Assessing the impact of roadkill on the persistence of wildlife populations: A case study on the giant anteater

SupportedbyInstitutoTecnol´ogicoVale

www.perspectecolconserv.com

Research Letters

Assessing the impact of roadkill on the persistence of wildlife populations: A case study on the giant anteater

Fernando Ascensão

, Arnaud L.J. Desbiez

acE3c-CenterforEcology,EvolutionandEnvironmentalChanges&CHANGE-GlobalChangeandSustainabilityInstitute.FaculdadedeCiências,UniversidadedeLisboa,Lisboa, Portugal

bInstitutodeConservac¸ãodeAnimaisSilvestres(ICAS),RuaLicuala622,79046150,CampoGrande,MatoGrossodoSul,Brazil

cInstitutodePesquisasEcológicas(IPÊ),RodoviaDomPedroI,km47,12960-000,NazaréPaulista,SãoPaulo,Brazil

dRoyalZoologicalSocietyofScotland(RZSS),Murrayfield,Edinburgh,EH126TS,UnitedKingdom

h i g h l i g h t s

•Few studies empirically estimated theimpactofroadkillonwildlifepop- ulations.

•Giantanteatersinhabitingroadvicin- ityareashavelowersurvivalrates.

•20%oftheanteatersinhabitingroad vicinityareasareroad-killed.

•Roads are sink areas for giant anteaters.

g ra p h i c a l a b s t r a c t

a rt i c l e i nf o

Articlehistory:

Received3January2022 Accepted27May2022 Availableonline13July2022

Keywords:

Populationviabilityanalysis Populationpersistence Roadecology

Transportationinfrastructures Myrmecophagatridactyla

a b s t ra c t

Humanactivityisdepletingbiodiversity,androadnetworksaredirectlycontributingtothistrenddueto roadkill.Nevertheless,fewstudiesempiricallyestimatedtheimpactofroadkillonwildlifepopulations.

Weintegratedinformationonroadkillrates,populationabundance,andanimalmovementtoestimate thesurvivalratesandtheproportionofthepopulationlikelytobeextirpatedduetoroadkill,usinggiant anteater(Myrmecophagatridactyla)asmodelspecies.Wethenassessedtheconsequentimplications ofroadkillonpopulationpersistenceusingpopulationviabilityanalysis(PVA).Theyearlysurvivalrate ofresidentanteatersinhabitingroadvicinityareas(0.78;CI:0.62−0.97)wasconsiderablylowerthan forthoselivingfarfromroads(0.95;CI:0.86–1.00).Therealnumberofanteatersbeingroad-killedis considerablyhigherthantheonerecordedinpreviousstudies(byafactorof2.4),withca.20%ofthe populationinhabitingroadvicinityareasbeingroad-killedeveryyear.AccordingtoPVAresults,roadkill cangreatlyaffectthepersistenceofthegiantanteaterpopulationsbyreducingthegrowthratedown tonullornegativevalues.Thisstudyconfirmsthatroadshavesignificantimpactsonlocalpopulation persistence.Suchimpactsarelikelytobecommontootherlargemammals,callingforeffectivemitigation toreduceroadkillrates.

©2022PublishedbyElsevierB.V.onbehalfofAssociac¸˜aoBrasileiradeCiˆenciaEcol ´ogicae Conservac¸˜ao.ThisisanopenaccessarticleundertheCCBY-NC-NDlicense(http://creativecommons.

org/licenses/by-nc-nd/4.0/).

∗Correspondingauthor:cE3c—CentreforEcology,EvolutionandEnvironmentalChanges,FaculdadedeCiênciasdaUniversidadedeLisboa,EdifícioC2,5^oPiso,Sala2.5.46, CampoGrande,1749-016Lisboa,Portugal.

E-mailaddress:[email protected](F.Ascensão).

https://doi.org/10.1016/j.pecon.2022.05.001

2530-0644/©2022PublishedbyElsevierB.V.onbehalfofAssociac¸˜aoBrasileiradeCiˆenciaEcol ´ogicaeConservac¸˜ao.ThisisanopenaccessarticleundertheCCBY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).

F.AscensãoandA.L.Desbiez PerspectivesinEcologyandConservation20(2022)272–278 Introduction

Billionsofanimalsofvariousspecies arekilledeveryyearin wildlifevehiclecollisions(WVC),withimpressiveestimatesreach- ing2.2millionmammalsonBrazilianroads(González-Suárezetal., 2018)and29millionmammalsinEurope(Griloetal.,2020).Vir- tuallyallspeciesinhabitingroadvicinityareascanbeimpactedby WVC,andsuchpervasiveimpactisreflectedinthegrowingnumber ofroadkillstudiesacrosstheglobe(Barrientosetal.,2021;Schwartz et al., 2020).However, todate there is still scarceinformation regardingtheeffectsofroadkillonwildlifepopulationpersistence (Barrientosetal., 2021), andthekeyquestionremains:canthe addedmortalityfromWVCrepresentathreatthatmayleadtolocal populationextinctions?

