Tiger Predation

[Acinonyx sp.]

Prey preferences of the tiger Panthera tigris


Abstract 
Tigers Panthera tigris continue to decline despite the best efforts of the worldwide scientific and conservation communities. Prey depletion has been linked to this decline, but a clear definition of what constitutes preferred prey and preferred prey weight range does not exist. This is critical information if we are to assess tiger reintroduction potential, monitor unforeseen poaching of predators and prey, and successfully conserve the species. Here we reviewed the available literature on tiger diet and prey availability and calculated Jacobs’s electivity index scores from 3187 kills or scats of 32 prey species. We found that wild boar and sambar deer are significantly preferred by tigers, with red deer and barasingha likely to be significantly preferred also with a larger sample size. Prey body mass was the only variable that related to tiger prey preference with species weighing between 60 and 250 kg preferred by tigers yielding a ratio of predator to preferred prey of 1:1, which is similar to other solitary felids. This information can be used to predict tiger diet, carrying capacity and movement patterns, as it has been for Africa’s large predator guild, and has important implications for tiger conservation throughout its distribution.

Introduction
Tigers Panthera tigris are the largest of the felids with a distribution that once spanned Asia and much of the diverse array of habitats therein (Nowell & Jackson, 1996). Today, tiger numbers have plummeted and its range has vastly contracted (Fig. 1) such that the species is listed as endangered (IUCN, 2009) despite long-term conservation actions, such as the 40-year-old Project Tiger initiative in India (Panwar, 1987) and intensive research. The causes of this decline have been illegal hunting, often for tiger bone, a declining prey base, and habitat loss and degradation (Karanth & Stith, 1999; Carroll & Miquelle, 2006; Chapron et al., 2008; Wilson & Mittermeier, 2009; Miquelle et al., 2010). In light of the importance that prey has for tiger conservation (Karanth & Stith, 1999; Miquelle et al., 2010), it is critical to understand tiger predation to ensure an adequate prey base. There are numerous excellent studies that have documented the local diet of tigers (see examples in Tables 1 and 2) and their local preferences (Støen & Wegge, 1996; Miquelle et al., 2010); however, these are invariably affected by site-specific factors such as variable prey abundance and habitat features that do not afford conclusions to be drawn on the species as a whole compared with a broader meta-analysis. These studies provide data to holistically review the diet and prey selection of the tiger – information that can ultimately be applied to predicting diet, carrying capacity and home range at unstudied sites, as has been done with Africa’s large carnivore guild (Hayward, O’Brien & Kerley, 2007b; Hayward et al., 2007a, 2009). Tigers are solitary, ambush hunters that also actively search for prey (Wilson & Mittermeier, 2009) and require >5 kg of meat daily to maintain body condition (Sunquist, 1981). They are capable of capturing and killing large prey, including adult male gaur Bos gaur (Karanth & Sunquist, 1995), but also take prey as small as hares (Johnsingh, 1983). Tiger diet is biogeographically diverse (Miquelle et al., 1996, 1999; Sunquist, Karanth & Sunquist, 1999). Based on the breadth of their diet across their range, one might infer that tigers are non-selective predators; however, it seems likely that the species’ morphology and solitary hunting strategy imposes limitations on the prey it can capture most efficiently with minimal risk. Solitary leopards P. pardus, cheetahs Acinonyx jubatus and Eurasian lynx Lynx lynx preferentially prey on species at or just below their own body mass (Jedrzejewski et al., 1993; Hayward et al., 2006a,b), whereas group hunters, like lions P. leo and African wild dogs Lycaon pictus preferentially prey on species much larger than themselves (Hayward & Kerley, 2005; Hayward et al., 2006c). Sunquist (2010) has suggested that the size of prey preferred by tigers is approximately their own size, as has been found with other solitary carnivores. Consequently, we aimed to determine the preferred prey of tigers from throughout the species’ range and then test whether these preferred prey species conformed to the prediction that solitary hunters preferentially prey on species of similar body mass to themselves.

