Evolutionary Applications. 2018;11:1231–1244.  | 1231wileyonlinelibrary.com/journal/eva

Received:25July2017  |  Accepted:22January2018DOI: 10.1111/eva.12618

O R I G I N A L A R T I C L E

How do phenology, plasticity, and evolution determine the fitness consequences of climate change for montane butterflies?

Joel G. Kingsolver1  | Lauren B. Buckley2

ThisisanopenaccessarticleunderthetermsoftheCreativeCommonsAttributionLicense,whichpermitsuse,distributionandreproductioninanymedium,providedtheoriginalworkisproperlycited.©2018TheAuthors.Evolutionary ApplicationspublishedbyJohnWiley&SonsLtd

1DepartmentofBiology,UniversityofNorthCarolina,ChapelHill,NC,USA2DepartmentofBiology,UniversityofWashington,Seattle,WA,USA

CorrespondenceJoelG.Kingsolver,DepartmentofBiology,UniversityofNorthCarolina,ChapelHill,NC,USA.Email:jgking@bio.unc.edu

Funding informationUSNationalScienceFoundation,Grant/AwardNumber:DEB-1120062andIOS-152767

AbstractSpecieshave responded toclimatechangevia seasonal (phenological) shifts,mor-phologicalplasticity,andevolutionaryadaptation,buthowtheseresponsescontrib-utetochangesandvariationinpopulationfitnessarepoorlyunderstood.Weassesstheinteractionsandrelativeimportanceoftheseresponsesforfitnessinamontanebutterfly,Colias eriphyle,alonganelevationalgradient.Becauseenvironmentaltem-peraturesaffectdevelopmentalratesofeachlifestage,populationsalongthegradi-entsdifferinphenologicaltimingandthenumberofgenerationseachyear.Ourfocalphenotype, wing solar absorptivity of adult butterflies, exhibits local adaptationacrosselevationandrespondsplasticallytodevelopmentaltemperatures.Weinte-grateclimaticdataforthepasthalf-centurywithmicroclimate,developmental,bio-physical, demographic, and evolutionary models for this system to predict howphenology,plasticity,andevolutioncontribute tophenotypicand fitnessvariationalongthegradient.Wepredictthatphenologicaladvancementsincompletelycom-pensate for climatewarming, and also influencemorphological plasticity. Climatechangeispredictedtoincreasemeanpopulationfitnessinthefirstseasonalgenera-tionathighelevation,butdecreasemeanfitnessinthesummergenerationsatlowelevation.Phenologicalshifts reducethe interannualvariation indirectionalselec-tionandmorphology,butdonothaveconsistenteffectsonvariationinmeanfitness.Morphologicalplasticityanditsevolutioncansubstantiallyincreasepopulationfit-nessandadaptationtoclimatechangeatlowelevations,butenvironmentalunpre-dictability limits adaptive plastic and evolutionary responses at high elevations.Phenological shiftsalsodecrease the relative fitnessadvantagesofmorphologicalplasticityandevolution.Ourresultsillustratehowthepotentialcontributionsofphe-nologicalandmorphologicalplasticityandofevolutiontoclimatechangeadaptationcanvaryalongenvironmentalgradientsandhowenvironmentalvariabilitywilllimitadaptiveresponsestoclimatechangeinmontaneregions.

K E Y W O R D S

adaptation,climatechange,ectotherms,evolution,fitnessconsequences,phenology,plasticity

1232  |     KINGSOLVER aNd BUCKLEY

1  | INTRODUC TION

Organismshave respondedecologically to recentandongoingcli-matechangesvia shifts inpopulation size, geographicdistributionand range boundaries, seasonal timing, and species interactions(Parmesan, 2006). Many of these changes are mediated by phe-notypic plasticity in ecologically important traits, including devel-opment time, body size, body condition, and coloration (Chevin,Lande, & Mace, 2010; Merila & Hendry, 2014). Evolutionary re-sponses include changes in body size, coloration, diapause cues,and thermal sensitivity (Anderson, Inouye, McKinney, Colautti, &Mitchell-Olds,2012;vanAsch,Salis,Holleman,vanLith,&Visser,2013;Bradshaw&Holzapfel,2001;Higgins,MacLean,Buckley,&Kingsolver, 2014; Karell, Ahola, Karstinen, Valkama, & Brommer,2011;Leal&Gunderson,2012).Inmostcases,however,thefitnessconsequencesoftheseplasticandevolutionaryresponsestoclimatechangeareunknown(Sgro,Terblanche,&Hoffmann,2016;Vedder,Bouwhuis,&Sheldon,2013).

Changesinseasonaltiming(phenology)areacommonresponseto recent climate changes in many systems (Parmesan, 2006;Thackeray etal., 2016). The fitness consequences of phenologicalshiftscanbeheterogeneousanddependontheenvironmentalandcommunitycontext(Forrest,2016;Forrest&Miller-Rushing,2010;Pau etal., 2011). For example, seasonal changes associated withmeanclimatewarminghavebeen shown tohavepositive impactson energybalance, survival, or reproduction in some systemsbutnegativeimpactsinothers(Andersonetal.,2012;Charmantieretal.,2008; Inouye,2008;Ozgul etal., 2010;Tafani,Cohas,Bonenfant,Gaillard,&Allaine, 2013). The negative impacts often result fromseasonalmismatchesbetweenapopulationanditsfoodresourcesornaturalenemies,andcanleadtoselectionandevolutionoftraitsresponsive to seasonal environmental conditions (Anderson etal.,2012;vanAschetal.,2013;Boggs&Inouye,2012;Charmantier&Gienapp,2014).

Developmental plasticity in morphological and physiologicaltraits is alsocentral toorganisms in seasonalenvironments,espe-ciallyforpopulationsthatcompletemultiplegenerationseachyear.For example in many insects, thermal conditions for generationsemergingintheearlyspringcanbedrasticallydifferentfromthosefor generations emerging inmidsummer, and plastic responses todevelopmental environments can lead to different phenotypes indifferentseasonalgenerations(Brakefield&Larsen,1984;Shapiro,1976).Developmentalplasticityiscentraltoadaptationtoseasonalenvironments in both tropical and temperate systems (Brakefield,1987;Tauber&Tauber,1976;Watt,1969).However,theextenttowhichdevelopmentalplasticityandevolutionareadaptiveandinflu-encefitnessinvariable,seasonalenvironmentsispoorlyunderstood.Willtheirfitnessconsequencesbesufficienttoenablepopulationstotracksustainedclimatechangesinthecomingdecades(Sgroetal.,2016)?

We explore these issues using the butterflyC. eriphyle, whichexhibitspopulationvariation inphenology, thenumberofgenera-tions per year, andmorphology along elevational gradients in the

westernUnitedStates.Detailed,mechanisticmodels enableus topredictthefitnessimplicationsofakeythermoregulatorytrait,wingabsorptivity(determinedbytheproportionofmelanicscales),basedon environmental conditions (Buckley & Kingsolver, 2012). Wingabsorptivity varies among populations due to both adaptive, her-itablegeneticdifferences(Ellers&Boggs,2004;Kingsolver,1983b)anddevelopmentalplasticity,wherebywingabsorptivitydecreaseswith increaseddevelopmental temperatures.Ourmodeling in thissystempredictsthatrecentclimatechangehasgeneratedselectionfordecreasedmeanandincreasedplasticityinwingabsorptivityandthattheimportanceofplasticityvarieswithelevation(Kingsolver&Buckley,2015,2017).

Hereweutilizethissystemtoinvestigatehowphenologicalplas-ticity interactswithmorphological plasticity and evolution to de-termine seasonal,phenotypic, and fitness consequencesof recentclimatechangeforC. eriphylepopulationsalonganelevationgradi-ent.First,howdodevelopmentaltemperaturesinfluencephenologyacrosstheelevationalgradient,andwhataretheconsequencesformorphologicalplasticity?Second,howdoesplasticityalterseasonalandelevationalpatternsofselectionandexpressionofwingabsorp-tivity? Third, we quantify the fitness consequences of phenologyandmorphologicalplasticityandevolutionalongthegradient,andhowtheseenhanceor limitadaptive responses toclimatechange.By comparing these fitness consequences to those in a best-casescenarioof“perfect”(optimal)plasticity,wealsoillustratehowun-predictable environmental variability limits adaptive responses toclimatechangeandtestthehypothesisthatphenologicalandmor-phologicalplasticitywillbufferfitnessvariationinseasonalandun-predictableenvironments(Merila&Hendry,2014;Parmesan,2006).

