Latitudinal variation in biophysical characteristics of avian eggshells to cope with differential effects of solar radiation

Ecology and Evolution. 2018;8:8019–8029.   | 8019 www.ecolevol.org Received:15February2018 | Revised:8June2018 | Accepted:13June2018 DOI:10.1002/ece3.4335 ORIGINAL RESEARCH Latitudinal variation in biophysical characteristics of avian eggshells to cope with differential effects of solar radiation Jesús Gómez1 | Cristina Ramo1 | Martin Stevens2 | Gustavo Liñán-Cembrano3 | Miguel A. Rendón1 | Jolyon T. Troscianko2 | Juan A. Amat1 ThisisanopenaccessarticleunderthetermsoftheCreativeCommonsAttributionLicense,whichpermitsuse,distributionandreproductioninanymedium, providedtheoriginalworkisproperlycited. ©2018TheAuthors.Ecology and EvolutionpublishedbyJohnWiley&SonsLtd. 1DepartamentodeEcologíade Humedales,EstaciónBiológicadeDoñana (EBD-CSIC),Sevilla,Spain 2CentreforEcologyand Conservation,CollegeofLifeand EnvironmentalSciences,Universityof Exeter,Penryn,UK 3InstitutodeMicroelectrónicadeSevilla (IMSE-CNMCSIC/UniversidaddeSevilla), Sevilla,Spain Correspondence JuanA.Amat,DepartamentodeEcologíade Humedales,EstaciónBiológicadeDoñana (EBD-CSIC),Sevilla,Spain. Email:[email protected] Funding information Financialsupportwasreceivedfrom EstaciónBiológicadeDoñanaEBD- CSICthroughSeveroOchoaProgramme forCentresofExcellence(grantSEV- 2012-0262,MinisteriodeEconomíay CompetitividadofSpain),andpartlyby grantsCGL2011-24230andCGL2017- 83518-PfromtheSpanishGovernment, withEU-ERDFfinancialsupport.JG wassupportedbyanFPUpredoctoral fellowship(FPU12-01616)fromMinisterio deEducación,CulturayDeporte,Spain. MSandJTwerefundedbyaBiotechnology andBiologicalSciencesResearchCouncil (BBSRC)grantBB/J018309/1toMS. Abstract Solarradiationisanimportantdriverofanimalcoloration,notonlybecauseofthe effectsofcolorationonbodytemperaturebutalsobecausecolorationmayprotect fromthedeleteriouseffectsofUVradiation.Indeed,darkcolorationmayprotect fromUV,butmayincreasetheriskofoverheating.Inaddition,theeffectofcolorationonthermoregulationshouldchangewitheggsize,assmallereggshavehigher surface-volumeratiosandgreaterconvectivecoefficientsthanlargereggs,sothat small eggscandissipate heat quickly. Wetestedwhether the reflectanceof eggshells,eggspottiness,andeggsizeoftheground-nestingKentishploverCharadrius alexandrinus is affected by maximum ambient temperature and solar radiation at breedingsites.Wemeasuredreflectance,bothintheUVandhumanvisiblespectrum,spottiness,andeggsizeinphotographsfromamuseumcollectionofplover eggshells. Eggshells of lower reflectance (darker) were found at higher latitudes. However,insouthernlocalitieswheresolarradiationisveryhigh,eggshellsarealso ofdarkcoloration.Eggshellcolorationhadnosignificantrelationshipwithambient temperature. Spotiness was site-specific. Small eggs tended to be light-colored. Thermalconstraintsmaydrivetheobservedspatialvariationineggshellcoloration, whichmaybelighterinlowerlatitudestodiminishtheriskofoverheatingasaresult ofhigherlevelsofsolarradiation.However,insouthernlocalitieswithveryhighlevelsofUVradiation,eggshellsareofdarkcolorationlikelytoprotectembryosfrom moreintenseUVradiation.Eggsizeexhibitedvariationinrelationtocoloration,likely throughtheeffectofsurfacearea-to-volumeratiosonoverheatingandcoolingrates ofeggs.Therefore,differentialeffectsofsolarradiationonfunctionsofcoloration andsizeofeggshellsmayshapelatitudinalvariationsineggappearanceintheKentish plover. KEYWORDS biogeographicalpattern,biophysicalmechanisms,Charadrius alexandrinus,eggcoloration,egg size,latitudinalgradient,UVprotection 8020 | GÓMEZ Et al. 1 | INTRODUCTION Oneofthemostthoroughlystudiedtopicsofanimalcolorationis that of melanism, the occurrence of individuals that are darker in pigmentation. Several hypotheses have been advanced to explain the existence of melanism, the most prominent including links to camouflageandsolarradiation,suchasthermoregulationandprotectionfromultraviolet(UV)radiation(seeClusella-Trullas,vanWyk, &Spotila,2007).Thesedifferentfunctionsofpigmentationshould placeconsiderableselectionpressureonanimalappearance,which inturnshouldalsobeinfluencedbyenvironmentalfactorssuchas solarradiationandhabitat. Thesteady-statetemperaturethatanorganismreachesinthe absence of metabolic heating and evaporative cooling is referred toasoperativetemperature,whichdependsonabsorbedradiation, airtemperature,andwindspeed(Bakken,1992).Giventhatcolorationaffectstheamountofenergyabsorbedorreflectedatdifferentwavelengths,theoperativetemperatureofanorganismmaybe affectedbyitscolor.Underthethermoregulationhypothesis,melanismisadvantageousin cold climates,and lighter individualsare expectedtooccurinhotterareasgiventhelowerabsorptanceby lightcolors(Bishopetal.,2016).Consistentwiththishypothesis,a geographicalvariationinthedegreeofmelanismhasbeenfoundin