Transcript
Jaguar( Panthera onca )feedingecology:distributionofpredatorandpreythroughtimeandspace M. Weckel 1 , W. Giuliano 2 & S. Silver 3 1 Fordham University, Bronx, NY, USA2 University of Florida, Gainesville, FL, USA3 Wildlife Conservation Society, Bronx, NY, USA Keywords jaguar; Panthera onca ; opportunistic preyselection; camera traps. Correspondence Mark Weckel, Fordham University, Bronx,NY 10458, USA.Email:
[email protected] 12 May 2005; accepted14 November 2005doi:10.1111/j.1469-7998.2006.00106.x Abstract Jaguars Panthera onca inhabiting tropical or subtropical evergreen moist foresthave often been classified as opportunistic predators because they consume preyrelative to its availability. However, these studies failed to address simultaneouslythe distribution of predator and prey through time and space, which may lead toan incomplete or erroneous understanding of jaguar foraging strategies. In thisstudy, we reconstructed jaguar diet from scat, and used camera traps to investigate jaguar prey availability and the distribution of jaguar and its prey through spaceand time. We focused our examination of predator–prey temporal and spatialrelations on forest infrastructure comprising man-made paths, small mammaltrails, tapir Tapirus bairdi trail and trail-less, forested areas as they representdistinct habitats for prey selection. Overall, we observed high overlap between theprey used and available, suggestive of opportunistic foraging. However, jaguarsexhibited selective tendencies in discriminating between larger prey. Jaguars usedcollared peccary Tayassu tajacu greater than its availability, while preying uponthe equally abundant and similarly distributed white-lipped peccary Tayassu pecari and tapir less than predicted based upon availability. Armadillo Dasypusnovemcinctus and paca Agouti paca , 56.6% of total consumption, were consumedrelative to availability but exhibited low spatial overlap with jaguar. Armadilloand paca used trail-less, forested areas and small mammal trails not used by jaguarand were photographed more frequently at greater distances from man-madepaths, major thoroughfares for jaguars. This study suggests that although forest jaguars use prey relative to its abundance, jaguars may rely on foraging strategiesother than chance encounters for exploiting prey. Introduction Jaguars Panthera onca inhabiting tropical or subtropicalevergreen moist forest have been classified as opportunisticpredators because they consume prey relative to its abun-dance (Rabinowitz & Nottingham, 1986; Aranda &S ´anchez-Cordero, 1996; Garla, Setz & Gobbi, 2001). It hasbeen suggested that low visibility and fewer predictable sitesfor prey aggregation in tropical forests may limit highlyselective foraging decisions (Emmons, 1987) characteristicof jaguars of more open habitat (Scognamillo et al ., 2003),and therefore diet largely reflects chance encounters. How-ever, studies have shown that even forest jaguars exhibitspecies-specific prey selection (Emmons, 1987; Novack,2003) and how these feeding preferences can be site specific(Novack, 2003). A possible cause for this disparity may bethe failure to address simultaneously the distribution of predator and prey through time and space. Temporal andspatial overlap among predator and prey may be as impor-tant as the relative abundance of prey in predicting encoun-ter probabilities and defining opportunistic tendencies(Griffiths, 1975); however, these parameters are often over-looked in jaguar dietary studies.The primary objective of this study was to assess thepossibility of opportunistic prey selection among jaguars inthe Guam Bank region of the Cockscomb Basin WildlifeSanctuary (CBWS), Belize by addressing the followingpredictions: (1) jaguars select prey in proportion to therelative abundance of grounddwelling prey items; (2) ja-guars and their prey are positively associated through time;and (3) jaguars are positively associated with their preythrough space. We focused our investigations on elementsof the forest infrastructure consisting of man-madepaths, animal trails and trail-less, forested areas as spatialcomponents offering distinct habitats for possible preyencounters. In addition, we investigated how habitat typeand proximity to man-made