TocorrectlyassesstheimpactofWVConwildlifepopulation persistence, one requires a detailed knowledge of localspecies abundanceandonroadkillrates(Visintinetal.,2016).However, compilingsuchinformationforthesamespecieshasseldombeen achieved(e.g.,D’Amicoetal.,2015;Santosetal.,2016),preclud- ingourunderstandingontheroadkillimpactonwildlifespecies (Barrientosetal.,2021).Here,weusedthegiantanteater(Myrme- cophagatridactyla)asamodelspeciestoassesstherealimpactof WVConlocalpopulationsofalarge-sizedmammal.Thisspecies hasabody massof 33kg,low recruitment,withaboutonepup per year (Gaudin et al., 2018), and low densities of <1in./km² (Bertassonietal.,2021),traitsthatmake themparticularlyvul- nerabletoadditionalnon-naturalmortality(FahrigandRytwinski, 2009;RytwinskiandFahrig,2012).Despiteitswidedistribution throughoutCentralandSouthAmerica,thisspeciesislistedas“Vul- nerabletoExtinction”,andoneoftherecognizedthreatsisbeing recurrentlyinvolvedinWVCthroughoutitsdistribution(Miranda etal.,2014),namelyintheBrazilianstateofMatoGrossodoSul (Ascensãoetal.,2021,2017).

Weintegratedandupdateddifferentsourcesofempiricaland bibliographicinformationongiantanteaters,namelyroadkillrates, localpopulation,andtelemetrydata.Weaimedtoobtaincredi- bleestimateson(i)theprobabilityofsurvivalofgiantanteaters inhabitingroadvicinityareasincomparisontothoselivingfarfrom roads;andon(ii)theproportionofanimalsthatareannuallyroad- killedonmainroads.Wethenusedthisinformationonpopulation viability analysisto(iii)assess theconsequencesof roadkillfor anteaterpopulationpersistenceinroadvicinityareas.Wehypoth- esized that thoseanteaters living nearpavedroadshave lower survival rates,given thehighroadkillrates,and consequentlya largeproportionofanteaterswasexpectedtoberoad-killed.Also, weexpectedtoestimateahighproportionofindividualsthatare annuallyroad-killed,overallhavingasignificantimpactontheper- sistenceofanteatersbyloweringtheirpopulationgrowthrate,and consequentlytheirresiliencetoovercomestochasticeventsinclud- ingnaturalcatastrophes.

Methods

Studyareaanddatasets

Theempiricaldataanalyzedherewascollectedbythe‘Anteater

&HighwaysProject’(www.giantanteater.org),whichhasbeenrun- ningforthepastsixyearsinthestateofMatoGrossodoSul(MS), intheCerradobiome(savannah)ofBrazil.Theprojecthasfocused onfourhighwayssections(alltwo-laneroads,total1160km)and adjacent areas(Fig.1A).Traffic volumesofstudiedroadsrange approximatelybetween600vehiclesperdayinMS040(Fig.1D;

pers.obs)and4300vehiclesperdayinBR267(Fig.1E)(DNIT,2020).

In BR262,thetraffic volumeisca. 3350vehicles perdayinthe westsection(Fig.1B)and3850vehiclesperdayintheeastsec-

tion(Fig.1C)(DNIT,2020)(seeforfurtherdetailsonstudyareain Ascensãoetal.,2021andNoonanetal.,2021).

Roadkillrates

Roadkilldatahasbeensystematicallycollectedthroughoutthe fourroadsections(Fig.1B–E)(Ascensãoetal.,2017,2019b,2021).

Here,weusedthedatasetfromAscensãoetal.(2021).Briefly,road- killsurveyswerecarriedoutbycar(40−60km/h),onthefourroads, searchingforroad-killedanimalsonbothlanesandshoulders,for threeyears,onafortnightlybasis.Atotalof420surveyeventswere performed,totaling84,673kmofsurveyeffort.Duringthisstudy, 608road-killedgiantanteaterswererecorded,whichrepresenteda meanannualroadkillrateof∼0.19in./km/year(rawvalues)across surveyedroads.Inparallelwithroadkillsurveys,thedatasetalso hasinformationoncarcasspersistenceexperiments,carriedoutto estimatetheroadkillratemoreaccuratelybyaccountingforcarcass persistenceanddetectabilitybias(Ascensãoetal.,2021).Exper- imentswereperformedonthreeofthesurveyedroad sections, onewithlow trafficvolume(MS040)and twowithhigher traf- ficvolume(BR267andBR262-east),withdailysurveysfor15–30 days(mean25.9days).Monitoringended whenthecarcasswas notvisibleorafterthesurveyperiod (seeAscensãoetal.,2021 forfurtherdetails).Thecarcassespersistenceinformationfrom21 anteatersplusof96individualspertainingtomediumandlarge mammalspecies(including67Cerdocyonthous,21Hydrochoerus hydrochaeris,7Tapirusterrestrisand1Mazamagouazoubira)were usedtocorrecttheroadkillrates(seeAscensãoetal.,2021formore detailsonanalyticalproceduresforestimatingthenumberofroad kills).

Animalmovementinformation

Movementdataofgiantanteatershasbeencollectedforthe pastfiveyears(2017–2021)inthevicinityofthreeofthepaved roads(Fig.1C–E)(Noonanetal.,2021);currentlytotaling46adult wildgiantanteaters(25femalesand21males).Inthisdataset,four anteaters(ca.9%)areknowntohavedispersedduringtheirtrack- ingperiod,ofwhichtwodiedwhiledispersing(oneroad-killedon adirtroadandonefromunknowncause),andtheothertwoset- tledinanewterritoryafterdispersing.Otherseven(allresident) individualswerealsoroad-killedduringthetrackingperiod,oneof whichinhabitedaterritoryfarfrompavedroadsbutwashitona dirtroad.Overall,of46animalsmonitored,atleasteight(17%)died duetovehiclecollisions,withtwocollisionsoccurringondirtroads (farfrompavedroads),andsixonpavedroads.Theanteaterswere trackedonaverage395±235days(range:46–1089,fixesobtained every20min).Alldetailsaboutcaptureandmovementdataare providedinNoonanetal.(2021).

