J Appl Ecol. 2019;56:2057–2068. wileyonlinelibrary.com/journal/jpe | 2057

Received:12November2018  |  Accepted:10May2019DOI: 10.1111/1365-2664.13455

R E S E A R C H A R T I C L E

Predicting the fundamental thermal niche of crop pests and diseases in a changing world: A case study on citrus greening

Rachel A. Taylor1,2  | Sadie J. Ryan3,4,5  | Catherine A. Lippi3,4 | David G. Hall6 | Hossein A. Narouei‐Khandan4,7 | Jason R. Rohr1,8 | Leah R. Johnson9

1DepartmentofIntegrativeBiology,UniversityofSouthFlorida,Tampa,Florida;2DepartmentofEpidemiologicalSciences,AnimalandPlantHealthAgency(APHA),Weybridge,UK;3QuantitativeDiseaseEcologyandConservation(QDEC)Lab,DepartmentofGeography,UniversityofFlorida,Gainesville,Florida;4EmergingPathogensInstitute,UniversityofFlorida,Gainesville,Florida;5SchoolofLifeSciences,UniversityofKwaZulu,Natal,SouthAfrica;6USDA‐ARS,FortPierce,Florida;7DepartmentofPlantPathology,UniversityofFlorida,Gainesville,Florida;8DepartmentofBiologicalSciences,UniversityofNotreDame,NotreDame,Indianaand9DepartmentofStatistics,VirginiaPolytechnicInstituteandStateUniversity(VirginiaTech),Blacksburg,Virginia

©2019Crowncopyright.JournalofAppliedEcology©2019BritishEcologicalSocietyThisarticleispublishedwiththepermissionoftheControllerofHMSOandtheQueen’sPrinterforScotland.

CorrespondenceRachelA.TaylorEmail:rachel.taylor@apha.gov.uk

Funding informationDivisionofMathematicalSciences,Grant/AwardNumber:1750113;DivisionofIntegrativeOrganismalSystems,Grant/AwardNumber:IOS‐1754868;U.S.DepartmentofAgriculture,Grant/AwardNumber:2009‐35102‐0543;DivisionofEnvironmentalBiology,Grant/AwardNumber:DEB‐1518681;DivisionofEmergingFrontiers,Grant/AwardNumber:EF‐1241889;NationalInstitutesofHealth,Grant/AwardNumber:R01AI136035‐01,R01GM109499andR01TW010286‐01

HandlingEditor:LukeFlory

Abstract1. Predictingwherecroppestsanddiseasescanoccur,bothnowandinthefutureunderdifferentclimatechangescenarios, isamajorchallengeforcropmanage-ment.Onesolutionistoestimatethefundamentalthermalnicheofthepest/dis-easetoindicatewhereestablishmentispossible.Here,wedevelopmethodsforestimatinganddisplayingthefundamentalthermalnicheofpestsandpathogensandapplythesemethodstoHuanglongbing(HLB),avector‐bornediseasethatiscurrentlythreateningthecitrusindustryworldwide.

2. WederiveasuitabilitymetricbasedonamathematicalmodelofHLBtransmissionbetweentreehostsanditsvectorDiaphorina citri,andincorporatetheeffectoftem-peratureonvectortraitsusingdatafromlaboratoryexperimentsperformedatdiffer-enttemperatures.WevalidatethemodelusingdataonthehistoricalrangeofHLB.

3. OurmodelpredictsthattransmissionofHLBispossiblebetween16and33°Cwithpeaktransmissionat~25°C.Thegreatestuncertaintyinoursuitabilitymetricisassoci-atedwiththemortalityofthevectorsatpeaktransmission,andfecundityattheedgesofthethermalrange,indicatingthattheseparametersneedfurtherexperimentalwork.

4. Weproduceglobalthermalnichemapsbyplottinghowmanymonthseachloca-tion is suitable forestablishmentof thepest/disease.Thisanalysis reveals thatthehighestsuitabilityforHLBoccursneartheequatorinlargecitrus‐producingregions,suchasBrazilandSouth‐EastAsia.WithintheNorthernHemisphere,theIberianpeninsulaandCaliforniaareHLBsuitableforupto7monthsoftheyearandarefreeofHLBcurrently.

5. Policy implications.Wecreateathermalnichemapwhichindicatestheplacesatgreatestriskofestablishmentshouldacropdiseaseorpestentertheseregions.

2058  |    Journal of Applied Ecology TAYLOR eT AL.

1  | INTRODUC TION

The quality and quantity of yields formany crop systems can besignificantlyreducedbypestsanddisease.Forexample,thewheataphidreducesyieldsofgraincrops(Merrill,Holtzer,Peairs,&Lester,2009),Pierce'sdiseasevectoredbytheglassy‐wingedsharpshooterdiminishes grape yields (Bruening, Kirkpatrick, Esser, & Webster,2014), codling moth damages apple orchards (Rafoss & Sæthre,2003),andHuanglongbing(HLB)vectoredbytheAsiancitruspsyllid(ACP)decimatescitruscrops(daGraçaetal.,2016).Thepestsand/orvectorsofdiseaseareoftensmallarthropodsthataresensitiveto environmental conditions, including temperature and humidity(Mordecaietal.,2013;Tsai,Wang,&Liu,2002).Therefore,predict-ingwhenandwherethesepestsordiseasevectorswilloccur,andhencepotentiallossofcrops,isoftenhighlydependentonenviron-mentalconditions,andthusachangingclimate.Somecropsareal-readyexperiencing reducedyields associatedwith climate change(Challinor et al., 2014), anundesiredoutcome that couldbe exac-erbatedifitcoincideswithincreasesinpestpopulationsordiseasetransmission(Cammell&Knight,1992).However,predictingrealisticimpactsofclimatechangeonlivingsystems,suchaspestsanddis-easevectors,isamajorchallengeinecology(Rohretal.,2011).

Withthepressingneedtounderstandtheeffectsofclimatechangeonfoodproduction,weprovideamethodtoestimateanddisplaythefundamentalthermalnicheofacroppestordisease.Thefundamentalthermalnicheisthesetoftemperaturesunderwhichpopulationsofaspecieswouldbeexpectedtopersist,allelsebeingequal (Angilletta&Sears,2011). Itcanbeusedtopredictwherepestsordiseasecancurrentlyestablishoutsideoftheirexistingrange,aswellaspredictfu-turelocationsofpestanddiseaseoutbreaksassociatedwithachangingclimate.This,inturn,canfacilitatetargetingrisk‐basedsurveillanceandprophylactic interventions.Consequently,ourmethod forestimatingthefundamentalthermalnicheofacroppestordiseaseshouldbeanimportanttoolforcurrentandfutureagriculturalplanning.

Inthisstudy,weborrowapproachesestablishedforhumanvec-tor‐bornediseases (Mordecaietal.,2017,2013) tosimultaneouslyestimate the fundamental thermal niche of a crop pathogen andits insectvector.More specifically,we firstdevelopamechanisticmodelofdiseasetransmissionwheretheparametersofthemodelarefittedtodatafromtemperature‐dependentlaboratorystudies.

Ametricderivedfromthismodelisthenusedtoindicatehowsuit-ablelocationsarebasedonaveragemonthlytemperatures.Finally,we validate themodel by assessing howwell it correspondswithobservational dataon knowndiseaseoccurrence (Mordecai et al.,2013).Additionally,weuseBayesianinferencetoincorporateuncer-taintyarisingfromtheuseofmultipledatasourcesanduncertaintyintemperaturedependenceofeachofthevector'slifehistorytraitstoestimatetheoveralluncertaintyinthesuitabilitymetric(Johnsonetal.,2015).Ourmethodallowsfortheinclusionoftheintrinsicrea-sonsbehindwhyanorganismisfoundwhereitisandthepotentialinterplaywhendifferenttraitsoforganismsresponddisparatelytochangesinextrinsicfactors(Angilletta&Sears,2011).