paths affect predator and preydistribution. Journal of Zoology 270 (2006) 25–30 c 2006 The Authors. Journal compilation c 2006 The Zoological Society of London 25 Journal of Zoology. Print ISSN 0952-8369 Methods Studyarea We conducted the study from 1 June to 24 November 2002in the north-east region of the CBWS, Belize, subsequentlyreferred to as Guam Bank. The CBWS occupies 437km 2 and is delineated by two mountain ranges: the CockscombMountains to the north and the Maya Mountains to thewest. Elevation ranges from 50 to 1120m. On average, theregion receives 270cm of rain annually, with the majorityfalling during the wet season (June–December; Kamstra,1996). The mean annual temperature is 25 1 C, fluctuating anaverage of 5 1 between the warmest (April and June) andcoldest months (November and January; Kamstra, 1996).Ecotourism was the primary human disturbance in GuamBank during the study. Patrols conducted by the BelizeAudubon Society (BAS) failed to document evidence of humans hunting jaguar prey in Guam Bank; however,subsistence hunters inhabiting villages on the CBWS per-iphery established hunting routes and encampments to thesouth and west of the Guam Bank study area (E. Saqui,pers. comm., Belize Audubon). Cameraarray Camera traps were used to describe the abundance anddistribution of Guam Bank jaguars and potential prey. Weestablished five 1-km north–south transects, each starting ata man-made path and spaced at c . 2.5-km intervals to coversystematically the Guam Bank region of the CBWS. Wepositioned three cameras on each transect in the followingmanner: a single camera (Wildlife Pro Camera, ForestrySuppliers Inc.; Jackson, MS, USA) was permanently placedon the man-made path (MCAM) at the head of the transect.A second camera was set on a small mammal trail intersect-ing the transect (SMCAM) and rotated every 2weeksamong small mammal trails along the transect. A thirdcamera was placed along the transect at one of four forestedlocations lacking trails (FCAM) and corresponding to oneof four distances from the man-made path (275, 525, 775and 1025m). The FCAM camera was systematically rotatedamong the four distances every 2weeks. We sampled succes-sive distances to examine the impact of man-made paths on jaguar and prey distribution in forest interior areas, definedas areas set back from man-made paths. A total of 15cameras was operational at any given time across alltransects.We placed four additional cameras on four individualBaird’s tapir Tapirus bairdii trails intersecting man-madepaths (TCAM) as they represented discrete and uniqueelements of the forest infrastructure. Tapir trails weredistinguished from those of small mammals by their largerwidth and consistent presence of tapir tracks. The TCAMwas rotated along successive 25m distances from the man-made path to investigate the effect of man-made pathproximity on tapir trail use. However, tapir trails were tooshort and captures per trail too few to allow a distanceanalysis. All cameras across all locations were placedbetween 0.5 and 0.75m off the ground and 0.5–1.0m off the trail depending on the topography. Scatanalysis We constructed jaguar diet from scats collected concomi-tantly with the camera surveys. Jaguar scats were distin-guished from those of the sympatric puma by severalmethods: (1) the presence of jaguar hair (assumed ingestedwhile grooming), (2) associated pugmarks, (3) collection andidentification by trained indigenous BAS staff and (4) mole-cular techniques. Identification by BAS staff was based onthe presence of tracks; however, such information could notbe corroborated by the authors. Molecular classification of scat srcin required extraction of DNA from the scat sur-face, where the colon wall presents the possibility of extract-ing genetic material from sloughed epithelial cells (Foran,Crooks & Minta, 1997). We targeted the 16S mitochondrialgenome for DNA amplification using 16S universal primers(Hoezel & Green, 1992). The protocol for DNA amplifica-tion and sequencing followed those prescribed by Farrell,Roman & Sunquist (2000) with minor modifications toreactant concentrations and annealing temperatures (Weck-el, 2005). Following multiple attempts at molecular classifi-cation for all scat samples, scats were broken