Dataanalyses

Estimatingsurvivalrates

We used the GPS-tracking data to estimate giant anteater survivalrates. We consideredtwogroups ofresident anteaters, distinguishingthoseanteaterswhosecentroidofthehomerange wasfar(>2km)frompavedroads(G₁,n=20)and near(<2km) pavedroads(G2,n=24).Thisdistancethresholdwasselectedbased onNoonanetal.(2021),inwhichweobservedthatanimalsthat lived>2kmfromapavedroaddidnotinteractwithroadsoverthe courseofthetrackingperiod,makingitunlikelythattheywouldbe impactedbyroadkill.Forourcalculations,wedidnotconsiderthe dispersalperiodsbutincludedthedataofthetwoanteatersthat settledinnewterritories(1monthafterstoppingthedispersal).

Previousresearchshowedthattherearenosignificantdifferences inhabitatbetweenroadsidesoracrossroads(Noonanetal.,2021).

EstimatesofsurvivalwereperformedusingtheKaplan–Meieresti-

Fig.1. LocationoftheBrazilianstateMatoGrossodoSulandthemainroadnetwork(A).Highlightedroadsections(thickwhitelines)indicatetheroadsectionswhereroadkill datahasbeencollected(B–E),andsquaresindicatewheremovementdatawascollected(C–E),withintheresearchproject‘AnteaterandHighways’(www.giantanteater.org).

Greenareascorrespondtonativeforestpatchesandyellowareasaremostlypastureareas.InB–E,thewhitelinesindicatetheroadsBR-262(west),BR-262(east),MS-040, andBR-267,respectively.InFisshownthehomerangesof12anteaterstrackedinMS-040(D).Aphotoofareleaseofonecollaredgiantanteaterisshown(©A.D.).

mator(Richetal.,2010)throughtheRpackage‘survival’(Therneau, 2021).

Estimatingtheproportionofthepopulationbeingroad-killed Toestimatetheproportionofthepopulationthatisroad-killed annually,wecalculatedtherealnumberofroadkillsandthenum- berofanteatersinhabitingroadvicinityareas.

Estimatingtherealnumberofroadkills. InAscensãoetal.(2021), we providedan estimateof realroadkillnumberscorrectedfor persistenceanddetectabilitybiases,usingtheGENESTframework (Dalthorpetal.,2018).Therein,weusedvaluesfor the‘density weightedproportion’(DWP)i.e.,theproportionoftotalmortal- ityexpectedtobewithinthesearchedcorridorarea—pavement and shoulders(ca.12mwidth),rangingbetween0.80and0.95, whichseemedreasonableaccordingtoourexperienceinroadkill surveys.Here,weupdatedtheseestimatesforgiantanteaterusing less-conservativevaluesforDWP.Basedonourongoingmovement datacollection,wewereabletoobtainnewdataonthefateof anteatercarcasseslocationafterbeinghitbyavehicle.Threeout ofthesixconfirmedanteatersbeingtrackedthatwereroad-killed onpavedroadswerestillabletomoveawayfromthepavedroad (upto100m),whichrepresentsaDWPof0.50(i.e.,halfofthecar- casseswouldnotbefoundwithinthesearchcorridorarea).This datasuggeststhatalargerproportionofanteaterscanmoveaway fromtheroadaftercollidingwithavehicleonapavedroad.Hence, werelaxedtheDWPparameterandproducedestimatesofgiant anteater roadkillratesusinga DWP=0.50.For comparison pur- poses,wealsocalculatedtheroadkillrateswhenusingDWP=0.80 (Ascensãoetal.,2021).

Estimatingthepopulationabundance. Toestimatethepopulation sizepronetobeaffectedbyroadkill,weneededtoestimatethe populationdensityandtodelimittheroadvicinityareai.e.,thearea inwhichresidentanteatersweremorelikelytocrosstheroads.

Forthe populationdensity parameter,weperformed asystem- aticreviewofthepublishedliteraturetoobtainvaluesforgiant anteaterpopulationdensity.WesearchedtheISIWebofKnowl- edgeandGoogleScholardatabases,usingthesearchstring“ALL FIELDS:( ¨Giantanteater ¨OR¨MyrmecophagatridactylaOR ¨Tamanduá-¨ bandeira ¨OR¨Tamanduábandeira¨) AND ALLFIELDS: (Density OR Abundance OR Densidade OR Abundância)”, for the timespan 2000−2021.Thisresulted in22 scientific articlesin ISI Webof Knowledge,andca.1750resultsinGoogleScholarfromwhichwe screenedthefirst10pages.Wereviewedallabstractsandretained onlystudiesthatprovidedsomemeasureofgiantanteaterdensity.

Thereferencelistofeacharticlewasalsocheckedforotherrelevant publications.

