valuable advice throughout the project We thankS Bass M Criffield M Cunningham D Giardina D JansenA Johnson D Land M Lotz C McBride Rocky McBrideRowdy McBride Roy McBride L Oberhofer S Schulze staffsat Everglades National Park and Big Cypress National Preserveand others for assistance with fieldwork We also thank JAustin M Conroy J Hines J Laake S Mills J Nichols JHines M Sunquist J Fryxell and T Coulson for advice andassistance regarding data analysis and panther biology TheMuseum of Texas Tech University Natural Science ResearchLaboratory provided samples from pumas from west Texas forcomparative genetic analyses M Ben‐David D Land WMcCown E Hellgren and 4 anonymous reviewers providedmany helpful comments on the manuscript We thank NereaUbierna Lopez for completing the Spanish translation of theabstract We are grateful to the Department of ResearchComputing University of Florida for excellent high‐perfor-mance computing services We would like to thank US Fishand Wildlife Service Florida Fish and Wildlife ConservationCommission (FWC) US National Park Service QuantitativeSpatial Ecology Evolution and Environment (QSE3) In-tegrative Graduate Education and Research Traineeship(IGERT) program funded by the National Science Foundationand the University of Florida for financially supporting thisstudy Funding for panther research conducted by the FWC issupported predominantly via donations accrued with vehicleregistrations for ldquoProtect the Pantherrdquo license plates

LITERATURE CITEDAcevedo‐Whitehouse K T R Spraker E Lyons S R Melin F Gulland RL Delong and W Amos 2006 Contrasting effects of heterozygosity onsurvival and hookworm resistance in California sea lion pups MolecularEcology 151973ndash1982

Adams J R L M Vucetich P W Hedrick R O Peterson and J AVucetich 2011 Genomic sweep and potential genetic rescue during limitingenvironmental conditions in an isolated wolf population Proceedings of theRoyal Society B Biological Sciences 2783336ndash3344

Agresti A 2013 Categorical data analysis Third edition John Wiley amp SonsHoboken New Jersey USA

Agudo R M Carrete M Alcaide C Rico F Hiraldo and J A Donaacutezar2012 Genetic diversity at neutral and adaptive loci determines individualfitness in a long‐lived territorial bird Proceedings of the Royal Society BBiological Sciences 2793241ndash3249

Albeke S E N P Nibbelink and M Ben‐David 2015 Modeling behavior bycoastal river otter (Lontra canadensis) in response to prey availability in PrinceWilliam Sound Alaska a spatially‐explicit individual‐based approach PLOSOne 10e0126208

Alho J S K Vaumllimaumlki and J Merilauml 2010 Rhh an R extension for estimatingmultilocus heterozygosity and heterozygosity‐heterozygosity correlationMolecular Ecology Resources 10720ndash722

Alvarez K 1993 Twilight of the panther Myakka River Publishing SarasotaFlorida USA

Aparicio J M J Ortego and P J Cordero 2006 What should we weigh toestimate heterozygosity alleles or loci Molecular Ecology 154659ndash4665

Bakker V J D F Doak G W Roemer D K Garcelon T J Coonan S AMorrison C Lynch K Ralls and R Shaw 2009 Incorporating ecologicaldrivers and uncertainty into a demographic population viability analysis for theisland fox Ecological Monographs 7977ndash108

Balloux F W Amos and T Coulson 2004 Does heterozygosity estimateinbreeding in real populations Molecular Ecology 133021ndash3031

Barone M A M E Roelke J Howard J L Brown A E Anderson and DE Wildt 1994 Reproductive characteristics of male Florida panthersmdashcomparative studies from Florida Texas Colorado Latin‐America andNorth‐American Zoos Journal of Mammalogy 75150ndash162

Bateson Z W P O Dunn S D Hull A E Henschen J A Johnson and LA Whittingham 2014 Genetic restoration of a threatened population ofgreater prairie‐chickens Biological Conservation 17412ndash19

Bauer H G Chapron K Nowell P Henschel P Funston L T B HunterD W Macdonald and C Packer 2015 Lion (Panthera leo) populations aredeclining rapidly across Africa except in intensively managed areas Pro-ceedings of National Academy of Sciences of the United States of America11214894ndash14899

Beier P 1995 Dispersal of juvenile cougars in fragmented habitat Journal ofWildlife Management 59228ndash237

Benson J F J A Hostetler D P Onorato W E Johnson M E Roelke SJ OrsquoBrien D Jansen and M K Oli 2011 Intentional genetic introgressioninfluences survival of adults and subadults in a small inbred felid populationJournal of Animal Ecology 80958ndash967

Benson J F P J Mahoney J A Sikich L E K Serieys J P Pollinger H BErnest and S P D Riley 2016 Interactions between demography geneticsand landscape connectivity increase extinction probability for a small popu-lation of large carnivores in a major metropolitan area Proceedings of theRoyal Society B Biological Sciences 28320160957

Beverly F 1874 The Florida panther Forest and Stream 3290Boyce M S C V Haridas C T Lee and the NCEAS Stochastic Demo-graphy Working Group 2006 Demography in an increasingly variable worldTrends in Ecology amp Evolution 21141ndash148

Brook B W J J OrsquoGrady A P Chapman M A Burgman H R Akcakayaand R Frankham 2000 Predictive accuracy of population viability analysis inconservation biology Nature 404385ndash387

Brys R and H Jacquemyn 2016 Severe outbreeding and inbreeding de-pression maintain mating system differentiation in Epipactis (Orchidaceae)Journal of Evolutionary Biology 29352ndash359

Burke J M and M L Arnold 2001 Genetics and the fitness of hybridsAnnual Review of Genetics 3531ndash52

Burnham K P 1993 A theory for combined analysis of ring recovery andrecapture data Pages 199ndash213 in J‐D Lebreton and P M North editorsMarked individuals in the study of bird population Springer‐Verlag NewYork New York USA

Burnham K P and D R Anderson 2002 Model selection and multimodelinference a practical information‐theoretic approach Second edition SpringerVerlag New York New York USA

Caswell H 2001 Matrix population models construction analysis and in-terpretation Sinauer Associates Sunderland Massachusetts USA

Charpentier M J M Setchell F Prugnolle L A Knapp E J Wickings PPeignot and M Hossaert‐McKey 2005 Genetic diversity and reproductivesuccess in mandrills (Mandrillus sphinx) Proceedings of the NationalAcademy of Sciences of the United States of America 10216723ndash16728

Cooch E and G C White editors 2018 Program MARK a gentle in-troduction lthttpwwwphidotorgsoftwaremarkdocsbookgt Accessed 7Apr 2016

Cooley H S R B Wielgus G M Koehler H S Robinson and B TMaletzke 2009 Does hunting regulate cougar populations A test of thecompensatory mortality hypothesis Ecology 902913ndash2921

Coulson T E A Catchpole S D Albon B J Morgan J M Pemberton TH Clutton‐Brock M J Crawley and B T Grenfell 2001 Age sex densitywinter weather and population crashes in Soay sheep Science 2921528ndash1531

Coulson T J Pemberton S Albon M Beaumont T Marshall J Slate FGuinness and T Clutton‐Brock 1998a Microsatellites reveal heterosis in reddeer Proceedings of the Royal Society B Biological Sciences 265489ndash495

Coulson T N S D Albon J M Pemberton J Slate F E Guinness and TH Clutton‐Brock 1998b Genotype by environment interactions in wintersurvival in red deer Journal of Animal Ecology 67434ndash445

Cox D 1972 Regression models and life‐tables Journal of the Royal StatisticalSociety Series B Statistical Methodology 34187ndash220

Creel S 2006 Recovery of the Florida panthermdashgenetic rescue demographic rescueor both Response to Pimm et al (2006) Animal Conservation 9125ndash126

Crnokrak P and D A Roff 1999 Inbreeding depression in the wild Heredity83260ndash270

Crow J 1948 Alternative hypotheses of hybrid vigor Genetics 33477ndash487

van de Kerk et al bull Florida Panther Population and Genetic Viability 31

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 26: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

wane after 2 or 3 generations The WrightndashFisher model ofgenetic drift predicts a geometric decline in expected hetero-zygosity at the rate of (1 minus 12Ne) per generation where Ne is theeffective population size (Hartl and Clark 1997 Frankham et al2014) Thus it seems logical to assume that the duration of thebenefits of introgression is limited especially in a species per-sisting in a small population such as Florida panthersWe expected Florida panther survival in the most recent years

(2005ndash2013) post‐introgression to differ from that observed in thedecade following the introgression in 1995 because pantherabundance and heterozygosity factors known to influence panthersurvival (Hostetler et al 2010 Benson et al 2011) have changed(Figs 2 and 6) We found that subadult and adult survival rates inrecent years (2005ndash2013) were similar to those in the period im-mediately following introgression (1995ndash2004 Fig 5C) overallkitten survival was similar to estimates of Hostetler et al (2010) Infact the pattern of covariate effects on Florida panther survivalreported by Benson et al (2011) and Hostetler et al (2010) hasbarely changed The negative effects of population density andpositive effects of heterozygosity may counterbalance survival ratesreducing population fluctuations Our result that benefits of geneticrestoration have remained in the population nearly for 5 genera-tions post‐intervention was surprising but also encouraging Pos-sible explanations for this observation may include 1) genetic ero-sion occurs at slower rate than previously thought 2) introducing 5migrants in 1 genetic intervention may have beneficial effects si-milar to those from 1 migrant per generation over multiple gen-erations or 3) other and as yet unknown factors may reduce therate of genetic erosion and subsequent demographic effects Adensity‐dependent effect on kitten survival could also explain whynon‐F1 admixed kittens did not survive substantially better thandid canonical kittens most kittens sampled from 2005ndash2013 wereadmixed and their survival was lower possibly because of a density‐dependent effect as the Florida panther population continuouslygrew during that period Trinkel et al (2010) also found thatpopulation density and inbreeding coefficient interacted to affectdemographic parameters and growth rate of an African lion po-pulation Likewise population density interacted with winterweather to affect the survival and population growth rates in Soaysheep (Ovis aries Milner et al 1999 Coulson et al 2001)

Population Dynamics and PersistenceThe finite deterministic and stochastic growth rates were gt10reflecting that much of the data used in this study were collectedfollowing genetic introgression in 1995 during which time thepopulation increased substantially Because our demographicparameter estimates did not change markedly in comparison tothose of Hostetler et al (2013) the similarity between thepopulation growth estimates from these studies was expectedBut these estimates reflect population growth during thedata‐collection period and do not predict future growth In factestimates of population size reported by McClintock et al(2015) suggest that population growth may already be slowing(Fig 2) A similar pattern was observed in an introduced po-pulation of African lion which increased steadily from 1995until 2001 then fluctuated around the presumed carrying ca-pacity thereafter (Trinkel et al 2010) Several other African lionpopulations that experienced various degrees of recovery

subsequently became relatively stable or exhibited decliningtrends (Bauer et al 2015)Our estimates of quasi‐extinction probabilities were lower than

those reported by Hostetler et al (2013) using a 2‐sex age‐structured matrix population model Lower probabilities ofquasi‐extinction than those reported by the earlier study forcomparable scenarios were likely a consequence of the fact thatthe estimates of abundance (McBride et al 2008) and thus thedensity‐dependent effects on survival of kittens and old panthers(gt10 yr) have changed in recent years Additionally these earlyestimates of the probability of quasi‐extinction and of time toquasi‐extinction did not consider genetic erosion that will in-evitably occur without management interventionUnder the MVM scenario the equilibrium population size was

twice that attained under the MPC scenario The MVM popula-tion estimates are approximately twice the MPCs (Fig 2) as aresult the simulated population under the MVM scenario achieveda higher equilibrium size Probability of quasi‐extinction under theMVM scenario was generally greater than that under the MPCscenario This result is a consequence of the fact that the estimateddensity dependence on kitten survival based on the MPC scenario isstronger than that for theMVM scenario (regression coefficients forabundance for the MPC scenario= minus0068 vs minus0002 for theMVM scenario Appendix B Table B1) Consequently simulatedpopulations under the MPC scenario have a stronger tendency tomove toward the equilibrium population size which inherentlyreduces the quasi‐extinction probability (eg Royama 1992)As expected for long‐lived mammals (Heppell et al 2000 Oli

and Dobson 2003) the population growth rate was proportio-nately most sensitive to changes in survival of prime adultsfollowed by that in younger age classes Global sensitivity ana-lysis in contrast showed that under all scenarios extinctionparameters were particularly sensitive to changes in survival ofFlorida panther kittens This result is a consequence of the highsensitivity of the population growth rate to kitten survival(Fig 9) and a larger standard error of kitten survival (Table 3)Results of our sensitivity analyses indicate that future researchshould focus on acquiring additional information on early lifestages to improve the accuracy and precision of estimates ofkitten survival Our results suggest that demographic variables(or their environmental drivers) that strongly influence extinc-tion risks are not necessarily those with the largest elasticityvalues A study of island fox (Bakker et al 2009) found thatextinction risk was strongly influenced by survival modifierparameters that had low elasticity values