Todevelopourmethods,weleveragedataontheeffectoftempera-tureonbacteriaofcitrusthatcausethediseaseHLB(orcitrusgreening)andtheirprimaryvector,theACP(Diaphorina citriKuwayama).Citrusisacommerciallyimportantcropgrownthroughouttheworldinclud-ingSouth‐EastAsia,Australia,theMediterranean,SouthAfrica,SouthandCentralAmericaandsouthernstatesofUSA(FAO,2017).HLBisadevastatingdiseaseofcitrustreesthathasspreadgloballyfromitsorigininAsia(Bové,2006).Itaffectsthequalityandquantityofcitrusfruitonatree,forallcitrusspecies,leadingtomisshapenfruit,bittertaste, and fruit dropping early (Bové, 2006). The symptoms can bedifficulttodetect,andmaytakemonthstoappearonatree,but in-cludechlorosisofleaveswitheventualdiebackanddeathofthetree(Gottwald, 2010; Leeet al., 2015).HLB is causedby threebacteria:Candidatus Liberibacter asiaticus (CLas), Candidatus Liberibacter af-ricanus(CLaf),andCandidatusLiberibacteramericanus(CLam)(Bové,2006).Thepredominantbacterium,andthefocusinthisstudy,isCLas,which occurs in all HLB‐infected areas (Gottwald, 2010) apart fromSouthAfrica(whereCLafispresent).CLamoccursprimarilyinSouthAmericaalongsideCLas,andisresponsibleforonlyaminorityofcasesthere(Gottwald,2010).TransmissionofboththeCLasandCLambac-teriaoccursduetofeedingoftheACP(Grafton‐Cardwell,Stelinski,&Stansly,2013;Hall,Richardson,Ammar,&Halbert,2013).ACPandHLBhavespreadthroughouttheworldmostlyviaworldwidetrade(Byrneetal.,2018)andnowexistinnearlyallcitrusproducingregions(Halletal.,2013).Thecostofthisdiseasetothecitrusindustryishuge,andin-terventionstopreventitsspreadandreducethedeleteriouseffectsofthediseaseare,forthemostpart,ineffective(Hodges&Spreen,2012).Here,wemapthesuitabilitymetricofHLBaroundtheworldtoprovide

Thisindicateswheresurveillanceshouldbefocusedtopreventestablishment.OurmechanisticmethodcanbeusedtopredictnewareasforHuanglongbingtrans-missionunderdifferentclimatechangescenariosand iseasilyadapted toothervector‐bornediseasesandcroppests.

K E Y W O R D S

AsianCitrusPsyllid,Bayesianinference,cropmanagement,Huanglongbing,riskofestablishment,speciesdistributionmodels,transmissionsuitability,vector‐bornedisease

     |  2059Journal of Applied EcologyTAYLOR eT AL.

aplanningtoolforcitrusgrowers,whilstthemethoditselfisapplicabletoallcropdiseasesspreadbyvectorsortocroppests.

2  | MATERIAL S AND METHODS

2.1 | Formulation of S(T)

Todeterminethethermalnicheofcitrusgreening,wecharacterizethepossibilityofan introductionofcitrusgreeningpersistingatalocallevel,fordifferentlocationsaroundtheworld.Weuseanes-timateofthebasicreproductivenumberR0.WhenR0>1,thedis-easeis likelytospreadandleadtoanepidemic,whereasifR0<1,thediseasewill dieout.Wedetermine anequation forR0 for thespreadofcitrusgreeningonasinglegrovebyusingapreviouslyde-velopedmodelforcitrusgreening(Taylor,Mordecai,Gilligan,Rohr,&Johnson,2016).Inthismechanisticmodel,treesandpsyllidsaresplitinto different compartments based on their disease status. Treesare susceptible, asymptomatic (infected, but no symptoms) or in-fected(withsymptoms).Psyllidsaresusceptible,exposedorinfected(Figure1).Deathoftreesandroguing(removingtreesfromagroveduetohighlevelsofinfection)areincluded,aswellasreplacementof all removed treeswith susceptible trees. For full detailsof thismodel,seeAppendixS1.ThebasicreproductivenumberR0isthencalculatedas(Tayloretal.,2016):

ThisequationforR0canbeunderstoodbyconsideringhowdiseasepropagatesthroughthecitrussystem.Thenumberofpsyllids inthepopulationisdeterminedby:

whichincludesthefecundityofadultpsyllids(FE),theprobabilityofeggtoadultsurvival(pEA),thedevelopmentratefromeggtoadult(DP),themortalityrateofadultpsyllids(μ)andtheamountoftreesflushing(F)withinthegrovewhichvariesthroughouttheyear.These

adultpsyllidsare incontactwiththesingle infectedtreebasedonthebiterate(a)withaprobabilityoftransmissionfromtreetopsyllid(c).Thepsyllidsundergoanextrinsic incubationperiodbeforebe-cominginfectiousgivenbyrate(ϕ).Buttheycandieduringthistimewhichleadstotheterm

(

3�

3�+�

)3

forthenumberofpsyllidsthatsur-

vivetheextrinsicincubationperiod.Theseinfectiouspsyllidsareincontactwithsusceptibletrees(totalN),onceagainwithbiteratea andaprobabilityoftransmissionfrompsyllidtotree (b).Theterme−rτ represents theproportionof treesthatsurvivethe incubationperiod (τ)tobecomeinfectious.Thefinalcombinedtermintheequa-tiondetermineshowlongatreeisinfectious,duringboththeasymp-tomaticstageandtheinfectedstage,andincludestherateatwhichtreesdevelopsymptoms(γ),thedeathrateofasymptomatictrees(r)andthedeathrateofinfectedtrees(r1).

WedefineourmeasureofthermalsuitabilityforHLBasthevec-tor/infectioncomponentsofR0thatdependontemperature,T,only.Thatis,thesuitability,S(T),isgivenby:

whereCisaconstantthatscalesthemeansuitabilitytoliebetween0and1.Thus,thesuitabilityiszerowhentemperatureispredictedtofullyexcludetransmissionand1atmaximaltransmission.Intheresults,wewillprimarilyfocusonpredictionsbasedontwosuitabilityregimes:a “permissive” thermal niche corresponding to temperatures whereS(T)>0;anda“highlysuitable”thermalnichecorrespondingtotem-peraturessuchthatS(T)>0.75(i.e.thehighestquartileofsuitability).

2.2 | Bayesian fitting of thermal traits in S(T)

Ithasbeenwidelyrecognizedthatperformancetraitsofectotherms,such as survival, reproduction, and movement, exhibit unimodalresponses to temperature (Amarasekare & Savage, 2012, p. 134;Dell,Pawar,&Savage,2011).FollowingtheapproachintroducedinMordecaiet al. (2013),we fitunimodal temperature responses tolaboratorydatafortraitsofthevectorthatappearintheequationforS(T).ThesecurvescanthenbeinsertedintotheequationforS(T)to determine how transmission depends upon temperature.As in

(1)R0=

(

FEpEADPa2bcF

N�3

(

3�

3�+�

)3

e−r�(

1

�+ r+

�

(�+ r)r1

)

)1∕2

.

(2)V=FEpEADPF

�2

(3)S(T)=C

(

FE(T)pEA(T)DP(T)

�(T)3

)1∕2 (3�

3�+�(T)

)3∕2

,

F I G U R E 1  AschematicofthemodelforHLBtransmittedbetweentreesandpsyllids.Treesaresusceptible,asymptomatic,orinfected.Psyllidsaresusceptible,exposedorinfected.Deadandroguedtreesarereplacedbysusceptibletrees.Blackarrowsshowthetransitionsbetweenthecompartments.Orangedashedarrowsshowthenecessaryinteractionsbetweentreesandpsyllidstoobtaintransmission

Extrinsic Incuba�on period

Suscep�blePsyllid

InfectedTree

Suscep�bleTree

ExposedPsyllid

InfectedPsyllid

Naturaldeath RoguingReplacement

Transmission

Transmission

Death Death Death

Birth

Asymptoma�cTree

Naturaldeath

Naturaldeath

2060  |    Journal of Applied Ecology TAYLOR eT AL.