Methods 
We determined tiger prey preferences following the methods used to determine the prey preferences of Africa’s large predator guild (Hayward et al., 2006b,c). Specifically, we reviewed the literature on tiger ecology and extracted data on tiger diet and the abundance (actual or relative) of prey at each site. This literature survey revealed 24 published studies from 18 different conservation areas in 7 countries describing the diet of the tiger, which included some measure of prey abundance [either actual or relative (Table 1; Fig. 1)]. Several of these studies were conducted over a long term and these allowed temporally separated prey preferences to be calculated as prey abundance changed over time (Table 1). Such partitioning has been used previously in the study of carnivore ecology (see Creel & Creel, 2002). Therefore, a total of 28 assessments of prey preference were calculated from sites throughout the extant range of the tiger. We do not consider autocorrelation to be problematic by using data from the same area at different levels of prey abundance, as one of the fundamental rules of whether an individual is captured and killed is the probability of the predator encountering it, and this varies with prey density (Hayward & Kerley, 2005). The dietary data collected in these studies were largely derived from incidental observations of kills, although faecal analysis was also used (Table 1). Continuous observations are widely regarded as the superior method of ascertaining the diet of a large predator (Bertram, 1979; Mills, 1992); however, these are extremely difficult with such secretive and elusive predators as the tiger. Global positioning system telemetry locations are probably the best source for determining the diet of solitary carnivores (e.g. Swanepoel, Dalerum & Van Hoven, 2010; Tambling et al., 2010). While other methods may exhibit biases, we are not sure to what extent this may affect these results; however, the tested predictive accuracy of these methods suggest their effect is minor (Hayward et al., 2007a,b). Faecal analysis is another valuable method used in ascertaining predator diets, although if used alone may overemphasize the importance of small prey items (Mills, 1992).

Table 1 Sites and sources of prey preference data used in this study

*This image is copyright of its original author

*This image is copyright of its original author

Table 2 Preference status, mean Jacobs’s index value of each tiger prey species, number of studies recording it as potential prey (np) and actual prey (na), mean percentage abundance and kills of each prey species, body mass (three-fourths adult female), and categories of herd size and main habitat based on Nowak (1999) and references in Hayward et al. (2006a)

*This image is copyright of its original author

*This image is copyright of its original author

The inclusion of studies using all these methods ensured that the majority of prey species of the tiger were assessed in our analysis. Numerous studies provided excellent descriptive information on tiger diet but insufficient or no information on prey abundance (Sankar & Johnsingh, 2002; Talukdar, 2002; Avinandan, 2003), whereas others had no kill data (Reza, Feeroz & Islam, 2002; Shrestha, 2004; Adhikari & Khadka, 2009; Siddiqui, 2010) or were collected over such a long period that great fluctuations in prey abundance occurred (Stephens et al., 2006; Miquelle et al., 2010). This kind of analysis is reliant upon the quality of the underlying data and this may vary from source to source; however, the predictive ability of the methods we employ here and the fact that the studies we cite have been peer-reviewed, suggests the impact of this variation is minor. We used Jacobs’s index to determine the prey selectivity of tigers via the formula: D r p r p rp = ( ) − ( ) + − i i i i ii , 2 where, ri is the proportion of species i among the total kills at a site and pi is the proportion of species i in the available prey community (Jacobs, 1974). The resulting values range from +1 (maximum preference) to -1 (maximum avoidance) (Jacobs, 1974). The mean Jacobs’s index value for each prey species across studies was calculated [ 1 standard error (se) wherever the mean is shown], and these values were tested for significant preference or avoidance using t-tests against an expected value of 0 if they conformed to the assumptions of normality. Where transformation could not satisfy these assumptions, the sign test (Zar, (1999)) was used. There are three major benefits of this kind of meta-analysis. Firstly, it is not biased by the results from one particular area. Secondly, it is not influenced by the available prey community because for a species to be significantly preferred or avoided, it must be so at several sites. Lastly, it is not biased by predation of particular sexes or age classes as the data arise from hunting by tiger populations that consist of both sexes and all age classes. The number of species with relatively small sample sizes (i.e. few studies recording them as prey) means that significant preference and avoidance is less likely because at least five Jacobs’s index values are required to obtain a significant result using the sign test, hence it is a conservative measure of prey preference. Nonetheless, plots of Jacobs’s index with error bars illustrate which species are likely to be significantly preferred or avoided with a larger sample size, assuming the existing trend continues. Multiple regression was conducted on non-correlated, transformed variables to determine which factors influenced the prey preferences of the tiger. The variables used were prey relative abundance at a site, prey body mass, herd size and preferred habitat type (Table 2). Significant relationships were plotted using distance-weighted least-squares and linear regression. Linear regression was also used for testing the relationship between prey relative abundance with Jacobs’s index value and the proportion of kills at a site. We used three-quarters of the mean adult female body mass of prey species to take account of calves and sub-adults eaten. This value was used in previous studies (Hayward et al., 2006b,c), and we continue its use here to allow comparison between these studies. Prey weights were taken from Nowak (1999). We used herd size as an indicator of the ability of prey to detect predators and vice versa. This was a categorical variable, with 1 relating to solitary individuals, 2 to species that exist in pairs, 3 to small family grouping species, 4 to small herds (10–19) and 5 to large herds (20; Table 2). Obviously, this is a simplification as large herding species may also have solitary males among them and species normally existing in smaller groups may aggregate in certain habitats or following habitat disturbance (Schaller, 1967; Dhungel & O’Gara, 1991; Olson et al., 2005); however, this variable has been successfully used previously (Funston, Mills & Biggs, 2001; Hayward et al., 2006c). Habitat type may influence predation rates as the density of vegetation can affect the detectability of both predator and prey. Animals inhabiting dense vegetation generally adopt a silent, solitary, hider strategy to evade detection, whereas prey on open grasslands are detected by sight rather than sound and often exist in large herds (Geist, 1974; Leuthold, 1977). Furthermore, tigers are solitary ambush hunters reliant upon cover to stalk their prey (Wilson & Mittermeier, 2009). On this basis, we would expect solitary tigers to predominately hunt prey inhabiting denser habitat types. Although inherently difficult to classify (Sunquist & Sunquist, 1997), a categorical variable of habitat was used with 1 relating to species living in open environments, 2 to those in woodlands and 3 to those inhabiting dense forests (Table 2). Means are presented with standard errors throughout.

