ORIGINAL ARTICLE

Sachin Sridhara • Advait Edgaonkar

Ajith Kumar

Understorey structure and refuges from predators influence habitatuse by a small ungulate, the Indian chevrotain (Moschiola indica)in Western Ghats, India

Received: 30 June 2012 / Accepted: 27 January 2013� The Ecological Society of Japan 2013

Abstract The availability of refuges from predators andhigh quality food are thought to determine habitat use insmall ungulates. We tested this hypothesis on habitat useby the Indian chevrotain in a tropical rainforest in theWestern Ghats, using pellet-groups to infer habitat use.Between December 2009 and April 2010, we sampled204 grids of 50 m · 50 m with four spatial replicates ineach, using occupancy framework. We quantified ref-uges such as fallen logs and boulders, understoreycomplexity, and noted the presence of fruiting trees inthe grid. Detection probability, p, of pellet-groups wasestimated at 0.61. The naive estimate of occupancy was0.52, which increased to 0.73 when p was accounted for.Out of eight candidate models compared using AICC,the one with the number of refuges and understoreycomplexity was the best model. Both covariates hadnon-zero positive slopes. Fruiting trees occurred veryinfrequently and did not influence habitat use, perhapsbecause the chevrotain fed more on buds, shoots andyoung leaves during the dry period covered by thisstudy. The strong influence of understorey complexityon habitat use was perhaps also due to the abundance ofthese food items. These results highlight the need tocontrol human use that adversely impacts the avail-ability of refuges and understorey complexity such asremoval of fallen logs and rattans. This study alsodemonstrates the application of the occupancy approachin the study of small and elusive ungulates.

Keywords Occupancy models Æ Rainforest Æ Feedingecology Æ Someshwara Wildlife Sanctuary

Introduction

Small ungulates, with a body mass of 2–15 kg, are dis-tributed widely throughout the tropics (Wilson and Ree-der 2005). Although many species are threatenedthroughout the world, they are amongst the least studiedungulates (Corlett 2007; Sodhi et al. 2004) and the resul-tant lack of information is hindering their conservation(Baillie et al. 2004). Specifically, their cryptic nature (Geist1998; Jarman 1974) poses a challenge to quantitativelyassessing habitat needs. Due to the remarkable degree ofconvergence in morphological and behavioural traits ofsmall ungulates (Barrette 1987; Bodmer 1990), anunderstanding of their habitat requirements has consid-erable applicability across species.

Habitat requirements of small ungulates are thoughtto be driven by the constraints imposed by their bodysize and structure (Geist 1998; Jarman 1974). Smallungulates consistently have a highly arched back andlean hind limbs adapted for ‘saltation’, or successiverapid jumps. They are predominantly solitary or occurin pairs, and use stealth to avoid predators, often takingcover in buttresses, boulders and dense undergrowth,avoiding open areas (Geist 1998; Macdonald 2001). Thefact that metabolic rates increase with decreasing bodysize, whereas digestive capacity decreases, constrainssmall ungulates to consume food items like fruits,known to be low in fibre content and highly digestible(Demment and van Soest 1985; Hofmann 1989; Prinset al. 2006). Between 50 and 80 % of their diet consistsof fruits—a larger proportion compared to larger un-gulates (Gagnon and Chew 2000). However, they alsofeed on grasses, young leaves, shrubs, forbs, buds andshoots (Bodmer 1990; Branan and Marchinton 1985;Dubost 1984; Gautier-Hion et al. 1980). Thus, the

S. Sridhara (&)Post-graduate Program in Wildlife Biology and Conservation,WCS-India Program, National Centre for Biological Sciences,GKVK Campus, Bangalore, IndiaE-mail: sachin.sridhara@gmail.com

A. EdgaonkarIndian Institute of Forest Management, Bhopal, India

A. KumarWildlife Conservation Society, India Program, Bangalore, India

Ecol ResDOI 10.1007/s11284-013-1031-3

availability of cover to escape from predators and theavailability of high quality food such as fruits can behypothesised as the major determinants of habitat useby small ungulates.

In this study we used the Indian chevrotain (Mo-schiola indica) as a model species to test whether theavailability of cover from predators and the presence offruit resources significantly influenced habitat-use ofsmall ungulates. The Indian chevrotain is a smallungulate (�3 kg in body weight) distributed widely inpeninsular India (Groves and Meijaard 2005). Althoughits conservation status is categorised as ‘Least Concern’(Duckworth et al. 2008), it is among the most frequentlyhunted animals in its habitat (Madhusudan and Ka-ranth 2002; Kumara and Singh 2004).