Johnsonetal.(2015),wetakeaBayesianapproachtofitting.Thus,wecanquantifytheuncertaintyinthetemperaturerelationshipofindividualcomponentsandexploretheemergentuncertaintyinS(T)overallthatisduetotheuncertaintyinthecomponents.CompletedetailsoftheapproacharepresentedinAppendixS2.

Here,wefocusonfourpsyllidtraitsforwhichthereexistsdataacrosssufficient temperaturestoquantify theresponses: fecundity(FE);probabilityofeggtoadultsurvival(pEA);averagelongevity(theinverseofthemortalityrate,1/μ);andthedevelopmentrateofpsyl-lidsfromeggsintoadults(DP).ThemajorityofthedatacomesfromLiuandTsai(2000),whichhasarangeof15°Cupto30°Cforallfourpa-rameters,withadditionaldataat10and33°Cforsomeoftheparam-eters.Hall,Wenninger,andHentz(2011)andHallandHentz(2014)provide data on high and low extreme temperatures that preventdevelopmentofthepsyllidsand/orleadtomortality.However,Halletal.(2011)alsoprovideinformationonfecundityoffemalepsyllidsacrossatemperaturerangeof11–41°C.Weincludebothdatasetsforfecundity,andconsiderwhethertheygeneratedifferentpredictions.Hereafter,LiuandTsai(2000)andHalletal.(2011)willbereferredtoasLT00andH11,respectively,and,whenreferringtotheS(T)outputcreatedbyeitherdataset,LT00S(T)andH11S(T)willbeused.

Forallsetsofdata,wefittwofunctionstodescribethemeanrelationshipbetweenthetraitandtemperature:quadratic,givinga symmetric relationship (f

(

T)

=qn(T−T0)(T−TM)); Brière, givingan asymmetric relationship (f

(

T)

= cT(T−T0)(T−TM))1∕2, Brière,

Pracros,Roux,&Pierre,1999).All responseswere fittedusingaBayesianapproach.Afterspecifyingthemeanrelationship,prob-abilitydistributionsappropriatetodescribethevariabilityaroundthismeanwere chosen, and priors specified (Appendix S2).Wechosepriors to limit parameters for theminimumandmaximumtemperature thresholds to approximate known limits to psyl-lidsurvival.For instance, theprioronthethermal limitswassetuniformlyovertheintervalfrom30to50°Ctoacknowledgethatexposingpsyllidstotheveryhightemperatureskillsthemalmostimmediately, while being wide enough to allow the laboratorydata primacy in the analysis. Priors for other parameters of themeanwerechosen tobe relativelyuninformativebut scaledap-propriatelyfortheresponse.Priorsonvarianceparametersweretypicallyalsochosentoberelativelyuninformative,althoughtheprecisespecificationvariedduetodifferencesinthescaleoftheresponsesandtoimprovemixingandconvergenceoftheMarkovChainMonteCarlosamplingscheme.Allresponseswerefittedinr(RDevelopmentCoreTeam,2008)usingthejags/rjagspackages(Plummer,2003,2013).AfterfittingbothquadraticandBriérere-sponses toeachdataset, thepreferred responsewaschosenviadevianceinformationcriterionasimplementedinrjags.

Once samples of the posterior distributions of parameters forthepreferredmodelwereobtained,thesewereusedtocalculatetheposteriorsamplesoftheresponseacrosstemperature.Then,ateachtemperature themean/medianof the responseandthe95%high-estposteriordensity(HPD)intervalwerecalculated.Theposteriorsamplesofeachtraitresponsewerethencombinedtocreatesam-plesfromtheoverallresponseofS(T) totemperature.Aswiththe

individualthermalresponses,theseposteriorsamplesofS(T)acrosstemperaturewereusedtocalculatethemeanand95%HPDintervalaroundthemeanofS(T).

2.3 | Uncertainty in response of S(T) to temperature

TheuncertaintyforeachparameterinS(T)iscalculatedusingthevari-ationinS(T)ateachtemperaturewhenallparameters,apartfromtheoneofinterest,areheldconstantattheirmeanposteriorvaluesforthattemperature.Wecalculatethe2.5and97.5quantilesoftheS(Tm)posteriordistributionandplotthedifferencebetweenthequantilesagainstTm,whereweuseTm to represent the temperaturewithallbaroneparameterheldattheirmeanvalues.WedothisforvaluesofTm inourtemperaturerange,andthenforallparameters inS(T).Thismethodindicateswhichparametershavethegreatestvariationateachtemperatureandhencewhichparametershavethegreatestuncertaintyatthattemperature.Thisallowsustoknowwhenoures-timateofS(T)isuncertainandwhichparameteriscausingthis.

2.4 | Mapping suitability across space

Tocommunicatethepotentialsuitabilityoftheworld'slandsurfacefortransmissionofHLB,wemappedthemonthsofsuitabilityasafunction ofmeanmonthly temperatures.Weused theWorldClimdataset(version1.4,www.worldclim.org)(Hijmans,Cameron,Parra,Jones, & Jarvis, 2005), which corresponds to a climate period of1960–1990,usedtorepresenta longtermperiodinthe20thcen-tury;thisappropriatelyrepresentsthedaterangeofthevalidationoccurrencedata,from1956to2014.Usingthis“baseline”longtermclimatedatareducesanybiasesthatmayarisefromrecentwarmingsignalsmis‐representingtheearlierpartofthevalidationdata.WetooktheposteriormeanS(T)curvesfortheH11andLT00modelsacrosstemperaturesandextractedthetemperaturescorrespondingtothetop25thpercentileoftheS(T)curve(i.e.S(T)>0.75,AppendixS5) for the highly suitable thermal niche, and temperatures cor-responding to the transmission limits (S(T) >0) for thepermissivethermalniche(Ryanetal.,2015).Thesevalueswereusedwiththeclimate models (see Section 2.4.1) to determine, on average, thenumberofmonthseachyearthateachpixeliseitherpermissiveorhighlysuitable.

2.4.1 | Climate models

Wemappedthesuitabilitymeasuresontorastersofcurrentmeanmonthly temperature data at 0.1°C intervals. Data were derivedfromWorldClim version 1.4 dataset at 5 min resolution (roughly10km2attheequator).Thescaledsuitabilitymodelwasprojectedontotheclimatedatausingtherasterpackage(Hijmans,2016)inR(RDevelopmentCoreTeam,2008).VisualizationsweregeneratedinArcMap.Foreachofthescenarios,wecreatedglobalmapsandin-setsforareasofcitrusgrowingconcernintheNorthernHemisphere:California,FloridaandtheIberianpeninsula.

     |  2061Journal of Applied EcologyTAYLOR eT AL.

2.5 | Validation of suitability measure

Narouei‐Khandan,Halbert,Worner,andBruggen(2016)presentspatially explicit data on locations with confirmed presence ofthe ACP or HLB (CLas form specifically). These are presence‐onlydata,sowecannotexaminehowwellourmodelpartitionspredictions between suitable andunsuitable areas. Instead,wefocusonamorequalitativeassessmentofmodeladequacy.Foreachlocationinthedataset,wecalculatethenumberofmonthsthattheaveragetemperaturefallswithintheboundsoftheper-missive or the highly suitable thermal ranges of ACP from ourmodel. If our model can adequately capture temperature con-ditions related to transmission and vector establishment, thenmost locations of HLB presencewill havemanymonths in thesuitablerange.Werestrictourselvestolocationsfromthedata-setthatarenotfrommountainousregions(i.e.exclusionofACP[n=26]andHLB [n=12]points)by removing thosecoincidingwithnamedmountainranges.Thisisbecause,atthespatialscaleweuse for climate layers, theseareas tend tohavemuchmoretrue variation in temperature, leading to higher uncertainty inthetemperaturepredictorthan inother locations.Thisremoval

alleviatestheriskofintroducingbiasineitherwarmerorcolderdirections.