2  | METHODS

We integrate microclimate, developmental, biophysical, demo-graphic,andevolutionarymodelstoexaminethefitnessimplicationsofphenotypicandmorphologicalplasticity(Figure1).ThismodelingframeworkanddescriptionfollowsthatofKingsolverandBuckley(2017)(seeFigureS1).Amicroclimatemodelisusedtodeterminetheenvironmentalconditionsexperiencedbylarvae,pupae,andadultsatthreesitesalonganelevationalgradient.Bodytemperaturesde-terminedevelopmentalratesandultimatelylarval,pupal,andadultphenology.Thefocaltrait,wingabsorptivity,isinitiallydeterminedbyelevationdifferencesamongsites(genetics)andisalsoinfluencedbymicroclimateduetotheplasticeffectsofdevelopmental(pupal)temperatures.We incorporatemicroclimate and thermoregulatorytraits(includingwingabsorptivity) intoabiophysicalmodeltopre-dictadultbodytemperature,behavior,andperformance.Weuseademographicmodel to relate adult performance to fitness. Thesefitnessestimatesallowustoapplyaquantitativegeneticmodeltopredictphenotypicselectionandevolutionarychangesinwingab-sorptivity, and inplasticityof absorptivity, in thenextgeneration.Belowwebrieflyoutlineeachcomponentofthemodel(moredetailsareprovidedintheSMandin(Kingsolver&Buckley,2017)),andhow

     |  1233KINGSOLVER aNd BUCKLEY

themodelisthenusedtopredictphenology,plasticity,andfitnessacrosstheelevationalgradient.

2.1 | Study system

Colias eriphyleoccursinopenhabitatsinthewesternUnitedStatesatelevationsfrom1.4to3.0km.Larvaefeedonhostplants inthelegume family, including alfalfa, clover, and vetch. Diapause is in-ducedbylarvalphotoperiodandoccursoverthewinterinthe3rdin-star.Thetimingofonsetoflarvalfeedingandofthefirstadultflightseasonvariesstronglywithelevation(seebelow);andthenumberofadultgenerationsperyearrangesfrom4-5atlowerelevationsto2athigherelevations(Tabashnik,1980;Watt,Chew,Snyder,Watt,&Rothschild,1977;Watt,Han,&Tabashnik,1979).

We considered three sites spanning elevations of 1.8–3.0km.Our high-elevation site (3.0km, 40.03N, 105.55W) is C1 of theNiwot Ridge LTER (http://niwot.colorado.edu). We examined twoadditional sites that are part of the National Weather ServiceCooperative (COOP) Program: Cochetopa Creek (2.4km, station51713, 38.43N, 106.75W) and Montrose No. 2 (1.8km, station55722, 38.48N, 107.88W).Weather data were cleaned and filled(McGuire, Nufio, Bowers, & Guralnick, 2012).We averaged snowdepthdataacrossour studyperiodateachsiteandassumed thatlarvaecouldresumedevelopmentoncesnowmeltedandtempera-tureswerepermissive.WeestimatedsnowmeltasJuliandate,J=20at1.8kmandJ=105at2.4km(both1961–1990averagesofdatafrom the Western Regional Climate Center, http://www.wrcc.dri.edu/)andJ=141at3.0km(averagesof1960–2010dataavailablefromtheNiwotRidgeLTER).

Rates of larval and pupal development for C. eriphyle dependon temperature and are well described by linear reaction norms(Higgins,2014;MacLean,2015).Asaresult,wecancharacterizethereactionnormfordevelopmentrateDintermsoftwoparameters:

the developmental zero temperature (D0) belowwhichD=0; andthe accumulated degree-days (G) above D0 needed to completedevelopment.Forourmodels,weusedthreedifferentsetsofval-ues ofD0 and G based on recent data (Higgins, 2014; MacLean,2015): for postdiapause (4th and 5th instar) larval development(D0 =9.22°CandG=117.06°Cd);forpupaldevelopment(D0 =9.7°Cand G=101.9°Cd);andfortheentireperiodof(nondiapause)larvaldevelopment (D0 =11.5°C andG=270.39°C d). Studies with twopopulations ofC. eriphyle from different elevations yielded similarestimates,soweassumedthatD0 and Gdonotchangewitheleva-tioninourmodel.

Colias adults are strong flyers, and active flight is essentialfor courtship, mating, nectar-feeding, oviposition, and other ac-tivities (Kingsolver, 1983a; Stanton, 1984). Flight is tempera-ture dependent: We estimated the probability of flight as afunction of operative environmental temperatures, Te: Pflight= exp(−0.5*(abs(Te−33.5))/5)

3.5. The function and parameter val-ues are based on field flight data for C. eriphyle in Montrose(Kingsolver,1981).Adultsbehaviorallythermoregulatetoachievethebodytemperaturesneededforflightanddonotuseendoge-nousheatproductiontoelevatebodytemperatures(Watt,1968).Weassumedthatbutterfliesselectthebodytemperatureclosesttotheirthermaloptima(33.5°C)withavailablebodytemperaturesbracketedbythoseinfullsun(lateralbaskingposturewithwingsclosedandtheventralhindwingsurfacesorientedperpendiculartothesun)andfullshade(nodirectradiation).

Coliasadultsmaybeexposedtoshortintervalsofdeleteriouslyhighbodytemperatures(>40°C)evenathighelevations,wheremi-croclimaticvariationcanbesubstantial (Kingsolver&Watt,1983).Exposure to such high temperatures can reduce survival and fe-cundity: Daily heat shocks at 45°C reduce adult lifespan and eggproduction (Kingsolver &Watt, 1983). We modeled egg viabilityasanexponentiallydecayingfunctionofbodytemperaturefrom1

F IGURE  1 Flowdiagramforthemodelingframework.Climateandweatherateachelevationdeterminethemicroclimaticconditionsexperiencedbylarvae,pupae,andadultsateachsite.Microclimatedeterminesdevelopmentalratesoflarvaeandpupae,whichdeterminephenology.Thefocaltrait,wingmelanin,isinitiallydeterminedbyelevationdifferencesamongsitesandisalsoinfluencedbymicroclimateduetotheplasticeffectsofpupaltemperatures(Tpupal).Wemodelhowwingmelanininfluencesadulttemperatures(Tadult)andultimatelyperformanceingivenmicroclimatesandthenuseperformancetoestimatefitness.FitnessdifferencesamongindividualsexhibitingvariationinwingmelanincangenerateselectionandcauseevolutionarychangesinthemeanandplasticityofwingmelanininthenextgenerationbasedonKingsolverandBuckley(2017)

Climate

Microclimate

Performance Fitness

AbsorptivityαElevation

TpupalDevelopment rates Tadult

LOCAL ADAPTATION α20

PLA

STIC

ITY

B

PHENOLOGY

1234  |     KINGSOLVER aNd BUCKLEY

at40°Cto0.75at50°C(Buckley&Kingsolver,2012;Kingsolver&Buckley,2015).

BecauseColiaspopulationsandspecies(includingC. eriphyle)areadaptedtolocalclimatethroughdifferencesinsolarabsorptivity(α)oftheposteriorventralhindwings(Ellers&Boggs,2004;Kingsolver,1983b;Watt,1968),ouranalysesherefocusonvariation,plasticity,andevolutionofthistrait.Wingsolarabsorptivity(α=thefractionofincidentsolarradiativeenergythatisabsorbedbythewingsur-face)isdeterminedbytherelativeproportionsofpteridine(yellowororange)andmelanic(black)scalesandthusspanspossiblevaluesof0.4(allpteridinescales)to0.7(allmelanicscales)(Kingsolver,1983b).Twoothermorphologicaltraitsalsoinfluencetheheatbalanceandbodytemperatureofabutterfly:thelengthofsetaeonthethorax(furthickness)anddiameterofthethorax.Weusedafurthicknessof0.82mmandthoraxdiameterof3.6mminouranalyses,basedonmeasurementsforC. eriphyleatseveralsitesinColorado(Kingsolver,1983b).

Wing melanin in C. eriphyle is also phenotypically plastic:Increasing temperatureduringpupal developmentdecreaseswingmelanin (Higgins,2014;Hoffman,1978).Theslopeofthereactionnormrelatingwingmelanintopupal temperature issteeper (morenegative)forpopulationsatlower(1.5km)thanmiddle(2.1km)el-evations,andformalescomparedwithfemales(J.K.Higgins,2014;MacLean,2015).Wecharacterizedsolarabsorptivity (α) foran in-dividualbytwotraits:theslopeofthereactionnorm(B)relatingα tothemeantemperatureduringpupaldevelopment;andthemid-pointabsorptivity (α20), theabsorptivityata referencepupal tem-peratureof20°C(Kingsolver&Buckley,2017).Themeanvalueofα20increaseswithelevation(Ellers&Boggs,2002,2004;Kingsolver,1983b;Watt,1968);basedondatafromKingsolver(1983a),weusedthe initial mean starting value given by α20 =0.4226+0.06517*E,whereE=elevation inkm.Weestimated themean reactionnormslopeasB=−0.00083/°C,basedondataforC. eriphylemalesatourlowelevationsite (Higgins,2014).For simplicity,weassumed thattheslopedoesnotvarywithelevation.Weexplorebothfixedandevolving valuesof these two traits in our simulations at each site(seebelow).