Wealsoincludedourownestimateusingthetrackingdata.The densityestimationwasbasedonthedatafromoneofthestudy areas(MS040),inanareawhereaneffortwasmadetoattempt tocaptureallresidentindividuals(Fig.1F).Therein,therewere12 residentgiantanteaters,whichhomerangesoverlapped,arranged withinaminimumconvexpolygonof44.3km²area.Thisrepre- sentsca.0.3in.ividualsperkm²(assumingallresidentindividuals werecaptured;seeNoonanetal.(2021)fordetailsoncalculating homeranges).

Finally, we estimated the number of anteaters prone to be affectedbyroadkillbasedondensityestimates,usingabandwidth of8kmalongroads(4kmforeachside)asaconservativedistance uptowhichresidentanteaterswerelikelytocrosstheroadsand

F.AscensãoandA.L.Desbiez PerspectivesinEcologyandConservation20(2022)272–278

Fig.2.Survivalcurvesforthethreegroupsofgiantanteaterstrackedwithinthe

‘AnteaterandHighways’researchproject(www.giantanteater.org).Allanteaters (n=42)wereresidentsthroughoutthetrackingperiod.G1—anteaterslivingfar fromroads(n=18),G2—anteaterslivingnearroads(n=24).Survivalwasestimated usingtheKaplan–Meierestimator(Richetal.,2010).

beroad-killed.Thisisapproximatelythelongestdistancefromthe roadthatwehavedetectedanimalsbeingtrackedthatcrossedthe studiedroads.

Populationviabilityanalysis

WeusedtheVortexmodelpreviouslydescribedinDesbiezetal.

(2020) toassesstheimpactof roadkillmortalityonpopulation persistencethroughPopulationViabilityAnalysis(PVA).We ran different sets ofsimulations, varyingthe initialpopulationsize basedonresultsfromprevioussection(seeResults),andthenum- berofanimalsremovedbyroadkillperyear(basedonestimates ofrealnumberofroadkillsusingdifferentDWPvalues).Wemod- eled atotal of14 scenarios:3levelsof densityvalues (0.2,0.3, and0.4in./km²),2levelsofroadkillrates(usingDWPof0.80and 0.50)andincluding/notincludingcatastrophesinsimulations.We comparetheresultswiththebaselinescenariousingtheminimum populationabundance,withandwithoutcatastrophes,butwithout harvestingindividualsduetoroadkill(SupplementarymaterialS1).

AllinputparametersareshowninSupplementarymaterialS2.We usedtheVortexversion10.5.5(LacyandPollak,2021)togenerate thePVAmodels.

Results

Survivalrates

Theyearlysurvivalofgiantanteatersinhabitingareasfarfrom theroadi.e.,withhomerangecentroidsdistancing>2kmfromthe nearestpavedroad(G1)wasestimatedtobe0.95(CI:0.86–1.00) (Fig.2).For those anteaters inhabitingroad vicinityareas (G₂), theestimatedsurvivaldecreasedto0.78(0.62−0.97)(Fig.2).This suggeststhattheprobabilityofsurvivalofagiantanteateriscon- siderablylowerwheninhabitingareasclosetoroads.

Estimatingtherealnumberofroadkills

Themediansoftheestimatedrealnumberofcollisionsinvolving anteaterswhenusingDWP=0.50was1538(95%CI:1417–1683), representing ca. 48 (44–52)anteaters per 100km per year.For comparison, the estimates when using DWP=0.80 were 965

(894–1046),representingaroadkillrateofca.30(28–33)anteaters per 100km per year. In both cases, estimates were consider- ably higher than the roadkill rate obtained from raw values (19in./100km/year).

Estimatingthepopulationabundance

Sevenstudiesreported populationdensity values or equiva- lent (Table S3.1 in Supporting information S3). The density of giantanteaterasfoundinthesestudiesrangedbetween0.1and 2.9in./km². However, we considered as a reasonable bounding rangethedensityinterval of0.2−0.4in./km² (seeourreasoning inSupportinginformationS3).Thedensityofgiantanteaters(resi- dents)inhabitingthevicinityofthesurveyedroadswouldtherefore rangebetween160and320individualsper100kmofroads,when consideringdensitiesof0.2to0.4in.ividualsperkm²,respectively.

Asthemedianoftheestimatesacrossstudiesinsavannaareaswas 0.3in./km²,whichisthesamevaluewefoundbasedonthemove- mentdata,weconsidered0.3in./km²tobethemostprobablevalue ofdensityacrossthegiantanteaterdistributionrangeinsavanna areas.

Combiningtheroadkillestimateswiththeseabundanceesti- mates,weprojectedthatroadkillisremovingannuallyca.20.0%

(5–95%CI:18.5–21.9%)ofthelocalpopulationinthevicinityof theroad(consideringtheroadkillestimatesusingDWP=0.50and apopulationdensityof0.3in.ividualsperkm²)(Fig.3A).Theesti- matesusingtherangeofvaluesofDWPanddensityvariedbetween ca. 9.4% (8.7–10.2%), when using DWP=0.8 and density=0.4;

and30.0%(27.7–32.9%), whenusingDWP=0.4and density=0.2 (Fig.3A).