3  | RESULTS

3.1 | Posterior distributions of thermal traits

Theprobabilityofeggtoadultsurvival(pEA)andlongevity(1/μ)arebothfittedbestbyquadraticcurves,whiledevelopmentratefromeggtoadult(DP)isbestfittedbyaBriérecurve(Figure2).However,fecundity(FE)switchesfromaBriéretoaquadraticdependingonwhether we use the data from Liu and Tsai (2000) or Hall et al.(2011)respectively(Figure2d,e).Thedatasourcesalsopredictdif-ferentupperthermallimitsforfecundity:LT00predictsnofecun-dityabove31°C,whereasfecundityispossibleupto41°CaccordingtoH11.FullposteriorplotsoftheparametersareinAppendixS3.

3.2 | Posterior distribution of S(T)

ThelowerthermalboundoftheposteriordistributionsofLT00andH11S(T)areinagreement,predictedusingthetwodifferentdatasets

F I G U R E 2  Psyllidtraitdataagainsttemperature(°C)withthebestfitplottedasasolidlineand95%quantilesasdashedlines.In(a),theprobabilityofeggtoadultsurvival(pEA);in(b),thedevelopmentratefromeggtoadultpsyllid(DP);in(c),thelongevityofadultpsyllids(1/μ);in(d),thefecundityofadultpsyllids(FE)withonlyLiuandTsai(2000)dataused;andin(e)thefecundityofadultpsyllids(FE)withonlytheHalletal.(2011)dataisused

0 10 20 30 40 50

0.0

0.2

0.4

0.6

0.8

1.0

pE

A

0 10 20 30 40 50

0.00

0.02

0.04

0.06

0.08

0.10

DP

–10 0 10 20 30 40 50

020

4060

8010

0

1µ

0 10 20 30 40 50

05

1015

2025

30

F E

0 10 20 30 40 50

050

100

150

F E

(a)

(d) (e)

(b) (c)

2062  |    Journal of Applied Ecology TAYLOR eT AL.

for fecundity, although LT00 S(T) hasmore uncertainty (Figure 3,AppendixS4).ThetemperatureatwhichthepeakofS(T)occursisalso very closely aligned.However, the value ofS(T) at that peaktemperatureandtheupperthermal limitareverydifferent.WhenscaledsothattheLT00S(T)hasamaximumof1,thepeakofH11S(T)is1.35timeslarger.LT00S(T) isdrivento0atapproximately31°Cbecausefecundityisnotpossibleforhighertemperatures.However,H11S(T)stillhasatransmissionpredictedupto33°C.

3.3 | Sources of uncertainty in S(T)

Theuncertaintyof eachparameteronS(T) is plottedagainst tem-perature as all other parameters are held constant at theirmeans(Figure4).WecanuseFigure4 tounderstandwhatdrivesS(T) atdifferenttemperatures,andthereforeitindicateshowbesttoaffectS(T)at thosetemperatures, if theaim is intervention. InFigure4a,fecundity(FE)isthemainparameterdrivingvariabilityinS(T)duringtherange15–20°CaswellaswhenS(T)isdecreasingto0at31°C.However, adultmortality (μ) is themain proponent of uncertaintyduringthemidtohightemperaturesof20–30°C.Incomparison,inFigure4b,adultmortality(μ)leadstothemostuncertaintyoverthewholetemperaturerange.Whileadultfecundityisonceagainimpor-tantatlowtemperatures,itisthedevelopmentrate(DP)thatemergesasproducingthemostvariabilityinS(T)athightemperatures.

3.4 | Validation

Wepresenthistogramsof thenumberofmonths that theaver-agetemperature fallswithin theboundsofboththepermissive

orthehighlysuitablethermalranges(Figure5,basedontheH11model) at each location in the dataset from Narouei‐Khandan etal.(2016).

F I G U R E 3  PosteriordistributionofS(T)againsttemperature(°C)usingdatafromLiuandTsai(2000)(LT00,inblue)andHalletal.(2011)(H11,inpink).MeanS(T)forbothmodelsisplottedusingsolidlines,95%credibleintervalsareplottedwithdashedlines.Bothareindependentlyscaledsothattheirmaximumis1

10 15 20 25 30 35 40

0.0

0.2

0.4

0.6

0.8

1.0

1.2

Temperature (°C)

S (T

)

LT00H11

F I G U R E 4  TheuncertaintyofS(T)propagatedthrougheachparameters’relationtotemperature.In(a)uncertaintyinLT00S(T)andin(b)uncertaintyinH11S(T).ThisisproducedbasedupontheposteriorofS(T),holdingallparametersconstantattheirpredictivemeanapartfromtheparameterofinterest.The2.5%and97.5%quantilesoftheresultantestimationofS(T)arethencalculatedateachtemperatureandthedifferencebetweenthesequantilesisplotted

Rel

ativ

e w

idth

of q

uant

iles

0.0

0.2

0.4

0.6

0.8

1.0

pEADP

µFE (LT00)

15 20 25 30 35

0.0

0.2

0.4

0.6

0.8

1.0

Temperature (°C)

Rel

ativ

e w

idth

of q

uant

iles

pEADP

µFE (H11)

(a)

(b)

     |  2063Journal of Applied EcologyTAYLOR eT AL.

Most locationswithHLBorACPhavepermissive temperaturerangesforatleast6monthsoftheyear.Over82%oflocationsareinthepermissiverangefornineormoremonthsoftheyear,andmorethan50%havepermissivetemperaturesyearround.Locationswithyearroundhighlysuitableconditionsaccountfor12%and28%ofrecords forHLBandACP, respectively, andover70%of locationshavehighlysuitableconditionsfor6monthsormore.Thereareal-mostnolocationswithHLBorACPpresentthattheconditionsarepermissiveforlessthan3months.ResultsbasedonthedatabyLiuandTsai(2000)aresimilar,asaretheresultsthatincludemountain-ousareas(seeAppendixS6).

3.5 | Thermal niche of HLB

InFigures6and7,wepresentthemappedoutputsofthepermis-siveandhighly suitable regions, respectively, forH11S(T). Similarmaps for the LT00 S(T) model are presented in Appendix S7 andmapswiththeHLBandACPvalidationpointsincludedareprovidedinAppendixS8.ManylocationsintheSouthernHemisphereareper-missiveforHLBallmonthsoftheyear(Figure6),includinginSouthAmericaandSouthwestAsiawherethediseaseiscurrentlypresent.Australiaandmanycountries inAfrica,whichare largecitruspro-ducing regions, arepermissive forHLBall ormanymonthsof theyear,butCLasHLBisnotcurrentlypresent.TheinsetshighlightthatsouthernFloridaispermissiveforHLBallmonthsoftheyear,andfor

at least7months inthenorth.This isconfirmedonthegroundasHLBispresentthroughoutthewholestateofFlorida.CaliforniaandtheIberianpeninsulahavesimilarsuitabilityprofilestoeachother,withupto7monthspermissiveinthesouthoftheIberianpeninsula,and up to 8months at the very southofCalifornia.As expected,morenortherlyregionsoftheworld,whicharenotsuitableforgrow-ingcitrus,arealsonotabletomaintainACPpopulations.

Thepatternforsuitability issimilarforthehighlysuitablemap(Figure7)butwithlowernumbersofmonthssatisfyingthisstrictercriteria.SouthAmericamaintainsyearroundhighsuitabilityforthedisease across much of the continent. Similarly, Southwest Asiamaintains suitability year‐round, whereas Australia is reduced to7monthsorlessacrossthecountry.CaliforniaandtheIberianpen-insulaarehighlysuitablefor4–5monthsoftheyear,whereasFloridaisstillhighlysuitableforupto9monthseachyear.

