Introduction
Studies have shown that the density of carnivores is related to the availability of prey biomass (Fuller & Sievert, Reference Fuller, Sievert, Gittleman, Funk, Macdonald and Wayne2001; Carbone & Gittleman, Reference Carbone and Gittleman2002; Karanth et al., Reference Karanth, Nichols, Kumar, Link and Hines2004; Khorozyan et al., Reference Khorozyan, Malkhasyan and Abramov2008; Carbone et al., Reference Carbone, Pettorelli and Stephens2010). These analyses are based largely on studies of prey and predator species in natural or semi-natural ecosystems (Fuller & Sievert, Reference Fuller, Sievert, Gittleman, Funk, Macdonald and Wayne2001; Carbone & Gittleman, Reference Carbone and Gittleman2002; Andheria et al., Reference Andheria, Karanth and Kumar2007; Karanth & Nichols, Reference Karanth, Nichols, Tilson and Nyhus2010). However, more recent studies have found that large carnivores can also persist in human-dominated areas (Yirga et al., Reference Yirga, De Iongh, Leirs, Gebrihiwot, Deckers and Bauer2012; Athreya et al., Reference Athreya, Odden, Linnell, Krishnaswamy and Karanth2013) by relying fully or partially on food resources associated with humans (Gehrt et al., Reference Gehrt, Riley and Cypher2010). The potential carrying capacity of human-dominated landscapes for large carnivores must therefore be investigated in terms of the abundance and availability of domestic prey species as well as wild prey (Boitani & Powell, Reference Boitani and Powell2012).
The biomass of domestic animals in human-use landscapes can be higher than that of wild prey, as seen in Brazil, Nepal and Kenya (Schaller, Reference Schaller1983; Seidensticker et al., Reference Seidensticker, Sunquist, McDougal, Daniel and Serrao1990; Mizutani, Reference Mizutani1999). Anthropogenic food sources such as garbage and pet food can also contribute to the diet of wild carnivores (Gehrt et al., Reference Gehrt, Riley and Cypher2010). These food resources can be abundant, leading to densities of wild carnivores comparable to, or even greater than, their densities in the wild. For example, densities of urban red foxes Vulpes vulpes were 15 times higher (Bino et al., Reference Bino, Dolev, Yosha, Guter, King, Saltz and Kark2010), and of black bear Ursus americanus three times higher (Beckmann & Berger, Reference Beckmann and Berger2003), in semi-urban areas than in natural habitats because of better foraging opportunities from crops, garbage, livestock and artificial feeding.
In India several carnivore species, such as wolves Canis lupus (Jhala & Giles, Reference Jhala and Giles1991), Asiatic lions Panthera leo persica (Vijayan & Pati, Reference Vijayan and Pati2002; Meena et al., Reference Meena, Jhala, Chellam and Pathak2011) and striped hyaenas Hyaena hyaena (Shilpi et al., Reference Shilpi, Krishnendu, Shankar and Qamar2009; Singh et al., Reference Singh, Gopalaswamy and Karanth2010), also occur in human-dominated landscapes and feed on livestock. Tigers Panthera tigris are also known to attack livestock in and around protected areas (Sekhar, Reference Sekhar1998; Karanth & Gopal, Reference Karanth, Gopal, Woodroffe, Thirgood and Rabinowitz2005; Woodroffe et al., Reference Woodroffe, Thirgood, Rabinowitz, Woodroffe, Thirgood and Rabinowitz2005). Leopards Panthera pardus are adaptable, using a variety of habitats and feeding on a range of wild and domestic prey (Seidensticker et al., Reference Seidensticker, Sunquist, McDougal, Daniel and Serrao1990; Daniel, Reference Daniel2009; Hunter, Reference Hunter2011), which enables them to live close to human settlements (Nasik Gazetteer, 1883; Daniel, Reference Daniel2009). A high density of domestic animals (Thornton et al., Reference Thornton, Kruska, Henninger, Kristjanson, Reid and Atieno2002; FAO, 2005) could therefore constitute a stable and abundant prey base, facilitating the persistence of leopard populations in human-dominated landscapes far from protected conservation areas (Athreya et al., Reference Athreya, Odden, Linnell, Krishnaswamy and Karanth2013).