Results
 We calculated Jacobs’s index scores based on 3187 kills or scats of 32 species recorded as tiger prey in the literature (Table 2). These studies had a mean sample size of 120 25 scats or kills (Table 1). Excluding species with just one Jacobs’s index estimate, the mean number of preference estimates per prey species was 8.0 ( 1.6). The most frequently preyed upon species by tigers were wild boar, chital and sambar and these were killed wherever they were available (Table 2). Although less frequently present at study sites, red deer, roe deer and muntjac were also almost invariably killed by tigers (Table 2). Conversely, peafowl, musk deer, goral and Indian rhinoceros were generally avoided when they were present in a community (Table 2). Sambar and chital deer constituted the highest proportion of tiger prey in all sites where they occurred, both comprising 34 (4%) of kills where they occur (Table 2). Red deer, banteng and wild boar were the next most important prey in those locations where they occurred (Table 2). Chital and red deer were the most abundant prey where they occurred (Table 2). Tigers significantly prefer to prey upon wild boar (t54 = 4.38, P < 0.001) and sambar deer [t42 = 4.25, P < 0.001; (Table 2; Fig. 2)]. Red deer and, possibly also, barasingha are likely to be significantly preferred prey of tigers with a larger sample size (Table 2; Fig. 2). Tigers significantly avoided preying on peafowl (Z = 100, n = 6, P = 0.001), langur (t26 = -15.06, P < 0.001) and nilgai (t32 = -2.35, P = 0.025; Table 2; Fig. 2). Larger sample sizes of Indian rhinoceros, common goral, tahr, musk deer, livestock, macaque monkeys, elephants and chinkara gazelle are likely to see them significantly avoided also (Fig. 2). The remaining prey species are killed by tigers in accordance with their availability in the prey community (Fig. 2). A multiple linear regression on prey relative abundance, body mass (log transformed), herd size and main habitat (Table 2), was not significant (R2 = 0.34, F4, 15 = 1.89, P = 0.16). Prey body mass (log) was the only variable that was significantly related to the Jacobs’s index value (t15 = 2.32, P = 0.03). There were positive significant relationships or trends between the proportional abundance of common tiger prey species and the proportion they constituted among kills at a given study site (Fig. 3). There was no relationship between proportional abundance and the Jacobs’s index value of these species; however, there was a negative trend for the two significantly preferred prey species: sambar and wild boar (P < 0.10 for both; Fig. 3). 


Figure 2 Mean Jacobs’s index values ( 1 standard error) for species that were potential prey of tigers at two or more sites. Black fill illustrates significantly preferred prey species, dark grey species that are likely to be significantly preferred with larger samples sizes, pale grey species as killed in accordance with their proportional abundance by tigers and open bars represent significantly avoided species.

*This image is copyright of its original author

When all prey species with >2 estimates of Jacobs’s index were considered, tigers prefer prey weighing between 60 and 250 kg (Fig. 4); however, tigers increasingly prefer prey up to 200 kg (r2 = 0.77, n = 17, P = 0.001; Fig. 5). The mean body mass of preferred prey species, that is those species with two or more Jacobs’s index estimates where the mean (1 se) exceeded 0, was 140 13 kg, and the mean body mass of significantly preferred (ideal) prey species was 130 21 kg (Table 2). On the basis that an adult tigress weighs on average 125 kg (lower range of mean adult female body mass reported for Bengal tigers by Wilson & Mittermeier, 2009), the ratio of predator body mass to that of their preferred prey was 1:1.1 and that of their ideal prey was 1:1.