apart andwashed over a series of screens (1.7, 1.0, 0.5 and 0.1mmmesh). We identified food items to the lowest possible taxonby hair and nail identifications (Putnam, 1984). Diet wasexpressed as frequency of occurrence (per cent of total scatsin which a prey type was found). Preyselectivity We considered a species as potential jaguar prey if it hadbeen cited as such in jaguar literature. A relative abundanceindex (RAI) of available jaguar prey was generated fromspecies-specific capture rates (SSCRs) averaged across allcamera locations. No independent assessment of animaldensity was made; therefore, photographs could not becorrected to account for actual abundance (see Carbone et al ., 2001). As camera trapping may inflate the relativeabundance of some species (Silveira, Jacomo & Diniz-Filho,2003), largely a consequence of camera placement (Carbone et al ., 2001), the following measures were taken. Capturesfrom TCAMs were not included in the analysis, as one ortwo individual tapirs were believed to utilize frequently thesame trail and may have artificially inflated the relativecontribution of the perissodactyl. Captures of arboreal preytypes, such as northern tamandua Tamandua mexicana andkinkajou Potos flavus across all camera stations, were alsoexcluded from analysis. Semi-arboreal species (i.e. coat-imundi Nausa narica ) were included in the analysis, withcapture rates interpreted as a measure of terrestrial avail-ability, not actual abundance. Photographic captures of crested guan Crax rubra and three species of tinamou( Crypturellus boucardi , Crypturellus souri , Tinamus major )were aggregated into the single category, ‘birds,’ as remains Journal of Zoology 270 (2006) 25–30 c 2006 The Authors. Journal compilation c 2006 The Zoological Society of London 26 Jaguar feeding ecology M. Weckel, W. Giuliano and S. Silver found in scat were not identified to species level. Multipleexposures of the same species on the same trap night at agiven camera station were considered a single event exceptfor captures of white-lipped Tayassu pecari and collared Tayassu tajacu peccaries. Multiple exposures of Tayassu spp. were counted as multiple captures at a 1:1 ratio toaccount for their herding behavior (Robinson & Eisenberg,1985; Mayer & Wetzel, 1987; Sowls, 1997). Failure toaccount for herd size would underestimate the proportionalcontribution of these species to the prey base (Rabinowitz &Nottingham, 1986). In the subsequent analyses, individualcamera stations served as independent experimental units( n =63). To standardize capture rates among camera sta-tions, we calculated a SSCR for each location (defined as thenumber of capture events per species expressed per 100 trapnights).Horn’s similarity index (SI; Horn, 1966) was used todescribed the extent of overlap between prey used (scatanalysis) and prey available (camera survey), with valuesvarying from 0 (no shared species) to 1 (identical diets). Weused Ivlev’s electivity index (EI; Ivlev, 1961) to assessselection for individual prey types. Values for the EI varyfrom 1 (complete avoidance) to + 1 (complete preference).Zero and near-zero values indicate that prey were usedsimilar to availability. Temporaloverlap Predator–prey temporal associations were based on the timeof photographic capture. Captures were classified into threecategories (Schaik & Griffiths, 1996): nocturnal(18:31–05:00h), diurnal (06:31–17:00h) and crepuscular(17:01–18:30 and 05:01–06:30h). We used Horn’s SI todetermine the amount of temporal overlap between jaguarand its potential prey. Habitatoverlapandspatialdistribution We used Horn’s SI to describe the spatial overlap between jaguar and each prey type for the four habitats sampled(man-made path, small mammal trail, tapir trail and trail-less, forested areas). To further characterize jaguar and itsprey distribution, we compared habitat use by analysis of variance (ANOVA), blocked for the effect of transect (Sokal& Rohlf, 1995). Four treatments were usedcorresponding toman-made paths (MCAMs; n =5), small mammal trails(SMCAMs; n =24), tapir trails (TCAMs; n =14) and trail-less, forested areas (FCAMs; n =19). We performedBonferonni post hoc tests ( a =0.1) to compare amongtreatments (Sokal & Rohlf, 1995).To address the effect of man-made