Although widespread occurrence of leopards across India has been documented (Vijayan & Pati, Reference Vijayan and Pati2002; Daniel, Reference Daniel2009; Athreya et al., Reference Athreya, Odden, Linnell, Krishnaswamy and Karanth2013), few studies have assessed the leopard's diet and availability of prey in human-dominated landscapes (Punjabi et al., Reference Punjabi, Athreya and Linnell2012). To improve our understanding of resource usage by leopards in a rural, human-dominated landscape, we analysed the diet of leopards, and estimated prey densities and biomass, in an agricultural landscape in western Maharashtra.
Study area
The study was conducted in an irrigated valley (238 km2), dominated by crop lands, around the town of Akole (human population c. 20,000) in the Ahmednagar district of western Maharashtra, India (Fig. 1). The mean population density reported for Ahmednagar district in 2011 was 266 people per km2 (Census of India, 2011). Approximately 80% of the human population in the district is rural, with farming of millet, sugar cane and vegetables being the major sources of livelihood. Annual rainfall is 1,000–2,000 mm. The nearest protected area is the Kalsubai Harishchandragarh Wildlife Sanctuary (299 km2), 18 km from the western edge of the study area. There are no patches of natural forest within the study area.
The Ahmednagar district of western Maharashtra, India, where leopard Panthera pardus scats were collected during December 2007–April 2009 in a human-dominated agricultural landscape. The rectangle on the inset indicates the location of the main figure in India.
Camera-trapping surveys have recorded leopard, striped hyaena, golden jackal Canis aureus, Bengal fox Vulpes bengalensis, jungle cat Felis chaus and rusty-spotted cat Prionailurus rubiginosus (Athreya et al., Reference Athreya, Odden, Linnell, Krishnaswamy and Karanth2013) in the area. No wild ungulate species were recorded in this study, nor have any been reported by the Forest Department. There are various occupational groups in the area, the dominant one being settled farmers who own land and livestock. Pastoral, migratory shepherds arrive annually in the dry season, with herds of sheep and goats (each herder has >100 animals), to pasture on crop-residues in the fields.
Methods
Government records indicate an overall livestock density of 176 head of livestock per km2 across the district (Livestock Census, 2003). This includes domestic and feral/semi-feral dogs Canis familiaris, pigs Sus scrofa and cats, which may also constitute part of the leopard's diet. To get a more detailed overview of the prey available in the study area we interviewed a random sample of 77 households of resident farmers to assess the number and species of domestic animals owned. The density of houses was obtained by digitally mapping all residential houses, using 2007 imagery from Google Earth v. 6.1.0.5001 (Google, Mountain View, USA), and then ground-truthing a sample of 200 homesteads to obtain the percentage of houses (as distinct from shops, schools and other buildings). The interview data and estimated density of houses were used to estimate densities of domestic animals in the study area. Although the urban area of Akole has a large number of pigs, they are restricted to the town area and are not present in the wider crop-land landscape. The Wadhari people, who claim ownership of the feral pigs, state that there are at least several hundred pigs in the town of Akole.
Initial scat surveys in the region indicated that leopards used trails such as foot paths, edges of fields, paved roads and dry stream beds. These trails were identified using Google Earth and surveyed using three methods: (1) 130 km of trails that were marked across the entire study area were walked a total of three times each during December 2007–April 2008, with a 3-week interval between each sampling session, (2) 85 randomly selected 1-km2 grid cells were overlaid on a Google Earth map and a mean of 2.37 ± SD 0.86 km of road/path within each cell was searched, on foot, for scats, and (3) scats were collected opportunistically during December 2007–April 2009. We used all three methods because we lacked a priori information on where to locate leopard scats in a human-dominated landscape. The scats represent a sample from the dry season.
Two trained surveyors walked on either side of the trail and collected all scats judged to be of carnivore origin, based on size, shape and ancillary evidence such as scrape signs and tracks. The scats were measured and stored in zip-lock bags, and the geographical coordinates of each location were recorded using a geographical positioning system. The scats were later transferred to polythene bags and part of each scat was transferred into vials for storage in absolute alcohol for subsequent DNA analysis to identify leopard scats (Mondol et al., Reference Mondol, Ramesh, Athreya, Sunagar, Selvaraj and Ramakrishnan2009, Reference Mondol, Thatte, Yadav and Ramakrishnan2011; Navya et al., unpubl. data). Visual identification of scats is not always reliable (Farrell et al., Reference Farrell, Roman and Sunquist2000), and therefore diet analysis was conducted only on scats confirmed to be of leopard origin, either using DNA methods or having been collected from scrapes characteristic of those made by large felids (as the leopard is the only large felid present in the study area).