Populationviabilityanalysis

RegardingtheresultsfromPVA,themean±SDstochasticgrowth rate(stoch-r)acrossthebaselinescenariowithoutharvestofindi- viduals from thepopulation by roadkill and in the absence of catastropheswas 0.59±0.50. Whenconsidering themost likely density value of 0.3in.ividuals per km², and a DWP=0.50, we obtainedameanstoch-rof0.10±0.50(Fig.3B).Thedifferentscenar- iosoftheproportionofindividualsroad-killedandoftheresident populationsizeledtostoch-rraging between-0.50±0.50 when densityis0.2in./km²andDWP=0.5,and0.37±0.44whendensity is0.4in./km² andDWP=0.8.Inallcases,thestoch-rwhencon- sidering roadkillharvestingis considerably lower than theone in thebaseline scenario (Fig.3B).When including catastrophic eventsin simulations,allstoch-r values decreased substantially togetherwithhigheramplitudeintheconfidenceintervals,denot- ingahighervulnerabilitytolocalextinctionsaswellasahigher uncertaintyonpopulationtrend(Fig.3B).Importantly,allsimula- tions–evenwithoutcatastrophes–showedasteadyreductionin populationsize,exceptthesimulationsusingpopulationdensities of0.4in./km²andDWP=0.80(Fig.S4.1inSupplementarymaterial S4).

Discussion

Ourresultssuggestthatthesurvivalofresidentanteatersiscon- siderablyreducedwheninhabitingroadvicinityareas.Moreover, weestimatedthatca.20%ofthepopulationinhabitingroadvicinity areasislikelytobeextirpatedduetovehiclecollisionseveryyear.

Finally,whenmodelingtheviabilityofanteaterpopulationsnear pavedroads,weobtainedasignificantdecreaseinthestochastic growthrate,downtonearlyzero(whenusingdensity0.3in./km² andDWP=0.5).Therefore,ourresultsclearlysupportthatroads aresinkareas,withpopulationslivinginroadvicinityareasbeing

Fig.3.Estimatesontheproportionofthepopulationroad-killed(A)andofthestochasticgrowthrate±SD(B)accordingtodifferentdensityscenarios(0.2–0.4in./km²)and valuefortheparameterDWP(0.5and0.8,forcalculatingtotalmortality;seetextfordetails).In(B)thegreyareasanddottedlinesrepresentthestochasticgrowthrateand SD,respectively,forthebaselinescenario(usingadensityof0.2in./km²).InpanelA,dotsareestimatesofmedianandbarstheCI5-95%.InB),dotsaremeanvaluesacross VORTEXsimulations(n=1000)andbarsrepresentSD.

significantlydepleted,requiringthecontinuedrecruitmentofindi- vidualsfromotherterritories(sources).

Withtheexpectedexpansionandincreasingdensityoftheroad network(Meijeretal.,2018),thecumulativeeffectofthesource- sinkdynamicsacrosstheregionislikelytobetranslatedintoa significantdecreaseofanimalabundance.Moreover,thehabitat ofgiantanteatersisbeingrapidlyreducedintheCerrado,namely duetotheexpansionofsojacultureandothercashcrops(Green etal.,2019).Thismeansthatthesourceofareastoreplenishpop- ulationsinthevicinityofroadsisalsodecreasing.Suchdepletion impactmayfurtherreducegiantanteaterpopulationresilienceand abilitytowithstandorrecoverfromotheranthropogenicthreats, suchasfires,impactofpesticidesandagrochemicals,conflictwith dogs,persecutionbypeopleordiseaseoutbreak(Bertassonietal., 2019;Garciaetal.,2021;Mirandaetal.,2014).Althoughwefocused ourstudyonthisspecies,itisreasonabletoassumethatasim- ilarpatternmaybefoundforothermediumandlargemammals, knowntobehighlyimpactedandvulnerabletoroadkill(Fahrigand Rytwinski,2009;RytwinskiandFahrig,2012).Assuch,roadsmust beaccountedinpopulationandlandscapeconservationmanage- mentofcurrentandfutureroadnetworks,aimingtoreduceroadkill throughouttheroadnetwork.

Tothebestofourknowledge,theinstallationoffencingconnect- ingtoexistingroadpassagesconstitutesthesinglebestroadkill mitigationmeasures(Denneboomet al.,2021; Rytwinskiet al., 2016), promotingthesafecrossingmovementformanyspecies, includingthosevulnerabletoroadkillandthoseposingariskfor humanswheninvolvedinwildlife-vehiclecollisions(Lesbarrères andFahrig,2012).Theimplementationofroadfencingshouldpri- oritizesites of highermortality i.e.,hotspots (Spanowiczet al., 2020). However, for low density and broad distribution range species, suchasanteaters, detecting hotspots of mortalitymay not be feasiblegiven thehigh dispersion of occurrences along roads(Ascensãoetal.,2017).Analternativeapproachis topri- oritizeplacesthatcrossareasofgreaterconnectivityi.e.,where animalsaremore likely tocrosstheroad intheirdaily orsea- sonaldisplacements(Ascensãoetal.,2019a).Suchapproachhas the advantage of requiring few base information, while being unaffectedbyshort-termpopulationdynamicsthatmayshiftthe locationofconcentrationofroadkill(Santosetal.,2017;Teixeira etal.,2017).