We examined whether habitat use by the Indianchevrotain was influenced by two factors: the availabilityof cover and the presence of fruiting trees. Leopards areknown to predate heavily on small mammals such assmall carnivores, hare, porcupine and even murid ro-dents either when larger prey have been depleted, oftendue to human impacts, or are rare, as in tropical rain-forests (Ramakrishnan et al. 1999; Hayward et al. 2006).In fact, the chevrotain was the second most frequentprey in leopard scats in a tropical rainforest in theWestern Ghats (Sidhu et al. 2011). The Nilgiri marten(Martes gwatkinsi) has also been observed to feed onIndian chevrotain in tropical rainforest (Divya Mudap-pa, personal communication). The feeding ecology ofthe Indian chevrotain is very poorly understood. TheIndian chevrotain is known to feed on fallen fruit(Prasad and Sukumar 2010); however, it might also feedon young leaves, buds and shoots, which are likely to befound as small patches in the understorey.

Materials and methods

Study area

We conducted the study in Someshwara Wildlife Sanc-tuary (88.4 km2), located in the Western Ghats moun-tain range in Karnataka, India (Fig. 1). The altitude inthe Sanctuary ranges from 75 to 870 m and the annualrainfall is �4,000 mm, mostly from the southwestmonsoon during June to September (Pascal 1988).Tropical rainforest is the major forest type, with Di-pterocarpus indicusm, Diospyros candolleana, Diospyrosoocarpa association (Pascal 1988). Large mammals inthe Sanctuary include barking deer (Muntiacus muntjac),sambar (Rusa unicolor), gaur (Bos gaurus), lion-tailmacaque (Macaca silenus), tiger (Panthera tigris) andleopard (Panthera pardus).

Field methods

We conducted the study between December 2009 andApril 2010—the dry period in the Western Ghats (Pascal

1988). Due to the cryptic and elusive nature of the spe-cies, we used pellet-groups as an indicator of habitat-use,as has been done by other studies on ungulates (Gopa-laswamy et al. 2012; Krishna et al. 2008). In order toaccount for imperfect detections of pellets we usedspatial replicates within an occupancy framework(MacKenzie et al. 2002). Nearly 70 ha of least disturbedtropical rainforest in the Sanctuary were overlaid withgrids of size 50 m · 50 m using QGIS 1.6.0 (Fig. 1),generating about 250 grids, although some were notsampled. Grids that fell on a road passing through thesanctuary and human settlements on the roadside wereexcluded from the survey. Each grid was further sub-divided into four sub-grids (25 m · 25 m), the diagonalsof which formed the spatial replicates (see Fig. 1). Thegrids were realised on the ground using a Garmin eTrexVista GPS (http://sites.garmin.com/etrex/). One ob-server walked along the four spatial replicates andcarefully scanned the forest floor for pellet-groups andrecorded all sightings. Pellet-groups could be unambig-uously identified as that of Indian chevrotain since theywere much smaller in size and globular (4–5 mm indiameter) while that of the next large ungulate, Indianmuntjac, were cylindrical with a length of about 10 mm(Fig. 2). In each grid we measured the following fivehabitat covariates expected to influence the detection ofpellet-groups and habitat-use by the chevrotain: (1)number of potential food tree species in fruit (henceforthFRUIT); (2) number of refuges from predators (REF-UGE) measured as the total number of buttresses, fallenlogs, and crevices in rocks and boulders; (3) an index ofunderstorey complexity (USI) was computed by com-bining the height and ground cover of the undergrowth.USI could provide cover from predators and foragingopportunities for high quality food such as fallen fruits,grass and young leaves. The index was computed frommeasurements at two circular plots of 3 m radius in eachspatial replicate, thus eight in a grid. Height of theundergrowth was categorized as 0–50 cm, 51–100 cm,or >100 cm, while the percentage ground cover wasestimated visually. The index USI was computed foreach grid as:

USI ¼ 1=nXn

i¼1mi:hi

where n is the number of points in the grid, mi is themidpoint of the height category and hi is percentage ofundergrowth; (4) visibility in the understorey (VISIB): arod 1 m in height and marked at 10 cm intervals wasplaced at four locations in each 3 m circular plot in agrid. VISIB was estimated as the average number ofmarks that the observer could see from 2 m away; (5)percentage canopy cover was estimated as the mean ofcanopy cover measured using a standard spherical den-sitometer (Forestry Supplies) at two points in everyspatial replicate. All the above measurements were madeby the first author.