varioustaxaofectotherms,withdarkerindividualsbeingfoundat higherlatitudesandlighteronesatlowerlatitudes(Alhoetal.,2010; Brakefield,1984;MorenoAzócaretal.,2015;Rapoport,1969).The UVresistancehypothesispositsthatdarkcolorsreducethetransmissionofUVlightthroughbodylayers(Bastide,Yassin,Johanning, &Pool,2014;Thompson,1955).Therefore,changesincolortocope with temperatureand UVradiationarelikelywhen differencesin bodytemperatureandUVprotectionaffectfitness(Clusella-Trullas, Terblanche, Blackburn, & Chown, 2008; Umbers, Herbestein, & Madin,2013). Thesameprinciplesthatareappliedtoectothermsregardingthe relationshipbetweentheradiativeenvironmentandcolorationmight beappliedtoeggshells.Therefore,asinectotherms,changesincolorationwithsolarradiationshouldbeexpectedinavianeggshells. Theeggsmayremainexposedtoenvironmentalconditionsduring absences of adultsfromnests, whichmay becriticalforembryos ofspeciesnestingatgroundlevelinexposedsites,asunattended eggsreceivingsolarradiationmayoverheatinaveryfewminutes (Amat, Gómez, Liñán-Cembrano, Rendón, & Ramo, 2017; Amat & Masero,2007;Grant,1982;Wilson-Aggarwal,Troscianko,Stevens, &Spottiswoode,2016).Avoidingoverheatingisimportantbecause high body temperatures denature proteins. Likely because of this therearesomebiophysicalmechanismswithwhichbirdsmaycounteractthenegativeeffectsofhightemperatureonembryos’overheating.Oneofsuchmechanismsiseggshellcolor,aslightereggs overheatlessquicklythandarkereggswhenexposedtodirectsolar radiation (Gómez etal., 2016; Lahti & Ardia, 2016; Montevecchi, 1976).However,theeggshellspottingtypicalofmanyspeciesmakes theeggsdarker(Gómezetal.,2016;Troscianko,Wilson-Aggarwal, Stevens, & Spottiswoode, 2016), so that the degree of spottiness may affect the rates of egg overheating. At least during the first momentsaftereggexposuretodirectsolarradiation,itmaybeexpected that spots overheat quicker than the eggshell background (seeWacker,McAllan,Körtner,&Geiser,2016),althoughafterlongerexposure,theeggsmayreachequilibriumtemperaturesacross theirsurfaces. Eggshell color may also affect embryo viability through the probabilityofUVtransmittancethroughouttheeggshell(Veterány, Hluchý, & Veterányová, 2004). UV-B is strongly mutagenic, and mostmutationsareharmful(e.g.,deGruij&Forbes,1995).Because theeggscontaintheDNA,harmfulmutationscouldbeinherited bythenextgeneration(e.g.,Flenley,2011).Darkercolorsreduce lighttransmittancethrougheggshells,thusprotectingtheembryo from UV radiation (Abram etal.,2015; Brulez, Pike, & Reynolds, 2015; Gaudreau, Abram, & Brodeur, 2017; Lahti & Ardia, 2016; Maurer etal., 2015; Shafey, Ghannam, Al-Batshan, & Al-Ayed, 2004). Because the absorptance by darker colors is higher than that of lighter colors,thismay resultinatrade-offbetweenthe riskofeggoverheatingandtheriskof UVradiationonembryos if ambient temperature and solar radiation are positively related (Lahti&Ardia,2016):Darkereggshellswouldprotecttheembryo fromUVradiation,butinturnwouldincreasetheriskofoverheating.However,thethermoregulationandUVprotectionhypothesesarenotnecessarilymutuallyexclusive,forinstancewhenthe relationshipbetweensolarradiationandtemperatureisnonlinear, asitmayoccurintropicalmountainswhereradiationishighbut temperaturelow. The effect of coloration on thermoregulation should change witheggsize,sothattherewouldbeapositiverelationshipbetween eggsizeanddarkcoloration(Gates,1980),apredictionsupported bytheoreticalmodelsandempiricalevidence(Clusella-Trullasetal., 2008; Schweiger & Beierkuhnlein, 2016). Therefore, another biophysicalmechanismwithwhichbirdsmaycounteractthenegative effectsofhightemperatureonembryos’overheatingiseggsize,as smallereggshavehighersurface-volumeratiosandgreaterconvectivecoefficientsthanlargereggs,sothatsmalleggscandissipate heatquickly(e.g.,Porter&Gates,1969). Giventheaboveconsiderations,dependingontheenvironment, theimportanceofselectivedriversoneggshellcolorationmayvary spatially,leadingtocomplexinteractionsamongfactorsthataffect variationsincolorpatterns(Ahlgren,Yang,Hansson,&Brönmark, 2013;Bastideetal.,2014; Lindstedt,Linström,&Mappes, 2008), aswellasineggsize,tocopewithsolarradiationindifferentenvironments. The resolution of the above trade-offs would depend on which effects are more limiting on embryos’ viability. In this study, we tested whether the reflectance of Kentish plover Charadrius alexandrinuseggshells,bothinthehumanvisible(VIS) and UV spectrum, is affected by maximum ambient temperature (whichmay affect egg temperatures through thermal convection) andsolarradiation(includingtotalmaximumradiationincidenton thesurfaceofEarth,whichmayaffecteggtemperaturesthrough thermalradiation)acrossalargegeographicalrange.Kentishplovers aresmallshorebirdsthatusuallynestinexposedsitesthatreceive | 8021 GÓMEZ Et al. directsolarradiation(Figure1).Wealsotestedtherelationshipbetweenbothcolorationandspottingpatternsofeggshellsandegg size.Weexpectedthateggshellswithhigherreflectance(i.e.,those oflightercolors)andlessspottingshouldbefoundinsiteswhere solarradiationishigher.However,ifUVradiationisveryhigh,the impactofUVradiationonembryoviabilitymaybestrongerthanthe riskofeggoverheating,inwhichcaseeggshellswithlowerreflectanceandmorespottingshouldbeexpectedinsiteswithveryhigh solarradiationlevels,eveniftheriskofeggoverheatingishigh(see Bastideetal.,2014).Lastly,wetestedwhethertherearevariations ineggsizeinrelationtoeggshellcolor,sothatlightereggsshould be smaller, given the thermal advantages of such eggs in hotter environments. 