2.2 | Microclimate

Weestimatedairtemperatures(Ta)at10-minintervalsbasedondailymaximumandminimumtemperaturesusingadiurnal temperaturevariation function incorporating sineandexponential components(Parton&Logan,1981).Globalhorizontalsolarradiationwascalcu-latedasafunctionofelevation,latitude,andlongitudebydiscount-ing global extraterrestrial radiation (Campbell & Norman, 1998).Radiationwasthenpartitionedintodirectanddiffusecomponentsasafunctionoftheatmospherictransmissivitytau[ratioofglobalhorizontalsolarradiationatsurfaceandcalculatedglobalextrater-restrial (top of atmosphere) horizontal solar radiation] (see FigureS1).Variationincloudinesswithinandamongdayswasmodeledintermsofvariationintau,usingastochasticweathergenerationap-proach(Kingsolver&Buckley,2015).

We implemented a microclimate model (Mitchell, Beckman,Bailey, & Porter, 1975; Porter & James, 1979; Porter, Mitchell,Beckman,&DeWitt,1973)usingfinite-differencemethodstosolveheatbalanceequationsdescribingsoiltemperaturesatthesurfaceandspecifieddepths(Kingsolver,1979;Kingsolver&Buckley,2015).Wescaledmicroclimatevariablestoplantheightbyestimatingtem-peratureandwindspeedprofiles(Campbell&Norman,1998)usingdata collectedatheights spanning0.05–1.5m in July2012at thesubalpinesite.BasedonweatherstationdatafromJuly2011atthissite,themeanwindspeedat0.5mheightwas0.4m/s.

2.3 | Developmental rates and phenological timing

BecauseColiaslarvaeandpupaetypicallyoccurontheshadyunder-sidesofleavesonthehostplant,weassumedthatlarvalandpupaltemperatures were equal to air temperatures in the sun at plantheight(1.8km=50cm;2.4and3.0km:20cm).Weusedasinglesinewaveapproximation (seehttp://www.ipm.ucdavis.edu/WEATHER/ddss_tbl.html)tocalculatedegree-days(G)basedondailymaximumandminimumtemperaturesateachsite.Fortheoverwinteringgen-eration,weestimatedwhenlarvaldevelopmentresumes(asthefirstdatewithdegree-day accumulation after snowmelt, see above) aswellastheonsetandcompletionofpupation.Forsubsequentgen-erations,weassumedadurationof7daysfromadultemergencetoegg laying,and5additionaldaysuntil larvaehatch (Higgins,2014;MacLean, 2015). Field observations indicate (and our simulationscorrectlypredict)thattwogenerationsarecompletedbeforeover-winteringeachyearat3.0km,threegenerationsat2.4km,andfour(sometimes more) generations at 1.8km (Tabashnik, 1980; Wattetal., 1977, 1979). For comparison, we modeled two generationseachyearat3.0kmandthreegenerationsattheothertwosites(seeResultsandFigure2).

2.4 | Heat balance, performance, and fitness

Weusedasteady-stateheatfluxmodelforColiasadultsthatwasde-velopedandfieldvalidatedbyKingsolver(1983a)topredictthoracicbodytemperature(operativeenvironmentaltemperature,Te)basedonthermoregulatorytraits(bodysize,basalventralhindwingsolarabsorptivity, and thoracic fur thickness), behavioralposture (bask-ing andheat-avoidance), andenvironmental conditions (Buckley&Kingsolver, 2012). Themodel successfully predicts patterns ofTe,flight activity time and heat-avoidance in the field for C. eriphyle andotherColias speciesalonganelevationalgradient inColorado(Kingsolver,1983b;Kingsolver&Watt,1984).PredictionsofTe are updatedevery10min.

KingsolverandBuckley(2015,2017)usedthisbiophysicalmodeltogetherwith a demographicmodel to connectmicroclimate andthermoregulatorytraitstofitnessestimates(netreproductiverate;seeFigureS1fordetails).Ourfitnessestimatesarebasedon500femalespergeneration.Wesimulatedadateofadultemergenceforeachindividualusinganormaldistributionwithastandarddeviationof2days,truncatedsevendaysbeforeandafterourestimateddate

     |  1235KINGSOLVER aNd BUCKLEY

ofadultemergencefortheyearandgeneration (Tabashnik,1980;Watt etal., 1977, 1979). We calculated daily egg production foreachfemaleastheproductofavailableflighttime (assuming50%

of available time is spent ovipositing) and the rate of oviposition(0.73eggs/min, asestimated forColoradoColias (Stanton,1984)).Wemultiplieddailyeggproductionbytheaverageofhourlyviability

F IGURE  2 Predictedseasonalphenologyandpupaltemperaturesacrossyears.Climateandweatherdifferencesalongtheelevationgradient(toprow=panelsa–c:3.0km;middlerow=d–f:2.4km;bottomrow=g–i:1.8km)determinephenologyandpupaltemperatures(dashedlines:fixedphenology,varyingphenology:solidlines).Firstcolumn(panelsa,d,g):themeanJuliandateofappearanceforlarvae(orange),pupae(green),andadults(purple;grayshading:pupalduration).Theshortdashedlinedepictsthefixedphenology.Secondcolumn(panelsb,e,h):Theannualmeanpupaltemperature(Tpupal,in°C)duringthefirst(blue),second(orange),andthird(red)generationsdiffersbetweenthevaryingandfixedphenologyscenarios.Thirdcolumn(panelsc,f,i):phenologicalshiftscancounterincreasesinpupaltemperaturesinwarmyears.Thex-axisdepictsthetemperatureanomalyeachyearduringtheaveragedatesofpupation(i.e.,iftheaverageJuliandatesforpupationare150–155,averageTfordays150–155eachyear–averageTfordays150–155acrossallyears).Thus,positivevaluesindicatethattemperaturesduringthefixedphenologyarewarmerthanaverage.They-axisdepictstheshiftinpupaltemperaturesresultingfromthevaryingphenology(i.e.,Tpupalforvaryingphenology–Tpupalforfixedphenology).Agraylinecorrespondstophenologyperfectlytrackingpupaltemperatures(slope=−1).Wedepictsignificant(p<.05)temporaltrends

1960 1980 2000

100

150

200

3.0 km

LarvaePupaeAdults

1960 1980 2000

1015

2025

3035 gen 1

gen 2gen 3

phenologyFixedVariable

–4 –2 0 2 4

–4–2

02

4

1960 1980 2000

100

150

200

Julia

n da

te

2.4 km

1960 1980 2000

1015

2025

3035

Tpup

al (°

C)

–3 –2 –1 0 1 2 3

–3–2

–10

12

3Tp

upal

shi

ft by

pla

stic

ity (°

C)

1960 1980 2000

100

150

200

1.8 km

1960 1980 2000

1015

2025

3035

–6 –4 –2 0 2 4

–6–4

–20

24

Year Year Tpupal anomaly (fixed phenology, °C)

(a) (b) (c)

(d) (e) (f)

(g) (h) (i)

1236  |     KINGSOLVER aNd BUCKLEY

estimates.Weestimatedλbysummingoverdaystoeitheradura-tionof5days,reflectingthemeanadultlifespaninthefield(Wattetal.,1977,1979),orreachingamaximumlifetimeeggproductionof700(Kingsolver,1981)astheproductofsurvivaltomaturity,dailysurvival,andeggproduction(averagedacrossthe500females).Intheabsenceofother information,weassumethat juvenile (eggtoadult)survivalisconstantacrossseasonsandelevationinoursim-ulations (seeFigureS1).This strongassumptionwill clearlyaffectthevalidityofpredictionsaboutabsolutemeanfitnessacrosssea-sons and sites, butwill not alter our qualitative results about thefitnessconsequencesofdevelopmentalplasticityandevolution(seeResults).

2.5 | Selection and evolutionary response

Ourmicroclimate,biophysical,anddemographicmodelsallowustopredictthefitnessλofanindividualColiasasafunctionofclimatevariablesandourfocaltraitsolarabsorptivity(α).Foragivensetofweather conditions at a site and time period, the relationship be-tweenabsorptivityαand fitnessλ (the fitnesssurface) is typicallyquadratic (Kingsolver & Buckley, 2015), and fitness is greatest atsome intermediate, “optimal”valueαopt.At lowervaluesofα,but-terflieswithvaluesofα below αoptarelessabletoachievebodytem-peraturesneededforflightandhavelessflightandovipositiontime,lowereggproductionandreducedfitness;butterflieswithvaluesofα above αoptexperiencedeleteriouslyhighbodytemperaturemorefrequently,andhavegreatermortality,lessflighttimeandeggpro-duce,andhencereducedfitness.Differencesinweatherconditionsamongsitesand timeperiodschange thevalueofαopt and curva-tureof thefitnesssurfacequadratic (Kingsolver&Buckley,2015).Variationinabsorptivityinapopulationcausesvariationinfitness,resultinginphenotypicselection(Kingsolver&Buckley,2017).

