• Embed Doc
  • Copy Link
  • Readcast
  • Collections
  • CommentGo Back
 
 Journal of Tropical Ecology
(2005)
21
:349–353. Copyright © 2005 Cambridge University Pressdoi:10.1017/S0266467405002397 Printed in the United Kingdom
SHORT COMMUNICATION
Ocelot (
Felis pardalis
) population densities, activity, and ranging behaviour in the dry forests of eastern Bolivia: data from camera trapping
Leonardo Maffei
∗
†
1
, Andrew J. Noss
†
, Erika Cu´ellar 
†
and Dami´an I. Rumiz
†‡
∗
Capitan´ıa del Alto y Bajo Isoso, Casilla 3800, Santa Cruz, Bolivia
†
Wildlife Conservation Society-Bolivia, Casilla 6272, Santa Cruz, Bolivia
‡
Museo de Historia Natural Noel Kempff Mercado, Santa Cruz, Bolivia(
 Accepted 6 November 2004
)
Key Words:
activity, capture–recapture, Chaco,
Felis (Leopardus) pardalis
, population density, ranging patterns
In comparison with the Neotropical big cats, jaguar(
Panthera onca
L.) and puma (
Felis concolor 
L.), mediumandsmallfelidsarepoorlystudied.Furthermore,studyingwild felids in forest habitats is extremely difficult usingdirect methods given that most species are principallynocturnalandsecretive(Gittleman1996).Indirectmeth-ods are therefore particularly important, e.g. radio-tele-metry (Emmons 1987, 1988; Konecny 1989, Ludlow &Sunquist 1987) or camera trapping (Maffei
et al
. 2002,Trolle & K´ery 2003). Using systematic camera trapsurveys, we compare the population density of ocelots(
Felis pardalis
L.) across five Bolivian dry-forest sites withdifferent habitat types and/or annual rainfall regimes(Table 1). We hypothesize that ocelot densities will de-cline as rainfall declines. In addition, we estimate thepopulationofocelotsinthe34400-km
2
Kaa-IyadelGranChaco National Park. Finally, we describe and evaluateadditional ecological information provided by cameratrapping: activity patterns relative to seasonality andmoon phase, sex ratios, ranging patterns and relativeabundance compared with sympatric felids.The Kaa-Iya del Gran Chaco National Park protectsthe northern end of the Gran Chaco ecoregion, withchaco savannas in the park’s driest south-west corner,and characteristic dry chaco vegetation with a low forestcanopy(4–6m)andemergentsupto10mtall,numerouscactus species and terrestrial bromeliads (Navarro &Maldonado 2002). To the north and east, as annual pre-cipitationincreasesthevegetationgradesintochiquitanodryforest,anecoregionendemictoBoliviaandforminga
1
Correspondingauthor,atCapitan´ıadelAltoyBajoIsoso,Casilla3800,Santa Cruz, Bolivia. Email: lmaffei@wcs.org
transition area between chaco and cerrado (Killeen
et al
.1998). Transitional chaco–chiquitano dry forests withintheKaa-IyaNationalParkincludealow–mediumcanopy(12–15m), with emergents 18–20m tall (Navarro &Maldonado 2002).The initial focus of our study was jaguars (Maffei
et al
.2002,2004;Silver
etal
.2004),butpilotstudiesindicatedthat we could simultaneously collect valuable data onocelots. Methods for camera trapping and estimatingpopulation densities follow Karanth & Nichols (2002) inadditiontothereferencesabove.Weconductedseven60-dsystematic camera trap surveys, utilizing 16–31 sets of camera traps per survey, at five dry-forest sites (Table 1).Individual ocelots are relatively easy to distinguish,usingblack-and-whiteimages,basedonrosetteandstripepatterns (see also Trolle & K´ery 2003). In the roughly10% of cases (incomplete photos) where we were unableto positively identify an individual, either matching itwith previously identified individuals or confirming thatit did not match any of them, we tentatively consideredthe photos to be repeat observations of previouslyidentified individuals at the same location. While doingso increased the number of recaptures in some cases, wedid not consider them to be new individuals, so neitherabundance nor buffer estimates changed. In 75–80% of cases across sites, we could confirm sex of the individualsas males present prominent testicles.We used capture-recapture sampling methods (Otis
et al
. 1978), namely the CAPTURE program (Rexstad &Burnham 1991), to estimate total abundance based onthe number of individuals identified (M
t+1
) and capturefrequency statistics, generating estimates of captureprobability (p) and population size (N). We consideredeachdaytobeaseparatesamplingoccasion,thussurveys
 