4  | DISCUSSION

Inthispaper,wepresentedamodelofsuitabilityforaspecificcropdisease,HLB.Thismethodofdemonstratingdurationofriskofpo-tentialdiseaseemergenceandtransmission,asafunctionofther-malsuitability,hasbeensuccessfullyusedforhumanvector‐bornediseasessuchasdengue,Zikaandmalaria(Mordecaietal.,2017,2013).Thisclimate‐predictivemappingframeworkprovidesatool

F I G U R E 5  ThenumberofmonthsthateverylocationwithcurrentHLBorACPpresenceiseitherpermissiveorhighlysuitableaccordingtoourH11S(T)model.Toprow:locationsinthevalidationdatasetwhereHLBispresent.Bottomrow:locationsinthevalidationdatasetwhereACPispresent.WedefinepermissivesuitabilityasS(T)>0andhighsuitabilityasS(T)>0.75

HLB − permissive

Months

Den

sity

0 2 4 6 8 10 12

0.0

0.1

0.2

0.3

0.4

0.5 99% 6+ mo

0.5% <4 mo

HLB − highly suitable

Months

Den

sity

0 2 4 6 8 10 12

0.00

0.10

0.20

72.8% 6+ mo

9.4% <4 mo

ACP − permissive

Months

Den

sity

0 2 4 6 8 10 12

0.0

0.1

0.2

0.3

0.4

0.5 97.4% 6+ mo

0% <4 mo

ACP − highly suitable

Months

Den

sity

0 2 4 6 8 10 120.

000.

100.

20

70.7% 6+ mo

12% <4 mo

2064  |    Journal of Applied Ecology TAYLOR eT AL.

for planning and intervention, and is adaptable to multiple sys-tems,includingvector‐bornediseasesofcropsandthermallysen-sitivecroppests.Whilewehavedemonstratedthisapproachforacoupledvector–pathogensystem,andthususedR0asastartingpointtocreateatransmission‐basedsuitabilitymetric,themethodtocreatea fundamental thermalniche for invasivecroppests is

readilypossibleusingpopulationmodels (inwhichR0 representsthe likelihoodofpopulationpersistencerather thandiseaseper-sistence).Giventheavailabilityofrigorouslaboratoryexperimentson the thermal responsesofother croppestordisease systems(Deutschetal.,2008),thisapproachisbroadlyapplicabletomanysystems.

F I G U R E 6  ThenumberofmonthsayearthatlocationshavepermissivetemperaturesaccordingtoourH11S(T)model.InsetplotsofCalifornia,FloridaandtheIberianpeninsula,respectively,areincluded.WedefinepermissivetemperaturesforsuitabilityasS(T)>0.LocationsingreyhavezeromonthssuitableforHLBtransmission

F I G U R E 7  ThenumberofmonthsayearthatlocationshavehighlysuitabletemperaturesaccordingtoourH11S(T)model.InsetplotsofCalifornia,FloridaandtheIberianpeninsula,respectively,areincluded.WedefinehighlysuitabletemperaturesasS(T)>0.75.LocationsingreyhavezeromonthssuitableforHLBtransmission

     |  2065Journal of Applied EcologyTAYLOR eT AL.

Ourmodeloutputsa suitabilitymetricS(T) for transmissionofHLBdependentontemperature.ConditionalonthedataweusetoparameterizeS(T),wepredictthattransmissioncanoccurbetween16°Cand30–33°Cwithpeaktransmissionatapproximately25°C.Whilethelowerboundandpeaktemperaturepredictionsaresimilarregardless ofwhichof the two fecundity datasetsweuse for pa-rameterization,i.e.thosefromHalletal.(2011)(H11)orLiuandTsai(2000) (LT00), thepredictedupper limitofsuitabilityof thepeaksdifferdependingonthedataused(Figure3).Morespecifically,therangeiswiderfortheH11data.Thus,basedonthesuitabilityindex,theremightbemoreareasstrictlysuitablefortransmissionifweas-sumetheH11dataaremore representativeofpsyllidpopulationsinthefield.Furthermore,whenweconsiderouruncertaintyanaly-sis,thereareadditionaldifferencesbetweenthetwodatasources(Figure4).ThemainuncertaintyintheLT00modelarisesfromfe-cundity at lower temperatures andmortality for higher tempera-tures,whereas for theH11model,mortality is themain driver ofuncertaintyoverall.Inbothcases,mortalityofACPisthemostim-portantparameterwhenS(T)isnearitspeakat25°C.Together,theseindicatethatfocusshouldbeonfurtherexperimentalunderstandingofmortalityratesandfecundityneartheedgesofthethermaltoler-ancesofpsyllidstorefineourestimatesofthethermalniche.

Weuseoursuitabilitymetrictopredictregionswhicharepermis-siveorhighlysuitableforHLBaroundtheworld.OurmapsindicatethatregionsclosetotheequatorhavethegreatestnumberofmonthspermissiveforHLBtransmission,especiallyinSouthAmerica,AfricaandSouthEastAsia.SouthAmericaandSouthEastAsiaare,inpar-ticular,largecitrusproducingregions,andHLBisalreadypresentinboth(Coletta‐Filhoetal.,2004;Garnier&Bové,1996,2000;Torres‐Pacheco et al., 2013). The fact thatHLB is not only permissive all12monthsoftheyear,butisactuallyhighlysuitableyear‐round,indi-catesthescaleofthepotentialprobleminthoseareas.However,theHLBepidemicinSãoPauloStateinBrazil,themajorcitrus‐producingregioninthecountry,issuccessfullymanagedinthoseareasthathaveadheredtostrictrecommendationsforcontrol(Belasqueetal.,2010).Thediseaseenteredthestatein2004(Coletta‐Filhoetal.,2004).By2012,incidenceofdiseasewasestimatedaslowas1%forsymptom-atictreesacrossathirdofthecitrusacreageinthestate(Bové,2012).Incomparison, inFlorida,wherethediseasewasfirstdiscoveredin2005,growersinasurveyin2015wereaskedtoestimateboththepercentageof theircitrusacreswithat leastonetree infectedandthepercentageofalltheircitrustreesinfected,withresultsindicat-ing90%and80%,respectively(Singerman&Useche,2016).Thisisdespite the fact that our suitabilitymetric estimates the northernregionsofFloridatobehighlysuitableforHLBonly6monthsoftheyear.WhileBrazilmighthavemanagedtocontrolHLBsuccessfullyinsomeregions, thedisease is still spreading throughout thecountryin those regionswhichhavenotbeen asproactive in their control(Belasqueetal.,2010);areminderthat,withoutstrictcontrolmea-sures,thediseasecanspreadquicklyanddevastatingly,with100%HLBincidencepossibleininfectedgroves(Bové,2012).

Within the Northern Hemisphere, we highlight the suitabil-ityofCalifornia and the Iberianpeninsula forHLB transmission.

Although California has had incursions of the disease, with thefirstoccurringin2012(Kumagaietal.,2013),allhavebeenintreesatresidentialpropertiesandhencethecitrusindustryinCaliforniais currently free from disease (Byrne et al., 2018). Similarly, theIberianpeninsulahashadnocasesofHLB(Cocuzzaetal.,2017).However,both regionshavehighcitrusproductionand thus thepotential consequencesof incursionofHLBarehigh.TheyhavesimilarsuitabilityprofileswithpermissivesuitabilityofHLBonav-erageabout6monthsoftheyear inbothregions.Whilediseasetransmissionmightnotbepermissibleallyearround,treescanre-maininfectedwithHLBindefinitelyunlesstheyarerogued,thusallowing over‐wintering of the disease during the seasons thatACPwillnotbeactive(Gottwald,2010).Therefore,forthesetworegionstomaintainHLB‐freestatus,theyneedtodealwithincur-sions promptly to ensure infected trees are removed. However,Californiahastheaddeddisadvantagethatitsneighbouringcoun-tryandmanyneighbouringstateshave thediseaseorhaveACPpresent(Torres‐Pachecoetal.,2013).Indeed,theinitialincursionofACPintoCaliforniaismostlikelytohaveoccurredfromMexico(Bayles,Thomas,Simmons,Grafton‐Cardwell,&Daugherty,2017).ThismakesithardertoreducethelikelihoodofHLBincursionasitisdifficulttocontrolthemovementofvectorsacrossborders.FortheIberianpeninsula,ithasbeensuggestedthatincursionofHLBismostlikelythroughcontaminatedtrade,suchasinfectedplantmaterials(Cocuzzaetal.,2017).