Weusedasimplequantitativegeneticmodeltopredictselectionandevolutionofthetwotraitsthatdeterminethesolarabsorptiv-ityofabutterfly—α20 and B—in response toclimate (Kingsolver&Buckley,2017).Weusedestimatesofthephenotypicstandardde-viationofαforC. eriphyle(0.062)reportedbyKingsolver(1983a)forpopulationsamplestakenin1980.EllersandBoggs(2002)usedpar-ent–offspringbreeding experiments to estimate thenarrow-senseheritabilityh2ofwingmelaninforC. eriphyle,yieldingh2=0.43formalesand0.36forfemales(Ellers&Boggs,2002);weusedah2 value of0.40forα20inoursimulations.Weuseddataforfull-sibfamiliesofC. eriphyle from amidelevation population (2.3km) to estimatethephenotypicstandarddeviationofBas0.083(Higgins,2014).IntheabsenceofinformationaboutheritabilityofBoraboutthephe-notypicorgeneticcovariancebetweenα20 and B,weassumedthath2=0.4forBandthatα20 and Bareuncorrelated.Wealsoassumedthatselectionissufficientlyweaksothattheheritabilitiesandphe-notypic and genetic variances do not change with time (Lynch &Lande,1993).Finally,weassumednogeneflowamongpopulations.

Ourmodels predict the fitness function relating solar absorp-tivity α to fitness for a given site and year. Combined with esti-mates of the phenotypic mean and variance, we estimated the

(unstandardized)directional selectiongradientsβ forbothα20 and B,andusedtheheritabilityh2topredicttheevolutionaryresponsestoselectioninthenextgenerationinthesetraits(Lande&Arnold,1983).Notethattheevolutionaryresponsetoselection isdirectlyproportionaltoh2inthismodel(e.g.,lettingh2 =0fortraitBresultsinnoevolutionofplasticity);thesimulationswedescribeinthenextsectionexplorehow this andother scenarios influenceourmodelpredictions(seebelow).

2.6 | Model predictions

Weusedourmodelsimulationstoexplorethreemainissuesforthethree sites along the elevational (climatic) gradient over the past50years (1960–2010). First, we considered how developmentaltemperaturesinfluenceseasonalphenologyandthetemperaturesexperiencedbypupaeandadultsineachgeneration.Inparticular,we compared the effects of varying phenology to those of fixedphenology, inwhich the timing of each life stage and generationoccursonthesamecalendardateeachyear(fixedatthemeanpre-dicteddateforeachstageandgenerationatthesiteacrossyears).Second,weexploredtheeffectsofdevelopmentalplasticityonsea-sonalpatternsofsolarabsorptivity (wingmelanin),and itsconse-quencesforvariationinselectionandmeanpopulationfitness.Weusedtheseto identify theoptimalvalueofabsorptivity (αopt)andthemeanpupaltemperatureforeachgenerationwithinayearforeachsite.Therelationshipbetweenαoptandmeanpupaltempera-tureforagivenyearrepresentstheoptimalreactionnormforthatyear,whichwecall “perfectplasticity.”Weassessedhowoptimalreactionnormsvary amongyears and sites, andhow these com-paretotheevolvingreactionnormspredictedbyourevolutionarysimulations.

Third,weexploreddifferentscenariosforphenology,plasticity,andevolutionalongthegradient,andhowtheseinfluencethegeo-metricmeanpopulation fitness (acrossgenerations) foreachyear.Weconsideredsixscenarios:

1. Constant absorptivity: α20 remains constant at its initial value,and there is neither plasticity nor evolution.

2. Observedplasticity:Plasticityoccursat its initialvalue,butnei-therα20 nor B evolves.

3. Evolutionofabsorptivity:α20evolves,butthereisneitherplastic-itynorevolutionofplasticity.

4. Evolutionwithobservedplasticity:α20evolves,andplasticity isfixedatitsinitialvalue.

5. Evolutionofplasticity:Bothα20 and B evolve.6. Perfectplasticity:Theoptimalabsorptivity (αopt) isexpressedineachgenerationeachyear.

Weconsideredeachofthesescenariosforboththevariableandfixedphenologycases.Bycomparingtheresultstotheperfectplas-ticitycase(whichbydefinitionwillhavethehighestmeanfitness),wecanquantifyhowphenology,plasticity,evolution,andtheevolutionofplasticitycontributetoadaptationtochangingclimates.

     |  1237KINGSOLVER aNd BUCKLEY

3  | RESULTS

3.1 | Effects of phenological plasticity on seasonal timing and temperatures

The predicted seasonal patterns of emergence vary across lifestages, generations, and the three sites along the elevational gra-dient(Figure2a,d,g).Atthehigh-elevationsite,postdiapauselarvalandpupaldevelopmentisdelayedbythedateofsnowmeltandonlytwogenerationseachyeararepossible.Atmiddleandlowersites,postdiapausedevelopmentandadultemergenceoccurmuchearlier,andthree(ormore)generationscanoccureachyear.Thedurationoflarvalandpupaldevelopment(Figure2,grayshading)decreasesastemperatureswarmthroughtheseason.Thepredicteddateofadultemergence has advanced significantly (p<.05, linear regression)since1960foreachgenerationatthe2.4kmand3.0kmsites.Therelationshipisweakduetoseasonalandannualvariationintempera-ture.Thisvariationisespeciallypronouncedatthelowsite,wherethereisnophenologicaltrend.

Variation in phenology generates variation in the predictedmeanpupaltemperaturesexperiencedeachgenerationateachsite(Figure2b,e,h).Unsurprisingly,pupaltemperaturesarelowerforthefirstthanforsubsequentgenerationsateachsite,andthereissub-stantialannualvariabilityinpupaltemperaturesatallsites.However,mean pupal temperatures in the first generation actually increasewithincreasingelevationduetothedifferencesinphenologyamongsites. There are larger differences in pupal temperature betweengenerationsatlowerthanhigherelevationsites.

Predictedpupaltemperaturesatthe3.0kmsitehaveincreasedsignificantly (p<.05, linear regression) since 1960 for all genera-tions,regardlessofwhetherphenologyvaries.Theadvancingphe-nology at the 3.0km sites has not substantially altered the ~2°Cincreaseinpupaltemperaturessince1960.Environmentaltrackinghasbeenmoreeffectiveatthe2.4kmsite:Pupaltemperatureshaveincreasedsignificantly(p<.05,linearregression)acrossallgenera-tionswhenphenology is fixed,but the increase isonlysignificantforthethirdgenerationifphenologyisvariable.Pupaltemperatureshavenotshiftedovertimeatthethermallyvariable1.8kmsite.

We next consider how phenological shifts influence pre-dicted pupal temperatures by examining interannual variability(Figure2c,f,i). This addresseswhether phenological shifts enabletracking of the thermal niche.Warm years (positive anomaly inpupaltemperatureatfixeddates)acceleratedevelopmentandre-sult in pupation occurring earlier. Consequently, pupal tempera-tures can be cooler (relative to pupal temperatures at the fixeddates)inwarmyears,whichmayresultinthethermalplasticityofwingmelaninbeingmaladaptive.Thisphenologicaltrackingofthethermalnicheisincomplete:Slopesareallshallowerthanthecaseof perfect phenological tracking (Figure2c,f,i). The tendency forpupaltemperaturestobecoolinwarmyearsduetophenologicaltracking is particularly pronounced at1.8km sites. Thenegativeslopesaresignificantandsteepacrossallgenerationsat1.8km(es-timates±SEbygeneration:1st=−0.37±0.07;2nd=−0.46±0.09;

3rd=−0.44±0.10;allp<.001,linearregressions).Slopesdeclinewithelevation,butremainsignificantfortheearlygenerationsatthe 2.4km (1st=−0.33±0.08; 2nd=−0.35±0.07; all p<.001)and3.0km (1st=−0.15±0.02;p<.001, linear regressions) sites.At 3.0km, the second generation exhibits a significant contrarytrendbecausephenologicaladvancementsinthefirstgenerationare limited by snowmelt at this site, with slightly warmer pupaltemperaturesinwarmyears.Thecoolpupaltemperaturesinwarmyears predicted by our simulations illustrate the importance ofconsideringphenologicalshiftsandhowtheycanalterpatternsofplasticityandselectiononsubsequentlifestages.

3.2 | Consequences for plasticity and selection

Thesepatternsofphenologyandpupaltemperaturehaveimportantconsequencesforthepredictedpatternsoftraitexpressionandse-lectiononwingmelanin(solarabsorptivity;Figure3).Herewefocusonpredictionsforthecasewithobservedplasticitywithnoevolutionofeithermeanorplasticityinabsorptivity.Atthehighsite,meanab-sorptivityishigh,butthedifferenceinabsorptivitybetweengenera-tionsismodest.Absorptivitytendstodecreaseacrossyearsduetoincreasesinpupaltemperature(Figures2–3).Differencesinabsorp-tivityamonggenerationsaregreateratthemiddleandespeciallythelowsite,reflectinggreaterseasonalvariationinpupaltemperatures.Asaresult,thepredictedphenotypicconsequencesofplasticityaregreateratlowerthanathigherelevations(Figure3b,e,h)(Kingsolver&Buckley,2017).Theeffectsofvaryingphenologyonmeanabsorp-tivityarerelativelymodestandareinconsistentindirectionacrossgenerations and years (Figure3b,e,h). Overall, varying phenologytendstodecreaseinterannualvarianceinmeanabsorptivityatthelowelevation site,buthasnoconsistenteffectonvariance inab-sorptivityatthemid-orhigh-elevationsites.