Genetic Management When and How to InterveneThe use of genetic introgression as a management tool has al-ways been controversial and the probable effectiveness it mighthave in preventing the imminent demise of the panther popu-lation was initially debated (Creel 2006 Maehr et al 2006Pimm et al 2006) These debates notwithstanding the pantherpopulation had suffered from genetic morphological and bio-medical correlates of inbreeding before this management in-tervention (Johnson et al 2010 Onorato et al 2010) whereasthe post‐introgression period has been characterized by a sub-stantial reduction in the biomedical correlates of inbreeding andimprovements in demographic vigor and abundance (McBride

28 Wildlife Monographs bull 203

et al 2008 Hostetler et al 2010 2013 Johnson et al 2010Benson et al 2011 McClintock et al 2015) Nevertheless thepopulation remains small and isolated habitat is still being lostand there is no current possibility of natural gene flow betweenthis and other North American puma populations thus therecurrence of inbreeding‐related problems and subsequent po-pulation declines are inevitable The question therefore is notwhether genetic management of the Florida panther populationis needed but when and how it should be implementedWithout genetic introgression probabilities of quasi‐extinc-

tion were substantially greater than those from models thatincorporated periodic genetic introgression events (Fig 15)highlighting the importance of genetic management in smallisolated populations When 5 female pumas are added to theFlorida panther population every 10 years genetic variationincreased substantially under the MPC and MVM scenariosThe IBM predicted that the equillibrium population size wouldbe substantially larger when 5 pumas were released into thepopulation every 10 years than populations without intervention(ie no introgression) Releasing 15 Texas females did not af-ford substantially greater benefit to the equilibrium populationsize than did releasing only 5 Texas females Instead addingmore individuals caused increasingly large fluctuations in po-pulation size due to the numerical increases and density de-pendence (Fig 14) possibly diminishing the benefits associatedwith increased genetic variationCompared with the no‐intervention scenario (ie no genetic

introgression implemented) the introduction of 5 female pumasevery 10 years reduced quasi‐extinction probabilities from 13to 4 and from 17 to 6 for the MPC and MVM scenariosrespectively The benefit of genetic‐introgression strategies de-creased as the interval between successive management inter-ventions increased and no benefit was evident once the intervalreached 80 years (Fig 15) Introgression reduced the averagetime until quasi‐extinction (Fig 18) This somewhat counter‐intuitive result can be explained by the fact that repeated geneticintrogression makes the population more robust over timewhich will reduce the probability of extinction Consequentlyfew simulated population trajectories that fall below the criticalthreshold do so early reducing the mean time to extinctionAlso density dependence has a stabilizing effect on populations(Royama 1992) populations tend toward equilibrium popula-tion sizes which reduces the probability over time of popula-tions falling below quasi‐extinction threshold However timeuntil quasi‐extinction must be interpreted in conjunction withthe probability of quasi‐extinction which is very low for allscenarios with genetic introgression (Fig 15)Cost is a significant impediment to the implementation of a

genetic introgression program because captures relocations andmonitoring efforts before and after introgression are expensive(Edmands 2007 Frankham 2010b 2015 2016) Costs may beacceptable to society if the management actions result in sub-stantial fitness benefits especially if the species of concern ispopular and charismatic (Frankham 2015) We estimated the100‐year cost of introgression strategies modeled here to rangefrom $50000 to $12 million More expensive strategies gen-erally led to larger decreases in the probability of quasi‐extinc-tion but cheaper strategies also substantially improved

population viability (ie reduced quasi‐extinction probabilityTable 6) One of the cheaper strategies (5 pumas released every20 years) which cost an estimated $200000 over 100 yearsreduced the probability of quasi‐extinction by 59 and 44 forthe MPC and MVM scenarios respectively But these pre-liminary cost estimates are based on expenses incurred duringthe 1995 introgression and include costs of capture quarantinetransportation release and post‐release monitoring of femalesonly Although the cost for future genetic interventions wouldlikely be higher than those reported here the relative differencesin cost between alternative intervention strategies should remainapproximately the sameOur ability to simulate genetics and link it to the viability of

the Florida panther population is based on the empirically es-timated relationship between individual heterozygosity andsurvival Such heterozygosity and fitness correlations have beenstudied for decades (David 1998) but have only recently beenused to predict population viability (Benson et al 2016) noneto our knowledge have incorporated cost into their modelingefforts The mechanisms underlying heterozygosity and fitnesscorrelations remain unclear (David 1998 Szulkin et al 2010)and their significance as a proxy for inbreeding depression hasbeen debated (Balloux et al 2004 Miller and Coltman 2014)Nevertheless evidence continues to mount demonstratingpositive relationships between heterozygosity and survival(Coulson et al 1998ab Hansson et al 2004 Silva et al 2005Acevedo‐Whitehouse et al 2006 Jensen et al 2007 Velandoet al 2015) and between heterozygosity and fecundity (Seddonet al 2004 Charpentier et al 2005 Agudo et al 2012 Velandoet al 2015) Although the reported correlations are generallyweak their consequences can be substantial if population growthand persistence parameters are highly sensitive to the affectedvital rates (Velando et al 2015) Compared with scenarios thatignore genetics incorporating the effects of inbreeding on sur-vival increased quasi‐extinction probabilities within a 100‐yeartimeframe from 15 to 13 and from 14 to 17 for theMPC and MVM scenarios respectively These results high-light the importance of incorporating genetics when analyzingthe viability of small isolated populationsAlthough conservation biologists have recognized the im-

portance of genetics in population viability many still fail toincorporate genetic factors into PVA models (Reed et al 2002)Explicit consideration of genetics in PVAs requires long‐termgenetic and demographic data This permits the assessment ofthe functional relationship between genetic variability and de-mographic parameters Population viability analysis softwarepackages such as VORTEX (Lacy 1993 2000) offer a powerfulframework for individual‐based simulations but they often donot adequately capture a speciesrsquo life history or genetics Basedon a thorough review of the importance of genetic factors inwildlife conservation Frankham et al (2014) concluded thatgenetic factors can strongly influence the dynamics and persis-tence of small andor fragmented populations They furthernoted that ldquoMost PVA‐based risk assessments ignore or in-adequately model genetic factorsrdquo and recommended that ldquoPVAshould routinely include realistic inbreeding depressionhelliprdquo(Frankham et al 201457) Our custom‐built IBM permitted usto develop a model that adequately captured the Florida panther

van de Kerk et al bull Florida Panther Population and Genetic Viability 29

life history and we could parameterize the model with our long‐term genetic and demographic dataThe explicit consideration of the effects of inbreeding on

Florida panther population dynamics and persistence representsan important step forward Nonetheless our model remainsimperfect First we neglected the possible effects of habitat lossdetrimental anthropogenic activities climate change the prob-ability of immigration of pumas from western populationsdisease or catastrophes on demographic rates and populationviability Although immigration from puma populations outsideFlorida is possible its probability is very low given the extensivechallenges the anthropogenically altered landscape would posefor such a migrant to make it successfully to South Florida Weomitted mutations from our model because we have no in-formation on mutation rates though we expect that they are lowbased on values reported for other mammals (Kumar and Sub-ramanian 2002) Finally we only considered possible effects ofinbreeding on age‐specific survival rates because there was noevidence that heterozygosity affected female reproduction Lossof heterozygosity can negatively affect reproduction becauseinbred male panthers can suffer from cryptorchidism which canadversely affect their fecundity However given the polygynousmating system we expect the Florida panther population growthis primarily female‐limitedAlthough it is generally accepted that genetic introgression

only temporarily relieves a population from inbreeding depres-sion we know of no cases in which it has been implementedmore than once to conserve a threatened wildlife populationOur study provides an example of how demographic and mo-lecular data can be used within an IBM framework to evaluatethe efficacy of alternative genetic management strategies via theFlorida panther as a case study Our model can be adjusted forand applied to other species and offers great potential as a toolfor genetic management of small isolated wildlife populations

MANAGEMENT IMPLICATIONS

Our results and those of Hostetler et al (2013) and Johnsonet al (2010) provide persuasive evidence that the genetic in-tervention implemented in 1995 prevented the demise of theFlorida panther and restored demographic vigor to the popu-lation The Florida panther population remains small and iso-lated from other puma populations the closest puma populationis located gt1000 km away in Texas and the intervening land-scape is highly developed and fragmented with many interstate(and other high‐traffic) highways and major urban centersTheory suggests that 1 migrant per generation is sufficient tominimize the loss of genetic diversity (eg Mills and Allendorf1996) However the possibility of successful migration of apuma to South Florida followed by successful mating every ~5years is very low considering the distance between Texas andFlorida (Hedrick 1995) and the many risks and barriers thatdispersing pumas face (Beier 1995 Maehr et al 2002 Thatcheret al 2006) Consequently the pantherrsquos long‐term persistencewill likely depend on subsequent timely genetic introgressionsgiven the near‐impossibility of natural gene flow from otherpuma populations To that end our study offers considerableinsights into benefits of different genetic management strategiesand associated costs

Without genetic management the Florida panther populationfaces a substantial risk of quasi‐extinction within 100 years (10ndash46 depending on quasi‐extinction threshold and scenarios)Twenty‐three years have passed since genetic introgression wasimplemented and the population appears healthy as indicatedby measures of genetic variation increases in the populationsize and improvements in age‐specific survival rates Ourfindings suggest that releasing more pumas more frequently doesnot necessarily improve persistence probability because such astrategy would cause larger fluctuations in population size(Fig 14) which negatively affects persistence probability Thusextinction risks faced by inbred populations of large carnivorescan be substantially reduced by releasing approximately 5 im-migrants every 2 decades We suggest that such a managementstrategy be followed only in conjunction with continued long‐term monitoring of the population Our analyses demonstratethat population growth and persistence are highly sensitive tokitten and female (adult and subadult) survival Continuing tocollect data on these demographic variables along with bio-metric and genetic sampling will remain important to pantherrecovery and vigilance in identifying issues associated with in-breeding depression The collection of these monitoring datashould allow managers to detect any demographic or geneticchanges in a timely fashion and respond with genetic in-trogression or other management actions if warrantedRecovery objectives as defined by the USFWS in its current

recovery plan (USFWS 2008) delineate the need for additionalpopulations in the historical range to downlist or delist theFlorida panther If the only extant population of Floridapanthers continued to grow it could serve as a source ofpanthers to be translocated to suitable habitat to seed addi-tional populations Maintaining current information on thegenetic health of the population and precise estimates of itssize will be important in assessing whether it can withstand theremoval of a subset of animals for translocation Although the2017 documentation of a female panther with kittens north ofthe current breeding range is encouraging in terms of thepotential for population expansionmdashgiven that no female hadbeen recorded north of the Caloosahatchee River since 1972mdashthe re‐establishment of separate new populations will mostlikely require translocations by wildlife managers and a lengthysociopolitical processConservation of threatened or endangered species such as the

Florida panther is inherently challenging For panthers and otherlarge carnivores that require vast extents of natural habitat topersist habitat is typically the most limiting factor that will de-termine whether recovery efforts succeed Although geneticmanagement invariably plays a key role in the long‐term persis-tence of the panther population even a healthy population caneventually succumb to the impacts of habitat loss The humanpopulation of Florida continues to be one of the fastest‐growingin the nation the United States Census Bureau currently ranksFlorida as the third most populous Therefore habitat loss isgoing to continue to be a key issue for panthers Diligence towardpreserving panther habitat in its historical range via conservationeasements or other means and trying to keep such areas inter-connected via corridors is a goal stakeholders such as governmentagencies sportsmen non‐governmental organizations ranchers