2 | METHODS 2.1 | Study species and photography TheKentishploverisashorebird widelydistributedacrossthe Palearctic.Itnestsatgroundlevelinexposedsites.Lackofnest coverallows aquickdetectionofapproaching predators by incubating birds (Amat & Masero, 2004a), but eggs overheat to temperatures that may be lethal for embryos in <1min when left unattended by adults and receiving direct solar radiation (Amat&Masero,2007;Amatetal.,2017;Grant,1982).Theabsencesofshorebirdsfromnestsmaylast3–10min,butthereare caseslasting>4hr(Pedler,Weston,&Bennett,2016),suggesting that there may be selective pressure on eggs to have pigmentationthatwouldprotectthemagainstadverseeffectsofsolar radiation. Forthisstudy,weusedtheKentishplovereggshellsinthecollectionoftheNaturalHistoryMuseumatTring(UnitedKingdom). We noted the date and locality of collection of every clutch. Using GoogleTMEarth,weobtainedthecoordinatesofthoselocalities. The eggs had been collected between 1,858 and 1,972, across a geographical range encompassing 11.35°N–54.52°N and 13.86°W–77.25°E(Figure2a). EggshellswerephotographedfollowingtheprotocolsofStevens, Párraga, Cuthill, Partridge, and Troscianko (2007), Troscianko and Stevens(2015),andGómezandLiñán-Cembrano(2017).Onlyone eggshellperclutch(n=110)wasphotographedusingstandardized lightingconditionsprovidedbyaUVlamp(EYEColorARCMT70D, IwasakiElectricCo.,Ltd.,Tokyo,Japan)diffusedwithasilverphotographicumbrella,usingaNikkonD7000camerawitha105mm Micro-Nikkor lens that transmits UV (see Troscianko etal., 2016). ThecamerahadundergoneaquartzconversiontoallowthesensortodetectUVlight.UsingaBaaderUV-IRblockingfilter(Baader Planetarium,Mammendorf,Germany),wetookphotographsinthe human visible spectrum (400–700nm, which reflects light in the RGB),andwithaBaaderUV-passIRblockingfilter,wecouldtakeimagesintheUVspectrum(approximately315–400nm).Thecamera wasmountedonacamerastand,andphotographsweretaken,with ashuttercable,atf/4inRAWformat.Weplacedbesidetheeggs aSpectralongreystandard(40%UV-visible,Labsphere,Congleton, UK)thatreflectslightintheUVandhumanvisiblespectra.Ametric scalewasincludedinallphotographs. 2.2 | Image analysis Allimageswerelinearizedwithrespecttoradianceusingthetool SpotEgg (Gómez & Liñán-Cembrano, 2017). Values from the Spectralon grey standard were employed for normalization to standardizeimageswithregardtotheilluminatinglightandconvert thedatatopercentreflectance(Gómez&Liñán-Cembrano,2017; Stevensetal.,2007;Troscianko&Stevens,2015).Oncethereflectanceimagesweregenerated,thenextstepwastomanuallydraw closedpolygonallinesdefiningtheregionsofinterest(RoIs),where spottinessandcoloranalyseswereexecuted.Finally,weanalyzed colorandspottiness.Forthis,SpotEggemploysanimage-processing algorithmtosegmentthespotsfromthebackgroundineachofthe RoIsonanimage.Thespotdetectionroutineproduceddetailed.CSV reportswithinformationaboutreflectance(inthered,green,and blue camera’s color bands) for spots (S) and eggshell background (B),aswellasthesizeofeachdetectedspot,thenumberofspots, andtotalspottiness(percentageoftheareaofeggshellcoveredby spots). For the UV images, the sensitivity of the camera system (filter+lens+imagesensor) inthe UVis significantly lowerthanfor thevisiblewavelengths,requiringlongerexposuretimes,suchthat imagessignal-to-noiseratioisdegraded.Thesetwofactsmaypose difficultiestodetectthespotsintheUVimages.Consequently,we optedforusingtheimagesofspotsdetectedinthevisibleimageas amaskforevaluatingreflectanceofbothspotsandeggshellbackgroundintheUVband.Althoughwetriedtokeepthecamerapositionwhenthefilterswerechanged,itwasinevitablethattheUV andvisibleimageswereinpracticetakenfromslightlydifferentpositions,sothatbothimageswerenotcoincident.Hence,usingspot informationfromthevisibleimageasamasktomeasurereflectance FIGURE1 FemaleKentishplover(Charadrius alexandrinus) besideitsnest(photocredit:XavierFerrer) 8022 | GÓMEZ Et al. in the UV images required applying a space-variant geometrical transformationtothemaskimagetoensurepropermatching.We implementedapluginforSpotEggthatinferredthespatialtransformationbetweenthevisibleandUVimagesfromasetofatleast13 correspondingpoints(manuallymarkedasfeaturematchingdidnot