Realizedseasonalityacrosstheelevationgradientalsoinfluencesvariationindirectionalselectiononabsorptivityacrossgenerations(Figure3a,d,g). At all sites, cool conditions in the first generationgenerally select for increasedabsorptivity (positivedirectional se-lection),withsubstantialinterannualvariability.Selectionduringthefirstgenerationatthehighsiteshiftsfrompositivetolittleornega-tiveselectionoverthetimeperiod1960–2010.Incontrast,selectionisconsistentlynegativeinthesecondandthirdgenerationsthroughthisperiod,especiallyatthemiddleandlowsites.Thissuggeststhat,despiteadaptiveplasticityinabsorptivity,thedirectionofselectionvaries consistently between generations in response to seasonalvariation,favoringtheevolutionaryofgreaterplasticity(Kingsolver& Buckley, 2017). Differences in selection resulting from varyingphenologyareinconsistentindirectionacrossyears(Figure3a,d,g).Overall,varyingphenologytendstodecrease interannualvarianceinselectionatthelowelevationsite,withlittleeffectonvarianceinselectionatthemid-orhigh-elevationsites.

Byfindingtheabsorptivityvaluethatyieldsthehighestfitnessineachgeneration,wecanalsopredicttheoptimalreactionnorm(op-timalabsorptivityasafunctionofmeanpupaltemperature)foreachyearateachsite(Figure4).Theslopesoftheoptimalreactionnorms

1238  |     KINGSOLVER aNd BUCKLEY

arehighlyvariableamongyears,especiallyatthehigherelevationsites:Indeedinsomeyears,theslopeoftheoptimalreactionnormispositiveratherthannegative.Atthelowsite,theslopesoftheoptimalreactionnormsaremoreconsistent,but there is substantialvariationamongyears in the position (i.e.,α20) of the reaction norms (Figure4a,c,e).Importantly,theoptimalreactionnormsareconsistentlysteeper(morenegativeslope)thantheobservedreactionnorm(dashedblack line),especiallyatthelowersites.Asaresult,thereisconsistentselectionfor evolutionary increases in themagnitude of developmental plas-ticityinthissystem(Kingsolver&Buckley,2017).Theeffectsofvari-ablevsfixedphenologyonoptimalreactionnormsareheterogenous,especiallyat thehigherelevationsites (Figure4).Varyingphenologydoesnothaveconsistenteffectsoneithertheslopeorpositionoftheoptimalreactionnormsacrossyearsatanyofthesites.

3.3 | Fitness consequences of phenology, plasticity, and evolution

Thepredictedchangesinmeanpopulationfitnessineachgenerationrevealseveralinterestingpatterns(Figure3c,f,i).First,meanfitness

isgenerallylowerinthefirstthaninsubsequentgenerations,espe-ciallyatthelowsite.Severalpredictedtemporaltrendsinfitnessareevident:Meanfitnessincreasesovertimeinthefirstgenerationatthehighsiteanddecreasesovertimeinthesecondgenerationatthelowsite(Figure3c,f,i).Thesetrendsaretheconsequencesofwarmerspringconditionsathigherelevations,andhottersummerconditionsat lowelevations (Figure2).Second,annualvariability inmean fit-nessislargeatallsites,especiallyinthefirstgeneration:Annualvari-abilityislargerthanthetemporaltrendsinmeanfitness.Theeffectsofvaryingphenologyonvarianceinmeanfitnessarerelativelysmallinmagnitudeandinconsistentindirection.

Finally,weinvestigatehowplasticity,evolution,andphenologyaltergeometricmeanpopulationfitnessforeachyearoverthetimeperiod1960–2010.Acrossallplasticityandevolutionscenarios,fit-ness varies substantially across years due to seasonal and annualvariation in environmental conditions (Figure5).We consider thescenariosrelativetothecaseofconstantabsorptivity(noplasticitynorevolution),whichgenerally(butnotalways)confersthelowestfitness.Wefirstconsiderthecaseoffixedphenology(Figure5b,e,h).Across the elevation gradient, perfect plasticity increases fitness

F IGURE  3 Predicteddirectionalselection,meanabsorptivity,andmeanfitnessacrossyears.Forthecasewithobservedplasticityandnoevolution,wedepictthedirectionalselectiongradient(β)onabsorptivity(wingmelanin,panelsa,d,g),meanabsorptivity(panelsb,e,h),andmeanfitness(panelsc,f,i)acrosselevations(toprow=panelsa–c:3.0km;middlerow=d–f:2.4km;bottomrow=g–i:1.8km).Wedepicteachelevationunderbothvariable(blue:first,orange:second,red:third)andfixedphenology(lightblue:first,gold:second,pink:third).Wedepictsignificant(p<.05)temporaltrends

1960 1970 1980 1990 2000 2010 1960 1970 1980 1990 2000 2010 1960 1970 1980 1990 2000 2010

1960 1970 1980 1990 2000 2010 1960 1970 1980 1990 2000 2010

1960 1970 1980 1990 2000 2010 1960 1970 1980 1990 2000 2010

1960 1970 1980 1990 2000 2010

1960 1970 1980 1990 2000 2010

–2–1

01

23

–2–1

01

23

–2–1

01

23

3.0 km gen 1gen 2gen 3

0.7

0.6

0.5

0.4

0.7

0.6

0.5

0.4

0.7

0.6

0.5

0.4

12.

21.

81.

41

2.2

1.8

1.4

12.

21.

81.

4

Dire

ctio

nal s

elec

tion

2.4 km

Abs

orpt

ivity

Mea

n fit

ness

1.8 km

Year

(a) (b) (c)

(d) (e) (f)

(g) (h) (i)

     |  1239KINGSOLVER aNd BUCKLEY

13.3%–23.3%, with greater augmentation of fitness occurring atlowerelevations.Atthelowelevationsite,evolutionofabsorptivitydidnotincreasefitnessintheabsenceofplasticity.Observedplas-ticity(bothincludingandomittingevolutionofabsorptivity)consis-tentlyincreasesfitness.

Wepredictthattheevolutionofplasticitywillhavevariablefit-ness implications.At themiddleelevationsite, fitness tendsto in-crease over time relative to the scenario of constant absorptivityfor the scenarios that include plasticity and evolution. Evolutionofabsorptivity(withobservedplasticityandincludingevolutionofplasticity)tendsto increasefitnessoverthescenariosofobservedplasticitywithoutevolutionandevolutionofabsorptivity.Incontrastatthehigh-elevationsite,thecontributionsofplasticity,evolutionofabsorptivity, andevolutionofplasticity to increasingmean fitnessarevariablewithpredictedfitnessfluctuatingnearzeroacrosstime.

Phenological shifts generally produce slight increases in pre-dicted relative mean fitness across years but also increase inter-annualvariance.Varyingphenologyhas little influenceon relativefitnessatthehigh-elevationsite(Figure5c,f,i).Atthemidelevationsites,relativefitnessissimilaramongscenariostothefixedphenol-ogycase,buttheincreaseinthefitnessadvantagerelativetocon-stant absorptivity over time is reduced.At the low elevation site,interannual variation is reduced for the scenario of evolution ofthemeanandplasticityofevolution.This scenarioexhibitshigherrelative fitness over time. These results suggest that adaptation

at higher elevations is strongly limited by climatic variability andunpredictability.

4  | DISCUSSION

4.1 | Elevational gradients in phenology and phenotypes

Meanenvironmentaltemperaturesdeclinewithincreasingelevation,andpopulationdivergenceinthermallyadaptivetraitsalongelevationalgradientshasbeenwidelydocumented.Fororganismswithmultiplegenerationsperyear,however,patternsofclimaticvariationandad-aptationalongelevationalgradientsarelessstraightforward.Becauseseasonal temperatures affect developmental rates, populations atlowerelevationcanreachkeylifestagesearlier,alteringbothseasonalphenologyandtheenvironmentalconditionsexperiencedwithinandacrossgenerations.OursimulationsforColiaspredictthat,inthefirstgenerationeachyear,meanpupaltemperaturesareactuallyloweratlowthanathigh-elevationsites,asaresultofearlierspringsnowmeltandemergenceatlowerelevations(Figure2).Thiselevationalpatternisreversedinsubsequentgenerations.Oneconsequenceisthatdiffer-encesinmeanpupaltemperaturesbetweengenerationsdeclinewithincreasing elevation: Thuswithin years, variation in pupal and adulttemperaturesismuchgreateratlowthanathighelevations(Kingsolver&Buckley,2017;MacLean,Kingsolver,&Buckley,2016).