Our analysis hasbeenperformed for the transmissionof theCLasformofHLBbythevectorACPandthusdoesnotquantifythe spread of CLaf HLB around the world from the African cit-ruspsyllid(AfCP).Therefore,thesuitabilityforHLBtransmissionmightbeunderestimated in someareasof theworld sinceAfCPhasadifferenttemperatureprofilethanACP,as itprefershigherelevationsandlowertemperaturesandthetwopsyllidshavenotbeenfoundinthesamelocations(daGraçaetal.,2016).In2014,AfCPwas first discovered inmainlandSpain, alarming the citrusindustry in Spain and Portugal (Cocuzza et al., 2017). Thus, it ispossiblethattheIberianpeninsulaispermissiveforHLBformoremonthsoftheyearthanwehavepredictedifAfCPistheprimaryvectorthere.

WehavevalidatedthismodelusingspatiallyexplicitrecordsofHLBandACPpresence.MostareaswithconfirmedHLBorACPareinregionsourmodelpredictsaspermissiveorhighlysuitableformostof theyear, indicatingthatourtemperature‐onlymodelcan capture an important component of the environment thatconstrainsthespatialdistributionofHLB.Narouei‐Khandanetal.(2016)usedspeciesdistributionmodellingwithclimaticvariablesto also create aHLB nichemodel, finding that annual precipita-tionlevels,resultinginhigherhumidity,arethegreatestpredictorofHLBpresencearound theworld.Unfortunately, there arenotenough laboratory experiments assessing the effect of humidityon psyllid life history traits for us to include this in our model.Whilst our validation results indicate that we have successfullycharacterizedasignificantcomponentofHLBtransmissionusingtemperature, our predictions could undoubtedly improve if we

2066  |    Journal of Applied Ecology TAYLOR eT AL.

alsoincludedtheeffectsofhumidity.Ourmodelproducessimilarresults to themodel ofNarouei‐Khandan et al. (2016), althoughitisdifficulttomakecomparisonsbetweennumberofpermissivemonths(ourmodel)versustheprobabilityofoccurrence(Narouei‐Khandan et al., 2016model). TheNarouei‐Khandan et al. (2016)modelpredictsgreaterareasinAustraliasuitableforHLBandACPestablishmentthanourmodelbutonlycoastalareasofCaliforniaare predicted to have a high probability ofHLP andACP occur-rencewitha lowprobabilityelsewhere inCalifornia. In contrast,ourmodelpredictsupto7monthsofpermissivesuitabilityacrossmuchofCalifornia.

Ourmethodforcreatingthethermalnicheisadaptabletoothercrop diseases and pests due to its strength of being built usingmechanisticmodels.Spatiallyexplicitdataofdiseasepresencearetypically used to build ecological niche models in other contexts(Gething et al., 2011; Peterson, Martínez‐Campos, Nakazawa, &Martínez‐Meyer,2005).Amechanisticmodelwithadditionalon‐the‐groundvalidationislikelytopredictmorerobustlyhowtemperatureconstrainstransmissionthanacorrelativeapproachbasedonpres-ence‐onlydata. Italsoenablescleanprojections for futureclimatescenarios,asitisnotlimitedbyunquantifiablechangesinlandcover.

Ouruseoftwodatasetsforoneparameterhighlightshowourunderstanding of a disease/pest population can change signifi-cantly depending on the datawe use to parameterize ourmod-els.Thebestway to avoid this and reduceouruncertainty is touse multiple data sources combined, but this requires multipleexperimentsfromdifferent laboratoriesestimatingthesamepa-rameteracrossarangeoftemperaturesusingthesameempiricalapproaches.Oftenmultipleexperimentslikethisdonottakeplacebecauseofaperceivedlackofnovelty,butasweshowhere,theyarepotentially important to fullyunderstandthe impactof tem-perature on the persistence and establishment of vector‐bornediseasesandpestpopulations.

Asourmappingofsuitabilityisperformedatthepixellevel,ap-proximately10km2attheequator,thisallowsustopredictsuitabil-itytoaveryfinescale.Therefore,ourmapforHLBsuitabilitycanbeusedasatooltodeterminesurveillanceandmanagementstrat-egiesatafinespatialscale.Furthermore,thegeneralmethodisap-plicableforothervector‐bornecropdiseasesorpests.Suitabilitydoes not indicate where incursions of the disease are likely tooccur,but itdoeshighlight the regionswhere it ismost likely toestablishandthereforewhereitismostnecessarytoavoidincur-sion. For example, for HLB in California, surveillance should betargetedmostlytowardsthesouthernpartofthestate(Figure6).Furthermore,forthosecountrieswithdiseasepresentalready,thesuitabilitymapscanindicatewhichregionsshouldhavedifferentmanagementaims:astrictmanagementpolicytokeep incidencelevelsloworcompleteeradicationinregionswithlowersuitabil-ity. A more focused surveillance and management strategy cansave time,moneyand resources,which isnecessaryconsideringtheeconomiccostscurrentlyinvolvedinmanagingcropdiseases(Challinoretal.,2014).SãoPauloState,Brazil,demonstratesthatitispossibletokeepincidenceofHLBlow,eveninaregionwhich

ishighlysuitableforHLBall12monthsoftheyear.Vera‐Villagránetal. (2016)estimated theeconomicbenefitsof implementingaBrazilianstrategyinMexicoandfoundthatitwascost‐effective,assuming all growers abide by the regulations. The costs of im-plementingsuchastrictcontrolstrategymaybeprohibitive,butitgiveshopefortheindustrythatcontrolispossible,especiallyifimplementedassoonasHLBisdiscovered(Belasqueetal.,2010).Similarly,forothervector‐bornediseasesofcropsandcroppests,successfulmanagement and control are possible if implementedquicklyandextensivelyafterdiseaseorpestemergence(Brueningetal.,2014;Enkerlinetal.,2015).Overall,oursuitabilitymapspro-videanadditionaltool,alongsidemodellingofinterventionstrate-gies,cost–benefitanalysis,experimentalstudies,developmentofdisease‐resistant trees andother inventions, in the fight againstvector‐bornecropdiseasesandpests.

ACKNOWLEDG EMENTS

L.R.J.,S.J.R.,C.A.L.andJ.R.R.weresupportedbytheNationalScienceFoundation (DEB‐1518681; https://nsf.gov/). S.J.R. was also sup-portedbytheNationalInstitutesofHealth(R01AI136035‐01;https://www.nih.gov/).L.R.J.wassupportedbyNSF(DMS/DEB‐1750113).J.R.R. was supported by theNSF (EF‐1241889 and IOS‐1754868),NationalInstitutesofHealth(R01GM109499andR01TW010286‐01),andUSDepartmentofAgriculture(2009‐35102‐0543;https://www.usda.gov/wps/portal/usda/usdahome).

AUTHORS' CONTRIBUTIONS

R.A.T., S.J.R., J.R.R. and L.R.J. conceived and designed the study.R.A.T. created the model and performed the Bayesian analysis.C.A.L. and L.R.J. performed the validation, S.J.R. performed thespatialmapping.H.A.N.andD.G.H.provideddata.R.A.T.,S.J.R.andL.R.J. wrote themanuscript. All authors reviewed themanuscriptandgaveapprovalforpublication.

DATA AVAIL ABILIT Y S TATEMENT

CodeavailableviaZenodohttps://doi.org/10.5281/zenodo.3235271(Tayloretal.,2019).