Toguaranteeitseffectiveness,fencesmustbecarefullyplanned andproperlyinstalled,alsorequiringregularinspection,mainte- nance,andrepair,whichwillraisetheassociatedcosts.Yet,when

F.AscensãoandA.L.Desbiez PerspectivesinEcologyandConservation20(2022)272–278

estimatingthemitigationcosts,onemustconsiderthatlargeani- mals represent a seriousa threat tohumans wheninvolved in collisionswithvehicles(Ascensãoetal.,2021;Conover,2019).The fact thatalargeproportionofanteatersareprobablyunnoticed in roadkillsurveys(this study)alsoimplies thatthedamagein vehicles andhumaninjuriesisalsohigher.Collisionswithlarge speciesrepresentaseriousriskforhumansafetymaterialdam- age(Ascensãoetal.,2021;Conover,2019).Forexample,inMato GrossodoSul,thosecollisionsinvolvingspecieswiththebodysize ofthegiantanteaterwereestimatedtohaveamaterial(vehicle) cost ofca. US$1150forcarsand US$530for trucks.Using the updatedestimatesofrealroadkillevents,theestimatedmaterial costonvehiclesofcollisionsinvolvinggiantanteatersascendsto US$2400perkm/year,significantlyhigherthanthepreviousworst estimate(US$985perkm/year;followingAscensãoetal.,2021).

Hence, investmentsto reducethenumber ofcollisions, namely properfencingconnectingtoexistingroadpassages,arelikelyto payoffinlessthan10yearsinsteadofthepreviouslyestimated 16–40yearsreportedinAscensãoetal.(2021).

Wenotethattherearestillsomeuncertaintiesinourestimates, namelyontheroadkillratesandpopulationabundance.Here,we relaxedtheparameterDWPfromtheGENEST,allowingahigher numberofundetectedcarcassestobeaccountedwhenestimating thetotalmortality.Theproportionofthoseanteatersbeingtracked andforwhichweknowthatthecauseofdeathwasroadkillwas considerablyhigh(17%),anditisremarkablethathalfoftheani- malshitonpavedroadswereabletomoveawayfromtheroad.

Ifconfirmed,thispatternsuggestthatthenumberofroad-killed anteatersislargelyunderestimatedinroadkillsurveys.Moreover, ourstudydidnotfocuseddirtroads.Yet,theroadnetworkofdirt roadsismuchmorewidespreadintherangeareaofgiantanteaters, andseveralanteatersarelikelytobekilledwhilemovingthrough them.Infact,recentresearchdemonstratedthatanteatersseem toactivelyseekfordirtroadsfordisplacements(Versianietal., 2021).Hence,theimpactofroadkillislikelytobemuchhigherthan ourestimatesindicate.Thisimpliesthattheconservationplanning ofthisspeciesmustintegratethemainactorsoftheruralsector, namelythosewhoplanandbuildnewdirtroads.Thesestakehold- ersshouldbeinformedoftheimpactoftheseroadsandshouldbe invitedtoparticipateinconservationplanning.

On theother hand, estimating the populationabundanceis alwayschallengingtoaccomplishandthereisa highuncertain- ityontherealdensityofmostspecies,especiallywhenestimates aremadeforsuchlargegeographicareas.Ourbibliographicrevi- sion,togetherwithourownestimateongiantanteaterdensity, allowed us toboundthe most likely density of this species to 0.2–0.4in./km².However,aneffortshouldbemadetoestimatethe densityofgiantanteatersandotherspeciesequallyvulnerableto WVC,asonlywiththisinformationwecanprovidesoundestimates ontheimpact ofroadkill onpopulationpersistence (Barrientos etal.,2021).

Additionalresearchshouldbecarriedouttoquantifytheimpact ofroadkillonthepersistenceofpopulationsfordifferentspecies withdiversetraitse.g.,movementbehaviororroadavoidance,to ensurethatmitigationstrategiesareeffectiveforavarietyoftaxo- nomicgroups.Ourstudyservesasanimportantbaselineforlarge mammalsinBrazil.Ifasubstantialpartoftheroadnetworksisnot mitigated,namelyindevelopingcountries,theoccurrenceoficonic speciessuchasthegiantanteaterwillnolongerbepossibleinmuch ofitsrange.

Ethicsdeclarations

Theauthorsdeclarethattheyhavenoconflictofinterest.

Funding

FAwasfundedbyFundac¸ãoparaaCiênciaeTacnologia(contract CEECIND/03265/2017).

Dataavailability

AlldataisavailableuponreasonablerequesttoArnaudDesbiez ([email protected]).

Authorcontribution

FAandADcontributedequallytothestudyconception,data analysisandmanuscriptwriting.Bothauthorsreadandapproved thefinalmanuscript.

DeclarationofCompetingInterest

Theauthorsreportnodeclarationsofinterest.

Acknowledgements

WewouldliketothankthedonorstotheAnteaters&Highways ProjectespeciallytheFoundationSegreaswellasNorthAmerican andEuropeanZooslistedat(http://www.giantanteater.org/).We wouldalsoliketothanktheownersofalltheranchesthatallowed ustomonitoranimalsontheirproperty,inparticularNhuveira, QuatroIrmãosandSantaLourdesranches.WearegratefultoD.R.

Yogui,M.Alves,D.KluyberandC.Lubaforfieldwork,andtoMar- celloD’AmicoandRafaelBarrientosfortheirvaluablecomments onanearlyversionofthemanuscript.Wealsothanktheinputs fromtwoanonymousreviewers.VortexPVAsoftware(Lacy&Pol- lak2021)isprovidedunderaCreativeCommonsAttribution-No DerivativesInternationalLicense,courtesyoftheSpeciesConser- vationToolkitInitiative(https://scti.tools).