path proximity oncapture rates, we compared SSCRs along the forest transect(FCAMs) by linear regression with replication (Sokal &Rohlf, 1995). Five distances were sampled: 0, 275, 525, 775and 1025m. Captures from MCAMs were recorded as 0mfrom the man-made path. SSCRs were rank transformed forall statistical tests to meet assumptions of normality andhomogeneity of variance. All tests were performed onSYSTAT (2001) Version 10 (SPSS Inc.) and consideredsignificant at P 0.1. Results Scatanalysis A total of 34 scats were collected in the Guam Bank studyregion of the CBWS. Twenty-three scats were identified as jaguar with pugmarks, ingested hair, indigenous identifica-tion and DNA assigning srcin to 7, 1, 14 and 12 scats,respectively. Four puma scats were identified by moleculartechniques and seven scats were excluded from analysis aswe could not confidently determine their srcin. Preyselection Nine-banded armadillo Dasypus novemcinctus , peccary andpaca Agouti paca comprised 83% of the jaguar diet (Weckel,2005). We observed high overlap between prey used andavailable (SI=0.83; Table 1). Those prey items consumedrelative to their availability (EI 0) included nine-bandedarmadillo, paca and coatimundi. Guam Bank jaguars ex-hibited moderate avoidance of white-lipped peccary Table 1 Jaguar Panthera onca prey use, availability and selection,Cockscomb Basin Wildlife Sanctuary, Belize, 1 June–24 November2002SpeciesPreyuse a Preyavailable b EI c Nine-banded armadillo Dasypus novemcinctus 33.3 27.2 0.10Collared peccary Tayassu tajacu 23.3 9.5 0.42Paca Agouti paca 23.3 20.3 0.07Red brocket deer Mazama americana 6.7 9.9 0.19White-lipped peccary Tayassu pecari 3.3 7.3 0.38Coatimundi Nasua narica 3.3 3.4 0.01Kinkajou Potos flavus 3.3 NA NABaird’s tapir Tapirus bairdi 0 9.9 1.00Common opossum Didelphis marsupialis 0 5.6 1.00Agouti Dasyprocta punctata 0 2.6 1.00Striped hog-nosed skunk Conepatus semistriatus 0 0.4 1.00Bird spp. d 0 3.4 1.00Raccoon Procyon sp. 0 0.4 1.00Reptile sp. 0 NA NAUnidentified mammal 3.3 NA NA a Diet constructed from 23 scats collected from Guam Bank simulta-neously to the camera survey expressed as frequency of occurrence(Weckel, 2005). b Species-specific photographic capture rates are expressed as thenumber of captures per 100 trap nights. c EI, Ivlev’s electivity index (Ivlev, 1961) where values range from 1(complete avoidance) to + 1 (complete preference) and was used toexamine prey selection. d Birds include Crypturellus boucardi , Crypturellus souri , Tinamus major and Crax rubra . Journal of Zoology 270 (2006) 25–30 c 2006 The Authors. Journal compilation c 2006 The Zoological Society of London 27 Jaguar feeding ecologyM. Weckel, W. Giuliano and S. Silver (EI= 0.38) and red brocket deer Mazama americana (EI= 0.19) and complete avoidance of Baird’s tapir, com-mon opossum Didelphis marsupialis , striped hog-nosedskunk Conepatus semistriatus , raccoon Procyon lotor andbird species (EI= 1). Jaguars were found to use only thecollared peccary greater than its availability (EI=0.42),comprising 23.3% of all prey items. Temporaldistribution Jaguars were found to be active largely at night, with con-siderable activity observed during dawn and dusk (Table 2).Armadillo, paca, common opossum and tapir were almostexclusively nocturnal. Both peccary species and red brocketdeer shared a common activity pattern (Table 2). However,the latter were photographed more frequently during theday. Jaguars exhibited high overlap in temporal activitywith armadillo, paca, tapir, common opossum and bothpeccary species (SI Z 0.80), while displaying only moderateoverlap with red brocket deer, coatimundi and grounddwel-ling birds (0.50 SI o 0.80). Spatialdistributionandhabitatoverlap Tapir, bird spp. and red brocket deer were observed havingthe highest spatial overlap with jaguar (Table 3). The jaguarand tapir preferred man-made paths and tapir trails( P o 0.001) and were never photographed using small Table 2 Activity periods of jaguar Panthera onca and potential prey based on timed photographic captures, Cockscomb Basin Wildlife Sanctuary,Belize, 1 June–24 November 2002SpeciesTotalcapturesCrepuscular a (%) b Diurnal(%)Nocturnal(%) SI c Jaguar Panthera onca 35 20.0 17.1 62.9 NACommon opossum Didelphis marsupialis 