30 Wildlife Monographs bull 203

and other private landowners must work on collaboratively toachieve

ACKNOWLEDGMENTS

We thank D Shindle D Jennings T OrsquoMeara D Land andC Belden for valuable advice throughout the project We thankS Bass M Criffield M Cunningham D Giardina D JansenA Johnson D Land M Lotz C McBride Rocky McBrideRowdy McBride Roy McBride L Oberhofer S Schulze staffsat Everglades National Park and Big Cypress National Preserveand others for assistance with fieldwork We also thank JAustin M Conroy J Hines J Laake S Mills J Nichols JHines M Sunquist J Fryxell and T Coulson for advice andassistance regarding data analysis and panther biology TheMuseum of Texas Tech University Natural Science ResearchLaboratory provided samples from pumas from west Texas forcomparative genetic analyses M Ben‐David D Land WMcCown E Hellgren and 4 anonymous reviewers providedmany helpful comments on the manuscript We thank NereaUbierna Lopez for completing the Spanish translation of theabstract We are grateful to the Department of ResearchComputing University of Florida for excellent high‐perfor-mance computing services We would like to thank US Fishand Wildlife Service Florida Fish and Wildlife ConservationCommission (FWC) US National Park Service QuantitativeSpatial Ecology Evolution and Environment (QSE3) In-tegrative Graduate Education and Research Traineeship(IGERT) program funded by the National Science Foundationand the University of Florida for financially supporting thisstudy Funding for panther research conducted by the FWC issupported predominantly via donations accrued with vehicleregistrations for ldquoProtect the Pantherrdquo license plates

LITERATURE CITEDAcevedo‐Whitehouse K T R Spraker E Lyons S R Melin F Gulland RL Delong and W Amos 2006 Contrasting effects of heterozygosity onsurvival and hookworm resistance in California sea lion pups MolecularEcology 151973ndash1982

Adams J R L M Vucetich P W Hedrick R O Peterson and J AVucetich 2011 Genomic sweep and potential genetic rescue during limitingenvironmental conditions in an isolated wolf population Proceedings of theRoyal Society B Biological Sciences 2783336ndash3344

Agresti A 2013 Categorical data analysis Third edition John Wiley amp SonsHoboken New Jersey USA

Agudo R M Carrete M Alcaide C Rico F Hiraldo and J A Donaacutezar2012 Genetic diversity at neutral and adaptive loci determines individualfitness in a long‐lived territorial bird Proceedings of the Royal Society BBiological Sciences 2793241ndash3249

Albeke S E N P Nibbelink and M Ben‐David 2015 Modeling behavior bycoastal river otter (Lontra canadensis) in response to prey availability in PrinceWilliam Sound Alaska a spatially‐explicit individual‐based approach PLOSOne 10e0126208

Alho J S K Vaumllimaumlki and J Merilauml 2010 Rhh an R extension for estimatingmultilocus heterozygosity and heterozygosity‐heterozygosity correlationMolecular Ecology Resources 10720ndash722

Alvarez K 1993 Twilight of the panther Myakka River Publishing SarasotaFlorida USA

Aparicio J M J Ortego and P J Cordero 2006 What should we weigh toestimate heterozygosity alleles or loci Molecular Ecology 154659ndash4665

Bakker V J D F Doak G W Roemer D K Garcelon T J Coonan S AMorrison C Lynch K Ralls and R Shaw 2009 Incorporating ecologicaldrivers and uncertainty into a demographic population viability analysis for theisland fox Ecological Monographs 7977ndash108

Balloux F W Amos and T Coulson 2004 Does heterozygosity estimateinbreeding in real populations Molecular Ecology 133021ndash3031

Barone M A M E Roelke J Howard J L Brown A E Anderson and DE Wildt 1994 Reproductive characteristics of male Florida panthersmdashcomparative studies from Florida Texas Colorado Latin‐America andNorth‐American Zoos Journal of Mammalogy 75150ndash162

Bateson Z W P O Dunn S D Hull A E Henschen J A Johnson and LA Whittingham 2014 Genetic restoration of a threatened population ofgreater prairie‐chickens Biological Conservation 17412ndash19

Bauer H G Chapron K Nowell P Henschel P Funston L T B HunterD W Macdonald and C Packer 2015 Lion (Panthera leo) populations aredeclining rapidly across Africa except in intensively managed areas Pro-ceedings of National Academy of Sciences of the United States of America11214894ndash14899

Beier P 1995 Dispersal of juvenile cougars in fragmented habitat Journal ofWildlife Management 59228ndash237

Benson J F J A Hostetler D P Onorato W E Johnson M E Roelke SJ OrsquoBrien D Jansen and M K Oli 2011 Intentional genetic introgressioninfluences survival of adults and subadults in a small inbred felid populationJournal of Animal Ecology 80958ndash967

Benson J F P J Mahoney J A Sikich L E K Serieys J P Pollinger H BErnest and S P D Riley 2016 Interactions between demography geneticsand landscape connectivity increase extinction probability for a small popu-lation of large carnivores in a major metropolitan area Proceedings of theRoyal Society B Biological Sciences 28320160957

Beverly F 1874 The Florida panther Forest and Stream 3290Boyce M S C V Haridas C T Lee and the NCEAS Stochastic Demo-graphy Working Group 2006 Demography in an increasingly variable worldTrends in Ecology amp Evolution 21141ndash148

Brook B W J J OrsquoGrady A P Chapman M A Burgman H R Akcakayaand R Frankham 2000 Predictive accuracy of population viability analysis inconservation biology Nature 404385ndash387

Brys R and H Jacquemyn 2016 Severe outbreeding and inbreeding de-pression maintain mating system differentiation in Epipactis (Orchidaceae)Journal of Evolutionary Biology 29352ndash359

Burke J M and M L Arnold 2001 Genetics and the fitness of hybridsAnnual Review of Genetics 3531ndash52

Burnham K P 1993 A theory for combined analysis of ring recovery andrecapture data Pages 199ndash213 in J‐D Lebreton and P M North editorsMarked individuals in the study of bird population Springer‐Verlag NewYork New York USA

Burnham K P and D R Anderson 2002 Model selection and multimodelinference a practical information‐theoretic approach Second edition SpringerVerlag New York New York USA

Caswell H 2001 Matrix population models construction analysis and in-terpretation Sinauer Associates Sunderland Massachusetts USA

Charpentier M J M Setchell F Prugnolle L A Knapp E J Wickings PPeignot and M Hossaert‐McKey 2005 Genetic diversity and reproductivesuccess in mandrills (Mandrillus sphinx) Proceedings of the NationalAcademy of Sciences of the United States of America 10216723ndash16728

Cooch E and G C White editors 2018 Program MARK a gentle in-troduction lthttpwwwphidotorgsoftwaremarkdocsbookgt Accessed 7Apr 2016

Cooley H S R B Wielgus G M Koehler H S Robinson and B TMaletzke 2009 Does hunting regulate cougar populations A test of thecompensatory mortality hypothesis Ecology 902913ndash2921

Coulson T E A Catchpole S D Albon B J Morgan J M Pemberton TH Clutton‐Brock M J Crawley and B T Grenfell 2001 Age sex densitywinter weather and population crashes in Soay sheep Science 2921528ndash1531

Coulson T J Pemberton S Albon M Beaumont T Marshall J Slate FGuinness and T Clutton‐Brock 1998a Microsatellites reveal heterosis in reddeer Proceedings of the Royal Society B Biological Sciences 265489ndash495

Coulson T N S D Albon J M Pemberton J Slate F E Guinness and TH Clutton‐Brock 1998b Genotype by environment interactions in wintersurvival in red deer Journal of Animal Ecology 67434ndash445

Cox D 1972 Regression models and life‐tables Journal of the Royal StatisticalSociety Series B Statistical Methodology 34187ndash220

Creel S 2006 Recovery of the Florida panthermdashgenetic rescue demographic rescueor both Response to Pimm et al (2006) Animal Conservation 9125ndash126

Crnokrak P and D A Roff 1999 Inbreeding depression in the wild Heredity83260ndash270

Crow J 1948 Alternative hypotheses of hybrid vigor Genetics 33477ndash487

van de Kerk et al bull Florida Panther Population and Genetic Viability 31

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 27: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

et al 2008 Hostetler et al 2010 2013 Johnson et al 2010Benson et al 2011 McClintock et al 2015) Nevertheless thepopulation remains small and isolated habitat is still being lostand there is no current possibility of natural gene flow betweenthis and other North American puma populations thus therecurrence of inbreeding‐related problems and subsequent po-pulation declines are inevitable The question therefore is notwhether genetic management of the Florida panther populationis needed but when and how it should be implementedWithout genetic introgression probabilities of quasi‐extinc-

tion were substantially greater than those from models thatincorporated periodic genetic introgression events (Fig 15)highlighting the importance of genetic management in smallisolated populations When 5 female pumas are added to theFlorida panther population every 10 years genetic variationincreased substantially under the MPC and MVM scenariosThe IBM predicted that the equillibrium population size wouldbe substantially larger when 5 pumas were released into thepopulation every 10 years than populations without intervention(ie no introgression) Releasing 15 Texas females did not af-ford substantially greater benefit to the equilibrium populationsize than did releasing only 5 Texas females Instead addingmore individuals caused increasingly large fluctuations in po-pulation size due to the numerical increases and density de-pendence (Fig 14) possibly diminishing the benefits associatedwith increased genetic variationCompared with the no‐intervention scenario (ie no genetic

introgression implemented) the introduction of 5 female pumasevery 10 years reduced quasi‐extinction probabilities from 13to 4 and from 17 to 6 for the MPC and MVM scenariosrespectively The benefit of genetic‐introgression strategies de-creased as the interval between successive management inter-ventions increased and no benefit was evident once the intervalreached 80 years (Fig 15) Introgression reduced the averagetime until quasi‐extinction (Fig 18) This somewhat counter‐intuitive result can be explained by the fact that repeated geneticintrogression makes the population more robust over timewhich will reduce the probability of extinction Consequentlyfew simulated population trajectories that fall below the criticalthreshold do so early reducing the mean time to extinctionAlso density dependence has a stabilizing effect on populations(Royama 1992) populations tend toward equilibrium popula-tion sizes which reduces the probability over time of popula-tions falling below quasi‐extinction threshold However timeuntil quasi‐extinction must be interpreted in conjunction withthe probability of quasi‐extinction which is very low for allscenarios with genetic introgression (Fig 15)Cost is a significant impediment to the implementation of a

genetic introgression program because captures relocations andmonitoring efforts before and after introgression are expensive(Edmands 2007 Frankham 2010b 2015 2016) Costs may beacceptable to society if the management actions result in sub-stantial fitness benefits especially if the species of concern ispopular and charismatic (Frankham 2015) We estimated the100‐year cost of introgression strategies modeled here to rangefrom $50000 to $12 million More expensive strategies gen-erally led to larger decreases in the probability of quasi‐extinc-tion but cheaper strategies also substantially improved

population viability (ie reduced quasi‐extinction probabilityTable 6) One of the cheaper strategies (5 pumas released every20 years) which cost an estimated $200000 over 100 yearsreduced the probability of quasi‐extinction by 59 and 44 forthe MPC and MVM scenarios respectively But these pre-liminary cost estimates are based on expenses incurred duringthe 1995 introgression and include costs of capture quarantinetransportation release and post‐release monitoring of femalesonly Although the cost for future genetic interventions wouldlikely be higher than those reported here the relative differencesin cost between alternative intervention strategies should remainapproximately the sameOur ability to simulate genetics and link it to the viability of

the Florida panther population is based on the empirically es-timated relationship between individual heterozygosity andsurvival Such heterozygosity and fitness correlations have beenstudied for decades (David 1998) but have only recently beenused to predict population viability (Benson et al 2016) noneto our knowledge have incorporated cost into their modelingefforts The mechanisms underlying heterozygosity and fitnesscorrelations remain unclear (David 1998 Szulkin et al 2010)and their significance as a proxy for inbreeding depression hasbeen debated (Balloux et al 2004 Miller and Coltman 2014)Nevertheless evidence continues to mount demonstratingpositive relationships between heterozygosity and survival(Coulson et al 1998ab Hansson et al 2004 Silva et al 2005Acevedo‐Whitehouse et al 2006 Jensen et al 2007 Velandoet al 2015) and between heterozygosity and fecundity (Seddonet al 2004 Charpentier et al 2005 Agudo et al 2012 Velandoet al 2015) Although the reported correlations are generallyweak their consequences can be substantial if population growthand persistence parameters are highly sensitive to the affectedvital rates (Velando et al 2015) Compared with scenarios thatignore genetics incorporating the effects of inbreeding on sur-vival increased quasi‐extinction probabilities within a 100‐yeartimeframe from 15 to 13 and from 14 to 17 for theMPC and MVM scenarios respectively These results high-light the importance of incorporating genetics when analyzingthe viability of small isolated populationsAlthough conservation biologists have recognized the im-

portance of genetics in population viability many still fail toincorporate genetic factors into PVA models (Reed et al 2002)Explicit consideration of genetics in PVAs requires long‐termgenetic and demographic data This permits the assessment ofthe functional relationship between genetic variability and de-mographic parameters Population viability analysis softwarepackages such as VORTEX (Lacy 1993 2000) offer a powerfulframework for individual‐based simulations but they often donot adequately capture a speciesrsquo life history or genetics Basedon a thorough review of the importance of genetic factors inwildlife conservation Frankham et al (2014) concluded thatgenetic factors can strongly influence the dynamics and persis-tence of small andor fragmented populations They furthernoted that ldquoMost PVA‐based risk assessments ignore or in-adequately model genetic factorsrdquo and recommended that ldquoPVAshould routinely include realistic inbreeding depressionhelliprdquo(Frankham et al 201457) Our custom‐built IBM permitted usto develop a model that adequately captured the Florida panther

van de Kerk et al bull Florida Panther Population and Genetic Viability 29

life history and we could parameterize the model with our long‐term genetic and demographic dataThe explicit consideration of the effects of inbreeding on