AppendixA. Supplementarydata

Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/

j.pecon.2022.05.001.

References

Ascensão,F.,Desbiez,A.L.J.,Medici,E.P.,Bager,A.,2017.Spatialpatternsofroad mortalityofmedium–largemammalsinMatoGrossodoSul,Brazil.Wildl.Res.

44,135–146,http://dx.doi.org/10.1071/WR16108.

Ascensão,F.,Mestre,F.,Barbosa,A.M.,2019a.Prioritizingroaddefragmentation usinggraph-basedtools.Landsc.UrbanPlan.192,103653,

http://dx.doi.org/10.1016/j.landurbplan.2019.103653.

Ascensão,F.,Yogui,D.,Alves,M.,Medici,E.P.,Desbiez,A.,2019b.Predicting spatiotemporalpatternsofroadmortalityformedium-largemammals.J.

Environ.Manage.248,109320,

http://dx.doi.org/10.1016/j.jenvman.2019.109320.

Ascensão,F.,Yogui,D.R.,Alves,M.H.,Alves,A.C.,Abra,F.,Desbiez,A.L.J.,2021.

Preventingwildliferoadkillcanoffsetmitigationinvestmentsin short-mediumterm.Biol.Conserv.253,108902,

http://dx.doi.org/10.1016/j.biocon.2020.108902.

Barrientos,R.,Ascensão,F.,D’Amico,M.,Grilo,C.,Pereira,H.M.,2021.Thelostroad:

dotransportationnetworksimperilwildlifepopulationpersistence?Perspect.

Ecol.Conserv.19(4),411–416,http://dx.doi.org/10.1016/j.pecon.2021.07.004.

Bertassoni,A.,Costa,R.T.,Gouvea,J.A.,Bianchi,R.,de,C.,Ribeiro,J.W.,Vancine, M.H.,Ribeiro,M.C.,2019.Land-usechangesandtheexpansionofbiofuelcrops threatenthegiantanteaterinsoutheasternBrazil.J.Mammal.100,435–444, http://dx.doi.org/10.1093/jmammal/gyz042.

Bertassoni,A.,Bianchi,R.D.C.,Desbiez,A.L.J.,2021.Giantanteaterpopulation densityestimationandviabilityanalysisthroughmotion-sensitivecamera records.J.Wildl.Manag.85(8),1554–1562,

http://dx.doi.org/10.1002/jwmg.22123.

Conover,M.R.,2019.Numbersofhumanfatalities,injuries,andillnessesinthe UnitedStatesduetowildlife.Human–WildlifeInteract13,12,

http://dx.doi.org/10.26077/r59n-bv76.

D’Amico,M.,Román,J.,delosReyes,L.,Revilla,E.,2015.Vertebrateroad-kill patternsinMediterraneanhabitats:who,whenandwhere.Biol.Conserv.191, 234–242,http://dx.doi.org/10.1016/j.biocon.2015.06.010.

Dalthorp,D.,Madsen,L.,Huso,M.M.,Rabie,P.,Wolpert,R.,Studyvin,J.,Simonis,J., Mintz,J.,2018.GenEstStatisticalModels—AGeneralizedEstimatorof Mortality(No.2328–7055).USGeologicalSurvey.

Denneboom,D.,Bar-Massada,A.,Shwartz,A.,2021.Factorsaffectingusageof crossingstructuresbywildlife—asystematicreviewandmeta-analysis.Sci.

TotalEnviron.777,http://dx.doi.org/10.1016/j.scitotenv.2021.146061.

Desbiez,A.L.J.,Bertassoni,A.,Traylor-Holzer,K.,2020.Populationviabilityanalysis asatoolforgiantanteaterconservation.Perspect.Ecol.Conserv.18,124–131, http://dx.doi.org/10.1016/j.pecon.2020.04.004.

DNIT,http://servicos.dnit.gov.br/dadospnct,2020(Accessed6August2020).

Fahrig,L.,Rytwinski,T.,2009.Effectsofroadsonanimalabundance:anempirical reviewandsynthesis.Ecol.Soc.14.

Garcia,L.C.,Szabo,J.K.,deOliveiraRoque,F.,deMatosMartinsPereira,A.,Nunesda Cunha,C.,Damasceno-Júnior,G.A.,Morato,R.G.,Tomas,W.M.,Libonati,R., Ribeiro,D.B.,2021.Record-breakingwildfiresintheworld’slargestcontinuous tropicalwetland:integrativefiremanagementisurgentlyneededforboth biodiversityandhumans.J.Environ.Manage.293,112870,

http://dx.doi.org/10.1016/j.jenvman.2021.112870.

Gaudin,T.J.,Hicks,P.,DiBlanco,Y.,2018.Myrmecophagatridactyla(pilosa:

myrmecophagidae).Mamm.Species50,1–13, http://dx.doi.org/10.1093/mspecies/sey001.

González-Suárez,M.,ZanchettaFerreira,F.,Grilo,C.,2018.Spatialand species-levelpredictionsofroadmortalityriskusingtraitdata.Glob.Ecol.

Biogeogr.27,1093–1105,http://dx.doi.org/10.1111/geb.12769.