Discussion
This study provides information on the key requirements necessary for a source population of tigers to persist; namely, a viable population of large deer (e.g. sambar, barasingha or red deer) and/or wild boar. Tigers prefer preying upon species similar in body mass to themselves akin to the preferences of other solitary predators (Gittleman, 1985; Vezina, 1985; Jedrzejewski et al., 1993; Carbone et al., 1999; Hayward et al., 2006a,b). These preferences are reinforced throughout the predatory behavioural sequence (Creel & Creel, 2002; Hayward et al., 2011) and have theoretically evolved, through optimal foraging (Goss-Custard, 1977; Bertram, 1978; Krebs, 1980), to maximize energetic returns while minimizing injury risk (Hayward & Kerley, 2005). The preferred prey species supports the theory that tiger evolution followed the Pleistocene radiation of cervids in south-eastern Asia as the evolution of large ungulates created a niche for a large-bodied forest-edge predator (Sunquist et al., 1999) and largely quantifies Sunquist’s (2010) preferred prey weight range of tigers. This preferred prey weight range also includes the young of larger species, which are frequently preyed upon, such as rhinoceros calves (Talukdar, 2002) as well as Pere David’s deer Elaphurus davidianus whose behaviour suggests it may have been an important prey item of tigers before its 1200 years of captivity (Li et al., 2011). The preferred prey weight range also accords with that found at well-studied individual sites, like Nagarhole (Karanth & Sunquist, 1992, 1995). It is becoming increasingly evident that preferential predation on suids is practised solely by the largest felids, possibly because of their possession of retractile claws (this study; Hayward & Kerley, 2005; Miquelle et al., 2010). Suids, including wild boar and warthogs Phacochoerus africanus, possess defensive weapons that can inflict serious injuries to predators (Miquelle et al., 2010). This means that predators must avoid these weapons in the act of killing them. Felids are able to effectively do this by using their claws and forepaws to manipulate the suid to keep the tusks for goring them while obtaining a safe killing bite. Suids also possess a low centre of gravity that reduces the effectiveness of coursing predators 2009).who tend to rely on tripping fleeing prey and killing them before they can defend themselves. The tiger’s preference for wild boar is likely to be increased with the inclusion of data from Sumatra, which lacked abundance data, but where wild boar composed a large proportion of the diet (Table 1). The importance of suids to tigers has not previously been identified outside the Russian Far East, but has important implications for conservation actions in all tiger range states. Megaherbivores appear to avoid top-down limitation through preferential predation by large predators, as they do in Africa (Sinclair, Mduma & Brashares, 2003; Hayward, Tigers also provide more evidence (Fig. 3) that prey species that are rare in the environment avoid top-down limitation from predators through their scarcity in the prey community (Hayward, 2011). Karanth & Sunquist (1995) concluded that selective predation in Nagarhole facilitated coexistence of large predators. Leopards in Asia preferentially prey on chital Axis axis and prefer prey weighing between 10 and 40 kg (Hayward et al., 2006a), which is much smaller than prey preferred by tigers (this study). Thus, there is clear separation between leopard and tiger dietary preferences enabling their coexistence. Prey availability and body mass were key determinants of prey preferences of tigers in Nepal’s Royal Bardia National Park (Støen & Wegge, 1996). Wild boar live solitarily in the dry season and were more vulnerable to a stalking tiger (ambush predator) than the large herds of chital (Støen & Wegge, 1996). Johnsingh (1992) supported the theory that the grouping behaviour of chital reduced predation from tigers; however, the results of this present study question this explanation and suggest that chital are too small to be optimal prey of tigers. Tiger density has previously been predicted based on the biomass or density of all potential prey species present at a site, although the confidence intervals are wide (Miquelle et al., 1999; Sunquist et al., 1999; Karanth et al., 2004). Similar relationships exist for large African carnivores and all large carnivores (van Orsdol, Hanby & Bygott, 1985; Laurenson, 1995; Stander et al., 1997; Carbone & Gittleman, 2002). These relationships have been refined for African carnivores by relating predator density with the biomass of their preferred prey rather than all available prey, which has greatly reduced the amount of variation in the data (Hayward et al., 2007b). Similar attempts have been recently made for tigers; however, these are based on a limited number of local studies (Miquelle et al., 2010). It is highly likely that using the results of our study would further improve the accuracy of tiger density estimates over a broader area and more accurate predictions of tiger density could allow the identification of undetected poaching keeping tigers below carrying capacity or be used to improve the accuracy of population viability analyses (Hayward et al., 2007b; Bach et al., 2010). The use of preferred prey biomass could also be used to predict tiger diet and home range size, as has occurred with lions, leopards and spotted hyaenas Crocuta crocuta (Hayward et al., 2007a, 2009). Reply