Florida panther population dynamics and persistence representsan important step forward Nonetheless our model remainsimperfect First we neglected the possible effects of habitat lossdetrimental anthropogenic activities climate change the prob-ability of immigration of pumas from western populationsdisease or catastrophes on demographic rates and populationviability Although immigration from puma populations outsideFlorida is possible its probability is very low given the extensivechallenges the anthropogenically altered landscape would posefor such a migrant to make it successfully to South Florida Weomitted mutations from our model because we have no in-formation on mutation rates though we expect that they are lowbased on values reported for other mammals (Kumar and Sub-ramanian 2002) Finally we only considered possible effects ofinbreeding on age‐specific survival rates because there was noevidence that heterozygosity affected female reproduction Lossof heterozygosity can negatively affect reproduction becauseinbred male panthers can suffer from cryptorchidism which canadversely affect their fecundity However given the polygynousmating system we expect the Florida panther population growthis primarily female‐limitedAlthough it is generally accepted that genetic introgression

only temporarily relieves a population from inbreeding depres-sion we know of no cases in which it has been implementedmore than once to conserve a threatened wildlife populationOur study provides an example of how demographic and mo-lecular data can be used within an IBM framework to evaluatethe efficacy of alternative genetic management strategies via theFlorida panther as a case study Our model can be adjusted forand applied to other species and offers great potential as a toolfor genetic management of small isolated wildlife populations

MANAGEMENT IMPLICATIONS

Our results and those of Hostetler et al (2013) and Johnsonet al (2010) provide persuasive evidence that the genetic in-tervention implemented in 1995 prevented the demise of theFlorida panther and restored demographic vigor to the popu-lation The Florida panther population remains small and iso-lated from other puma populations the closest puma populationis located gt1000 km away in Texas and the intervening land-scape is highly developed and fragmented with many interstate(and other high‐traffic) highways and major urban centersTheory suggests that 1 migrant per generation is sufficient tominimize the loss of genetic diversity (eg Mills and Allendorf1996) However the possibility of successful migration of apuma to South Florida followed by successful mating every ~5years is very low considering the distance between Texas andFlorida (Hedrick 1995) and the many risks and barriers thatdispersing pumas face (Beier 1995 Maehr et al 2002 Thatcheret al 2006) Consequently the pantherrsquos long‐term persistencewill likely depend on subsequent timely genetic introgressionsgiven the near‐impossibility of natural gene flow from otherpuma populations To that end our study offers considerableinsights into benefits of different genetic management strategiesand associated costs

Without genetic management the Florida panther populationfaces a substantial risk of quasi‐extinction within 100 years (10ndash46 depending on quasi‐extinction threshold and scenarios)Twenty‐three years have passed since genetic introgression wasimplemented and the population appears healthy as indicatedby measures of genetic variation increases in the populationsize and improvements in age‐specific survival rates Ourfindings suggest that releasing more pumas more frequently doesnot necessarily improve persistence probability because such astrategy would cause larger fluctuations in population size(Fig 14) which negatively affects persistence probability Thusextinction risks faced by inbred populations of large carnivorescan be substantially reduced by releasing approximately 5 im-migrants every 2 decades We suggest that such a managementstrategy be followed only in conjunction with continued long‐term monitoring of the population Our analyses demonstratethat population growth and persistence are highly sensitive tokitten and female (adult and subadult) survival Continuing tocollect data on these demographic variables along with bio-metric and genetic sampling will remain important to pantherrecovery and vigilance in identifying issues associated with in-breeding depression The collection of these monitoring datashould allow managers to detect any demographic or geneticchanges in a timely fashion and respond with genetic in-trogression or other management actions if warrantedRecovery objectives as defined by the USFWS in its current

recovery plan (USFWS 2008) delineate the need for additionalpopulations in the historical range to downlist or delist theFlorida panther If the only extant population of Floridapanthers continued to grow it could serve as a source ofpanthers to be translocated to suitable habitat to seed addi-tional populations Maintaining current information on thegenetic health of the population and precise estimates of itssize will be important in assessing whether it can withstand theremoval of a subset of animals for translocation Although the2017 documentation of a female panther with kittens north ofthe current breeding range is encouraging in terms of thepotential for population expansionmdashgiven that no female hadbeen recorded north of the Caloosahatchee River since 1972mdashthe re‐establishment of separate new populations will mostlikely require translocations by wildlife managers and a lengthysociopolitical processConservation of threatened or endangered species such as the

Florida panther is inherently challenging For panthers and otherlarge carnivores that require vast extents of natural habitat topersist habitat is typically the most limiting factor that will de-termine whether recovery efforts succeed Although geneticmanagement invariably plays a key role in the long‐term persis-tence of the panther population even a healthy population caneventually succumb to the impacts of habitat loss The humanpopulation of Florida continues to be one of the fastest‐growingin the nation the United States Census Bureau currently ranksFlorida as the third most populous Therefore habitat loss isgoing to continue to be a key issue for panthers Diligence towardpreserving panther habitat in its historical range via conservationeasements or other means and trying to keep such areas inter-connected via corridors is a goal stakeholders such as governmentagencies sportsmen non‐governmental organizations ranchers

30 Wildlife Monographs bull 203

and other private landowners must work on collaboratively toachieve

ACKNOWLEDGMENTS

We thank D Shindle D Jennings T OrsquoMeara D Land andC Belden for valuable advice throughout the project We thankS Bass M Criffield M Cunningham D Giardina D JansenA Johnson D Land M Lotz C McBride Rocky McBrideRowdy McBride Roy McBride L Oberhofer S Schulze staffsat Everglades National Park and Big Cypress National Preserveand others for assistance with fieldwork We also thank JAustin M Conroy J Hines J Laake S Mills J Nichols JHines M Sunquist J Fryxell and T Coulson for advice andassistance regarding data analysis and panther biology TheMuseum of Texas Tech University Natural Science ResearchLaboratory provided samples from pumas from west Texas forcomparative genetic analyses M Ben‐David D Land WMcCown E Hellgren and 4 anonymous reviewers providedmany helpful comments on the manuscript We thank NereaUbierna Lopez for completing the Spanish translation of theabstract We are grateful to the Department of ResearchComputing University of Florida for excellent high‐perfor-mance computing services We would like to thank US Fishand Wildlife Service Florida Fish and Wildlife ConservationCommission (FWC) US National Park Service QuantitativeSpatial Ecology Evolution and Environment (QSE3) In-tegrative Graduate Education and Research Traineeship(IGERT) program funded by the National Science Foundationand the University of Florida for financially supporting thisstudy Funding for panther research conducted by the FWC issupported predominantly via donations accrued with vehicleregistrations for ldquoProtect the Pantherrdquo license plates

LITERATURE CITEDAcevedo‐Whitehouse K T R Spraker E Lyons S R Melin F Gulland RL Delong and W Amos 2006 Contrasting effects of heterozygosity onsurvival and hookworm resistance in California sea lion pups MolecularEcology 151973ndash1982

Adams J R L M Vucetich P W Hedrick R O Peterson and J AVucetich 2011 Genomic sweep and potential genetic rescue during limitingenvironmental conditions in an isolated wolf population Proceedings of theRoyal Society B Biological Sciences 2783336ndash3344

Agresti A 2013 Categorical data analysis Third edition John Wiley amp SonsHoboken New Jersey USA

Agudo R M Carrete M Alcaide C Rico F Hiraldo and J A Donaacutezar2012 Genetic diversity at neutral and adaptive loci determines individualfitness in a long‐lived territorial bird Proceedings of the Royal Society BBiological Sciences 2793241ndash3249

Albeke S E N P Nibbelink and M Ben‐David 2015 Modeling behavior bycoastal river otter (Lontra canadensis) in response to prey availability in PrinceWilliam Sound Alaska a spatially‐explicit individual‐based approach PLOSOne 10e0126208

Alho J S K Vaumllimaumlki and J Merilauml 2010 Rhh an R extension for estimatingmultilocus heterozygosity and heterozygosity‐heterozygosity correlationMolecular Ecology Resources 10720ndash722

Alvarez K 1993 Twilight of the panther Myakka River Publishing SarasotaFlorida USA

Aparicio J M J Ortego and P J Cordero 2006 What should we weigh toestimate heterozygosity alleles or loci Molecular Ecology 154659ndash4665

Bakker V J D F Doak G W Roemer D K Garcelon T J Coonan S AMorrison C Lynch K Ralls and R Shaw 2009 Incorporating ecologicaldrivers and uncertainty into a demographic population viability analysis for theisland fox Ecological Monographs 7977ndash108

Balloux F W Amos and T Coulson 2004 Does heterozygosity estimateinbreeding in real populations Molecular Ecology 133021ndash3031

Barone M A M E Roelke J Howard J L Brown A E Anderson and DE Wildt 1994 Reproductive characteristics of male Florida panthersmdashcomparative studies from Florida Texas Colorado Latin‐America andNorth‐American Zoos Journal of Mammalogy 75150ndash162

Bateson Z W P O Dunn S D Hull A E Henschen J A Johnson and LA Whittingham 2014 Genetic restoration of a threatened population ofgreater prairie‐chickens Biological Conservation 17412ndash19

Bauer H G Chapron K Nowell P Henschel P Funston L T B HunterD W Macdonald and C Packer 2015 Lion (Panthera leo) populations aredeclining rapidly across Africa except in intensively managed areas Pro-ceedings of National Academy of Sciences of the United States of America11214894ndash14899

Beier P 1995 Dispersal of juvenile cougars in fragmented habitat Journal ofWildlife Management 59228ndash237

Benson J F J A Hostetler D P Onorato W E Johnson M E Roelke SJ OrsquoBrien D Jansen and M K Oli 2011 Intentional genetic introgressioninfluences survival of adults and subadults in a small inbred felid populationJournal of Animal Ecology 80958ndash967

Benson J F P J Mahoney J A Sikich L E K Serieys J P Pollinger H BErnest and S P D Riley 2016 Interactions between demography geneticsand landscape connectivity increase extinction probability for a small popu-lation of large carnivores in a major metropolitan area Proceedings of theRoyal Society B Biological Sciences 28320160957

Beverly F 1874 The Florida panther Forest and Stream 3290Boyce M S C V Haridas C T Lee and the NCEAS Stochastic Demo-graphy Working Group 2006 Demography in an increasingly variable worldTrends in Ecology amp Evolution 21141ndash148

Brook B W J J OrsquoGrady A P Chapman M A Burgman H R Akcakayaand R Frankham 2000 Predictive accuracy of population viability analysis inconservation biology Nature 404385ndash387

Brys R and H Jacquemyn 2016 Severe outbreeding and inbreeding de-pression maintain mating system differentiation in Epipactis (Orchidaceae)Journal of Evolutionary Biology 29352ndash359

Burke J M and M L Arnold 2001 Genetics and the fitness of hybridsAnnual Review of Genetics 3531ndash52

Burnham K P 1993 A theory for combined analysis of ring recovery andrecapture data Pages 199ndash213 in J‐D Lebreton and P M North editorsMarked individuals in the study of bird population Springer‐Verlag NewYork New York USA

Burnham K P and D R Anderson 2002 Model selection and multimodelinference a practical information‐theoretic approach Second edition SpringerVerlag New York New York USA

Caswell H 2001 Matrix population models construction analysis and in-terpretation Sinauer Associates Sunderland Massachusetts USA

Charpentier M J M Setchell F Prugnolle L A Knapp E J Wickings PPeignot and M Hossaert‐McKey 2005 Genetic diversity and reproductivesuccess in mandrills (Mandrillus sphinx) Proceedings of the NationalAcademy of Sciences of the United States of America 10216723ndash16728

Cooch E and G C White editors 2018 Program MARK a gentle in-troduction lthttpwwwphidotorgsoftwaremarkdocsbookgt Accessed 7Apr 2016

Cooley H S R B Wielgus G M Koehler H S Robinson and B TMaletzke 2009 Does hunting regulate cougar populations A test of thecompensatory mortality hypothesis Ecology 902913ndash2921