19 10.5 0 89.5 0.90Baird’s tapir Tapirus bairdi 34 2.9 0 97.1 0.84White-lipped peccary Tayassu pecari 29 20.7 58.6 20.7 0.83Nine-banded armadillo Dasypus novemcinctus 55 3.6 0 96.4 0.83Collared peccary Tayassu tajacu 25 32.0 52.0 16.0 0.82Paca Agouti paca 49 4.1 0 95.9 0.82Coatimundi Nasua narica 8 0 87.5 12.5 0.63Red brocket deer Mazama americana 35 22.9 68.6 8.6 0.68Bird spp. d 39 20.5 79.5 0 0.52 a Crepuscular (17:01–18:30 and 05:01–06:30h); diurnal (06:31–17:00h); nocturnal (18:31–05:00h). b Per cent of captures during that period. c Similarity index (SI) (Horn, 1966), where values range from 0 (no overlap) to 1 (complete overlap) with jaguar. d Birds include Crypturellus boucardi , Crypturellus souri , Tinamus major and Crax rubra . Table 3 Species-specific capture rates (SSCRs) a of jaguar and potential prey among habitat types, Cockscomb Basin Wildlife Sanctuary, Belize,1 June–24 November 2002Habitat types Man-made Forest Small mammal Tapir P SI c Trap stations 5 19 24 14Trap nights 505 553 378 325Species Mean b SE Mean SE Mean SE Mean SE Jaguar Panthera onca 4.9 A 0.1 0 B 0 0 B 0 1.3 C 0.2 o 0.001 NABaird’s tapir Tapirus bairdi 1.7 A 0.4 0 B 0 0 B 0 9.5 A 0.9 o 0.001 0.73Birds spp. d 2.8 1.1 1.4 0.1 3.3 0.5 0.9 0.17 0.647 0.67Red brocket deer Mazama americana 1.5 0.3 1.5 0.2 1.8 0.1 4.6 0.8 0.489 0.61Collared peccary Tayassu tajacu 0.4 0.1 2.5 0.4 1.8 0.2 3 0.4 0.869 0.40White-lipped peccary Tayassu pecari 0.9 0.3 3.7 0.5 1.0 0.2 0.5 0.1 0.310 0.36Coatimundi Nasua narica 0 0 0.7 0.1 1.1 0.1 0.8 0.2 0.740 0.22Paca Agouti paca 0 A 0 2.1 A 0.3 9.4 B 0.5 1.9 A 0.3 0.003 0.13Nine-banded armadillo Dasypus novemcinctus 0 A 0 3.6 AB 0.4 13.0 B 0.7 0.6 A 0.1 0.032 0.08Common opossum Didelphis marsupialis 0 0 2.8 0.3 1.1 0.1 0 0 0.141 0 a SSCRs are expressed as the number of captures per 100 trap nights. b For each species, means followed by the same letter are not significantly different (Bonferroni post hoc test; P 0.1). c Similarity index (Horn, 1966), where values range from 0 (no overlap) to 1 (complete overlap) with jaguar. d Birds include Crypturellus boucardi , Crypturellus souri , Tinamus major and Crax rubra . Journal of Zoology 270 (2006) 25–30 c 2006 The Authors. Journal compilation c 2006 The Zoological Society of London 28 Jaguar feeding ecology M. Weckel, W. Giuliano and S. Silver mammal trails or forested areas. In contrast, armadillo andpaca preferred small mammal trails and trail-less, forestedareas, were never photographed on man-made paths, andexhibited very low spatial overlap with jaguar (Table 3).Both collared and white-lipped peccary demonstrated mod-erate habitat overlap with jaguar (SI 0.40) and failed toshow preferential habitat use. Only three species exhibiteddifferential capture rates as a function of man-made pathproximity. Capture rates for armadillo, common opossumand paca increased with greater distances from man-madepaths (Table 4). Discussion Similar to the earlier observations of Rabinowitz & Notting-ham (1986), Guam Bank jaguars behaved largely as opportu-nisticpredators–inthetraditionalinterpretation–consumingtheir prey relative to its numerical availability. The highoverlap between prey taken and available reflects the pre-valence of armadillo and paca in the jaguar diet (56.6%) andstudy area. However, as observed for forest jaguars in otherstudies (Emmon, 1987; Novack, 2003) Guam Bank jaguarsdid exhibit selective tendencies in discriminating amonglarger prey types. Jaguars consumed collared peccary morefrequently than expected while avoiding the comparablyabundant and similarly distributed white-lipped peccaryand tapir. Large foraging risks and long handling timesprobably explain why jaguars did not prey on tapir, despitea high probability of consistent encounters along man-madepaths, and may explain why tapirs are rarely reported asimportant jaguar prey. White-lipped peccary may alsorepresent a costly foraging option (Roosevelt, 1920; Kiltie& Terborgh, 1983) as they travel in large, aggressive herds(Mayer & Wetzel, 1987); however, unlike tapir, white-lippedpeccaries represent a larger portion of jaguar diet thancollared peccary in both Costa Rica (E. Carrillo, pers.comm., Wildlife Conservation Society) and Brazil (Garla et al ., 2001).Guam Bank jaguars