F IGURE  4 Optimalreactionnorms.Optimalreactionnorms(optimalabsorptivityasafunctionofmeanpupaltemperatureineachgeneration)acrossyearsforthefixed(panelsa,c,e)andvariable(panelsb,d,f)phenologyforeachelevation(toprow=panelsa–b:3.0km;middlerow=c–d:2.4km;bottomrow=e–f:1.8km).Theobservedreactionnorm(dashedblackline)isalsoincluded

Fixed phenology

0.4

0.5

0.6

0.7

10 15 20 25 30

Varying phenology

0.4

0.5

0.6

0.7

10 15 20 25 30

0.4

0.5

0.6

0.7

10 15 20 25 30

Abs

orpt

ivity

0.4

0.5

0.6

0.7

10 15 20 25 30

Abs

orpt

ivity

0.4

0.5

0.6

0.7

10 15 20 25 30Pupal temperature (°C)

0.4

0.5

0.6

0.7

10 15 20 25 30Pupal temperature (°C)

19601970198019902000

year

(a)

(c)

(e)

(b)

(d)

(f)

1240  |     KINGSOLVER aNd BUCKLEY

Phenologicalshiftsresultingfromthetemperaturedependenceofdevelopmentmay limittheeffectivenessofplasticresponsestoclimatechange(Thackerayetal.,2016).Phenologicaltrackingcanre-sultinearlydevelopmentalstagesexperiencingcoolertemperaturesinwarmyears (Figure2).Thismay result in colddamage (Boggs&Inouye,2012; Inouye,2008). Inoursystem,coolerpupaltempera-tures in the first generationmay lead to increased absorptivity in

relativelywarmeryears;thismaladaptiveresponseresultsfromthetime lagbetween theenvironmental cue (pupal stage) and the se-lectiveenvironment(adult)(Kingsolver&Huey,1998;Moran,1992).Moregenerally,developmentalplasticitythatevolvedinmorecon-stantenvironmentspriortoacceleratingclimatechangemaybemal-adaptiveinthevariableenvironmentsassociatedwithclimatechange(Chevinetal.,2010;Lynch&Lande,1993;Merila&Hendry,2014).

F IGURE  5 Geometricmeanpopulationfitnessasafunctionofyear.Thepopulationfitnessunderfixedphenologyforeachsite(toprow=panelsa–c:3.0km;middlerow=d–f:2.4km;bottomrow=g–i:1.8km)variesamongscenarios(seemethodsandlegend,leftcolumn).Weadditionallydepictgeometricmeanfitnessrelativetothecaseofconstantabsorptivityforthecasesoffixedphenology(panelsb,e,h)andvaryingphenology(panelsc,f,i)

1960 1980 2000

1.4

1.6

1.8

2.0

2.2

3.0 km

Constant absorptivityPerfect plasticityObserved plasticityEvol of absorptivityEvol with plasticityEvol of plasticity

1960 1980 2000

1.4

1.6

1.8

2.0

2.2

Geo

met

ric m

ean

fitne

ss

2.4 km

1960 1980 2000

1.4

1.6

1.8

2.0

2.2

1.8 km

1960 1980 2000

–0.2

0.0

0.2

0.4

1960 1980 2000

–0.2

0.0

0.2

0.4

Rel

ativ

e ge

omet

ric m

ean

fitne

ss

1960 1980 2000

–0.2

0.0

0.2

0.4

1960 1980 2000

–0.2

0.0

0.2

0.4

1960 1980 2000

–0.2

0.0

0.2

0.4

Rel

ativ

e ge

omet

ric m

ean

fitne

ss

1960 1980 2000

–0.2

0.0

0.2

0.4

Year

Fixed Phenology Varying Phenology

(a) (b) (c)

(d) (e) (f)

(g) (h) (i)

     |  1241KINGSOLVER aNd BUCKLEY

These patterns of phenology and pupal temperature have im-portantconsequencesforphenotypicvariationinwingmelanin.AsobservedinthefieldforC. eriphyle(MacLean,2015),predictedmeanabsorptivitydeclinesfromthefirsttolatergenerations,asaresultofdevelopmentalplasticity (Figure3).However, thisseasonaldeclinein absorptivity ismuch greater at lower elevations, such that theelevationcline inabsorptivity issteeper in later (mid-latesummer)generations(Kingsolver&Buckley,2017;MacLean,2015;MacLeanetal., 2016).As a result, the contributionof plasticity to seasonalvariation inwingmelanindeclineswith increasingelevation inthissystem.

Thepredictedeffectsofclimatewarmingonmeanabsorptivityalsovarywithelevationinthissystem.Forexampleathighandmide-levations, mean absorptivity declines in both spring and summergenerationsoverthe1960–2010timeperiod(Figure3),butthereisnotemporaltrendatthelowsite.Theseplasticityresponsesaretheresultoftheincreasingpupaltemperaturesoverthistimeperiodatthehighandmiddle(butnotlow)elevationsites(Figure2).Temporalpatterns inmeanwingmelanin in associationwith recent climatechangeshavebeenobserved inotherColias speciesbutnotbeenevaluatedinthisspecies(MacLean,2015;MacLeanetal.,2016).

4.2 | Elevational patterns in selection

Thepredictedseasonalandtemporalpatternsofselectiononwingabsorptivityalsovaryacross theelevationalgradient (Figure3). Inthefirstgenerationatthehigh-elevationsite,directionalselectionswitchedfrompositive(favoringincreasedabsorptivity)tonegative(favoringdecreasedabsorptivity)selectionoverthepast50yearsasa result of increasing spring temperatures during this time period(Figure2).Incontrastatlowelevations,oursimulationspredictcon-sistentpositiveselectiononabsorptivityinthefirstgeneration,butnegativeselectionduringthesecondandthirdgenerations.Despitetheinfluenceofadaptiveplasticityonmeanabsorptivityacrossgen-erations,thereremainsaconsistentseasonalpatternofalternatingselectionacrossgenerationsatthissite.Thissuggeststhatneitherphenologicalnormorphologicalplasticityissufficienttomatchsea-sonalphenotypes toenvironmentalconditions in this system.Ourresultsalsoindicatethatclimatechangewillgeneratedifferentpat-ternsofselectionandevolutiononmeananddevelopmentalplas-ticityofwingmelaninatdifferentelevations(Kingsolver&Buckley,2017).

Variationindirectionalselectiononabsorptivityamonggenera-tionsandyearsimpliesthattheoptimalreactionnorm(theoptimalabsorptivityasafunctionofmeanpupaltemperatureacrossgener-ationsforagivenyearandsite)mustvaryamongyears (Figure4).Interannual variation in the optimal reaction norm is particularlystrikingathigherelevation:Theoptimalreactionnormslopevariesfromstronglynegativetostronglypositiveinsomeyears.Atlowel-evation,theoptimalslopeismoreconsistent,butthereisvariationintheposition(optimalα20)ofthereactionnorm.Thesepredictedpatterns suggest thatpupal temperature is a less accuratepredic-torofclimaticconditionsexperiencedbyadultsathigherelevations,

because of greater stochastic climatic variability and a narrowerrangeofpupaltemperaturesexperiencedathigherelevations.Thelack of predictable environmental cues in this system limits theevolution of adaptive developmental plasticity at higher elevation(Hoffman,1978;Moran,1992).

Surprisingly, varying phenology does not reduce interannualvariationintheoptimalreactionnorms(Figure4).Wehadexpectedthat varying phenology would generate stronger correlations be-tweenpupalandadultconditionsandenhancetheadaptivenessofdevelopmentalplasticity,butourresultsdonotsupportthisexpec-tation.We believe that a key reason for this disconnect betweenpupal and adult conditions is that adult body temperatures (andflightandfitness)aremuchmorestronglydrivenbysolarradiativeconditionsandcloudcoverthanpupaltemperatures:Intheabsenceofhighsolarradiativeintensities,butterfliessimplycannotachievethebodytemperaturesneededforactiveflight,especiallyathigherelevations(Ellers&Boggs,2004;Kingsolver,1983b;Watt,1968).

4.3 | Plasticity, evolution, and fitness variation in changing environments

Oursimulationspredictthatclimatechangeduringthepast50yearshasdifferingeffectsonmeanfitnessforpopulationsatdifferentel-evations:Predictedmeanfitnessincreasedinthespringgenerationat high elevation, but decreased in the summer generation at lowelevation(Figure3).Ourhigh-elevationsite(3.0kmelevation)isneartheupperdistributionallimitofC. eriphyleinColorado,andpopula-tionsaresmallandwidelyscatteredattheseelevations.Asaresult,recent and future climatewarmingmay increase theupper eleva-tionallimitsofthisspecies,regardlessofitsphenologicalresponses.Theinterannualvariationinpredictedmeanfitnessisstrikingatallsites, especially for the first (spring) generation, and the variationamong years is substantially larger than themean trend over thistimeperiod.Theeffectsof variable springweatheronpopulationabundanceandfitnesshavebeendocumentedformanytemperateandmontanebutterflies (Boggs& Inouye,2012;Dennis,Kemp,&Beckwith,1986).