Data analysis

Habitat use and detection probability were modelled asa function of covariates in an occupancy framework(MacKenzie et al. 2002). We constructed detection his-tories for each grid based on whether pellet-groups weredetected in the four spatial replicates. For example, adetection history of ‘‘0110’’ means that the pellet-groupswere detected in second and third spatial replicates butnot in the first and fourth spatial replicates of the grid.These detection histories were incorporated in a cap-ture–recapture framework, using likelihood functions

(MacKenzie et al. 2002) to model detection probabilityof pellet-groups and consequently habitat-use by Indianchevrotain. We used a logistic model with logit link andbinomial error to model the effects of measured vari-ables on detection probability and habitat-use. Wemodelled detection probability as a function of VISIBand USI. We also modelled detection probability inde-pendent of any measured covariates to examine whetherany other factor influenced it (MacKenzie et al. 2006;Williams et al. 2002).

Based on our a priori expectation that the presence ofcover from predators (REFUGE, USI and VISIB) and

Fig. 1 Map of study area (Someshwara Wildlife Sanctuary). Thetwo parts of the sanctuary are separated by roads and denselypopulated human settlements. Although several grids were overlaidon the area of interest, only 204 grids were sampled, as shown in

the map. Right panel Schematic representation of each samplingunit (grid of dimension 50 m · 50 m), indicating four sub grids,spatial replicates and the locations at which some of the covariateswere measured

Fig. 2 Comparison of pellet groups of the Indian chevrotain (a) and Indian muntjac (b)

fruiting trees (FRUIT) would influence habitat-use, weconstructed eight candidate models. These models wereevaluated on AICC values, which correct the AkaikeInformation Criteria (AIC) value for small sample size(Burnham and Anderson 1998). Models were arrangedin ascending values of AICC and those with a difference(DAICC) of >10 from lowest AICc were rejected(Burnham and Anderson 1998). For covariates in thebest model, estimated parameters were averaged overmodels with DAICC <10. If the 95 % CI of the esti-mated parameter did not include zero, then we con-cluded that the covariate had a significant influence onhabitat use. We also used evidence ratio (w1/wj, where w1

is the estimate best model, and wj is another competingmodel) to evaluate the relative strength of the best modelcompared to another model (Burnham and Anderson1998). To determine whether detections in spatial repli-cates were independent we estimated the over-dispersionparameter c for our most complicated model (Mac-Kenzie and Bailey 2004). All modeling was performedusing the software PRESENCE (Hines 2006).

The home range of the lesser mouse-deer (Tragulusjavanicus), a smaller species closely related to the Indianchevrotain, varies from 3 to 6 ha (Matsubayashi et al.2003), while that of the larger Natal duiker (Cephalophusnatalensis) varies from 7 to 11 ha (Bowland and Perrin1995). The sampling unit of 0.25 ha used in this studywas very likely smaller than the home range of the In-dian chevrotain. Therefore, the parameter w estimatedusing software PRESENCE has to be interpreted asprobability of use rather than occupancy (MacKenzieet al. 2006).

Results

In all, we surveyed 204 grids. The naive occupancy (i.e.the proportion of grids in which pellet-groups werefound, without accounting for imperfect detection) was0.52. The overall estimated detection probability, p, was0.61. After accounting for detection probability theestimated parameter w or habitat-use was 0.73 ± 0.06(SE). Among the predictor variables, canopy cover wasnearly 100 % with very little variation (mean = 98.64%, SE ± 0.79) and was therefore dropped from furtheranalyses. USI influenced detection probability whileVISIB did not (Table 1). The constant p model (w(.),p(.)) performed poorly in comparison. Akaike weights,wi, of the models suggested good support for USIcompared to VISIB. Therefore, for modeling of habitat-use, detection probability was modelled as a function ofUSI.