workduetothelow-qualityUVimages).Oncethecontrolpairsof points had been selected, SpotEgg employed the Local Weighted MeanmethodproposedbyGoshtasby(1998)(especiallywellsuited whenthedistortionvarieslocally)tofindthegeometricaltransformationbetweentheUVandvisibleimages.Thistransformationwas appliedtotheblackandwhiteimagewiththelocationofthespots that were produced for the visible image. Theresult was another blackandwhiteimagecorrespondingtospotsthatwerecorrectly alignedwiththeircorrespondingUVimage.ThisUV-spotimagewas thenusedasamasktoobtainthereflectanceofbotheggshellspots andbackgroundintheUVband. As measuresof color, we used the mean values of thethree camerabands(red,greenandblue)intheVIS,andtheredband intheUV.Inaddition,usingSpotEgg(Gómez&Liñán-Cembrano, 2017), we also estimated egg volume (mm3) and surface (mm2), whichresultedfromintegratingRoIsasarevolvingsurfaceshape generator. 2.3 | Environmental variables WeobtaineddataonthemonthlyaverageamountofthetotalradiationincidentonahorizontalsurfaceatthesurfaceoftheEarth atnoon(averageof3hratthetimeclosesttothelocalsolarnoon), aswellasmaximumdailyEarth’ssurfacetemperaturefromNASA (https://eosweb.larc.nasa.gov/cgi-bin/sse/grid.cgi?email=skip@larc. nasa.gov).Themeteorologicaldatawereonascaleof1-degreelongitudeby1-degreelatitude,coveringtheentirerangefromwhich eggs were obtained, and were averaged on a monthly basis over a22-yearperiod(July 1983–January2005) for the corresponding quadratofeverylocality.Weonlyuseddataforthemonthsduring whichtheKentishploverisbreedingateverylocality(layingdates obtainedfromWiersma,Kirwan,&Boesman,2016).Thereisalatitudinalgradientinradiationandtemperatureacrossthelocalitiesat whichtheeggswerecollected(Figure2b). 2.4 | Statistical procedures Weperformedaprincipalcomponentsanalysis(PCA)fromacorrelation matrix on eggshell color and spotpatterning to obtain a smallersetofuncorrelatedcomponentsthatrepresentsmostofthe informationintheoriginalvariables.BeforePCA,log-transformation linearizedtherelationshipsbetweeneggcolorandspotvariables. Generalized additive mixed models (GAMMs) were applied toestimate spatialvariationinegg colorandpatterning(Wood& Augustin, 2002). In contrast to GAMs, GAMMs permit spatio- temporal correlationwithin blocksusing random effects. In order to model spatial patterns, we included the latitude of nests as a smoother.Furthermore,thespatialmodelincludedtheeffectofthe yearofcollection as linearcovariatetotest fortemporalchanges FIGURE2 (a)Mapshowingthespatial distributionofcollectionsitesforeggsof Kentishplover.Symbolsizeisproportional tosamplesize.(b)Relationshipsbetween latitudeandbothsolarradiationand temperatureinthelocalitieswherethe eggswerecollected | 8023 GÓMEZ Et al. ineggcolorationandpatterningduringlong-termstorage(Cassey, Maurer, Duval, Ewen, & Hauber, 2010; Navarro & Lahti, 2014). Finally,thecollectionsitenameswereconsideredasarandomeffect,resultinginasemiparametricmixedmodel. Theinfluenceoftheenvironment(temperatureandsolarradiation) on eggshell color and spot patterns was also analyzed using GAMMs, with collection site name as a random effect. Environmental models included year as a linear covariate if this variable was previously chosen in the spatio-temporal model. Furthermore,wetestedthelineareffectofeggsize(surfacearea- to-volumeratio)oneggcoloration(Gates,1980).Thesmall-sample- size corrected version of Akaike information criterion (AICc; Burnham, Anderson, & Huyvaert, 2011) was used to compare competingmodels(Zuur, Ieno, Walker, Savaliev,&Smith, 2009). Modelswith≤2AICofthetopmodelareconsideredascompetitive.However,uninformativeparameterswithmodels≤2ΔAICc (i.e.,donotexplainenoughvariation)wereinterpretedashaving noeffectontheresponse(Arnold,2010).Usingafullfixedeffect model,weassessedtheinfluenceoftherandomcomponentusing AICcbasedonREMLestimators.Theoptimalfixedeffectsstructure was alsoselected usingAICc,but based onML estimators. WerestrictedtheGAMMstoamaximumoffiveknotstoprevent over-fitting.Then,wecheckedifpolynomialorlinearmodelsfitted bettertodatathanGAMM.Fortheparametricmodels,therelativeexplanatorypoweroffixed(insolationandtemperature)and random (site identity) effects was determined using conditional andmarginalR2sforGeneralizedmixed-effectmodels(Nakagawa & Schielzeth, 2013). We measured for concurvity (the generalizedadditivemodelanaloguetocollinearity)betweencovariates (Hastie&Tibshirani, 1990),whichmayresultin inaccurateestimatesoftheGAMfunctions(Ramsay,Burnett,&Krewski,2003). The level of concurvity varies between 0, no concurvity, and 1, total lack of identifiability (Wood, 2017). In this study, pairwise values of observed concurvity between explanatory variables, both in spatio-temporal and environmentalmodels,rangedfrom low to moderatelyhigh(0.27–0.69).Residualsfromtheenvironmentalmodelsforeggshellcolorandspotpatternsdidnotpresentalarge-scalelatitudinaltrend(PC1:edf(equivalentdegreesof freedom)=2.66,p=0.167;PC2:edf=1.94,p=0.374)norspatial patterns, as determined by semivariance analysis and Mantel’s test(resultsnotshown).GAMMswereconductedwithRsoftware version3.3.3(RCoreTeam,2017)usingthepackagesmgcv1.8–17 (Wood,2006,2017)andMuMIN1.15.6(Barton,2016). 