350 LEONARDO MAFFEI
ET AL.
Table 1.
Location and characteristics of systematic camera-trap surveys at the five study sites.Forest type [and Precipitation SurveySurvey site Location landscape system]
a
(mm y
−
1
) season Dates Trap nightsGuanacos 20
◦
03
S Chaco woodland on 400 dry December 2003–February 2004 96062
◦
26
W sand dunes [Alluvialplains]Cerro 19
◦
31
S Chaco woodland on 500 dry April–May 2002 2280Cortado 62
◦
18
W sandstone outcrops/ wet December 2002–January 2003 1680sands [Alluvial plains]Tucavaca 18
◦
31
S Transitional chaco- 800 wet January–March 2002 192060
◦
49
W chiquitano forest dry April–June 2003 1560[Precambrian shield]Ravelo 19
◦
17
S Transitional chaco- 800 wet February–April 2003 216060
◦
37
W chiquitano forest dry September–November 2003 1320[Precambrian shield]San 17
◦
05
S Chiquitano forest and 1200 dry September–November 2002 1695Miguelito 61
◦
47
W transition to chacoforest [plains andshield contact]
a
Forest types and landscape systems are described by Navarro & Maldonado (2002).
included 60 sampling periods, and multiple observationsof the same individual during a single day to be a singleobservation. The CAPTURE program tests the dataagainst several capture–recapture models, and recom-mends the model that best fits the data. Following Otis
et al
. (1978), we assume that M(o) is not appropriatefor the species, and use the second-best model whenCAPTURE recommends M(o). We did not use the test forclosure in the CAPTURE program, because Rexstad &Burnham (1991) indicate that ‘the test has poor powerand is seldom capable of properly rejecting the nullhypothesis of closure’. Instead, we assumed that thepopulation of ocelots remained closed during the 60-dsurvey period.In order to estimate the effective sampled area wherethe population size (N) was estimated, we measuredthe mean maximum distance covered by all individualsphotographedattwoormorelocationsduringthesurveyperiod as a proxy for home-range diameter (Karanth &Nichols2002).Weusedhalfthemeanmaximumdistance(w) to buffer each camera trap location (following Silver
et al
. 2004). Animals whose home ranges overlap thebuffered area at least partially have a capture probabilitygreater than zero.Wedividedthepopulationsizeestimatebytheeffectivesampled area to estimate population density (individualskm
−
2
). In turn, we extrapolated population densities atthe four chaco study sites across the entire area coveredbysimilarforeststoroughlyestimatethetotalpopulationof ocelots within the Kaa-Iya National Park.Wederiveddailyactivitypatternsbysummingcapturesrecorded for each hour of the day: the camera trapsfunctioned 24h a day and the automatic camerasstamped the date and time on each photograph. Inturn, we compared them across sites, seasons, and moonphases. In addition to the rough estimate of home rangelengthprovidedbythemeanmaximumdistancedescribedabove, cumulative observations of individuals at three ormore locations provide a rough suggestion of rangingpatterns.Wepresenttheareaofthepolygonjoiningthesepoints as a minimum observed range.Individual ocelots were photographed as many as17timesduringthecumulative2-ysurveyperiod,thoughmany were recorded only once. Considering all indi-viduals that could be sexed over the entire period, sexratios varied from site to site: 44% males (N
=
45) atTucavaca, 47% males (N
=
19) at San Miguelito, 50%males (N
=
24) at Cerro and 65% males (N
=
59) atRavelo. Sex ratios did not vary by season except atTucavaca, where relatively fewer males were observedin the dry season. Across the three Kaa-Iya sites, moreindividualsofbothsexeswereidentifiedinthedryseasonas opposed to the wet season.We did not observe a single ocelot at Guanacos: this isthe driest site and the rainfall may be insufficient for oce-lots,whiletheopenchacograsslandsthatpredominateatthe site may not be suitable for the species. From the twosurveys at Cerro Cortado (chaco forest) and Tucavaca(transitional chaco-chiquitano forest) respectively, weestimatedanaveragepopulationdensityofapproximately0.3 ocelotskm
−
2
at each site (Table 2). Ocelots can be re-latively abundant even where rainfall is very low, if for-est cover is sufficiently dense and extensive. At Ravelo,transitional chaco–chiquitano forest with a rainfall re-gime similar to Tucavaca’s, the average density of 0.6ocelotskm
−
2
isdoubletheestimatesfortheothertwosites. Despite the considerably higher rainfall at SanMiguelito(chiquitanoforest),averagepopulationdensityappears to be lower than at Ravelo. Together withdensity estimates from other Neotropical sites, theseestimates contribute to a weak pattern of higher ocelotdensities with increasing precipitation: Pearson’s
 