Since most of the grids had one or no fruiting trees,we treated FRUIT as a binomial variable. We evaluatedeight a priori models, with the model incorporatingREFUGE and USI having the lowest AICC and there-fore the best model (Table 2). All the other models hadDAICC of ‡7.0. Akaike weight, wi, for this model was

close to 1 (0.94) and had an evidence ratio of 31.3 withthe next best model, indicating a high level of certaintyfor this model. Support for the model that includedfruiting trees was poor, with DAICC of 23.4 compared tothe best model. The estimated slope parameters forREFUGE and USI, weighted averaged over modelswith DAICC <10 were 0.46 ± 0.16 95 % CI and0.18 ± 0.09 95 % CI, respectively, clearly showing thatthese had non-zero positive slopes. Thus, the number ofrefuges and USI both increased the occupancy or habi-tat use by the Indian chevrotain. The over-dispersionparameter c was estimated at 0.69 for our global modelincorporating all measured covariates, indicating lessvariation in the data than expected from the model.

Discussion

Small-bodied ungulates are socially, morphologicallyand physiologically dissimilar to their larger counter-parts. Their small body size requires greater selectivitytowards easily digestible diet (Barrette 1987). Addition-ally, their mostly solitary behaviour forces them to re-main concealed in habitat features that provide cover

Table 1 Model selection results for covariates influencing detectionprobability, p, of pellet-groups of the Indian chevrotain

Model AICC DAICC wi K

w(.), p(USI) 771.84 0.00 0.99 3w(.), p(.) 784.16 12.32 0.00 3w(.), p(VISIB) 785.61 13.77 0.00 5

AICc Small sample Akaike information criterion, DAICc the dif-ference between the least AICc value and the AICc value of amodel, wi Akaike model weight, K number of model parameters,USI index of understorey structure, VISIB measure of understoreyvisibility, w estimated probability of use (habitat-use), p detectionprobability

Table 2 The habitat use parameter w and AICC values of eightcandidate models predicted to influence habitat use in the Indianchevrotain

Modela w SE AICC DAICC wi K

w(REFUGE + USI), p(USI) 0.72 0.02 749.89 0.00 0.94 5w(REFUGE), p(USI) 0.81 0.01 756.89 7.00 0.03 4w(REFUGE + VISIB), p(USI) 0.80 0.01 757.79 7.90 0.02 5w(USI), p(USI) 0.79 0.02 758.31 8.42 0.01 4w(VISIB), p(USI) 0.98 0.01 765.52 15.63 0.00 4w(.), p(USI) 0.93 0.09 771.84 21.95 0.00 3w(FRUIT), p(USI) 0.93 0.01 773.29 23.40 0.00 4w(.), p(.) 0.82 0.08 782.30 32.41 0.00 2

w Estimated probability of use (habitat-use), SE standard erroraround w, FRUIT presence or absence of food trees in the grid,REFUGE estimate of cover from predators, USI index of under-storey structure, VISIB measure of understorey visibility, w esti-mated probability of use (habitat-use), p detection probabilityaThe first model with the lowest AICC value was the selected as bestmodel

from predators (Geist 1998). Our study examined theinfluence of cover from predators and the presence offruiting trees on use of habitat by the Indian chevrotainusing the occupancy based approach.

Detection probability, p, for Indian chevrotain pellet-groups was moderate at 0.61 in our study, higher thandetecting individual Indian chevrotains in other studies[0.48 (Nag 2008) 0.29 (Ramesh et al. 2012)] using cameratraps. Since Nag (2008) did not model p as a function ofsampled covariates, estimates of p based on camera trapsmay have been lower. Nevertheless, our study demon-strates that pellet-groups are a good indicator of Indianchevrotain presence. A p of 0.61 is also within the rangeof 0.3–0.7 for the occupancy framework to be useful forsuch a study (MacKenzie et al. 2006), which may not bemet using camera traps (Ramesh et al. 2012). Modelingp as a function of USI significantly altered the occu-pancy parameter w. USI was also found to stronglyinfluence the detection probability of pellet-groups of aforest dwelling bovid, four-horned antelope Tetracerusquadricornis (Krishna et al. 2008).