A central question is whether plasticity and evolution canimprovepopulation fitness in the faceof recentand futureenvi-ronmental change and thereby reduce the likelihood of popula-tion decline and extinction (Chevin etal., 2010; Lynch & Lande,1993).Our simulations address thisquestion forwingmelanin inColiasrelativetotwolimitingcases:aconstantwingabsorptivity,inwhichabsorptivity isnotdevelopmentallyplasticanddoesnotevolve;andperfectplasticity, inwhichtheoptimalabsorptivityisachieved every generation in each year (Figure5). Even for per-fectlyadaptedpopulations,thereissubstantialannualvariationinmean fitness at all elevations as a result of seasonal and annualvariation inweather.Conversely, constantwingmelanin stronglyreducesmeanpopulationfitness;andtheevolutionofwingmela-nin(withoutplasticity)causesonlymodestimprovementsinmeanfitness.Theseresultssuggestthatplasticityratherthanevolutionalone can make greater contributions to adaptive responses to

1242  |     KINGSOLVER aNd BUCKLEY

climatechangeinthissystem(Charmantieretal.,2008;Sgroetal.,2016;Vedderetal.,2013).

Oursimulationsalsopredict that theadaptiveeffectsofplas-ticity varywith elevation.At the lowand (to a lesser extent) themidelevationsites,plasticitysubstantially increasesmeanpopula-tionfitness(relativetoconstantwingmelanin)andallowingevolu-tionofthereactionnormfurtherincreasesmeanfitnessaswellasvarianceatthesesites.Incontrast,atthehigh-elevationsite,plas-ticityand reactionnormevolutionhave littleeffecton improvingpopulationfitness.Adaptiveplasticityathigherelevationsislimitedbytheshortactiveseasonandbystochasticvariation inweatherat these sites (Kingsolver & Buckley, 2015, 2017). Lack of suffi-cient genetic variation, environmental unpredictability, and geneflowcanalso limitevolutionofadaptiveplasticity, relative toour“perfect plasticity” scenario (Levins, 1968;Moran, 1992). Furthersimulationsinwhichweincreasephenotypicandgeneticvariancein reactionnorms show that even in theabsenceof genetic con-straints, environmental unpredictability strongly limits adaptiveplasticity,especiallyathighelevations(Kingsolver&Buckley,2017).Collectively, ourmodel results suggest that the predicted contri-butionsofplasticityandevolutionforpopulationadaptationtocli-matechangedeclinewithincreasingelevation,andthatadaptationat higher elevations is strongly limited by climatic variability andunpredictability.

ACKNOWLEDG EMENTS

WethankN.Bouzid,J.Higgins,L.Hamon,R.Huey,H.MacLean,M.Nielsen,R.Telemeco,andtwoanonymousreviewersforsuggestionsonpreviousversionsofthemanuscript,andtheWhiteleyCenterforprovidingtheidealsettingforworkingonthisproject.JGKwassup-portedinpartbyaSeniorFacultyResearchandStudyLeavefromtheUNCProvost’sOffice.ResearchsupportedinpartbyNSFawardDEB-1120062totheauthors.

CONFLIC T OF INTERE S T

Nonedeclared.

DATA ARCHIVING S TATEMENT

DataandR codeused in this studywill be available at theDryadDigital Repository at https://datadryad.org/resource/doi:10.5061/dryad.72mr1.

ORCID

Joel G. Kingsolver http://orcid.org/0000-0002-9839-0208

R E FE R E N C E S

Anderson,J.T.,Inouye,D.W.,McKinney,A.M.,Colautti,R.I.,&Mitchell-Olds, T. (2012). Phenotypic plasticity and adaptive evolution

contributetoadvancingfloweringphenologyinresponsetoclimatechange. Proceedings of the Royal Society B: Biological Sciences, 279,3843–3852.https://doi.org/10.1098/rspb.2012.1051

vanAsch,M.,Salis, L.,Holleman,L. J.M., vanLith,B.,&Visser,M.E.(2013).Evolutionaryresponseoftheegghatchingdateofaherbivo-rousinsectunderclimatechange.Nature Climate Change,3(3),244–248. https://doi.org/www.nature.com/nclimate/journal/v3/n3/abs/nclimate1717.html#supplementary-information

Boggs,C.L.,&Inouye,D.W.(2012).Asingleclimatedriverhasdirectandindirecteffectson insectpopulationdynamics.Ecology Letters,15,502–508.https://doi.org/10.1111/j.1461-0248.2012.01766.x

Bradshaw,W.E.,&Holzapfel,C.M.(2001).Geneticshiftinphotoperiodicresponsecorrelatedwithglobalwarming.Proceedings of the National Academy of Sciences, 98, 14509–14511. https://doi.org/10.1073/pnas.241391498

Brakefield,P.M.(1987).TropicaldryandwetseasonpolyphenisminthebutterflyMelanitis leda(Satyrinae):Phenotypicplasticityandclimaticcorrelates. Biological Journal of the Linnean Society, 31, 175–191.https://doi.org/10.1111/j.1095-8312.1987.tb01988.x

Brakefield, P. M., & Larsen, T. B. (1984). The evolutionary signifi-cance or dry and wet season forms in some tropical butterflies.Biological Journal of the Linnean Society, 22, 1–12. https://doi.org/10.1111/j.1095-8312.1984.tb00795.x

Buckley,L.B.,&Kingsolver,J.G.(2012).Thedemographicimpactsofshiftsinclimatemeansandextremesonalpinebutterflies.Functional Ecology,26,969–977.https://doi.org/10.1111/j.1365-2435.2012.01969.x

Campbell, G. S., & Norman, J. M. (1998). An introduction to environ-mental biophysics. New York, NY: Springer Science. https://doi.org/10.1007/978-1-4612-1626-1

Charmantier, A., & Gienapp, P. (2014). Climate change and timing ofavian breeding andmigration: Evolutionary versus plastic changes.Evolutionary Applications, 7, 15–28. https://doi.org/10.1111/eva.12126

Charmantier,A.,McCleery,R.H.,Cole,R.H.,Perrins,L.R.,Kruuk,L.E.B.,&Sheldon,B.C.(2008).Adaptivephenotypicplasticityinresponsetoclimatechangeinawildbirdpopulation.Science,320,800–803.https://doi.org/10.1126/science.1157174

Chevin,L.,Lande,R.,&Mace,G.M. (2010).Adaptation,plasticity,andextinctioninachangingenvironment:Towardsapredictivetheory.PLoS Biology,8,357–365.

Dennis,B.,Kemp,W.P.,&Beckwith,R.C. (1986).Stochasticmodelofinsectphenology:Estimationandtesting.Environmental Entomology,15,540–546.https://doi.org/10.1093/ee/15.3.540

Ellers,J.,&Boggs,C.L.(2002).EvolutionofwingcolorinColiasbutter-flies:Heritability,sex-linkage,andpopulationdivergence.Evolution,56,836–840.https://doi.org/10.1111/j.0014-3820.2002.tb01394.x

Ellers, J., & Boggs, C. L. (2004). Functional ecological implications ofintraspecific differences inwingmelanization inColias butterflies.Biological Journal of the Linnean Society, 82, 79–87. https://doi.org/10.1111/j.1095-8312.2004.00319.x

Forrest,J.R.(2016).Complexresponsesofinsectphenologytoclimatechange. Current Opinion in Insect Science, 17, 49–54. https://doi.org/10.1016/j.cois.2016.07.002

Forrest,J.,&Miller-Rushing,A.J.(2010).Towardasyntheticunderstand-ingoftheroleofphenologyinecologyandevolution.Philosophical Transactions of the Royal Society, 365, 3101–3112. https://doi.org/10.1098/rstb.2010.0145

Higgins,J.K.(2014).Rapid evolution and population divergence in response to environmental change in Colias butterflies. (PhD), University ofNorthCarolina,ChapelHill,NC.

Higgins,J.K.,MacLean,H.J.,Buckley,L.B.,&Kingsolver,J.G.(2014).Geographicdifferencesandmicroevolutionarychangesinthermalsensitivity of butterfly larvae in response to climate. Functional Ecology, 28(4), 982–989. https://doi.org/10.1111/1365-2435. 12218

     |  1243KINGSOLVER aNd BUCKLEY

Hoffman,R.J.(1978).Environmentaluncertaintyandevolutionofphys-iological adaptation in Colias butterflies. American Naturalist, 112,999–1015.https://doi.org/10.1086/283343

Inouye, D. W. (2008). Effects of climate change on phenology, frostdamage,andfloralabundanceofmontanewildflowers.Ecology,89,353–362.https://doi.org/10.1890/06-2128.1

Karell,P.,Ahola,K.,Karstinen,T.,Valkama,J.,&Brommer,J.E. (2011).Climate change drives microevolution in a wild bird. Nature Communications, 2, 208.https://doi.org/www.nature.com/ncomms/journal/v2/n2/suppinfo/ncomms1213_s1.html

Kingsolver, J. G. (1979). Thermal and hydric aspects of environmentalheterogeneityinthepitcherplantmosquito.Ecological Monographs,49,357–376.https://doi.org/10.2307/1942468

Kingsolver, J.G. (1981).Thermoregulatory strategies of Colias butterflies: adaptation in variable environments. (PhD),StanfordUniversity,PaloAlto,CA.