Coulson T E A Catchpole S D Albon B J Morgan J M Pemberton TH Clutton‐Brock M J Crawley and B T Grenfell 2001 Age sex densitywinter weather and population crashes in Soay sheep Science 2921528ndash1531

Coulson T J Pemberton S Albon M Beaumont T Marshall J Slate FGuinness and T Clutton‐Brock 1998a Microsatellites reveal heterosis in reddeer Proceedings of the Royal Society B Biological Sciences 265489ndash495

Coulson T N S D Albon J M Pemberton J Slate F E Guinness and TH Clutton‐Brock 1998b Genotype by environment interactions in wintersurvival in red deer Journal of Animal Ecology 67434ndash445

Cox D 1972 Regression models and life‐tables Journal of the Royal StatisticalSociety Series B Statistical Methodology 34187ndash220

Creel S 2006 Recovery of the Florida panthermdashgenetic rescue demographic rescueor both Response to Pimm et al (2006) Animal Conservation 9125ndash126

Crnokrak P and D A Roff 1999 Inbreeding depression in the wild Heredity83260ndash270

Crow J 1948 Alternative hypotheses of hybrid vigor Genetics 33477ndash487

van de Kerk et al bull Florida Panther Population and Genetic Viability 31

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 28: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

life history and we could parameterize the model with our long‐term genetic and demographic dataThe explicit consideration of the effects of inbreeding on

Florida panther population dynamics and persistence representsan important step forward Nonetheless our model remainsimperfect First we neglected the possible effects of habitat lossdetrimental anthropogenic activities climate change the prob-ability of immigration of pumas from western populationsdisease or catastrophes on demographic rates and populationviability Although immigration from puma populations outsideFlorida is possible its probability is very low given the extensivechallenges the anthropogenically altered landscape would posefor such a migrant to make it successfully to South Florida Weomitted mutations from our model because we have no in-formation on mutation rates though we expect that they are lowbased on values reported for other mammals (Kumar and Sub-ramanian 2002) Finally we only considered possible effects ofinbreeding on age‐specific survival rates because there was noevidence that heterozygosity affected female reproduction Lossof heterozygosity can negatively affect reproduction becauseinbred male panthers can suffer from cryptorchidism which canadversely affect their fecundity However given the polygynousmating system we expect the Florida panther population growthis primarily female‐limitedAlthough it is generally accepted that genetic introgression

only temporarily relieves a population from inbreeding depres-sion we know of no cases in which it has been implementedmore than once to conserve a threatened wildlife populationOur study provides an example of how demographic and mo-lecular data can be used within an IBM framework to evaluatethe efficacy of alternative genetic management strategies via theFlorida panther as a case study Our model can be adjusted forand applied to other species and offers great potential as a toolfor genetic management of small isolated wildlife populations

MANAGEMENT IMPLICATIONS

Our results and those of Hostetler et al (2013) and Johnsonet al (2010) provide persuasive evidence that the genetic in-tervention implemented in 1995 prevented the demise of theFlorida panther and restored demographic vigor to the popu-lation The Florida panther population remains small and iso-lated from other puma populations the closest puma populationis located gt1000 km away in Texas and the intervening land-scape is highly developed and fragmented with many interstate(and other high‐traffic) highways and major urban centersTheory suggests that 1 migrant per generation is sufficient tominimize the loss of genetic diversity (eg Mills and Allendorf1996) However the possibility of successful migration of apuma to South Florida followed by successful mating every ~5years is very low considering the distance between Texas andFlorida (Hedrick 1995) and the many risks and barriers thatdispersing pumas face (Beier 1995 Maehr et al 2002 Thatcheret al 2006) Consequently the pantherrsquos long‐term persistencewill likely depend on subsequent timely genetic introgressionsgiven the near‐impossibility of natural gene flow from otherpuma populations To that end our study offers considerableinsights into benefits of different genetic management strategiesand associated costs

Without genetic management the Florida panther populationfaces a substantial risk of quasi‐extinction within 100 years (10ndash46 depending on quasi‐extinction threshold and scenarios)Twenty‐three years have passed since genetic introgression wasimplemented and the population appears healthy as indicatedby measures of genetic variation increases in the populationsize and improvements in age‐specific survival rates Ourfindings suggest that releasing more pumas more frequently doesnot necessarily improve persistence probability because such astrategy would cause larger fluctuations in population size(Fig 14) which negatively affects persistence probability Thusextinction risks faced by inbred populations of large carnivorescan be substantially reduced by releasing approximately 5 im-migrants every 2 decades We suggest that such a managementstrategy be followed only in conjunction with continued long‐term monitoring of the population Our analyses demonstratethat population growth and persistence are highly sensitive tokitten and female (adult and subadult) survival Continuing tocollect data on these demographic variables along with bio-metric and genetic sampling will remain important to pantherrecovery and vigilance in identifying issues associated with in-breeding depression The collection of these monitoring datashould allow managers to detect any demographic or geneticchanges in a timely fashion and respond with genetic in-trogression or other management actions if warrantedRecovery objectives as defined by the USFWS in its current

recovery plan (USFWS 2008) delineate the need for additionalpopulations in the historical range to downlist or delist theFlorida panther If the only extant population of Floridapanthers continued to grow it could serve as a source ofpanthers to be translocated to suitable habitat to seed addi-tional populations Maintaining current information on thegenetic health of the population and precise estimates of itssize will be important in assessing whether it can withstand theremoval of a subset of animals for translocation Although the2017 documentation of a female panther with kittens north ofthe current breeding range is encouraging in terms of thepotential for population expansionmdashgiven that no female hadbeen recorded north of the Caloosahatchee River since 1972mdashthe re‐establishment of separate new populations will mostlikely require translocations by wildlife managers and a lengthysociopolitical processConservation of threatened or endangered species such as the

Florida panther is inherently challenging For panthers and otherlarge carnivores that require vast extents of natural habitat topersist habitat is typically the most limiting factor that will de-termine whether recovery efforts succeed Although geneticmanagement invariably plays a key role in the long‐term persis-tence of the panther population even a healthy population caneventually succumb to the impacts of habitat loss The humanpopulation of Florida continues to be one of the fastest‐growingin the nation the United States Census Bureau currently ranksFlorida as the third most populous Therefore habitat loss isgoing to continue to be a key issue for panthers Diligence towardpreserving panther habitat in its historical range via conservationeasements or other means and trying to keep such areas inter-connected via corridors is a goal stakeholders such as governmentagencies sportsmen non‐governmental organizations ranchers

30 Wildlife Monographs bull 203

and other private landowners must work on collaboratively toachieve

ACKNOWLEDGMENTS

We thank D Shindle D Jennings T OrsquoMeara D Land andC Belden for valuable advice throughout the project We thankS Bass M Criffield M Cunningham D Giardina D JansenA Johnson D Land M Lotz C McBride Rocky McBrideRowdy McBride Roy McBride L Oberhofer S Schulze staffsat Everglades National Park and Big Cypress National Preserveand others for assistance with fieldwork We also thank JAustin M Conroy J Hines J Laake S Mills J Nichols JHines M Sunquist J Fryxell and T Coulson for advice andassistance regarding data analysis and panther biology TheMuseum of Texas Tech University Natural Science ResearchLaboratory provided samples from pumas from west Texas forcomparative genetic analyses M Ben‐David D Land WMcCown E Hellgren and 4 anonymous reviewers providedmany helpful comments on the manuscript We thank NereaUbierna Lopez for completing the Spanish translation of theabstract We are grateful to the Department of ResearchComputing University of Florida for excellent high‐perfor-mance computing services We would like to thank US Fishand Wildlife Service Florida Fish and Wildlife ConservationCommission (FWC) US National Park Service QuantitativeSpatial Ecology Evolution and Environment (QSE3) In-tegrative Graduate Education and Research Traineeship(IGERT) program funded by the National Science Foundationand the University of Florida for financially supporting thisstudy Funding for panther research conducted by the FWC issupported predominantly via donations accrued with vehicleregistrations for ldquoProtect the Pantherrdquo license plates

LITERATURE CITEDAcevedo‐Whitehouse K T R Spraker E Lyons S R Melin F Gulland RL Delong and W Amos 2006 Contrasting effects of heterozygosity onsurvival and hookworm resistance in California sea lion pups MolecularEcology 151973ndash1982

Adams J R L M Vucetich P W Hedrick R O Peterson and J AVucetich 2011 Genomic sweep and potential genetic rescue during limitingenvironmental conditions in an isolated wolf population Proceedings of theRoyal Society B Biological Sciences 2783336ndash3344

Agresti A 2013 Categorical data analysis Third edition John Wiley amp SonsHoboken New Jersey USA

Agudo R M Carrete M Alcaide C Rico F Hiraldo and J A Donaacutezar2012 Genetic diversity at neutral and adaptive loci determines individualfitness in a long‐lived territorial bird Proceedings of the Royal Society BBiological Sciences 2793241ndash3249

Albeke S E N P Nibbelink and M Ben‐David 2015 Modeling behavior bycoastal river otter (Lontra canadensis) in response to prey availability in PrinceWilliam Sound Alaska a spatially‐explicit individual‐based approach PLOSOne 10e0126208

Alho J S K Vaumllimaumlki and J Merilauml 2010 Rhh an R extension for estimatingmultilocus heterozygosity and heterozygosity‐heterozygosity correlationMolecular Ecology Resources 10720ndash722

Alvarez K 1993 Twilight of the panther Myakka River Publishing SarasotaFlorida USA

Aparicio J M J Ortego and P J Cordero 2006 What should we weigh toestimate heterozygosity alleles or loci Molecular Ecology 154659ndash4665

Bakker V J D F Doak G W Roemer D K Garcelon T J Coonan S AMorrison C Lynch K Ralls and R Shaw 2009 Incorporating ecologicaldrivers and uncertainty into a demographic population viability analysis for theisland fox Ecological Monographs 7977ndash108

Balloux F W Amos and T Coulson 2004 Does heterozygosity estimateinbreeding in real populations Molecular Ecology 133021ndash3031

Barone M A M E Roelke J Howard J L Brown A E Anderson and DE Wildt 1994 Reproductive characteristics of male Florida panthersmdashcomparative studies from Florida Texas Colorado Latin‐America andNorth‐American Zoos Journal of Mammalogy 75150ndash162

Bateson Z W P O Dunn S D Hull A E Henschen J A Johnson and LA Whittingham 2014 Genetic restoration of a threatened population ofgreater prairie‐chickens Biological Conservation 17412ndash19

Bauer H G Chapron K Nowell P Henschel P Funston L T B HunterD W Macdonald and C Packer 2015 Lion (Panthera leo) populations aredeclining rapidly across Africa except in intensively managed areas Pro-ceedings of National Academy of Sciences of the United States of America11214894ndash14899

Beier P 1995 Dispersal of juvenile cougars in fragmented habitat Journal ofWildlife Management 59228ndash237

Benson J F J A Hostetler D P Onorato W E Johnson M E Roelke SJ OrsquoBrien D Jansen and M K Oli 2011 Intentional genetic introgressioninfluences survival of adults and subadults in a small inbred felid populationJournal of Animal Ecology 80958ndash967

Benson J F P J Mahoney J A Sikich L E K Serieys J P Pollinger H BErnest and S P D Riley 2016 Interactions between demography geneticsand landscape connectivity increase extinction probability for a small popu-lation of large carnivores in a major metropolitan area Proceedings of theRoyal Society B Biological Sciences 28320160957

Beverly F 1874 The Florida panther Forest and Stream 3290Boyce M S C V Haridas C T Lee and the NCEAS Stochastic Demo-graphy Working Group 2006 Demography in an increasingly variable worldTrends in Ecology amp Evolution 21141ndash148

Brook B W J J OrsquoGrady A P Chapman M A Burgman H R Akcakayaand R Frankham 2000 Predictive accuracy of population viability analysis inconservation biology Nature 404385ndash387

Brys R and H Jacquemyn 2016 Severe outbreeding and inbreeding de-pression maintain mating system differentiation in Epipactis (Orchidaceae)Journal of Evolutionary Biology 29352ndash359

Burke J M and M L Arnold 2001 Genetics and the fitness of hybridsAnnual Review of Genetics 3531ndash52

Burnham K P 1993 A theory for combined analysis of ring recovery andrecapture data Pages 199ndash213 in J‐D Lebreton and P M North editorsMarked individuals in the study of bird population Springer‐Verlag NewYork New York USA

Burnham K P and D R Anderson 2002 Model selection and multimodelinference a practical information‐theoretic approach Second edition SpringerVerlag New York New York USA

Caswell H 2001 Matrix population models construction analysis and in-terpretation Sinauer Associates Sunderland Massachusetts USA