By explicitly accounting for detection probability inour models, w increased by 40 %, from 0.52 to 0.73,underscoring the importance of incorporating detectionprobability in such studies. Our results largely matchestimates of occupancy of Indian chevrotain (0.45–0.79)estimated using camera traps (Nag 2008; Ramesh et al.2012). However, Ramesh et al. (2012) estimated occu-pancy at point locations instead of sampling areas orpatches, which may violate several assumptions of thecapture–recapture based occupancy approach outlinedin MacKenzie et al. (2006). The parameter w may havebeen over estimated since we used spatial replicates(Kendall and White 2009). Based on simulated data,spatial replicates inherently tended to overestimate wwhen compared to temporal replicates, but never ex-ceeded 5 % (Kendall and White 2009). Temporal repli-cates are suggested for longer study periods or whenlogistically feasible during short periods. In spite of thesmall size of the adjacently placed grids in our study,there was perhaps little spatial correlation in detectingsigns between and within grids as indicated by an over-dispersion parameter of less than one (MacKenzie et al.2006; Williams et al. 2002). A low value of over dis-persion parameter also indicates that detection historiesdid not vary much, perhaps due to the low number ofspatial replicates present in each grid.

The results show that availability of refuges to escapefrom predators and understorey complexity positivelyinfluenced the use of habitat by the Indian chevrotain.This is consistent with predictions of the anti-predatorybehaviour of small ungulates (Jarman 1974). They areknown to spring into thick cover when closely ap-proached, but otherwise remain concealed in buttresses,boulders, and snags (Eisenberg and Lockhart 1972;Geist 1998). Other small ungulates such as the lessermouse-deer (Tragulus javanicus) and the African chev-rotain (Hyemoschus aquaticus) are also known to useareas that provide cover (Dubost 2001; Matsubayashi

et al. 2003). The importance of understorey is perhapsnot limited to escape from predators. Small ungulates,presumably the Indian chevrotain also, feed on grasses,buds and shoots that might exist as small patches withinthe structurally more complex understorey (Bodmer1990). VISIB did not seem to influence habitat use,perhaps due to the consistently thinner understoreyduring the dry season and thus greater visibility. Nev-ertheless, two of the top three models contained visibilityas a covariate, indicating at least a weak effect. It ispossible that the Indian chevrotain prefers visible areasto better detect their predators. This was true of thelesser mouse deer in Malaysia where open areas wereused during the night yet remained in cover during theday (Matsubayashi et al. 2003). During the wet season,when visibility is generally low, the selection of refugesand USI might be weaker.

Contrary to our a priori expectation, we did not findevidence that the presence of fruiting trees influencedhabitat-use by the Indian chevrotain. The study period,from December to April, marks a trough in fruiting inthe Western Ghats (Sundarapandian et al. 2005). In-deed, <10 % of the sampled grids contained trees infruit, often only one tree, and we had to treat thiscovariate as a binomial variable. Fruit availability showsmarked seasonal fluctuation in various forest typesincluding rainforests (van Schaik and Wright 1993).During such periods of fruit scarcity, small ungulateslike the steenbok (Raphicerus campestris) and brocketdeer (Mazama gouzoubira, M. americana) include moreyoung leaves and shoots in their diet (Richard and Julia2001; Toit 1993). The Indian chevrotain might increasethe intake of buds and young shoots compared to otherseasons when their diet may be predominantly fruits.The high occupancy estimate of 0.73, compared to thepresence of fruits trees in only 10 % of the grids, indi-cates their relative independence from fruiting treesduring the dry season. It is likely that the significantinfluence of USI is more related to the presence of buds,shoots and young leaves in the dry season, than perhapsto its use as a cover from predator. However, a betterunderstanding of the feeding ecology of the species isnecessary to inform us about the relative importance offruits, buds and young leaves in determining habitat use.

The occupancy framework using signs such as dung isperhaps the only available method to examine habitatuse and occupancy status of cryptic species such as theIndian chevrotain. However, the decay rate of dung canhave major influence on the estimated occupancyparameter (Rhodes et al. 2011). A high decay rate cancause false-absences and a very low decay rate can causefalse-presences. The influence of the former on occu-pancy parameter is likely to be low if the decay is notvery rapid, even when dung accumulation rate is low(Rhodes et al. 2011). In contrast, if the dung decay rateis too low during which occupancy changes, then it canlead to major over estimation of the occupancy param-eter. Although there is no data, we believe that theformer was the case since the study was carried out

during the dry period when dung decay rates are notvery rapid (Rhodes et al. 2011) and are consistent. Thesmall dung piles of the chevrotain are also unlikely tohave very long decay period. This is, however, an issuethat needs to be addressed in future.