Green,J.M.H.,Croft,S.A.,Durán,A.P.,Balmford,A.P.,Burgess,N.D.,Fick,S.,Gardner, T.A.,Godar,J.,Suavet,C.,Virah-Sawmy,M.,Young,L.E.,West,C.D.,2019.

Linkingglobaldriversofagriculturaltradetoon-the-groundimpactson biodiversity.Proc.Natl.Acad.Sci.U.S.A.116,23202–23208.

Grilo,C.,Koroleva,E.,Andráˇsik,R.,Bíl,M.,González-Suárez,M.,2020.Roadkillrisk andpopulationvulnerabilityinEuropeanbirdsandmammals.Front.Ecol.

Environ.18,323–328,http://dx.doi.org/10.1002/fee.2216.

Lacy,R.,Pollak,J.,2021.Vortex:AStochasticSimulationoftheExtinctionProcess.

ChicagoZoologicalSociety,Brookfield,Illinois,USA.

Lesbarrères,D.,Fahrig,L.,2012.Measurestoreducepopulationfragmentationby roads:whathasworkedandhowdoweknow?TrendsEcol.Evol.27,374–380, http://dx.doi.org/10.1016/j.tree.2012.01.015.

Meijer,J.R.,Huijbregts,M.A.J.,Schotten,K.C.G.J.,Schipper,A.M.,2018.Global patternsofcurrentandfutureroadinfrastructure.Environ.Res.Lett.13, 064006,http://dx.doi.org/10.1088/1748-9326/aabd42.

Miranda,F.,Bertassoni,A.,Abba,A.,2014.Myrmecophagatridactyla[WWW document].In:IUCNRedListThreat.Species.

https://doi.org/10.2305/IUCN.UK.2014-1.RLTS.T14224A47441961.en,.

(Accessed2January2019).

Noonan,M.J.,Ascensão,F.,Yogui,D.R.,Desbiez,A.L.J.,2021.Roadsasecological trapsforgiantanteaters.Anim.Conserv.,http://dx.doi.org/10.1111/acv.12728.

Rich,J.T.,Neely,J.G.,Paniello,R.C.,Voelker,C.C.J.,Nussenbaum,B.,Wang,E.W., 2010.ApracticalguidetounderstandingKaplan–Meiercurves.Otolaryngol.

NeckSurg.143,331–336,http://dx.doi.org/10.1016/j.otohns.2010.05.007.

Rytwinski,T.,Fahrig,L.,2012.Dospecieslifehistorytraitsexplainpopulation responsestoroads?Ameta-analysis.Biol.Conserv.147,87–98, http://dx.doi.org/10.1016/j.biocon.2011.11.023.

Rytwinski,T.,Soanes,K.,Jaeger,J.A.G.,Fahrig,L.,Findlay,C.S.,Houlahan,J.,Ree,R.

vander,Grift,E.A.vander,2016.Howeffectiveisroadmitigationatreducing Road-kill?Ameta-analysis.PLoSOne11,e0166941,

http://dx.doi.org/10.1371/journal.pone.0166941.

Santos,S.M.,Mira,A.,Salgueiro,P.A.,Costa,P.,Medinas,D.,Beja,P.,2016.Avian trait-mediatedvulnerabilitytoroadtrafficcollisions.Biol.Conserv.200, 122–130,http://dx.doi.org/10.1016/j.biocon.2016.06.004.

Santos,R.A.L.,Ascensão,F.,Ribeiro,M.L.,Bager,A.,Santos-Reis,M.,Aguiar,L.M.S., 2017.Assessingtheconsistencyofhotspotandhot-momentpatternsof wildliferoadmortalityovertime.Perspect.Ecol.Conserv.15,56–60, http://dx.doi.org/10.1016/j.pecon.2017.03.003.

Schwartz,A.L.W.,Shilling,F.M.,Perkins,S.E.,2020.Thevalueofmonitoringwildlife roadkill.Eur.J.Wildl.Res.66,1–12,

http://dx.doi.org/10.1007/s10344-019-1357-4.

Spanowicz,A.G.,Teixeira,F.Z.,Jaeger,J.A.G.,2020.Anadaptiveplanforprioritizing roadsectionsforfencingtoreduceanimalmortality.Conserv.Biol.n/a, http://dx.doi.org/10.1111/cobi.13502.

Teixeira,F.Z.,Kindel,A.,Hartz,S.M.,Mitchell,S.,Fahrig,L.,2017.Whenroad-kill hotspotsdonotindicatethebestsitesforroad-killmitigation.J.Appl.Ecol.54, 1544–1551,http://dx.doi.org/10.1111/1365-2664.12870.

Therneau,T.M.,2021.APackageforSurvivalAnalysisinR.Rpackageversion3.3-1, https://CRAN.R-project.org/package=survival.

Versiani,N.F.,Bailey,L.L.,Pasqualotto,N.,Rodrigues,T.F.,Paolino,R.M.,Alberici,V., Chiarello,A.G.,2021.Protectedareasandunpavedroadsmediatehabitatuseof thegiantanteaterinanthropogeniclandscapes.J.Mammal.102(3),802–813, http://dx.doi.org/10.1093/jmammal/gyab004.

Visintin,C.,Ree,R.vander,McCarthy,M.A.,2016.Asimpleframeworkfora complexproblem?Predictingwildlife–vehiclecollisions.Ecol.Evol.6, 6409–6421,http://dx.doi.org/10.1002/ece3.2306.