appear to further deviate from strictopportunistic foraging strategies in their consumption of armadillo and paca. Despite consuming armadillo and pacain proportions comparable to their abundance, and sharinga common activity pattern with these prey, jaguars exhibitedlow spatial overlap with both armadillo and paca. Jaguarwere photographed most frequently on man-made paths,whereas armadillos and pacas primarily used small mammaltrails and forested, trail-less areas with only minor preda-tor–prey overlap in the use of tapir trails. In addition,armadillo and paca were captured most often in the forestinterior, areas set back from man-made paths and potentialrefugia from jaguars, further reducing the possibility forchance encounters between the jaguar and its prey. The useof cameras to document these small-scale differences intemporal and spatial occurrence, as a method to investigatepredator–prey relationships, is particularly novel.It is possible that jaguars may opportunistically prey onarmadillo and paca in discrete habitats not sampled in thisstudy, that is ridgelines and stream banks. Furthermore,armadillos may be overrepresented in this dietary study andothers, owing to the high percentage of indigestible remainscompared with other prey species (Zielinski et al ., 1999).However, the results of this study suggest that jaguarconsumption of armadillo and paca reflects foraging strate-gies other than chance encounter, raising the question as towhy the armadillo and paca are of principal importance to jaguar diet. Armadillos and pacas are poor-quality jaguarprey items. A single armadillo can meet the energetic needsof an adult male jaguar for only 1–3days (Novack, 2003),requiring that successful predation events be consistent. If jaguars and armadillos use different habitats within theforest, or if armadillos actively avoid jaguars, jaguars wouldhave to pursue, stalk or ambush its prey, increasing theforaging costs required to capture a relatively poor-qualityfood item.The CBWS supports a dense population of jaguars(8.8/100km 2 ; Silver et al ., 2004) and, therefore, we assumethat prey quality, including the large contribution of arma-dillo and paca to jaguar diet, is not compromising thepopulation’s health. Considering the high photographicrates of armadillo and paca on small-mammal paths, it ispossible that intersections of man-made paths with smallmammal trails may offer predictable, consistent and fre-quent trail crossings of both prey and may represent discretehabitats for jaguars to concentrate foraging activities at lowenergetic costs. High photographic rates of armadillo andpaca deeper in the forest interior further support the theoryof greater predation pressure near man-made paths. Asimilar foraging technique was observed among jaguarsambushing cattle at trail crossings in Brazil (de Almedia,1976), and should be explored as a potential foragingstrategy for smaller prey.Although Guam Bank jaguars used prey relative to itsoverall abundance, suggestive of opportunistic prey selec-tion, we found the relationship to be more complicatedwhen accounting for predator–prey distribution. Jaguarsdiscriminated between similarly available large prey, asreflected in their use of collared peccary, while they fre-quently used armadillo and paca, despite their low energeticreward and apparent low spatial overlap. A dietary analysisthat accounted only for relative prey abundance would failto uncover these distinctions, resulting in spurious conclu-sions on jaguar foraging strategies. This is arguably mostimportant in understanding the numerical importance of armadillo to jaguar diet. If the dominance of armadillo in jaguar diet does not reflect chance encounter, understandinghow jaguars capitalize on the numerical abundance of Table 4 Effect of distance from man-made path on capture rates ofjaguar Panthera onca prey, Cockscomb Basin Wildlife Sanctuary,Belize, June–November 2002Species r r 2 P Nine-banded armadillo Dasypus novemcinctus 0.65 0.42 0.030Paca Agouti paca 0.65 0.42 0.028Common opossum Didelphis marsupialis 0.60 0.36 0.062 Journal of Zoology 270 (2006) 25–30 c 2006 The Authors. Journal compilation c 2006 The Zoological Society of London 29 Jaguar feeding ecologyM. Weckel, W. Giuliano and S. Silver