Conservation implications

Although the Indian chevrotain has been categorized asof ‘Least Concern’, it suffers moderate to heavy huntingpressure in many parts of its range (Madhusudan andKaranth 2002; Kumara and Singh 2004). This studyshows that, in addition to controlling hunting, it is alsonecessary to retain habitat features that provide the In-dian chevrotain with refuges from predators (such asbuttress trees and fallen logs), and USI that providecover from predators as well as potential feeding sites.Many parts of the Indian chevrotain’s habitat sufferfrom livestock grazing, and extraction of small timber,fallen logs, leaf litter and non-timber forest produce suchas cane (Daniel 1991). A better control of these activitiesis thus obvious. Although fruiting trees may not play animportant role during the dry season, they may be cru-cial during other times.

Acknowledgements This study was funded by Department of Sci-ence and Technology, Government of India. We thank the Kar-nataka Forest Department for research permits and logisticalsupport. Discussions with Ullas Karanth, Samba Kumar, ArjunGopalaswamy and Devcharan Jathanna helped conceptualise thestudy. We are grateful to Srinivasa for his invaluable help in thefield. Meghna Krishnadas supported all stages of the study. Twoanonymous reviewers provided very useful comments.

References

Baillie JEM, Hilton-Taylor C, Stuart S (2004) 2004 IUCN Red Listof threatened species: a global assessment. IUCN, Switzerland

Barrette C (1987) The comparative behavior and ecology of chev-rotains, musk deer, and morphologically conservative deer. In:Wemmer CM (ed) Biology and management of the Cervidae.Smithsonian Institution Press, Washington, DC, pp 200–213

Bodmer RE (1990) Ungulate frugivores and the browser-grazercontinuum. Oikos 57:319–325

Bowland AE, Perrin MR (1995) Temporal and spatial patterns inblue duikers Philatomba monticola and red duikers Cephalophusnatalensis. J Zool 237:487–498

Branan WV, Marchinton RL (1985) Food habits of brocket andwhite-tailed deer in Suriname. J Wild Manage 49:972–976

Burnham KP, Anderson DR (1998) Model selection and multi-model inference. Springer, New York

Corlett RT (2007) The impact of hunting on the mammalian faunaof tropical Asian forests. Biotropica 39:292–303

Daniel JC (1991) Ungulate conservation in India—problems andprospects. Appl Anim Behav Sci 29:349–356

Demment MW, van Soest PJ (1985) A nutritional explanation forbody-size patterns of ruminant and non-ruminant herbivores.Am Nat 125:641–672

Dubost G (1984) Comparison of the diets of frugivorous forestruminants of Gabon. J Mammal 65:298–316

Dubost G (2001) Chevrotains. In: MacDonald D, Norris S (eds)The new encyclopedia of mammals. Oxford University Press,Oxford, pp 500–501

Duckworth JW, Hemsagar Baral, Timmins RJ, (2008) Moschiola in-dica. IUCN red list of threatened species. http://www.iucnredlist.org/apps/redlist/details/136585/0

Eisenberg JF, Lockhart M (1972) An ecological reconnaissance ofWilpattu National Park, Ceylon. Smithsonian Contrib Zool101:1–118

Gagnon M, Chew AE (2000) Dietary preferences in extant AfricanBovidae. J Mammal 81:490–511

Gautier-Hion A, Emmons LH, Dubost G (1980) A comparison ofthe diets of three major groups of primary consumers of Gabonprimates, squirrels and ruminants. Oecologia 45:182–189

Geist V (1998) Deer of the world: their evolution, behaviour, andecology. Stackpole Books, Mechanicsburg

Gopalaswamy AM, Karanth KU, Kumar NS, Macdonald DW(2012) Estimating tropical forest ungulate densities from signsurveys using abundance models of occupancy. Anim Conserv15:669–679. doi:10.1111/j.1469-1795.2012.00565.x

Groves CP, Meijaard E (2005) Interspecific variation in Moschiola,the Indian chevrotain. Raffles Bull Zool 12:413–421

Hayward MW, Henschel P, O’Brien J, Hofmeyr M, Balme G,Kerley GIH (2006) Prey preferences of the leopard (Pantherapardus). J Zool 270:298–313

Hines JE (2006) PRESENCE2—Software to estimate patch occu-pancy and related parameters. USGS-PWRC. http://www.mbr-pwrc.usgs.gov/software/presence.html