Kingsolver, J. G. (1983a). Ecological significance of flight activ-ity in Colias butterflies: Implications for reproductive strategyand population structure. Ecology, 64, 546–551. https://doi.org/10.2307/1939974

Kingsolver, J.G. (1983b).Thermoregulation and flight inColias butter-flies: Elevational patterns andmechanistic limitations. Ecology, 64,534–545.https://doi.org/10.2307/1939973

Kingsolver,J.G.,&Buckley,L.B. (2015).Climatevariabilityslowsevo-lutionary responses ofColias butterflies to recent climate change.Proceeding of the Royal Society B, 282, 1802–1809. https://doi.org/10.1098/rspb.2014.2470

Kingsolver,J.G.,&Buckley,L.B.(2017).Evolutionofplasticityandadap-tiveresponsestoclimatechangealongclimategradients.Proceeding of the Royal Society B,284,387–393.

Kingsolver,J.G.,&Huey,R.B. (1998).Selectionandevolutionofmor-phologicalandphysiologicalplasticityinthermallyvaryingenviron-ments. American Zoologist, 38, 545–560. https://doi.org/10.1093/icb/38.3.545

Kingsolver, J. G., & Watt, W. B. (1983). Thermoregulatory strategiesin Colias butteflies: Thermal stress and the limits to adaptation intemporally varying environments.American Naturalist,121, 32–55.https://doi.org/10.1086/284038

Kingsolver,J.G.,&Watt,W.B.(1984).Mechanisticconstraintsandop-timality models: Thermoregulatory strategies in Colias butterflies.Ecology,65,1835–1839.https://doi.org/10.2307/1937780

Lande, R., & Arnold, S. J. (1983). The measurement of selection oncorrelated characters. Evolution, 37, 1210–1226. https://doi.org/10.1111/j.1558-5646.1983.tb00236.x

Leal,M.,&Gunderson,A.R.(2012).Rapidchangeinthethermaltoler-anceofatropicallizard.American Naturalist,180,815–822.https://doi.org/10.1086/668077

Levins, R. (1968). Evolution in changing environments. Princeton, NJ:PrincetonUPress.

Lynch,M.,&Lande,R. (1993).Evolutionandextinction in response toenvironmental change. In P. M. Kareiva, J. G. Kingsolver, & R. B.Huey(Eds.),Biotic Interactions and Global Change.Sunderland,Mass:SinauerAssociates.

MacLean,H.J.(2015).Adaptation to climate and climate change in Rocky Mountain butterflies: Morphology, physiology and behavior. (PhD),UniversityofNorthCarolina,ChapelHill,NC.

MacLean, H. J., Kingsolver, J. G., & Buckley, B. A. (2016). Historicalchangesinthermoregulatorytraitsofalpinebutterfliesrevealcom-plexecologicalandevolutionaryresponsestorecentclimatechange.Climate Change Responses, 3, 13–22. https://doi.org/DOI 10.1186/s40665-016-0028-x

McGuire,C.R.,Nufio,C.R.,Bowers,M.D.,&Guralnick,R. P. (2012).Elevation-dependent temperature trends in the rocky mountainfront range: Changes over a 56-and 20-year record. PLoS ONE, 7,e44370.https://doi.org/10.1371/journal.pone.0044370

Merila,J.,&Hendry,A.P.(2014).Climatechange,adaptation,andphe-notypic plasticity: The problem and the evidence. Evolutionary Applications,7(1),1–14https://doi.org/10.1111/eva.12137

Mitchell, J.,Beckman,W.,Bailey,R.,&Porter,W. (1975).Microclimatemodelingofthedesert.InD.A.DeVries&N.H.Afgan(Eds.),Heat and mass transfer in the bioshpere. Part I, transfer processes in the plant environment(pp.275–286).Washington,DC:ScriptaBookCo.

Moran,N.A.(1992).Theevolutionarymaintenanceofalternativepheno-types.American Naturalist,139,681–706.

Ozgul, A., Childs, D. Z., Oli, M. K., Armitage, K. B., Blumstein, D. T.,Olson,L.E.,&Coulson,T. (2010).Coupleddynamicsofbodymassandpopulationgrowthinresponsetoenvironmentalchange.Nature,466(7305),482–485.https://doi.org/10.1038/nature09210

Parmesan, C. (2006). Ecological and evolutionary responses to re-cent climate change. Annual Reviews of Ecology, Evolution and Systematics, 37, 637–669. https://doi.org/10.1146/annurev.ecolsys.37.091305.110100

Parton,W.J.,&Logan,J.A.(1981).Amodelfordiurnalvariationinsoilandairtemperature.Agricultural Meteorology,23,205–216.https://doi.org/10.1016/0002-1571(81)90105-9

Pau, S.,Wolkovich, E. M., Cook, B. I., Davies, T. J., Kraft, N. J. B., &Bolmgrem,K. (2011).Predictingphenologyby integratingecology,evoluionandclimatescience.Global Change Biology,17,3633–3643.https://doi.org/10.1111/j.1365-2486.2011.02515.x

Porter,W.P.,&James,F.C.(1979).Behavioralimplicationsofmechanis-ticecology II:TheAfricanrainbowlizard,Agamaagama.Copeia,4,594–619.https://doi.org/10.2307/1443867

Porter,W.P.,Mitchell, J.W.,Beckman,W.A.,&DeWitt,C.B. (1973).Behavioral implications of mechanistic ecology. Thermal and be-havioralmodelingofdesertecothermsandtheirmicroenvironment.Oecologia,13,1–54.https://doi.org/10.1007/BF00379617

Sgro,C.M.,Terblanche,J.S.,&Hoffmann,A.A.(2016).Whatcanplas-ticity contribute to insect responses to climate change? Annual Review of Entomology, 61, 433–451. https://doi.org/10.1146/annurev-ento-010715-023859

Shapiro,A.M.(1976).Seasonalpolyphenism.InM.K.Hecht,W.C.Steere,&B.Wallace(Eds.),Evolutionary Biology,Vol.9 (pp.259–333).NewYork,NY:PlenumPress.https://doi.org/10.1007/978-1-4615-6950-3

Stanton,M. L. (1984). Short-term learning and the searching accuracyof egg-laying butterflies. Animal Behavior, 32, 33–40. https://doi.org/10.1016/S0003-3472(84)80321-8

Tabashnik, B. E. (1980). Population structure of Pierid butterflies.Oecologia,47,175–183.https://doi.org/10.1007/BF00346817

Tafani, M., Cohas, A., Bonenfant, C., Gaillard, J. M., & Allaine, D.(2013). Decreasing litter size ofmarmots over time: A life historyresponse to climate change? Ecology, 94, 580–586. https://doi.org/10.1890/12-0833.1

Tauber, M. J., & Tauber, C. A. (1976). Insect seasonality: Diapausemaintenance, termination, and postdiapause development. Annual Review of Entomology,21,81–107.https://doi.org/10.1146/annurev.en.21.010176.000501

Thackeray,S. J.,Henrys,P.A.,Hemming,D.,Bell, J.R.,Botham,M.S.,Burthe,S.,&Wanless,S.(2016).Phenologicalsensitivitytoclimateacrosstaxaandtrophiclevels.Nature,535(7611),241–245.https://doi.org/10.1038/nature18608

Vedder,O.,Bouwhuis,S.,&Sheldon,B.C.(2013).Quantitativeassess-mentoftheimportanceofphenotypicplasticityinadaptationtocli-mate change inwild bird populations.PLoS Biology,11, e1001605.https://doi.org/10.1371/journal.pbio.1001605

Watt, W. B. (1968). Adaptive significance of pigment polymor-phisms inColias butterflies. I. Variation ofmelanin pigment in re-lation to thermoregulation. Evolution, 22, 437–458. https://doi.org/10.1111/j.1558-5646.1968.tb03985.x

Watt, W. B. (1969). Adaptive significance of pigment polymorphismsin Colias butterflies, II. Thermoregulation and photoperiodically

1244  |     KINGSOLVER aNd BUCKLEY

controlledmelaninvariation inColias eurytheme. Proceedings of the National Academy of Sciences of the United States of America,63,767–774.https://doi.org/10.1073/pnas.63.3.767

Watt,W.B.,Chew,F.S.,Snyder,L.R.G.,Watt,A.G.,&Rothschild,D.E.(1977).PopulationstructureofPieridbutterflies.I.Numbersandmovements of some montane Colias species.Oecologia, 27, 1–22.https://doi.org/10.1007/BF00345682

Watt,W.B.,Han,D.,&Tabashnik,B.E.(1979).PopulationstructureofPierid butterflies II. A “native” population ofColias philodice eriph-yle in Colorado. Oecologia, 44, 44–52. https://doi.org/10.1007/BF00346396

SUPPORTING INFORMATION

AdditionalSupportingInformationmaybefoundonlineinthesup-portinginformationtabforthisarticle.

How to cite this article:KingsolverJG,BuckleyLB.Howdophenology,plasticity,andevolutiondeterminethefitnessconsequencesofclimatechangeformontanebutterflies?Evol Appl. 2018;11:1231–1244. https://doi.org/10.1111/eva.12618