3 | RESULTS 3.1 | Eggshell reflectance and spot patterns Eggshell reflectance and spot pattern variables were intercorrelated,exceptspotnumberandreflectances(Table1).Thetwo-first principal components accounted for 85% of the variance of egg colorandspotpattern,sothatweretainedthemforfurtherconsideration(Figure3).Thefirstprincipal component(PC1)accounted for63%ofthevarianceandwasrelatedtoeggshellreflectance,with negativevaluesindicatingeggswithlargespotsanddarkerbackgroundsandspots,aswellaseggshellsandspotsthatreflectedless intheUV(eigenvectors:spotnumber=0.229,spotsize=−0.360, spottiness=−0.400, background VIS=0.428, spot VIS=0.390, backgroundUV=0.406,andspotUV=0.398).Thesecondprincipalcomponent(PC2)accountedfor22%ofthevarianceandrelated to patterning; positive values of PC2 were related to decreasing spotnumberbutincreasingspotsize(spotnumber=−0.655,spot size=0.523, spotiness=0.181, background VIS=0.218, spot VIS=0.295,backgroundUV=0.220,spotUV=0.286). 3.2 | Spatiotemporal model TheAICcvaluesforthemodelsincludinglatitudeandyearasfixed effects indicate that adding the site as a random effect significantlyimprovedmodelperformance,bothforeggshellreflectance (∆AICc=3.68)andspotpatterning(∆AICc=5.82). The eggshell reflectance (PC1) showed a spatial and temporalvariation(Table2).DespitetheGAMMthatincludedyearasa linearpredictorandlatitudeasasmootherwashighlysupported (AICc=450.59),includinglatitudeasalineartermsignificantlyimprovedmodelperformance(AICc=448.35).Eggshellstendedtobe lighterandwithmorereflectanceintheUVtowardlowerlatitudes (Figure4a;β=−0.052±[SE]0.022,p=0.019).Oncespatialstructurewasaccountedfor,theyearofcollectionwasrelatedtoeggshellcolor,witholdereggsbeingdarkerthanthosecollectedlater (β=0.027±0.010,p=0.010).Thelinearmodelhad R2 m =0.24due tospatio-temporalpattern,andaR2=0.15duetorandomeffects. B- VIS B - U V S - U V S- number S - s i z e S- area S - V I S 0.79*** 0.63*** 0.84*** 0.12ns −0.39*** −0.59*** B - V I S 0.90*** 0.74*** 0.22* −0.49*** −0.66*** B - U V 0.77*** 0.21* −0.45*** −0.60*** S - U V 0.17ns −0.40*** −0.55*** S-number −0.88*** −0.43*** S-size 0.81*** Notes.ns,nonsignificant. *p<0.05;**p<0.01;***p<0.001. TABLE1 Pearson’scorrelation coefficientsbetweenvariablesdescribing eggshellreflectance,forbothbackground (B)andspots(S),inthevisible(VIS)and ultraviolet(UV)spectrum,andspot patterning(number,size,andarea)inthe Kentishplover 8024 | GÓMEZ Et al. Spotpattern(PC2)didnotshowsignificantspatialortemporal patterns(Figure4b).Anullmodelincludingonlyrandomintercepts forsiteprovidedthebestfit(Table2). 3.3 | Environmental models For eggshell reflectance (PC1), the lower values of AICc were achieved when the model included year, egg surface-to- volume ratio, and insolation as explanatory variables (Table3). Temperature had no significant relationship with PC1. Surface area-to-volume ratio of the eggs was also positively related to PC1 (β=69.850±23.191, p=0.004), and thus, larger surface FIGURE3 Biplotofthetwo-firstprincipalcomponentsforthe variablesusedtodescribebotheggshell(S,spots,B,background) colorinthehumanvisible(VIS)andultraviolet(UV)spectrum,and spotpattern(size,area,andnumber)intheKentishplover Dependent variable Fixed effects df AICc ∆AICc wi R2 m R2 c PC1 Year+s (latitude) 6450.59 2.24 0.206 – – Year+latitude 5448.35 0.00 0.632 0.24 0.39 Latitude 4453.16 4.81 0.057 0.18 0.39 Year 4452.00 3.61 0.104 0.17 0.37 Nullmodel 3 464.12 16.12 0.000 0.00 0.37 PC2 Year+s (latitude) 6361.12 4.24 0.061 – – Year+latitude 5358.88 2.00 0.167 0.03 0.27 Latitude 4358.68 1.80 0.168 0.01 0.27 Year 4358.11 1.22 0.223 0.01 0.27 Nullmodel 3 356.89 0.00 0.381 0.00 0.27 Note. Parametric coefficients for latitude were estimated when the degrees of freedom for the splineestimates(s)equals1.Collectionsitewasenteredasanintercept-onlyrandomeffectinboth models.Thedifference betweenthelowestAICandtheAICc scoreof eachmodel(ΔAICc),the Akaikeweight(wi),thevarianceexplainedbythefixedfactors( R2 m ),andthevarianceexplainedbythe entiremodel( R2 c )arepresented. TABLE2 Resultsofthespatio- temporalGAMMsfittedtoeggshell reflectance(PC1)andspotpatterning (PC2) FIGURE4 (a)Partialplots(95%CI)ofGAMMsforthe relationshipsbetweenlatitudeandeggshellreflectance(PC1).(b) Spotpatterning(PC2).Onlythesignificantenvironmentalvariables wereincludedineachmodel | 8025 GÓMEZ Et al. area-to-volumeratios(smallereggs)tendedtobeassociatedwith lighter-coloredeggshells. Eggshell reflectance responded in a nonlinear fashion to solar radiation(Figure5).Thenonparametricsmoothingfunctionofsolar radiationwassignificant(e.d.f.