ORCID

Rachel A. Taylor https://orcid.org/0000‐0002‐2739‐6944

Sadie J. Ryan https://orcid.org/0000‐0002‐4308‐6321

Leah R. Johnson https://orcid.org/0000‐0002‐9922‐579X

R E FE R E N C E S

Amarasekare,P.,&Savage,V. (2012).A framework forelucidating thetemperature dependence of fitness. The American Naturalist, 179,178–191.https://doi.org/10.1086/663677

     |  2067Journal of Applied EcologyTAYLOR eT AL.

Angilletta,M.J.Jr,&Sears,M.W.(2011).Coordinatingtheoreticalandempirical efforts to understand the linkages between organismsandenvironments.Integrative and Comparative Biology,51,653–661.https://doi.org/10.1093/icb/icr091

Bayles,B.R.,Thomas,S.M.,Simmons,G.S.,Grafton‐Cardwell,E.E.,&Daugherty,M.P. (2017).Spatiotemporaldynamicsof thesouthernCaliforniaAsian citrus psyllid (Diaphorina citri) invasion.PLoS ONE,12,e0173226.https://doi.org/10.1371/journal.pone.0173226

Belasque,J.,Bassanezi,R.B.,Yamamoto,P.T.,Ayres,A.J.,Tachibana,A.,Violante,A.R.,…Bové, J.M. (2010).Lessons fromHuanglongbingmanagementinSãoPaulostate,Brazil.Journal of Plant Pathology,92,285–302.

Bové,J.M.(2006).Huanglongbing:Adestructive,newly‐emerging,cen-tury‐olddiseaseofcitrus.Journal of Plant Pathology,88,7–37.

Bové,J.M.(2012).HuanglongbingandthefutureofcitrusinSãoPaulostate,Brazil.Journal of Plant Pathology,94,465–467.

Brière,J.F.,Pracros,P.,LeRoux,A.Y.,&Pierre,J.S.(1999).Anovelratemodeloftemperature‐dependentdevelopmentforarthropods.Environmental Entomology,28,22–29.https://doi.org/10.1093/ee/28.1.22

Bruening,G.,Kirkpatrick,B.,Esser,T.,&Webster,R. (2014).Managingnewly established pests: Cooperative efforts contained spread ofPierce’sdiseaseandfoundgeneticresistance.California Agriculture,68,134–141.https://doi.org/10.3733/ca.v068n04p134

Byrne,F.J.,Grafton‐Cardwell,E.E.,Morse,J.G.,Olguin,A.E.,Zeilinger,A. R.,Wilen, C., … Daugherty,M. P. (2018). Assessing the risk ofcontainerizedcitruscontributing toAsiancitruspsyllid (Diaphorina citri) spread inCalifornia:Residence timesand insecticide residuesat retail nursery outlets. Crop Protection, 109, 33–41. https://doi.org/10.1016/j.cropro.2018.02.024

Cammell,M.E.,&Knight,J.D.(1992).Effectsofclimaticchangeonthepopulationdynamicsofcroppests.Advances in Ecological Research,22,117–162.

Challinor,A.J.,Watson,J.,Lobell,D.,Howden,S.,Smith,D.,&Chhetri,N. (2014).Ameta‐analysisof cropyieldunder climate changeandadaptation.Nature Climate Change,4,287.https://doi.org/10.1038/nclimate2153

Cocuzza, G. E. M., Alberto, U., Hernández‐Suárez, E., Siverio, F.,Di Silvestro, S., Tena, A., & Carmelo, R. (2017). A review onTrioza erytreae (Africancitruspsyllid),nowinmainlandEurope,and its potential risk as vector ofHuanglongbing (HLB) in cit-rus. Journal of Pest Science, 90, 1–17. https://doi.org/10.1007/s10340‐016‐0804‐1

Coletta‐Filho,H.D.,Targon,M.L.P.N.,Takita,M.A.,DeNegri, J.D.,Pompeu,J.Jr,Machado,M.A.,…Muller,G.W.(2004).FirstreportofthecausalagentofHuanglongbing(“CandidatusLiberibacterasiati-cus’’)inBrazil.Plant Disease,88,1382–1382.

daGraça,J.V.,Douhan,G.W.,Halbert,S.E.,Keremane,M.L.,Lee,R.F., Vidalakis, G., & Zhao, H. (2016). Huanglongbing: An overviewof a complex pathosystem ravaging the world’s citrus. Journal of Integrative Plant Biology, 58, 373–387. https://doi.org/10.1111/jipb.12437

Dell,A.I.,Pawar,S.,&Savage,V.M.(2011).Systematicvariationinthetemperature dependence of physiological and ecological traits.Proceedings of the National Academy of Sciences of the United States of America,108,10591–10596.https://doi.org/10.1073/pnas.1015178108

Deutsch,C.A.,Tewksbury,J.J.,Huey,R.B.,Sheldon,K.S.,Ghalambor,C.K.,Haak,D.C.,&Martin,P.R.(2008).Impactsofclimatewarmingonterrestrialectothermsacrosslatitude.Proceedings of the National Academy of Sciences of the United States of America,105,6668–6672.https://doi.org/10.1073/pnas.0709472105

Enkerlin, W., Gutiérrez‐Ruelas, J. M., Cortes, A. V., Roldan, E. C.,Midgarden,D., Lira, E., …Arriaga, F. J. T. (2015). Area freedom inMexico from Mediterranean fruit fly (Diptera: Tephritidae): A re-viewofover30yearsofasuccessfulcontainmentprogramusingan

integrated area‐wide SIT approach. Florida Entomologist, 98, 665–681.https://doi.org/10.1653/024.098.0242

FAO (2017). Food and agriculture organization of theUnitedNations.FAOSTAT.Retrievedfromhttp://www.fao.org/faostat

Garnier, M., & Bové, J. M. (1996). Distribution of the Huanglongbing(greening) Liberobacter species in fifteenAfrican andAsian coun-tries. International Organization of Citrus Virologists ConferenceProceedings(1957–2010)(Vol.13).

Garnier, M., & Bové, J. M. (2000). Huanglongbing in Cambodia, Laosand Myanmar. International Organization of Citrus VirologistsConferenceProceedings(1957–2010)(Vol.14).

Gething, P. W., Patil, A. P., Smith, D. L., Guerra, C. A., Elyazar, I. R.,Johnston, G. L., … Hay, S. I. (2011). A new world malaria map:Plasmodiumfalciparumendemicityin2010.Malaria Journal,10,378.https://doi.org/10.1186/1475‐2875‐10‐378

Gottwald,T.R.(2010).CurrentepidemiologicalunderstandingofcitrusHuanglongbing.Annual Review of Phytopathology,48,119–139.https://doi.org/10.1146/annurev‐phyto‐073009‐114418

Grafton‐Cardwell,E.E.,Stelinski,L.L.,&Stansly,P.A. (2013).BiologyandmanagementofAsiancitruspsyllid,vectoroftheHuanglongbingpathogens.Annual Review of Entomology,58, 413–432. https://doi.org/10.1146/annurev‐ento‐120811‐153542

Hall, D. G., & Hentz, M. G. (2014). Asian Citrus Psyllid (Hemiptera:Liviidae) tolerance to heat. Annals of the Entomological Society of America,107,641–649.https://doi.org/10.1603/AN13169

Hall,D.G.,Richardson,M.L.,Ammar,E.D.,&Halbert,S.E.(2013).Asiancitrus psyllid,Diaphorina citri, vector of citrus Huanglongbing dis-ease.Entomologia Experimentalis et Applicata,146,207–223.

Hall,D.G.,Wenninger,E.J.,&Hentz,M.G.(2011).TemperaturestudieswiththeAsiancitruspsyllid,Diaphorina citri:Coldhardinessandtem-peraturethresholdsforoviposition.Journal of Insect Science,11,83.

Hijmans,R.J.(2016).raster:GeographicDataAnalysisandModeling.Rpackageversion2.5‐8.