Charpentier M J M Setchell F Prugnolle L A Knapp E J Wickings PPeignot and M Hossaert‐McKey 2005 Genetic diversity and reproductivesuccess in mandrills (Mandrillus sphinx) Proceedings of the NationalAcademy of Sciences of the United States of America 10216723ndash16728

Cooch E and G C White editors 2018 Program MARK a gentle in-troduction lthttpwwwphidotorgsoftwaremarkdocsbookgt Accessed 7Apr 2016

Cooley H S R B Wielgus G M Koehler H S Robinson and B TMaletzke 2009 Does hunting regulate cougar populations A test of thecompensatory mortality hypothesis Ecology 902913ndash2921

Coulson T E A Catchpole S D Albon B J Morgan J M Pemberton TH Clutton‐Brock M J Crawley and B T Grenfell 2001 Age sex densitywinter weather and population crashes in Soay sheep Science 2921528ndash1531

Coulson T J Pemberton S Albon M Beaumont T Marshall J Slate FGuinness and T Clutton‐Brock 1998a Microsatellites reveal heterosis in reddeer Proceedings of the Royal Society B Biological Sciences 265489ndash495

Coulson T N S D Albon J M Pemberton J Slate F E Guinness and TH Clutton‐Brock 1998b Genotype by environment interactions in wintersurvival in red deer Journal of Animal Ecology 67434ndash445

Cox D 1972 Regression models and life‐tables Journal of the Royal StatisticalSociety Series B Statistical Methodology 34187ndash220

Creel S 2006 Recovery of the Florida panthermdashgenetic rescue demographic rescueor both Response to Pimm et al (2006) Animal Conservation 9125ndash126

Crnokrak P and D A Roff 1999 Inbreeding depression in the wild Heredity83260ndash270

Crow J 1948 Alternative hypotheses of hybrid vigor Genetics 33477ndash487

van de Kerk et al bull Florida Panther Population and Genetic Viability 31

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 29: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

and other private landowners must work on collaboratively toachieve

ACKNOWLEDGMENTS

We thank D Shindle D Jennings T OrsquoMeara D Land andC Belden for valuable advice throughout the project We thankS Bass M Criffield M Cunningham D Giardina D JansenA Johnson D Land M Lotz C McBride Rocky McBrideRowdy McBride Roy McBride L Oberhofer S Schulze staffsat Everglades National Park and Big Cypress National Preserveand others for assistance with fieldwork We also thank JAustin M Conroy J Hines J Laake S Mills J Nichols JHines M Sunquist J Fryxell and T Coulson for advice andassistance regarding data analysis and panther biology TheMuseum of Texas Tech University Natural Science ResearchLaboratory provided samples from pumas from west Texas forcomparative genetic analyses M Ben‐David D Land WMcCown E Hellgren and 4 anonymous reviewers providedmany helpful comments on the manuscript We thank NereaUbierna Lopez for completing the Spanish translation of theabstract We are grateful to the Department of ResearchComputing University of Florida for excellent high‐perfor-mance computing services We would like to thank US Fishand Wildlife Service Florida Fish and Wildlife ConservationCommission (FWC) US National Park Service QuantitativeSpatial Ecology Evolution and Environment (QSE3) In-tegrative Graduate Education and Research Traineeship(IGERT) program funded by the National Science Foundationand the University of Florida for financially supporting thisstudy Funding for panther research conducted by the FWC issupported predominantly via donations accrued with vehicleregistrations for ldquoProtect the Pantherrdquo license plates

LITERATURE CITEDAcevedo‐Whitehouse K T R Spraker E Lyons S R Melin F Gulland RL Delong and W Amos 2006 Contrasting effects of heterozygosity onsurvival and hookworm resistance in California sea lion pups MolecularEcology 151973ndash1982

Adams J R L M Vucetich P W Hedrick R O Peterson and J AVucetich 2011 Genomic sweep and potential genetic rescue during limitingenvironmental conditions in an isolated wolf population Proceedings of theRoyal Society B Biological Sciences 2783336ndash3344

Agresti A 2013 Categorical data analysis Third edition John Wiley amp SonsHoboken New Jersey USA

Agudo R M Carrete M Alcaide C Rico F Hiraldo and J A Donaacutezar2012 Genetic diversity at neutral and adaptive loci determines individualfitness in a long‐lived territorial bird Proceedings of the Royal Society BBiological Sciences 2793241ndash3249

Albeke S E N P Nibbelink and M Ben‐David 2015 Modeling behavior bycoastal river otter (Lontra canadensis) in response to prey availability in PrinceWilliam Sound Alaska a spatially‐explicit individual‐based approach PLOSOne 10e0126208

Alho J S K Vaumllimaumlki and J Merilauml 2010 Rhh an R extension for estimatingmultilocus heterozygosity and heterozygosity‐heterozygosity correlationMolecular Ecology Resources 10720ndash722

Alvarez K 1993 Twilight of the panther Myakka River Publishing SarasotaFlorida USA

Aparicio J M J Ortego and P J Cordero 2006 What should we weigh toestimate heterozygosity alleles or loci Molecular Ecology 154659ndash4665

Bakker V J D F Doak G W Roemer D K Garcelon T J Coonan S AMorrison C Lynch K Ralls and R Shaw 2009 Incorporating ecologicaldrivers and uncertainty into a demographic population viability analysis for theisland fox Ecological Monographs 7977ndash108

Balloux F W Amos and T Coulson 2004 Does heterozygosity estimateinbreeding in real populations Molecular Ecology 133021ndash3031

Barone M A M E Roelke J Howard J L Brown A E Anderson and DE Wildt 1994 Reproductive characteristics of male Florida panthersmdashcomparative studies from Florida Texas Colorado Latin‐America andNorth‐American Zoos Journal of Mammalogy 75150ndash162

Bateson Z W P O Dunn S D Hull A E Henschen J A Johnson and LA Whittingham 2014 Genetic restoration of a threatened population ofgreater prairie‐chickens Biological Conservation 17412ndash19

Bauer H G Chapron K Nowell P Henschel P Funston L T B HunterD W Macdonald and C Packer 2015 Lion (Panthera leo) populations aredeclining rapidly across Africa except in intensively managed areas Pro-ceedings of National Academy of Sciences of the United States of America11214894ndash14899

Beier P 1995 Dispersal of juvenile cougars in fragmented habitat Journal ofWildlife Management 59228ndash237

Benson J F J A Hostetler D P Onorato W E Johnson M E Roelke SJ OrsquoBrien D Jansen and M K Oli 2011 Intentional genetic introgressioninfluences survival of adults and subadults in a small inbred felid populationJournal of Animal Ecology 80958ndash967

Benson J F P J Mahoney J A Sikich L E K Serieys J P Pollinger H BErnest and S P D Riley 2016 Interactions between demography geneticsand landscape connectivity increase extinction probability for a small popu-lation of large carnivores in a major metropolitan area Proceedings of theRoyal Society B Biological Sciences 28320160957

Beverly F 1874 The Florida panther Forest and Stream 3290Boyce M S C V Haridas C T Lee and the NCEAS Stochastic Demo-graphy Working Group 2006 Demography in an increasingly variable worldTrends in Ecology amp Evolution 21141ndash148

Brook B W J J OrsquoGrady A P Chapman M A Burgman H R Akcakayaand R Frankham 2000 Predictive accuracy of population viability analysis inconservation biology Nature 404385ndash387

Brys R and H Jacquemyn 2016 Severe outbreeding and inbreeding de-pression maintain mating system differentiation in Epipactis (Orchidaceae)Journal of Evolutionary Biology 29352ndash359

Burke J M and M L Arnold 2001 Genetics and the fitness of hybridsAnnual Review of Genetics 3531ndash52

Burnham K P 1993 A theory for combined analysis of ring recovery andrecapture data Pages 199ndash213 in J‐D Lebreton and P M North editorsMarked individuals in the study of bird population Springer‐Verlag NewYork New York USA

Burnham K P and D R Anderson 2002 Model selection and multimodelinference a practical information‐theoretic approach Second edition SpringerVerlag New York New York USA

Caswell H 2001 Matrix population models construction analysis and in-terpretation Sinauer Associates Sunderland Massachusetts USA

Charpentier M J M Setchell F Prugnolle L A Knapp E J Wickings PPeignot and M Hossaert‐McKey 2005 Genetic diversity and reproductivesuccess in mandrills (Mandrillus sphinx) Proceedings of the NationalAcademy of Sciences of the United States of America 10216723ndash16728

Cooch E and G C White editors 2018 Program MARK a gentle in-troduction lthttpwwwphidotorgsoftwaremarkdocsbookgt Accessed 7Apr 2016

Cooley H S R B Wielgus G M Koehler H S Robinson and B TMaletzke 2009 Does hunting regulate cougar populations A test of thecompensatory mortality hypothesis Ecology 902913ndash2921

Coulson T E A Catchpole S D Albon B J Morgan J M Pemberton TH Clutton‐Brock M J Crawley and B T Grenfell 2001 Age sex densitywinter weather and population crashes in Soay sheep Science 2921528ndash1531

Coulson T J Pemberton S Albon M Beaumont T Marshall J Slate FGuinness and T Clutton‐Brock 1998a Microsatellites reveal heterosis in reddeer Proceedings of the Royal Society B Biological Sciences 265489ndash495

Coulson T N S D Albon J M Pemberton J Slate F E Guinness and TH Clutton‐Brock 1998b Genotype by environment interactions in wintersurvival in red deer Journal of Animal Ecology 67434ndash445

Cox D 1972 Regression models and life‐tables Journal of the Royal StatisticalSociety Series B Statistical Methodology 34187ndash220

Creel S 2006 Recovery of the Florida panthermdashgenetic rescue demographic rescueor both Response to Pimm et al (2006) Animal Conservation 9125ndash126

Crnokrak P and D A Roff 1999 Inbreeding depression in the wild Heredity83260ndash270

Crow J 1948 Alternative hypotheses of hybrid vigor Genetics 33477ndash487

van de Kerk et al bull Florida Panther Population and Genetic Viability 31

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 30: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

Cunningham S C W B Ballard and H A Whitlaw 2001 Age structuresurvival and mortality of mountain lions in southeastern Arizona South-western Naturalist 4676ndash80

David P 1998 Heterozygosityndashfitness correlations new perspectives on oldproblems Heredity 80531ndash537

Davis J H Jr 1943 The natural features of southern Florida Florida Geo-logical Survey Tallahassee Florida USA

Derocher A E and O Wiig 1999 Infanticide and cannibalism of juvenilepolar bears (Ursus maritimus) in Svalbard Arctic 52307ndash310

DeYoung R W and R L Honeycutt 2005 The molecular toolbox genetictechniques in wildlife ecology and management Journal of Wildlife Man-agement 691362ndash1384

Dorcas M E J D Willson R N Reed R W Snow M R Rochford M AMiller W E Meshaka P T Andreadis F J Mazzotti C M Romagosaand K M Hart 2012 Severe mammal declines coincide with proliferation ofinvasive Burmese pythons in Everglades National Park Proceedings of theNational Academy of Sciences of the United States of America1092418ndash2422

Driscoll C A M Menotti‐Raymond G Nelson D Goldstein and S JOrsquoBrien 2002 Genomic microsatellites as evolutionary chronometers a testin wild cats Genome Research 12414ndash423

Duever M J J E Carlson J F Meeder L C Duever L H Gunderson LA Riopelle T R Alexander R L Myers and D P Spangler 1986 The BigCypress National Preserve National Audubon Society New York NewYork USA

Earl D A and B M VonHoldt 2011 Structure Harvester a website andprogram for visualizing Structure output and implementing the Evannomethod Conservation Genetics Resources 4359ndash361

Edmands S 2007 Between a rock and a hard place evaluating the relative risksof inbreeding and outbreeding for conservation and management MolecularEcology 16463ndash475

Evanno G S Regnaut and J Goudet 2005 Detecting the number of clustersof individuals using the software STRUCTURE a simulation study Mole-cular Ecology 142611ndash2620

Fenster C B and L F Gallaway 2000 Inbreeding and outbreeding de-pression in natural populations of Chamaecrista fasciculata (Fabaceae) Con-servation Biology 141406ndash1412

Florida Fish and Wildlife Conservation Commission [FWC] 2017 Annualreport on the research and management of Florida panthers 2016ndash2017 Fishand Wildlife Research Institute and Division of Habitat and Species Con-servation Florida Fish and Wildlife Conservation Commission NaplesFlorida USA

Foster M L and S R Humphrey 1995 Use of highway underpasses byFlorida panthers and other wildlife Wildlife Society Bulletin 2395ndash100

Frakes R A R C Belden B E Wood and F E James 2015 Landscapeanalysis of adult Florida panther habitat PLOS One 10e0133044