=3.715,p=0.002).Whenthesmooth termwasreplacedbyafourth-orderpolynomialtermthemodelfit improved (Table3). Eggshell color became lighter from low values of solar radiation until reaching 0.55kW/m2 (Figure5). Thereafter, eggshellcolorvariedslightlyuntilradiationatgroundlevelwasnear 0.7kW/m2, and then became even lighter until values of 0.8kW/ m2,butwithhigherradiationlevels,eggshellsbecamedarkeragain. MarginalR2indicatesthatthefixedtermsaccountedformostofthe explainedvarianceofPC1( R2 m =0.45)relativetorandomeffect(0.03). Forspottingpattern(PC2),twomodelswerewithin2AICcunits ofeachotherandcompetedforbestmodelforthevariationinspot pattern (Table3). The model with the lower AICc value for PC2 was the null model (wi=0.415). The second best-fit mixed model containedthelineareffectofsolarradiationbutitwasnotsignificant (β=−1.289±1.216, p=0.295), had a lower Akaike weight (wi=0.266),andonlyexplained2%ofthevariation.Therefore,we consideredradiationasauninformativevariableonspottingpattern. Theseresultssuggestthatthespatialvariationofeggshellpatterningwasrathersite-specific( R2 c =0.27). 4 | DISCUSSION Thestudyofgeographicalvariationsinphenotypictraitsisusefulin theunderstandingoftheselectiveagentsthatshapethephenotypic variation(Roulin&Randin,2016).Ourresultsindicatethateggshell reflectanceintheKentishploveracrossalatitudinalgradient,both inthehumanvisibleandUVspectrum,isprimarilydeterminedby spatial variations in solar radiation, and that this may also affect eggsizeinawaythathasnotbeenappreciatedinpreviousstudies (seeWilliams,1994).Althoughsomestudieshaveshownthatlarger eggsarelaidincolderenvironments(Azevedo,French,&Partridge, 1996;Liefting,Weerenbeck,vanDooremalen,&Ellers, 2010), we showforthefirsttimearelationshipbetweeneggcolorandsize,as smallereggs(i.e.,thosewithhighersurface-volumeratio)tendtobe lighterthanlargereggs.Relativetodarkereggs,lighteggsoverheat moreslowlywhenexposedtodirectsolarradiation(Gómezetal., 2016),andsmallereggsinhotenvironmentsmaycooldownrelatively quickly comparedwithlarger eggs becauseheat dissipation increaseswithsurface/volumeratio(Porter&Gates,1969).Allthis indicatesthatinformationofeggsizeinrelationtoeggshellcolorationmaybeimportanttoforecastenvironmentalchangesensitivity ofground-nestingbirds.Incontrasttothis,birdsnestingincupnests above ground level use nest-building techniques as an adaptive Model df AICc ∆AICc wi R2 m R2 c PC1 Year+sv+s(insolation)+s (temperature) 9451.83 22.85 0.000 – – Year+s(insolation)+s (temperature) 8452.36 23.38 0.000 – – Sv+s(insolation)+s (temperature) 8449.81 20.83 0.000 – – Year+sv+s(insolation) 7 447.14 18.16 0.000 – – Year+sv+s(temperature) 7 447.81 18.83 0.000 – – Year+sv+insolation38442.02 13.04 0.001 0.34 0.47 Year+sv+insolation49428.98 0.00 0.998 0.45 0.48 Nullmodel 3 464.12 35.14 0.000 0.00 0.37 PC2 s(insolation)+s (temperature) 7363.76 6.88 0.014 – – s(insolation) 5360.06 3.18 0.089 – – s(temperature) 5361.23 4.35 0.049 – – Insolation 4357.87 0. 98 0.266 0.02 0.26 Temperature 4359.04 2. 15 0.148 0.00 0.27 Nullmodel 3 356.89 0.00 0.434 0.00 0.27 Note.Parametriclineareffectswere appliedfor surface/volume(sv)ofeggs andcollectionyear. Smoothingsplinefunctions(s)wereusedbothfortemperatureandsolarradiation,thenlinearand polynomial(cubicandquartic)modelswerealsofitted.Collectionsitewasenteredasarandomeffectinbothmodels. ΔAICc,thedifferencebetweenthelowestAICcandtheAICcscoreofeachmodel;wi,Akaikeweight; R2 m ,marginalR2; R2 c ,conditionalR2. TABLE3 Resultsoftheenvironmental GAMMsfittedtoeggshellreflectance (PC1)andspotpatterning(PC2) 8026 | GÓMEZ Et al. external phenotype, which can mitigate some of these physical effects in hot environments (Crossman, Rohwer, & Martin, 2011; Heenan,Goodman,&White,2015). Analternativeexplanationfortheobservedassociationbetween eggsizeandpigmentationmaybeduetolightereggsbeingsmaller simply because small eggs and limited pigment may be two correlatedcharacteristicsofeggslaidbyfemalebirdsthatareinpoor condition. That is, individual females that cannot invest as much ineggsize(producinglargeeggsiscostly)arealsolimitedintheir ability to invest in pigment production (if pigment is costly) (e.g., Siefferman, Navara, & Hill, 2006). However, this could imply that therearegeographicalgradientsinthebodyconditionoffemales, whichseemsunlikely. Assolarradiationismainlyaffectedbylatitudinalvariations,it maybeconcludedthatthepatternineggshellcolorationislatitudinal,witheggshellsofdarkercolorandlowerUVreflectancefound in northern latitudes, although this is conditioned by high solar radiation levels (see below). Thermal constraints may partly drive the observed spatial variation in eggshell coloration, which may be lightertodiminishtherisk