The scat analysis was carried out as described by Mukherjee et al. (Reference Mukherjee, Goyal and Chellam1994), Mukherjee & Mishra (Reference Mukherjee and Mishra2001) and in other diet studies of large felids (Karanth & Sunquist, Reference Karanth and Sunquist1995; Andheria et al., Reference Andheria, Karanth and Kumar2007; Khorozyan et al., Reference Khorozyan, Malkhasyan and Abramov2008; Odden & Wegge, Reference Odden and Wegge2009). The scats were washed and dried and the prey species were identified from the presence of claws, hoofs or hair. For hair, we used a microscope to identify the origin of 25 randomly selected hair samples per scat. Prey species were identified based on comparison with reference slides of hair samples from domestic animals in the study area and from reference slides of hair samples from wild prey, from collections at the Centre for Wildlife Studies, Bangalore, and the Bombay Natural History Society, Mumbai. Scats that were highly degraded or had too few identifiable prey remains were not used in the analysis.
We calculated the frequencies of occurrence of the various prey species (the percentage of the total number of scats that contained a specific prey item). However, this variable can be misleading because smaller prey species contribute more to a scat than larger species (Karanth & Sunquist, Reference Karanth and Sunquist1995; Klare et al., Reference Klare, Kamler and Macdonald2011). Based on feeding trials on captive mountain lions Puma concolor, Ackerman et al. (Reference Ackerman, Lindzey and Hemker1984) derived a regression equation to calculate the relative biomass of different prey species consumed, based on their relative proportions in scats. Mountain lions are similar in size to leopards and this method has been used previously for leopards (Karanth & Sunquist, Reference Karanth and Sunquist1995; Andheria et al., Reference Andheria, Karanth and Kumar2007; Khorozyan et al., Reference Khorozyan, Malkhasyan and Abramov2008; Odden & Wegge, Reference Odden and Wegge2009; Wegge et al., Reference Wegge, Odden, Pokharel and Storaas2009).
The regression equation is in the form:
$$Y = 1.98+0.35x$$
where Y is the mass of prey consumed per scat and x is the mean mass of the prey. The relative biomass (D) and the relative numbers of each prey species consumed (E) were obtained using the equations
$$D = (A \times Y)/\Sigma (A \times Y) \times 100$$
$$E = (D/x)/\Sigma (D/x) \times 100$$
where A is the frequency of occurrence of the prey item in the scats. The prey biomass (B prey = D prey · W prey; Khorozyan et al., Reference Khorozyan, Malkhasyan and Abramov2008) in the study area was calculated using density estimates (D prey) for the four most common prey species (domestic goats, dogs, calves and cats), based on interview data. The mean weight of the domestic animal species (W prey) was estimated by a livestock veterinarian working in the region. Adult cattle were not included because compensation records indicate that only one cow was attacked in the study area in 3 years, whereas calves were preyed on in relatively higher numbers (Forest Department records).
We used Ivlev's (Reference Ivlev1961) electivity index to assess prey selection:
$${E} =\displaystyle{{{{r}_i} - {{p}_i}}\over{{r}_i} + {{p}_i}}$$
where r i is the relative proportion of prey item i in the diet and p i is the relative proportion of prey item i in the environment. E is in the range −1–+1, where negative values indicate that the prey item is avoided or inaccessible and positive values indicate that it is selected for.
Results
Leopard scats, confirmed using DNA analysis, were found throughout the sampled area, including close to houses, and on a variety of trail types, including dirt and paved roads. The mean distance from leopard scats to the nearest house was 213 m (range 10–850 m), and 140 m (range 0–815 m) from roads. Two scats were found on the main street of Akole town.