Frankham R 2010a Challenges and opportunities of genetic approaches tobiological conservation Biological Conservation 1431919ndash1927

Frankham R 2010b Inbreeding in the wild really does matter Heredity104124

Frankham R 2015 Genetic rescue of small inbred populations meta‐analysisreveals large and consistent benefits of gene flow Molecular Ecology242610ndash2618

Frankham R 2016 Genetic rescue benefits persist to at least the F3 generationbased on a meta‐analysis Biological Conservation 19533ndash36

Frankham R C J A Bradshaw and B W Brook 2014 Genetics in con-servation management revised recommendations for the 50500 rules RedList criteria and population viability analyses Biological Conservation17056ndash63

Garrison E P J W McCown and M K Oli 2007 Reproductive ecologyand cub survival of Florida black bears Journal of Wildlife Management71720ndash727

Goto S H Iijima H Ogawa and K Ohya 2011 Outbreeding depressioncaused by intraspecific hybridization between local and nonlocal genotypes inAbies sachalinensis Restoration Ecology 19243ndash250

Goudet J 1995 FSTAT (version 12) a computer program to calculate F‐statistics Journal of Heredity 86485ndash486

Grimm V 1999 Ten years of individual‐based modelling in ecology what havewe learned and what could we learn in the future Ecological Modelling115129ndash148

Grimm V U Berger F Bastiansen S Eliassen V Ginot J Giske J Goss‐Custard T Grand S K Heinz G Huse A Huth J U Jepsen CJoslashrgensen W M Mooij B Muumleller G Peer C Piou S F Railsback A

M Robbins M M Robbins E Rossmanith N Rueger E Strand SSouissi R A Stillman R Vaboslash U Visser and D L DeAngelis 2006 Astandard protocol for describing individual‐based and agent‐based modelsEcological Modelling 198115ndash126

Grimm V U Berger D L DeAngelis J G Polhill J Giske and S FRailsback 2010 The ODD protocol a review and first update EcologicalModelling 2212760ndash2768

Grimm V and S F Railsback 2005 Individual‐based modeling and ecologyPrinceton University Press Princeton New Jersey USA

Hansson B H Westerdahl D Hasselquist M Akesson and S Bensch 2004Does linkage disequilibrium generate heterozygosity‐fitness correlations ingreat reed warblers Evolution 58870ndash879

Haridas C V and S Tuljapurkar 2005 Elasticities in variable environmentsproperties and implications The American Naturalist 166481ndash495

Hartl D L and A G Clark 1997 Principles of population genetics Thirdedition Sinauer Sunderland Massachusetts USA

Hedrick P W 1995 Gene flow and genetic restoration the Florida panther asa case study Conservation Biology 9996ndash1007

Hedrick P W and R Fredrickson 2009 Genetic rescue guidelines with ex-amples from Mexican wolves and Florida panthers Conservation Genetics11615ndash626

Hedrick P W M Kardos R O Peterson and J A Vucetich 2017 Genomicvariation of inbreeding and ancestry in the remaining two Isle Royale wolvesJournal of Heredity 108120ndash126

Hedrick P W R O Peterson L M Vucetich J R Adams and J AVucetich 2014 Genetic rescue in Isle Royale wolves genetic analysis and thecollapse of the population Conservation Genetics 151111ndash1121

Heisey D M and B R Patterson 2006 A review of methods to estimatecause‐specific mortality in presence of competing risks Journal of WildlifeManagement 701544ndash1555

Heppell S C Pfister and H de Kroon 2000 Elasticity analysis in populationbiology methods and applications Ecology 81605ndash606

Hogg J T S H Forbes B M Steele and G Luikart 2006 Genetic rescue ofan insular population of large mammals Proceedings of the Royal Society BBiological Sciences 2731491ndash1499

Hostetler J D Onorato B Bolker W Johnson S OrsquoBrien D Jansen andM Oli 2012 Does genetic introgression improve female reproductiveperformance A test on the endangered Florida panther Oecologia 168289ndash300

Hostetler J A D P Onorato D Jansen and M K Oli 2013 A catrsquos tale theimpact of genetic restoration on Florida panther population dynamics andpersistence Journal of Animal Ecology 82608ndash620

Hostetler J A D P Onorato J D Nichols W E Johnson M E Roelke SJ OBrien D Jansen and M K Oli 2010 Genetic introgression and thesurvival of Florida panther kittens Biological Conservation 1432789ndash2796

Hunter C M H Caswell M C Runge E V Regehr S C Amstrup and IStirling 2010 Climate change threatens polar bear populations a stochasticdemographic analysis Ecology 912883ndash2897

Hwang A S S L Northrup D L Peterson Y Kim and S Edmands 2012Long‐term experimental hybrid swarms between nearly incompatible Tigriopuscalifornicus populations persistent fitness problems and assimilation by thesuperior population Conservation Genetics 13567ndash579

Jensen H E M Bremset T H Ringsby and B‐E Saeligther 2007 Multilocusheterozygosity and inbreeding depression in an insular house sparrow meta-population Molecular Ecology 164066ndash4078

Johnson H E L S Mills J D Wehausen T R Stephenson and G Luikart2011 Translating effects of inbreeding depression on component vital rates tooverall population growth in endangered bighorn sheep Conservation Biology251240ndash1249

Johnson W E D P Onorato M E Roelke E D Land M CunninghamR C Belden R McBride D Jansen M Lotz D Shindle J Howard D EWildt L M Penfold J A Hostetler M K Oli and S J OBrien 2010Genetic restoration of the Florida panther Science 3291641ndash1645

Kardos M H R Taylor H Ellegren G Luikart and F W Allendorf 2016Genomics advances the study of inbreeding depression in the wild Evolu-tionary Applications 91205ndash1218

Keller L and D Waller 2002 Inbreeding effects in wild populations Trendsin Ecology amp Evolution 17230ndash241

Keyfitz N and H Caswell 2005 Applied mathematical demography Thirdedition Springer New York New York USA

Kohira M H Okada M Nakanishi and M Yamanaka 2009 Modeling theeffects of human‐caused mortality on the brown bear population on theShiretoko Peninsula Hokkaido Japan Ursus 2012ndash21

32 Wildlife Monographs bull 203

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 31: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

Kotz S N Balakrishnan and N L Johnson 2000 Dirichlet and inverteddirichlet disributions Wiley New York New York USA

Kumar S and S Subramanian 2002 Mutation rates in mammalian genomesProceedings of the National Academy of Sciences of the United States ofAmerica 99803ndash808

Laake J L 2013 RMark an R interface for analysis of capture‐recapture datawith MARK Alaska Fish Scientific Center NOAA National Marine FisheryService Seattle Washington USA

Lacy R C 1993 VORTEX a computer simulation model for populationviability analysis Wildlife Research 2045ndash65

Lacy R C 2000 Structure of the VORTEX simulation model for populationviability analysis Ecological Bulletins 48191ndash203

Lacy R C A Petric and M Warneke 1993 Inbreeding and outbreeding incaptive populations of wild animals Pages 352ndash374 in N W Thornhilleditor The natural history of inbreeding and outbreeding theoretical andempirical perspectives University of Chicago Press Chicago Illinois USA

Lambert C M S R B Wielgus H S Robinson D D Katnik H SCruickshank R Clarke and J Almack 2006 Cougar population dynamicsand viability in the Pacific Northwest Journal of Wildlife Management70246ndash254

Land E D D B Shindle R J Kawula J F Benson M A Lotz and D POnorato 2008 Florida panther habitat selection analysis of concurrent GPSand VHF telemetry data Journal of Wildlife Management 72633ndash639

Laundre J W L Hernandez and S G Clark 2007 Numerical and demo-graphic responses of pumas to changes in prey abundance testing currentpredictions Journal of Wildlife Management 71345ndash355

Liberg O H Andreacuten H‐C Pedersen H Sand D Sejberg P Wabakken MÅkesson and S Bensch 2005 Severe inbreeding depression in a wild wolf(Canis lupus) population Biology Letters 117ndash20

Logan K A and L L Sweanor 2001 Desert puma evolutionary ecology andconservation of an enduring carnivore Island Press Washington DC USA

Lotka A J 1924 Elements of physical biology Williams and Wilkins Balti-more Maryland USA

Madsen T R Shine M Olsson and H Wittzell 1999 Restoration of aninbred adder population Nature 40234ndash35

Madsen T B Ujvari and M Olsson 2004 Novel genes continue to enhancepopulation growth in adders (Vipera berus) Biological Conservation120145ndash147

Maehr D S 1990 Florida panther movements social organization and habitatuse Final Performance Report 7502 Florida Game and Freshwater FishCommission Tallahassee Florida USA

Maehr D S and G B Caddick 1995 Demographics and genetic in-trogression in the Florida panther Conservation Biology 91295ndash1298

Maehr D S P Crowley J J Cox M J Lacki J L Larkin T S Hoctor LD Harris and P M Hall 2006 Of cats and Haruspices genetic interventionin the Florida panther Response to Pimm et al (2006) Animal Conservation9127ndash132

Maehr D S E D Land D B Shindle O L Bass and T S Hoctor 2002Florida panther dispersal and conservation Biological Conservation106187ndash197

Maehr D S J C Roof E D Land and J W McCown 1989 First re-production of a panther (Felis concolor coryi) in southwestern Florida USAMammalia 53129ndash131

Mainguy J S D Cocircteacute and D W Coltman 2009 Multilocus heterozygosityparental relatedness and individual fitness components in a wild mountaingoat Oreamnos americanus population Molecular Ecology 182297ndash306

Marino S I B Hogue C J Ray and D E Kirschner 2008 A methodologyfor performing global uncertainty and sensitivity analysis in systems biologyJournal of Theoretical Biology 254178ndash196

Marr A B L F Keller and P Arcese 2002 Heterosis and outbreedingdepression in descendants of natural immigrants to an inbred population ofsong sparrows (Melospiza melodia) Evolution 56131ndash142

Marshall T C J Slate L E B Kruuk and J M Pemberton 1998 Statisticalconfidence for likelihood‐based paternity inference in natural populationsMolecular Ecology 7639ndash655

MathWorks 2014 MATLAB the language of technical computing Version14a Math Works Inc Natick Massachusetts USA

Mazerolle M J 2017 AICcmodavg model selection and multimodel inferencebased on (Q)AIC(c) R package version 21‐1 lthttpscranr‐projectorgpackage=AICcmodavggt Accessed 14 Oct 2016

McBride R T R T McBride R M McBride and C E McBride 2008Counting pumas by categorizing physical evidence Southeastern Naturalist7381ndash400

McClintock B T D P Onorato and J Martin 2015 Endangered Floridapanther population size determined from public reports of motor vehiclecollision mortalities Journal of Applied Ecology 52893ndash901

McCown J W D S Maehr and J Roboski 1990 A portable cushion as awildlife capture aid Wildlife Society Bulletin 1834ndash36

McKelvey K S and M K Schwartz 2005 DROPOUT a program to identifyproblem loci and samples for noninvasive genetic samples in a capture‐mark‐recapture framework Molecular ecology notes 5716ndash718

McLane A J C Semeniuk G J McDermid and D J Marceau 2011 Therole of agent‐based models in wildlife ecology and management EcologicalModelling 2221544ndash1556

Menotti‐Raymond M V A David L A Lyons A A Schaffer J F TomlinM K Hutton and S J OBrien 1999 A genetic linkage map of micro-satellites in the domestic cat (Felis catus) Genomics 579ndash23

Miller B K Ralls R P Reading J M Scott and J Estes 1999 Biologicaland technical considerations of carnivore translocation a review AnimalConservation 259ndash68

Miller J M and D W Coltman 2014 Assessment of identity disequilibriumand its relation to empirical heterozygosity fitness correlations a meta‐analysisMolecular Ecology 231899ndash1909

Mills L S and F W Allendorf 1996 The one‐migrant‐per‐generation rule inconservation and management Conservation Biology 101509ndash1518

Mills L S K L Pilgrim M K Schwartz and K McKelvey 2000 Identifyinglynx and other North American felids based on mtDNA analysis Conserva-tion Genetics 1285ndash288

Milner J M S D Albon A W Illius J M Pemberton and T H Clutton‐Brock 1999 Repeated selection of morphometric traits in the Soay sheep onSt Kilda Journal of Animal Ecology 68472ndash488

Min Y and A Agresti 2005 Random effect models for repeated measures ofzero‐inflated count data Statistical Modelling 51ndash19

Moritz C 1999 Conservation units and translocations strategies for conservingevolutionary processes Hereditas 130217ndash228

Morris W F and D F Doak 2002 Quantitative conservation biology Si-nauer Sunderland Massachusetts USA