ofoverheating (Gómezetal.,2016; Lahti&Ardia,2016).Inground-nestingbirds,darkereggscansometimeshavebettercamouflagethanlightereggs(Gómezetal.,2016; Troscianko etal., 2016), which suggests that ground-nesting birds mayfacetrade-offsbetweenthelevelofeggcamouflageanddealingwithhighlevelsofsolarradiation(Gómezetal.,2016;Wilson- Aggarwal etal., 2016). However, in sites where solar radiation is very high (>0.8kW/m2), lighter colors may not be advantageous becauseofthehighertransmittanceoflight-coloredeggshells,assumingnovariationineggshellthickness,asthisincreasestherisk of UV radiation reaching the embryo (Abram etal., 2015; Brulez etal.,2015;Gaudreauetal.,2017;Lahti&Ardia,2016;Maureretal., 2015).SuchhighUVradiationlevelsarefoundatlowlatitudes,and thismaybeareasonfortheoccurrenceofeggsofdarkcolorationat suchlatitudesinspiteofhighlevelsofsolarradiation.Studieswith Drosophilaflieshaveshownlatitudinalclinesinpigmentation,with darkerphenotypesbeingfoundathigherlatitudes.However,therelationshipisreversedatsouthernlatitudesneartheEquator,with UVradiationbeingthemainfactorresponsibleforthisreversedlatitudinalpattern(Bastideetal.,2014),similartoourresults. We expected that ambient temperatures could have affected eggshellcoloration.However,therewasnosignificantrelationship betweenthesevariables,likelybecauseincubatingploversmaycontrol egg temperature using several behavioral strategies (Amat & Masero,2004b;Vinczeetal.,2017),andthus,therisksofeggoverheatingbecomesrelevantonlyaftertheeggsremainexposedtodirectsolarradiation(i.e.,whenincubatingbirdsdepartfromnests). The Kentishplover,as other polygamous plovers,exhibitslow geneticstructureandweakisolation-by-distance(D’UrbanJackson etal.,2017;Küpperetal.,2012;Oyler-McCance,St.John,Kysela,& Knopf,2008;Vidal,Hernández,Luis,&Domínguez,2015).Despite gene flow,theremay be geographicphenotypic divergencein response to changes in local conditions (Bech etal., 2016; Ritcher- Boix, Teplitsky, Rogell, & Laurila, 2010), which may explain why eggshellspottingpatternsaresite-specific.Althoughtherepeatabilityofeggshellspottinessofindividualfemalesseemshigh(Skrade& Dinsmore,2013;Wheelwright,Graff,&Norris,2012),femalesmay exhibitplasticityineggshellcoloration(i.e.,inthequantitiesofpigmentsassignedtoeggshells;e.g.,Avilés,Stokke,Moksnes,Røskaft, &Møller,2007;Wheelwrightetal.,2012)dependingonlocalfactors, asthiscouldoptimizefitnessinresponsetosimultaneouschanging threats(Hansson,2004).PopulationsofAfricanvillageweaverbirds Ploceus cucullatuschangedineggappearanceafterbeingintroduced in sites where theyexperienced differentlevels ofsolarradiation comparedtothoseintheiroriginalsites(Lahti,2008).However,the variationdescribedinthesestudiescanalsobeinterpretedeitheras localadaptation(evolution),orenvironmentalconstraintsaffecting eggcolorationbymodifyingthestate(nutritionalstatus/stresslevel) ofthefemalebird,anditremainstobedemonstratedwhetherbirds plasticallyadjusteggcolorationtomatchenvironmentalconditions, asithasbeenshownininsects(Abrametal.,2015;Torres-Campos, Abram, Guerra-Grenier, Boivin, & Brodeur, 2016). Individual phenotypicplasticitymaybeadaptive,astheremaybedifferencesin theeffectsofsolarradiationthroughoutthelongnestingseasonof theKentishplover(individualfemalesmaylayuptofourclutchesin aseason[Fraga&Amat,1996]),aswellasbetweensitesatwhich individual plovers may breed during successive nesting attempts. Indeed,individualshorebirdsmaymovethousandsofkilometersto breed, not only between nesting seasons but also within seasons (Figuerola,2007;Stenzeletal.,1994).Inaddition,underaclimate- warming scenario, the local adaptations may be altered, in which casephenotypicplasticitymightbeanimportantmechanismtorespond to those changes. Therefore, historical and novel stressors FIGURE5 Partialplots(95%CI)ofafour-orderpolynomial modelfortherelationshipsbetweensolarradiationandeggshell color(PC1).Onlythesignificantenvironmentalvariableswere includedineachmodel.Rugplotrepresentsthespreadofthedata | 8027 GÓMEZ Et al. oftheenvironmentmayinteracttoshapespatialgradientsineggshellreflectance, and perhapsultimatelypatternsofnest survival througheffectsonnestcamouflage(e.g.,Skrade&Dinsmore,2013; Trosciankoetal.,2016). In conclusion, our study indicates that the latitudinal trends in eggshellcoloration inthe Kentish plover are affected by levels of solar radiation at the nesting sites, thus supporting the color- mediatedheatinghypothesis.Butsolarradiationmayaffecttheperformanceofembryosthroughatwo-wayinteractionbetweenthe riskofoverheatingandthedeleteriouseffectsofUVradiation.We havestudiedasingleshorebirdspecies,butdifferentspeciesmay show different phenotypic responses to the same environmental gradient,meaningthatdifferencesbetweenspeciesinrelativecosts andbenefits,especially duetodifferencesineggsize, may affect