Of the 265 scats collected 80 were confirmed as leopard scats based on DNA analysis and 43 were identified based on the presence of tracks or scrapes. Of these 123 leopard scats 85 had usable remains for diet analysis. Thirty-eight scats could not be used because the remains were degraded and, although they contained hair and bone, they could not be identified visually or under the microscope. Of 110 intact leopard scats the mean diameter at the thickest section was 25.15 ± SD 5.2 mm (range 11.5–38.3 mm). The scats contained a total of 131 prey items, comprising 11 prey species (Table 1). Fifty-six percent of the scats contained one prey species, 21% two species, 8% three species, 3% four species and only one scat contained five species. Domestic animals (pig, sheep, cat, dog, goat, cow) constituted 87% of the prey biomass consumed by leopards. The wild prey present in the scats were civet Viverricula indica, rodent, primate, bird and mongoose Herpestes spp. (Table 1). In the case of rodents a mean mass of 500 g (Table 1) was considered because of the presence of the bandicoot rat Bandicota spp. in the crop lands. Dogs were the most significant constituent of the leopard's diet, accounting for 39% of the biomass consumed. Ivlev's index indicates that despite the higher biomass of goats and calves available, dogs and cats were preyed upon to a greater extent than expected, which could be attributable to preference or greater accessibility (Fig. 2; Table 2).
Ivlev's index (Ivlev, Reference Ivlev1961) for goat Capra aegagrus, calf Bos taurus, cat Felis catus and dog Canis lupus familiaris, based on scat analysis of leopard prey in Ahmednagar district of western Maharashtra. The index is based on the frequency of prey species in scats relative to the availability of these species in the area. Species with positive index values are more selected for/more available than species with negative values.
Prey species identified in the diet of leopards Panthera pardus in the Ahmednagar district of western Maharashtra, India (Fig. 1), from analysis of 85 scats collected from a human-dominated landscape during December 2007–April 2009. Y is the correction factor from Ackerman et al. (Reference Ackerman, Lindzey and Hemker1984).
Density (km−2) of domestic animals in the study area, from interviews of a random sample of households (n = 77) in the town of Akole in the Ahmednagar district of western Maharashtra (Fig. 1). The interviews were conducted during September 2007–September 2009.
Discussion
The relationship between large felids and humans is complex and the spectrum of interactions ranges from fascination to fear (Boomgaard, Reference Boomgaard2001; Loveridge et al., Reference Loveridge, Wang, Frank, Seidensticker, Macdonald and Loveridge2010). Large felids are often portrayed as flagship species for conservation (Treves & Karanth, Reference Treves and Karanth2003) but conflict occurs at the local level, where the presence of a large carnivore can result in damage to property and loss of human life (Treves et al., Reference Treves, Wallace, Naughton-Treves and Morales2006). Retaliatory killings are a significant cause of mortality of large felids, and studies have focused on human–felid conflict (Inskip & Zimmermann, Reference Inskip and Zimmermann2009). Although depredation of livestock is a widespread occurrence (Linnell et al., Reference Linnell, Odden, Mertens, Boitani and Powell2012), domestic livestock usually constitute only a small part of the diet of large felids and complete dependency on domestic species has rarely been observed. The biomass of livestock can be high in human-dominated areas, exceeding that of wild prey in surrounding forest areas in Brazil, Nepal and Kenya (Schaller, Reference Schaller1983; Seidensticker et al., Reference Seidensticker, Sunquist, McDougal, Daniel and Serrao1990; Mizutani, Reference Mizutani1999). Domestic animals are easier to attack because they lack anti-predatory behaviour, unlike their wild counterparts (Diamond, Reference Diamond2002). However, availability does not always indicate accessibility as livestock may be guarded by day and enclosed in predator-proof enclosures at night.
Our results show that a large predator such as a leopard can attain relatively high densities in a rural landscape (Athreya et al., Reference Athreya, Odden, Linnell, Krishnaswamy and Karanth2013) while subsisting almost entirely on a diet of domestic animals. Despite the density of goats being seven times that of domestic dogs, goats constituted only 11% of the prey biomass of leopards, whereas dogs constituted 39%. This is probably because goats are less accessible than dogs, being actively herded by day and enclosed in sheds at night. The results show that dogs are an important food resource for leopards and they occur at high density in the study area. The proclivity of leopards towards killing and eating dogs has been noted in anecdotal, historical literature (Daniel, Reference Daniel2009). Two studies carried out within protected areas in the states of Maharashtra and Jammu & Kashmir have reported the importance of dogs as prey for leopards (Edgaonkar & Chellam, Reference Edgaonkar and Chellam2002; Shah et al., Reference Shah, Jan, Bhat, Ahmad and Ahmad2009).