Morrison C C Wardle and J G Castley 2016 Repeatability and reprodu-cibility of population viability analysis (PVA) and the implications for threa-tened species management Frontiers in Ecology and Evolution 4art98

Murchison E P O B Schulz‐Trieglaff Z Ning L B Alexandrov M JBauer B Fu M Hims Z Ding S Ivakhno and C Stewart 2012 Genomesequencing and analysis of the Tasmanian devil and its transmissible cancerCell 148780ndash791

Nichols J D M C Runge F A Johnson and B K Williams 2007Adaptive harvest management of North American waterfowl populations abrief history and future prospects Journal of Ornithology 148 Supplement2S343ndashS349

Nowak R M and R T McBride 1974 Status survey of the Florida pantherDanbury Press Danbury Connecticut USA

OrsquoBrien S J 1994 Genetic and phylogenetic analyses of endangered speciesAnnual Review of Genetics 28467ndash489

OBrien S J M E Roelke N Yuhki K W Richards W E Johnson W LFranklin A E Anderson O L Bass R C Belden and J S Martenson1990 Genetic introgression within the Florida panther Felis concolor coryiNational Geographic Research 6485ndash494

Oli M K and F S Dobson 2003 The relative importance of life‐historyvariables to population growth rate in mammals Colersquos prediction revisitedThe American Naturalist 161422ndash440

Onorato D C Belden M Cunningham D Land R McBride and MRoelke 2010 Long‐term research on the Florida panther (Puma concolorcoryi) historical findings and future obstacles to population persistence Pages453ndash469 in D Macdonald and A Loveridge editors Biology and con-servation of wild felids Oxford University Press Oxford United Kingdom

Onorato D P M Criffield M Lotz M Cunningham R McBride E HLeone O L Bass and E C Hellgren 2011 Habitat selection by criticallyendangered Florida panthers across the diel period implications for landmanagement and conservation Animal Conservation 14196ndash205

Ortego J G Calabuig P J Cordero and J M Aparicio 2007 Egg production andindividual genetic diversity in lesser kestrels Molecular Ecology 162383ndash2392

Peakall R and P E Smouse 2006 GenAlEx 6 genetic analysis in ExcelPopulation genetic software for teaching and research Molecular EcologyResources 6288ndash295

Peakall R and P E Smouse 2012 GenAlEx 65 genetic analysis in ExcelPopulation genetic software for teaching and researchmdashan update Bioinfor-matics 282537ndash2539

van de Kerk et al bull Florida Panther Population and Genetic Viability 33

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 32: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

Peterson R O 1977 Wolf ecology and prey relationships on Isle Royale USNational Park Service Scientific Monograph Series 11 WashingtonDC USA

Peterson R O and R E Page 1988 The rise and fall of Isle Royale wolves1975ndash1986 Journal of Mammalogy 6989ndash99

Peterson R O N J Thomas J M Thurber J A Vucetich and T A Waite1998 Population limitation and the wolves of Isle Royale Journal of Mam-malogy 79828ndash841

Pickup M D L Field D M Rowell and A G Young 2013 Source po-pulation characteristics affect heterosis following genetic rescue of fragmentedplant populations Proceedings of the Royal Society B Biological Sciences28020122058

Pierson J C S R Beissinger J G Bragg D J Coates J G B OostermeijerP Sunnucks N H Schumaker M V Trotter and A G Young 2015Incorporating evolutionary processes into population viability models Con-servation Biology 29755ndash764

Pimm S L L Dollar and O L Bass 2006 The genetic rescue of the Floridapanther Animal Conservation 9115ndash122

Pollock K H S R Winterstein C M Bunck and P D Curtis 1989Survival analysis in telemetry studies the staggered entry design Journal ofWildlife Management 537ndash15

Pritchard J K M Stephens and P Donnelly 2000 Inference of populationstructure using multilocus genotype data Genetics 155945ndash959

R Core Team 2016 R a language and environment for statistical computing RFoundation for Statistical Computing Vienna Austria

Railsback S F and V Grimm 2011 Agent‐based and individual‐basedmodeling a practical introduction Princeton University Press Princeton NewJersey USA

Reed J M L S Mills J B Dunning E S Menges K S McKelvey R FryeS R Beissinger M‐C Anstett and P Miller 2002 Emerging issues inpopulation viability analysis Conservation Biology 167ndash19

Reinert H K 1991 Translocation as a conservation strategy for amphibiansand reptiles some comments concerns and observations Herpetologica47357ndash363

Riley S P D L E K Serieys J P Pollinger J A Sikich L Dalbeck R KWayne and H B Ernest 2014 Individual behaviors dominate the dynamicsof an urban mountain lion population isolated by roads Current Biology241989ndash1994

Robinson H S R B Wielgus H S Cooley and S W Cooley 2008 Sinkpopulations in carnivore management cougar demography and immigration ina hunted population Ecological Applications 181028ndash1037

Robinson J A D Ortega‐Vecchyo Z Fan B Y Kim B M vonHoldt C DMarsden K E Lohmueller and R K Wayne 2016 Genomic flatlining inthe endangered island fox Current Biology 261183ndash1189

Roelke M E J S Martenson and S J OBrien 1993 The consequences ofdemographic reduction and genetic depletion in the endangered Floridapanther Current Biology 3340ndash349

Royama T 1992 Analytical population dynamics Chapman amp Hall LondonUnited Kingdom

Ruiz‐Loacutepez M J N Gantildean J A Godoy A Del Olmo J Garde G EspesoA Vargas F Martinez E R S Roldaacuten and M Gomendio 2012 Het-erozygosity‐fitness correlations and inbreeding depression in two criticallyendangered mammals Conservation Biology 261121ndash1129

Runge M C 2011 An introduction to adaptive management for threatened andendangered species Journal of Fish and Wildlife Management 2220ndash233

Ruth T K M A Haroldson K M Murphy P C Buotte M G Hornockerand H B Quigley 2011 Cougar survival and source‐sink structure onGreater Yellowstonersquos Northern Range Journal of Wildlife Management751381ndash1398

Ruth T K K A Logan L L Sweanor M Hornocker and L J Temple1998 Evaluating cougar translocation in New Mexico Journal of WildlifeManagement 621264ndash1275

Seal U S 1994 A plan for genetic restoration and management of the Floridapanther (Felis concolor coryi) Report to the Florida Game and Freshwater FishCommission IUCN Conservation Breeding Specialist Group Apple ValleyMinnesota USA

Seddon N W Amos R A Mulder and J A Tobias 2004 Male hetero-zygosity predicts territory size song structure and reproductive success in acooperatively breeding bird Proceedings of the Royal Society B BiologicalSciences 2711823ndash1829

Shull G H 1908 The composition of a field of maize Journal of Heredity4296ndash301

Sikes R S 2016 Guidelines of the American Society of Mammalogists for theuse of wild mammals in research and education Journal of Mammalogy97663ndash688

Silva A D G Luikart N G Yoccoz A Cohas and D Allaineacute 2005 Geneticdiversity‐fitness correlation revealed by microsatellite analyses in European al-pine marmots (Marmota marmota) Conservation Genetics 7371ndash382

Smith T B and R K Wayne 1996 Molecular genetic approaches in con-servation Oxford University Press New York New York USA

Stoner D C M L Wolfe and D M Choate 2010 Cougar exploitationlevels in Utah implications for demographic structure population recoveryand metapopulation dynamics Journal of Wildlife Management701588ndash1600

Szulkin M N Bierne and P David 2010 Heterozygosity‐fitness correlationsa time for reappraisal Evolution 641202ndash1217

Taberlet P S Griffin B Goossens S Questiau V Manceau N EscaravageL P Waits and J Bouvet 1996 Reliable genotyping of samples with verylow DNA quantities using PCR Nucleic Acids Research 243189ndash3194

Tallmon D A G Luikart and R S Waples 2004 The alluring simplicityand complex reality of genetic rescue Trends in Ecology amp Evolution19489ndash496

Terrell K A A E Crosier D E Wildt S J OBrien N M Anthony LMarker and W E Johnson 2016 Continued decline in genetic diversityamong wild cheetahs (Acinonyx jubatus) without further loss of semen qualityBiological Conservation 200192ndash199

Thatcher C A F T Van Manen and J D Clark 2006 Identifying suitablesites for Florida panther reintroduction Journal of Wildlife Management70752ndash763

Therneau T M and P M Grambsch 2000 Modeling survival data ex-tending the Cox model Springer New York New York USA

Thiele J C 2014 R marries NetLogo introduction to the RNetLogo packageJournal of Statistical Software 581ndash41

Thiele J C W Kurth and V Grimm 2012 RNetLogo an R package forrunning and exploring individual‐based models implemented in NetLogoMethods in Ecology and Evolution 3480ndash483

Thiele J C W Kurth and V Grimm 2014 Facilitating parameter estimationand sensitivity analysis of agent‐based models a cookbook using NetLogo andR Journal of Artificial Societies and Social Simulation 17art11

Thirstrup J C Pertoldi P Larsen and V Nielsen 2014 Heterosis in thesecond and third generation affects litter size in a crossbreed mink (Neovisonvison) population Archives of Biological Sciences 661097ndash1103

Trinkel M D Cooper C Packer and R Slotow 2011 Inbreeding depressionincreases susceptibility to bovine tuberculosis in lions an experimental testusing an inbredndashoutbred contrast through translocation Journal of WildlifeDiseases 47494ndash500

Trinkel M P Funston M Hofmeyr D Hofmeyr S Dell C Packer and RSlotow 2010 Inbreeding and density‐dependent population growth in asmall isolated lion population Animal Conservation 13374ndash382

Tuljapurkar S D 1990 Population dynamics in variable environmentsSpringer New York New York USA

US Federal Register 1967 Native fish and wildlife endangered species324001

US Fish and Wildlife Service [USFWS] 1994 Final environmental assess-ment genetic restoration of the Florida panther US Fish and WildlifeService Atlanta Georgia USA

US Fish and Wildlife Service [USFWS] 2008 Florida panther recovery plan(Puma concolor coryi) third revision US Fish and Wildlife Service AtlantaGeorgia USA

Velando A Aacute Barros and P Moran 2015 Heterozygosityndashfitness correla-tions in a declining seabird population Molecular Ecology 241007ndash1018

Vickers W T J N Sanchez C K Johnson S A Morrison R Botta TSmith B S Cohen P R Huber E B Ernest and W M Boyce 2015Survival and mortality of pumas (Puma concolor) in a fragmented urbanizinglandscape PLOS One 10e0131490

Vila C A K Sundqvist Oslash Flagstad J Seddon S Bjoumlrnerfeldt I Kojola ACasulli H Sand P Wabakken and H Ellegren 2003 Rescue of a severelybottlenecked wolf (Canis lupus) population by a single immigrant Proceedingsof the Royal Society of London B Biological Sciences 27091ndash97

Waller D M 2015 Genetic rescue a safe or risky bet Molecular Ecology242595ndash2597

Waser N M M V Price and R G Shaw 2000 Outbreeding depressionvaries among cohorts of Ipomopsis aggregata planted in nature Evolution54485ndash491

34 Wildlife Monographs bull 203

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35

Page 33: Dynamics, Persistence, and Genetic ... - University of Florida de Kerk_et_al-2019-Wildlife_Monographs(1).pdfFlorida panther population would face a substantially increased risk of

White G C and K P Burnham 1999 Program MARK survival rate estimationfrom both live and dead encounters Bird Studies 46(Suppl) S120ndashS139

Whiteley A R S W Fitzpatrick W C Funk and D A Tallmon 2015Genetic rescue to the rescue Trends in Ecology amp Evolution 3042ndash49

Wilensky U 1999 NetLogo Center for connected learning and computer‐based modeling Northwestern University Evanston Illinois USA

Wilkins L J M Arias‐Reveron B Stith M E Roelke and R C Belden1997 The Florida panther a morphological investigation of the subspecieswith a comparison to other North and South American cougars Bulletin ofthe Florida Museum of Natural History 40221ndash269

Willi Y M van Kleunen S Dietrich and M Fischer 2007 Genetic rescuepersists beyond first‐generation outbreeding in small populations of a rareplant Proceedings of the Royal Society B Biological Science 2742357ndash2364

Williams B K and E D Brown 2012 Adaptive management the USDepartment of the Interior applications guide US Department of the InteriorWashington DC USA

Williams B K and E D Brown 2016 Technical challenges in the applicationof adaptive management Biological Conservation 195255ndash263

Williams B K J D Nichols and M J Conroy 2002 Analysis and man-agement of animal populations Academic Press San Diego CaliforniaUSA

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupporting Information section at the end of the article

van de Kerk et al bull Florida Panther Population and Genetic Viability 35