Ocelots in dry forests
351
Table2.
OcelotpopulationdensitiesfromcameratrapsurveysinthedryforestofSantaCruz.Photos:Numberofindependentocelotphotosduringtheentire survey. Individuals: Number of individuals identified during each survey. w: Buffer equivalent to half the mean maximum distance coveredby animals observed at more than one location during the survey. A: Effective survey area estimated from the buffered camera trap locations.Model: model that best fits the results selected by the program, M(h) is heterogeneity model using jackknife estimator and assuming heterogeneityin capture probability among individual ocelots, M(o) is null model using null estimator and assuming no variation in capture probability amongindividuals, M(b) is behaviour model using constant probability removal estimator and assuming behavioural variation in capture probabilityamong individuals. N: Abundance estimated by the CAPTURE software. D: Abundance divided by effective sampled area. P: capture probability.D
±
SESurvey Photos Individuals w (m) A (km
2
) Model N
±
SE (individuals km
−
2
) PGuanacos (dry) 0Cerro Cortado (dry) 65 25 1440 99 M(h) 30
±
4.53 0.34
±
0.04 0.03Cerro Cortado (wet) 53 18 1340 80 M(h) 20
±
3.35 0.25
±
0.03 0.05Tucavaca (dry) 61 18 1340 80 M(b) 20
±
2.84 0.24
±
0.03 0.03Tucavaca (wet) 82 34 1140 84 M(o) 39
±
3.11 0.34
±
0.02 0.03Ravelo (dry) 91 42 1440 117 M(h) 61
±
11.3 0.52
±
0.05 0.02Ravelo (wet) 102 40 1015 98 M(h) 56
±
6.67 0.66
±
0.07 0.03San Miguelito (dry) 71 23 1350 52 M(h) 29
±
4.25 0.56
±
0.07 0.04
correlation coefficient
=
0.576, N
=
9, df 
=
7, P
=
0.10(Figure 1). Across these sites, the Venezuelan llanos andBelize rain-forest sites have unexpectedly low populationdensities. Considering only the five Bolivian dry forestsites, the relationship is stronger: Pearson’s correlationcoefficient
=
0.826, N
=
5, df 
=
3, P
=
0.07.We estimate the total population of ocelots for theKaa-Iya National Park to exceed 10000 adult animals(range
=
9300–11300), assuming an average popula-tion density of 0.3 ocelots km
−
2
based on our surveysof four sites in landscape systems that extend over25300km
2
of the protected area. One extensive land-scape system remains to be surveyed, chaco transitionalforest covering 8600km
2
in Kaa-Iya, whereas chacoriverine and chaco savannas each cover less than500km
2
. The absence of ocelots in the latter habitat(Guanacos)doesnotgreatlyaffectthepopulationestimate
Figure1.
Ocelotpopulationdensitiesandannualprecipitation.Datafromtelemetry studies at Manu National Park, Peru (1, Emmons 1988) andLlanos, Venezuela (2, Ludlow & Sunquist 1987). Data from cameratrapping studies in Chiquibul National Park, Belize (3, M. J. Kelly,unpublished);Pantanal,Brazil(4,Trolle&K´ery2003);KaaIyaNationalPark, Bolivia (5–8, this study); and San Miguelito Private Reserve,Bolivia (9, this study).
for the protected area. We prefer to extrapolate fromthe relatively low density estimates at Tucavaca andCerro, rather than the higher Ravelo estimate (0.6 oce-lotskm
−
2
), in order to err on the conservative side.Ocelots exhibited strongly nocturnal behaviour acrossall sites, with 89% of records at night, and activitypeaking at 21h (Figure 2). These results are consistentwithdatafromtelemetryresearchattropicalmoistforestand savanna sites elsewhere in South America (Emmons1988, Emmons
et al
. 1989, Konecny 1989, Ludlow &Sunquist 1987). The highest frequency of day-timephotographs(Ravelo)didnotexceed21%.Ocelotactivitydid not vary significantly with lunar phase or by season,though we did not survey during the coldest months(June–July).Werecordedcumulativeobservationsatthreeormorelocationsfor22animals:threeatCerroCortado,13atTu-cavaca, and six at San Miguelito. The average minimumobserved range was 3.0km
2
(range
=
0.2–13.2km
2
,
Figure2.
OcelotactivitypatternsintheKaa-IyadelGranChacoNationalPark and San Miguelito Private Reserve (per cent of observations byhour).
of 00

Leave a Comment

You must be to leave a comment.
Submit
Characters: ...
You must be to leave a comment.
Submit
Characters: ...