Traditional Versus Multivariate Methods for Identifying Jaguar
Total Page:16
File Type:pdf, Size:1020Kb
Journalof WildlifeManagement 74(5):1141-1153; 2010; DOI: 10.2193/2009-293 111/nu>* ρ ll ■ να τ η, Toolsand TechnologyArticle '%^% TraditionalVersus Multivariate Methods forIdentifying Jaguar, Puma, and Large Canid Tracks CARLOS DE ANGELO,1 NationalResearch Council of Argentina (CONICET) andAsociación Civil Centrode Invéstigaciones del Bosque Atlântico (CelBA),Yapeyû 23, CP 3370, PuertoIguazú, Misiones, Argentina AGUSTIN PAVIOLO, NationalResearch Council of Argentina (CONICET) andAsociación Civil Centrode Investigacionesdei Bosque Atlântico (CelBA),Yapeyú 23, CP 3370, PuertoIguazú, Misiones, Argentina MARIO S. DI BITETTI, NationalResearch Council of Argentina (CONICET) andAsociación Civil Centrode Investigacionesdel Bosque Atlântico (CelBA),Yapeyú 23, CP 3370, PuertoIguazú, Misiones, Argentina ABSTRACT The jaguar{Panthern onca) and puma {Puma concolor) are the largest felids of the American Continent and live in sympatry alongmost of their distribution. Their tracks are frequently used for research and management purposes, but tracks are difficult to distinguish fromeach other and can be confusedwith those of big canids. We usedtracks from pumas, jaguars, large dogs, and maned wolves {Chrysocyon brachyurus)to evaluate traditional qualitative and quantitative identification methods and to elaboratemultivariate methods to differentiatebig = canidsversus big felids and puma versus jaguar tracks (n 167tracks from 18 zoos).We testedaccuracy of qualitative classification through an identificationexercise with field-experienced volunteers. Qualitative methods were useful but there was highvariability in accuracyof track identification.Most of the traditional quantitative methods showed an elevatedpercentage of misclassified tracks (>20%). We usedstepwise discriminantfunction analysis to develop3 discriminantmodels: 1 forbig canid versus big felid track identification and 2 alternativemodels for jaguarversus puma track differentiation using 1) bestdiscriminant variables, and 2) size-independentvariables. These modelshad high classificationperformance, with <10% of errorin thevalidation procedures. We used simplerdiscriminant models in the elaborationof identificationkeys to facilitatetrack classification process. We developedan accuratemethod for track identification, capable of distinguishing betweenbig felids (puma and jaguar) and large canids (dog and manedwolf) tracks and between jaguar and puma tracks. Application of our methodwill allow a morereliable use of tracks in pumaand jaguar research and it will help managers using tracks as indicatorsof these felids' presencefor conservation or managementpurposes. KEY WORDS canids,discriminant function analysis, identification keys, jaguar, Panthern onca, puma, Puma concolor, track differentiation. Sign surveysare usefulnoninvasive tools to assess and and onlyfew tracks are foundin each event(Fjelline and monitorelusive or rare species (Wemmer et al. 1996, Mansfield1988, Wemmer et al. 1996, Grigioneet al. 1999, Gompperet al. 2006). Spoor countshave been used as Lewisonet al. 2001). Multivariateanalyses have been used indicatorsof presence,relative abundance, and density to reduce subjectivityand improve accuracy in sign estimationof differentspecies (Van Dyke et al. 1986, recognition(Zielinski and Truex 1995, Zalewski 1999, Stander1998, Crooks 2002, Wiltinget al. 2006). Likewise, Harringtonet al. 2008, Steinmetzand Garshelis2008). In manystudies involving endangered species have used sign addition,multivariate analyses of felinetracks have been surveysto obtain basic ecological information(e.g., used formore demanding objectives like sex determination distributionand habitatuse; Perovic and Herran 1998, and individualidentification (Smallwood and Fitzhugh Potvinet al. 2005,Markovchick-Nicholls et al. 2008) and as 1993, Riordan 1998, Sharma et al. 2003, Wiltinget al. preliminaryor complementaryassessment in ecological 2006, Isasi-Cataláand Barreto2008; but see Karanthet al. researchor conservationplans (Schaller and Crawshaw 2003 and Gordonet al. 2007). 