Hijmans,R.J.,Cameron,S.E.,Parra,J.L.,Jones,P.G.,&Jarvis,A.(2005).Very high resolution interpolated climate surfaces for global landareas. International Journal of Climatology, 25, 1965–1978. https://doi.org/10.1002/joc.1276

Hodges, A. W., & Spreen, T. H. (2012). Economic impacts of citrusgreening(HLB)inFlorida:2006/07–2010/11.Tech.rep.,FoodandResource Economics Department, Florida Cooperative ExtensionService, Institute of Food and Agricultural Sciences, University ofFlorida.

Johnson,L.R.,Ben‐Horin,T.,Lafferty,K.D.,McNally,A.,Mordecai,E.,Paaijmans,K.P.,…Ryan,S.J.(2015).Understandinguncertaintyintemperatureeffectsonvector‐bornedisease:ABayesianapproach.Ecology,96,203–213.https://doi.org/10.1890/13‐1964.1

Kumagai, L.B., LeVesque,C. S.,Blomquist,C. L.,Madishetty,K.,Guo,Y., Woods, P. W., … Polek, M. (2013). First report of Candidatus Liberibacter asiaticus associated with citrus Huanglongbing inCalifornia.Plant Disease,97,283.

Lee,J.A.,Halbert,S.E.,Dawson,W.O.,Robertson,C.J.,Keesling,J.E.,&Singer,B.H.(2015).AsymptomaticspreadofHuanglongbingandimplicationsfordiseasecontrol.Proceedings of the National Academy of Sciences of the United States of America,112,7605–7610.https://doi.org/10.1073/pnas.1508253112

Liu, Y.H.,& Tsai, J.H. (2000). Effects of temperature on biology andlife table parameters of the Asian citrus psyllid, Diaphorina citri Kuwayama (Homoptera: Psyllidae). Annals of Applied Biology, 137,201–206.https://doi.org/10.1111/j.1744‐7348.2000.tb00060.x

Merrill,S.C.,Holtzer,T.O.,Peairs,F.B.,&Lester,P.J.(2009).Modelingspatial variation of Russian wheat aphid overwintering populationdensitiesinColoradowinterwheat.Journal of Economic Entomology,102,533–541.https://doi.org/10.1603/029.102.0210

Mordecai, E. A., Cohen, J. M., Evans, M. V., Gudapati, P., Johnson,L. R., Lippi, C.A.,…Weikel,D. P. (2017).Detecting the impact of

2068  |    Journal of Applied Ecology TAYLOR eT AL.

temperatureontransmissionofZika,dengue,andchikungunyausingmechanisticmodels.PLoS Neglected Tropical Diseases,11,e0005568.https://doi.org/10.1371/journal.pntd.0005568

Mordecai,E.A.,Paaijmans,K.P.,Johnson,L.R.,Balzer,C.,Ben‐Horin,T.,Moor,E.,…Lafferty,K.D.(2013).Optimaltemperatureformalariatransmissionisdramaticallylowerthanpreviouslypredicted.Ecology Letters,16,22–30.https://doi.org/10.1111/ele.12015

Narouei‐Khandan,H.A.,Halbert,S.E.,Worner,S.P.,&vanBruggen,A.H.C.(2016).GlobalclimatesuitabilityofcitrusHuanglongbinganditsvector,theAsiancitruspsyllid,usingtwocorrelativespeciesdis-tributionmodelingapproaches,withemphasisontheUSA.European Journal of Plant Pathology, 144, 655–670. https://doi.org/10.1007/s10658‐015‐0804‐7

Peterson,A.T.,Martínez‐Campos,C.,Nakazawa,Y.,&Martínez‐Meyer,E. (2005). Time‐specific ecological nichemodeling predicts spatialdynamicsofvectorinsectsandhumandenguecases.Transactions of the Royal Society of Tropical Medicine and Hygiene,99,647–655.https://doi.org/10.1016/j.trstmh.2005.02.004

Plummer,M.(2003).Jags:Aprogramforanalysisof{B}ayesiangraphicalmodelsusingGibbssampling.In:Proceedingsofthe3rdInternationalWorkshoponDistributedStatisticalComputing(DSC2003).March.pp. 20–22.

Plummer,M. (2013). rjags: Bayesian graphicalmodels usingMCMC.Rpackageversion3.10.

R Development Core Team (2008). R: A language and environment for statistical computing. Vienna, Austria: R Foundation for StatisticalComputing.

Rafoss,T.,&Sæthre,M.G.(2003).Spatialandtemporaldistributionofbioclimaticpotential for thecodlingmothandthecoloradopotatobeetle in norway:Model predictions versus climate and field datafromthe1990s.Agricultural and Forest Entomology,5,75–86.https://doi.org/10.1046/j.1461‐9563.2003.00166.x

Rohr, J.R.,Dobson,A.P., Johnson,P.T.,Kilpatrick,A.M.,Paull,S.H.,Raffel,T.R.,…Thomas,M.B.(2011).Frontiersinclimatechange–dis-easeresearch.Trends in Ecology & Evolution,26,270–277.https://doi.org/10.1016/j.tree.2011.03.002

Ryan,S. J.,McNally,A., Johnson,L.R.,Mordecai,E.A.,Ben‐Horin,T.,Paaijmans,K.,&Lafferty,K.D. (2015).Mappingphysiological suit-ability limits for malaria in Africa under climate change. Vector‐Borne and Zoonotic Diseases,15,718–725.https://doi.org/10.1089/vbz.2015.1822

Singerman,A.,&Useche,P. (2016). Impactofcitrusgreeningoncitrusoperations in Florida. University of Florida/Institute of Food andAgriculturalSciencesExtension.FE983.

Taylor,R.A.,Mordecai,E.A.,Gilligan,C.A.,Rohr,J.R.,&Johnson,L.R.(2016).MathematicalmodelsareapowerfulmethodtounderstandandcontrolthespreadofHuanglongbing.PeerJ,4,e2642.

Taylor,R.A.,Ryan,S. J., Lippi,C.A.,Hall,D.G.,Narouei‐Khandan,H.A.,Rohr,J.R.,&Johnson,L.R. (2019).Predictingthefundamentalthermalnicheofcroppestsanddiseasesinachangingworld:Acasestudyoncitrusgreening:Rcode.Zenodo,https://doi.org/10.5281/zenodo.3235271

Torres‐Pacheco, I., López‐Arroyo, J. I., Aguirre‐Gómez, J.A.,Guevara‐González,R.G.,Yänez‐López,R.,Hernández‐Zul,M. I.,&Quijano‐Carranza,J.A.(2013).PotentialdistributioninMexicoofDiaphorina citri (Hemiptera: Psyllidae) vector of Huanglongbing pathogen.Florida Entomologist,96,36–47.

Tsai, J.H.,Wang, J. J.,&Liu,Y.H. (2002). Seasonal abundanceof the{A}sian citrus psyllid, Diaphorina citri (Homoptera: Psyllidae) insouthern Florida. Florida Entomologist, 85, 446–451. https://doi.org/10.1653/0015‐4040(2002)085[0446:SAOTAC]2.0.CO;2

Vera‐Villagrán,E.,Sagarnaga‐Villegas,L.M.,Salas‐González,J.M.,Leos‐Rodríguez,J.A.,DeDeMiranda,S.H.G.,&Adami,A.C.D.O.(2016).Economic impact analysis for combating HLB in key lime citrusgrovesinColima,Mexico,assumingtheBrazilianapproach.Custos E Agronegócio Online,12,344–363.

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupportingInformationsectionattheendofthearticle.

How to cite this article:TaylorRA,RyanSJ,LippiCA,etal.Predictingthefundamentalthermalnicheofcroppestsanddiseasesinachangingworld:Acasestudyoncitrusgreening.J Appl Ecol. 2019;56:2057–2068. https://doi.org/10.1111/1365‐2664.13455