Domestic dogs are ubiquitous in the Indian landscape and density estimates from an adjoining human-use landscape in Maharashtra range from 23 (farmland) to 113 km−2 (village area; Punjabi et al., Reference Punjabi, Athreya and Linnell2012; Hughes & Macdonald, Reference Hughes and Macdonald2013). Domestic cats also appear to be an important component of the leopard's diet in our study area, contributing 12% of the biomass consumed. Based on interviews, the densities of goats and cows in the study area were 174 and 162 km−2, respectively. Thus, rural landscapes in India can be prey-rich areas for wild carnivores because of the importance of animal husbandry in the livelihoods of rural people.
In natural ecosystems predator density is correlated with prey biomass (Carbone & Gittleman, Reference Carbone and Gittleman2002; Karanth et al., Reference Karanth, Nichols, Kumar, Link and Hines2004; Marker & Dickman, Reference Marker and Dickman2005; Beckmann & Lackey, Reference Beckmann and Lackey2008; Khorozyan et al., Reference Khorozyan, Malkhasyan and Abramov2008; Bino et al., Reference Bino, Dolev, Yosha, Guter, King, Saltz and Kark2010). If we consider the density of the four common prey species (goats, calves, cats and dogs) identified in the scat analysis, the potential prey biomass for leopards in the study area is 733,000 kg per 100 km2 (Table 2). The regression equation of Carbone & Gittleman (Reference Carbone and Gittleman2002) estimates that 10,000 kg of prey is required per 90 kg of predator, irrespective of predator size. Based on this equation the total amount of prey biomass (including goats and cattle) in our study area could, in theory, support >10 times the number of leopards that are present. The biomass of the owned dogs and cats alone constitutes c. 54,000 kg per 100 km2, which would be predicted to support a 45 kg predator at a density of 10.8 individuals per 100 km2, which is close to the combined density of leopards and striped hyaenas in the study area, based on estimates from a camera-trap study (Athreya et al., Reference Athreya, Odden, Linnell, Krishnaswamy and Karanth2013). The reason for the relatively low density of leopards despite the high biomass of prey is probably related to the low availability of domestic stock because they are protected by farmers (Athreya et al., unpubl. data) and the fact that large stock are not predated by leopards. Most cattle in the study area are hybrid varieties that are larger than the indigenous breeds.
The selection of domestic dogs and cats as prey means that the economic impact of predation by the leopard on valuable livestock is lower than expected. Thus human–leopard conflict is largely driven by people's fear of leopards foraging in the proximity of their houses, and the sentimental value of dogs as pets.
Our findings are similar to those of studies on rural and urban carnivores in western Europe and North America, where wild carnivores reside in modified human-dominated landscapes and are totally dependent on anthropogenic sources of food (Gehrt et al., Reference Gehrt, Riley and Cypher2010). Our work contributes to a growing awareness of the potential conservation value of private lands and non-protected areas in the tropics (Negrões et al., Reference Negrões, Revilla, Fonseca, Soares, Jácomo and Silveira2011).
Acknowledgements
We thank Ashok Ghule, Abhijit Kulkarni, Avinash Kulkarni and Kiran Rahalkar for taking part in collection of scats, Uma Ramakrishnan, R. Navya and Chandrima Home for carrying out the DNA analysis and standardization of the scat DNA procedures, the Maharashtra Forest Department for their support, and the Centre for Wildlife Studies and the Asian Nature Conservation Foundation, Bangalore, for institutional support.
Biographical sketches
Vidya Athreya's research has focused on understanding the ecology of large carnivores when they reside in human-use landscapes, and translating her findings into informed management actions. Morten Odden has worked on large carnivores in Nepal and on human–leopard interactions in India. He is interested in predator–prey interactions and in space use of large carnivores. John Linnell conducts multi-disciplinary research on the relationships between wildlife and humans, with a focus on large carnivores and herbivores. He works in Europe, South America, India and South-east Asia. Jagdish Krishnaswamy is interested in ecohydrology, landscape ecology and applications of Bayesian statistics in ecology and environmental science. Ullas Karanth studies the ecology of large carnivores, focusing on population modelling and estimation, and applies the results to practical conservation efforts.