1980, Rabinowitz and Nottingha 1986, Soisalo and Jaguars{Panthern onca) live in sympatrywith pumas (Puma Cavalcanti2006, Paviolo et al. 2008). Tracksare not only concolor)along mostof theirdistribution and both species one of themost commonly used signsbut theymay also be are the focusof manyresearch and conservationprograms usefulin identifyingother associated signs such as fecal (Nowelland Jackson 1996, Sanderson et al. 2002, Conroyet samples(Wemmer et al. 1996, Scognamilloet al. 2003, al. 2006, Shaw et al. 2007). These felidsalso sharemost of Shaw et al. 2007, Azevedo 2008). However,one of the theirrange with some large canids, such as pumaswith gray problemsof employingtracks and othersigns is correct wolves (Canis lupus) and coyotes(C. latrans) in North identificationof the species,particularly in areaswhere >2 America,both felids with the maned wolf (Chrysocyon similarspecies are likelyto be foundand wherelittle or no brachyurus)in centralSouth America, and both felidswith informationabout their presence is available. the domesticdog in mostof theirdistribution. Because trackshape is affectedby many factors(like Tracks of pumas and jaguars have been employedfor substratequality or the pace of the animal) it is often research,monitoring, and management(Smallwood and difficultto distinguishtracks of similarspecies, particularly Fitzhugh 1995, Hoogesteijn 2007, Shaw et al. 2007). in areaswhere soil conditionsmake track printing difficult However,it is oftendifficult to differentiatebetween puma and tracksbecause are similarin size and 1 jaguar they shape E-mail:[email protected] and both are frequentlyconfused with big canids' tracks De Angelo et al. · Large Felid and Canid Track Identification 1141 This content downloaded on Fri, 25 Jan 2013 17:25:26 PM All use subject to JSTOR Terms and Conditions (Smallwoodand Fitzhugh1989, Childs 1998, Shaw et al. foras manyindividuals and zoos as possible,looking for all 2007). For this reason, many authors have described kind of tracks(front, rear, right, and left feet),diverse qualitativetraits to differentiatepuma fromdog tracks substrates,and differentanimal size and behavior(to favor (Smallwoodand Fitzhugh 1989, Shaw et al. 2007; Appendix includingtrack variability caused by animal pace andwt). We A), and to distinguishjaguar from puma tracks (Schaller and collectedall tracksfrom free-walking animals (i.e., the paw Crawshaw1980, Aranda1994, Childs 1998; AppendixA). was notpressed into the substrate by handlers). We obtained Nevertheless,due to thesubjectivity and exceptionsfound in all dog tracksand 78% of tracksfrom zoos fromunprepared applicationof qualitativetraits, some authorsproposed substrates,whereas remaining zoo trackswere from sleekly quantitativeidentification measures (Beiden 1978, Small- preparedsurfaces on. which animals walked freely to yield wood and Fitzhugh1989, Aranda 1994, Childs 1998, Shaw betterprints (mainly in zoos lackingan adequatesurface for et al. 2007; AppendixΒ). However, these quantitative printingtracks). Substrate conditions varied among zoos and criteriawere createdand evaluatedbased on eitherfew among the siteswhere we collecteddog tracks,including knownindividuals or unknownfield tracksclassified by mud,dust, and sand (wetor dry),for all groupsconsidered. qualitativetraits. We triedto collectas manytracks as possiblefrom each In view of these shortcomings,we used a diverseset of individual.However, because we obtainedfew tracks in most tracksfrom definitely known origin with the aim of 1) of the zoos, we triedto select <1 track/footfrom each evaluatingperformance of qualitativeand quantitative individual,in orderto haveall individualsrepresented by the methodologiesdescribed to differentiatetracks of big canids, same numberof tracks.We used 3 methodsfor recording pumas,and jaguars;and 2) developingaccurate and robust tracks:plaster molds, digital photos with a metricrule as size quantitativemultivariate criteria to classifytracks of these reference,and trackcontour in acetatepaper. Multiplicity of groupsand species. zoos, individuals,collectors, and methodsgenerated a variety oftracks of different fromwhich we discarded STUDY AREA qualities, only for furtheranalysis those tracks that had conspicuous Between2004 and 2008, we obtainedtrack records from deformationsprecluding an accuratemeasurement. jaguars,pumas, and manedwolves kept in captivity.Our To givedigital format to collecteddata, we photographed taskwas carriedout in collaborationwith 18 zoos from6 the plastermolds with digitalcameras and scannedtrack countries:Zoo Batán(Batán, Argentina), Zoológico Bosque tracings,including a metricrule. Then, the firstauthor Guarani(Foz do Iguaçu, Brazil), RefugioBiológico Bela digitalizedall tracksusing the splinetool in AutoCAD® Vista (Foz do Iguaçu, Brazil), Zoológico Itaipú Paraguay 2004 (AutoDesk,Inc., Fresno,CA). For digitization and (Hernandarias,Paraguay), Parque Ecológico Urbano de Rio measurement,we scaled tracks accordingto the size- Cuarto(Rio Cuarto,Argentina), Zoológico de la Ciudad de referencemetric rule. Subsequently, we rotatedand placed BuenosAires (Buenos Aires,Argentina), Temaiken (Bue- tracksover a base linedefined by a tangentline between the nos Aires,Argentina), Zoológico de Roque Sáenz Pena outerheel lobes (Smallwoodand Fitzhugh1989; Fig. 1). Tatu (Roque Sáenz Pena, Argentina),Zoológico Carreta We analyzed167 tracks,including plaster molds (n = 98), de Ia Ciudad de (La Cumbre,Argentina), Jardin Zoológico photographs(n = 57), and