Hofmann RR (1989) Evolutionary steps of ecophysiologicaladaptation and diversification of ruminants: a comparativeview of their digestive system. Oecologia 78:443–457

Jarman PJ (1974) The social organisation of antelope in relation totheir ecology. Behaviour 48:215–267

Kendall WL, White GC (2009) A cautionary note on substitutingspatial subunits for repeated temporal sampling in studies ofsite occupancy. J Appl Ecol 46:1182–1188

Krishna YC, Krishnaswamy J, Kumar NS (2008) Habitat factorsaffecting site occupancy and relative abundance of four-hornedantelope. J Zool 276:63–70

Kumara HN, Singh M (2004) The influence of differing huntingpractices on the relative abundance of mammals in two rain-forest areas of the Western Ghats, India. Oryx 38:321–327

Macdonald DW (2001) The new encyclopedia of mammals. OxfordUniversity Press, Oxford

MacKenzie DI, Bailey LL (2004) Assessing the fit of site occupancymodels. J Agric Biol Environ Stat 9:300–318

MacKenzie DI, Nichols JD, Lachman GB, Droege S, Royle AJ,Langtimm CA (2002) Estimating site occupancy rates whendetection probabilities are less than one. Ecology 83:2248–2255

MacKenzie DI, Nichols JD, Royle JA, Pollock KH, Bailey LA,Hines JE (2006) Occupancy modeling and estimation. Elsevier,San Diego

Madhusudan MD, Karanth KU (2002) Local hunting and theconservation of large mammals in India. Ambio 31:49–54

Matsubayashi H, Bosi E,Kohshima S (2003)Activity and habitat useof lesser mouse-deer (Tragulus javanicus). J Mammal 84:234–242

Nag K (2008) Assessing animal abundance from photographiccapture data using an occupancy approach. MSc thesis, Man-ipal University

Pascal JP (1988) Wet evergreen forests of the Western Ghats ofIndia: ecology, structure, floristic composition and succession.French Institute of Pondicherry, Pondicherry

Prasad S, Sukumar R (2010) Context-dependency of a complexfruit-frugivore mutualism: temporal variation in crop size andneighborhood effects. Oikos 119:514–523

Prins HH, de Boer WF, van Oeveren H, Correia A, Mafuca J, OlffH (2006) Co-existence and niche segregation of three smallbovid species in southern Mozambique. Afr J Ecol 44:186–198

Ramakrishnan U, Coss RG, Pelkey N (1999) Leopard diets insouthern India: management implications for tiger conserva-tion. Biol Conserv 89:113–120

Ramesh T, Kalle R, Sankar K, Qureshi Q (2012) Dry season fac-tors determining habitat use and distribution of mouse deer(Moschiola indica) in the Western Ghats. Eur J Wildl Res. doi:10.1007/s10344-012-0676-5

Rhodes JR, Lunney D, Moon C, Mattews A, McAlpine CA (2011)The consequences of using indirect signs that decay to deter-mine species’ occupancy. Ecography 34:141–150

Richard E, Julia JP (2001) Dieta de Mazama gouazoubira (Mam-malia, Cervidae) en un ambiente secundario de Yungas,Argentina. Iheringia Ser Zool 90:147–156

Sidhu S, Mudappa D, Raman TRS (2011) People and predators:leopard diet and interactions with people in a tea plantationdominated landscape in the Anamalai Hills, Western Ghats.NCF Technical Report #18, Nature Conservation Foundation,Mysore

Sodhi NS, Koh LP, Brook BW, Ng PKL (2004) Southeast Asianbiodiversity: an impending disaster. Trends Ecol Evol19:654–660

Sundarapandian SM, Chandrasekaran S, Swamy PS (2005) Phe-nological behaviour of selected tree species in tropical forests atKodayar in the Western Ghats, Tamil Nadu, India. Curr Sci88:805–810

Toit JT (1993) The feeding ecology of a very small ruminant, thesteenbok (Raphicerus campestris). Afr J Ecol 31:35–48

van Schaik Terborgh J, Wright JS (1993) The phenology of tropicalforests: adaptive significance and consequences for primaryconsumers. Annu Rev Ecol Syst 24:353–377

Williams BK, Nichols JD, Conroy MJ (2002) Analysis and man-agement of animal populations. Academic, New York

Wilson DE, Reeder DM (2005) Mammal species of the World, 3rdedn. Johns Hopkins University Press, Baltimore