latitudinalvariationsineggshellcolor. ACKNOWLEDGMENTS OurthankstoDrs.DouglasRussellandRobertPruett-Jonesatthe NaturalHistoryMuseum,Tring,forgrantingaccesstotheeggscollectionandmanyfacilitiesduringourstayatthemuseum.Also,to XavierFerrerforthephotograph.DanielHanley,ananonymousrefereeandtheeditorscommentedonanearlierversion. CONFLICT OF INTEREST Theauthorsdeclarenoconflictofinterest. AUTHORS CONTRIBUTIONS JAAandGLCconceivedtheidea.JAA,JG,CR,andMSdesigned thestudy.MSandJTcontributedmaterials.JAAandCRcollected the images. JG, GLC, CR, MAR, and JT analyzed the images. CR retrieveddataoftemperatureandsolarradiationfromlargedatabases.MARranthestatisticalanalyses.JAA,JG,GLC,CR,andMAR wroteafirstversion,andallauthorscontributedtorevisionsofthe manuscript. DATA ACCESSIBILITY Data available from Digital.CSIC Repository: http://hdl.handle. net/10261/165880. ORCID Juan A. Amat http://orcid.org/0000-0003-1685-1056 REFERENCES Abram, P. K., Guerra-Grenier, E., Després-Einspenner, M.-L., Ito, S., Wakamatsu,K.,Biovin,G.,&Brodeur,J.(2015).Aninsectwithselective control of egg coloration. Current Biology, 25, 2007–2011. https://doi.org/10.1016/j.cub.2015.06.010 Ahlgren,J.,Yang,X.,Hansson,L.A.,&Brönmark,C.(2013).Camouflaged ortanned:Plasticityinfreshwatersnailpigmentation.Biology Letters, 9,20130464.https://doi.org/10.1098/rsbl.2013.0464 Alho, J. S., Herczeg, G., Söderman, F., Laurila, A., Jönsson, K. I., & Merilä, J. (2010). Increasing melanism along a latitudinal gradient in a widespread amphibian: Local adaptation, ontogenic or environmentalplasticity?BMC Evolutionary Biology,10,317.https://doi. org/10.1186/1471-2148-10-317 Amat, J. A., Gómez, J., Liñán-Cembrano, G., Rendón, M. A., & Ramo, C. (2017). Incubating terns modify risk-taking according to diurnal variations in egg camouflage and ambient temperature. Behavioral Ecology and Sociobiology,71,72.https://doi.org/10.1007/ s00265-017-2306-4 Amat, J. A., & Masero, J. A. (2004a). Predation risk on incubating adultsconstrainsthechoiceofthermallyfavourablesitesinaplover. Animal Behaviour, 67, 293–300. https://doi.org/10.1016/j. anbehav.2003.06.014 Amat,J.A., &Masero, J.A.(2004b).How Kentishplovers, Charadrius alexandrinus, cope with heat stress during incubation. Behavioral Ecology and Sociobiology, 56, 26–33. https://doi.org/10.1007/ s00265-004-0758-9 Amat, J. A., & Masero, J. A. (2007). The functions of belly-soaking in KentishploversCharadrius alexandrinus. Ibis,149,91–97. Arnold, T. W. (2010). Uninformative parameters and model selection usingAkaike’sInformationCriterion.Journal of Wildlife Management, 74, 1175–1178. https://doi.org/10.1111/j.1937-2817.2010.tb012 36.x Avilés,J.M.,Stokke,B.G.,Moksnes,A.,Røskaft,E.,&Møller,A.P. (2007). Environmental conditions influence egg color of reed warblersAcrocephalus scirpaceusandtheirparasite,thecommon cuckoo Cuculus canorus. Behavioral Ecology and Sociobiology, 61, 475–485. Azevedo,R.B.R.,French,V.,&Partridge,L.(1996).Thermalevolution of egg size in Drosophila melanogaster. Evolution, 50, 2338–2345. https://doi.org/10.1111/j.1558-5646.1996.tb03621.x Bakken, G. S. (1992). Measurement and application of operative and standardoperativetemperaturesinecology.American Zoologist,32, 194–216.https://doi.org/10.1093/icb/32.2.194 Barton,K.(2016).MuMIn: multi-model inference, R package version 1.15.6. Retrievedfromhttp://CRAN.R-project.org/package=MuMIn Bastide,H.,Yassin,A.,Johanning,E.J.,&Pool,J.E.(2014).Pigmentation in Drosophila melanogaster reaches its maximum in Ethiopia and correlates most with ultra-violet radiation in sub-Saharan Africa. BMC Evolutionary Biology, 14, 179. https://doi.org/10.1186/ s12862-014-0179-y Bech,N.,Foucart,T.,Fauvel,T.,Brischoux,F.,Bouchon,D.,&Bonnet, X.(2016).Phenotypicvariationcontrastswithgenetichomogeneity across scattered sea snakes colonies. Journal of Biogeography, 43, 1573–1582.https://doi.org/10.1111/jbi.12735 Bishop,T.R.,Robertson,M.P.,Gibb,H.,vanRensburg,B.J.,Braschler, B.,Chown,S.L.,…Parr,C.L.(2016).Antassemblageshavedarker and larger members in cold environments. Global Ecology and Biogeography,25,1489–1499.https://doi.org/10.1111/geb.12516 Brakefield,P.M.(1984).Ecologicalstudiesonthepolymorphicladybird Adalia bipunctataintheNetherlands.I.Populationbiologyandgeographicalvariationofmelanism.Journal of Animal Ecology,53, 761– 774.https://doi.org/10.2307/4658 Brulez,K.,Pike,T.W.,&Reynolds,S.J.(2015).Eggsignalling:Theuse ofvisual,auditory,and chemical stimuli. In D.C. Deeming, &S. J. Reynolds (Eds.), Nests, eggs, and incubation: New ideas about avian reproduction (pp. 127–141). Oxford, UK: Oxford University Press. https://doi.org/10.1093/acprof:oso/9780198718666.001.0001 Burnham, K. P., Anderson, D. R., & Huyvaert, K. P. (2011). AIC modelselectionandmultimodelinferenceinbehavioralecology: Some background, observations, and comparisons. Behavioral