Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Journal of Mammalogy, 97(3):919–927, 2016 DOI:10.1093/jmammal/gyw017 Published online February 22, 2016 Prey morphology and predator sociality drive predator prey preferences Hayley S. Clements,* Craig J. Tambling, and Graham I. H. Kerley Centre for African Conservation Ecology, Department of Zoology, Nelson Mandela Metropolitan University, Port Elizabeth 6031, South Africa * Correspondent: [email protected] Predator prey preferences shape the dynamics of predator–prey assemblages. Understanding the determinants of a predator’s prey preferences is therefore important. Trends in prey preferences of the large African predators have been described at a species scale, limiting our ability to assess the influence of prey morphology (size and horns) and predator social structure (sex and sociality) on prey preferences. An analysis of cheetah Acinonyx jubatus kill and prey abundance information from throughout South Africa shows that the presence of prey horns and the type of cheetah social group interact with prey size to influence cheetah prey preference. The size threshold above which prey is avoided by cheetah is lower for horned prey than non-horned prey, providing evidence that horns are a predation deterrent in medium-sized prey. Horned females occur significantly more frequently in antelope species on the cusp of being too large for cheetah predation, supporting the hypothesis that horns evolved as an antipredator defense in the females of medium-sized prey. Male coalition cheetah have access to a broader weight range of prey than solitary cheetah, which may infer fitness benefits by way of expanded resource options. The prey weight range accessible to solitary male cheetah is similar to that accessible to solitary female cheetah, suggesting that, in the absence of cooperative hunting, the slightly larger size of the male cheetah infers no hunting advantage. These findings provide insight into the predation pressures driving the evolutionary selection for large body size and horns in prey, and expanded resource access leading to predator sociality. Key words: cheetah, horns, large carnivore, predator sociality, prey preference, prey size © 2016 American Society of Mammalogists, www.mammalogy.org Predators display prey preferences that in turn influence ecosystem patterns and processes such as biomass fluxes (OwenSmith and Mills 2008), the abundance of predators relative to prey (Carbone and Gittleman 2002), and prey population composition and dynamics (Sinclair et al. 2003; Gervasi et al. 2015). Across a myriad of carnivore species, these prey preferences are influenced by the size of the prey relative to the size of the predator (Gittleman 1985; Sinclair et al. 2003; Owen-Smith and Mills 2008; Gervasi et al. 2015). Among terrestrial carnivores, there is a body mass limit (21.5 kg) above which a carnivore’s energy requirements necessitate predation on prey weighing more than 45% of the carnivore’s body mass (Carbone et al. 1999). The relationship between predator size and prey size is perhaps most critical for these larger predators, due to the challenges of capturing and subduing large prey (Gittleman 1985). Large African predators regularly predate ungulates double their mass (Gittleman 1985; Owen-Smith and Mills 2008), with the possibility of injury or death to the predator during prey capture (Hayward and Kerley 2005). According to optimal foraging theory, a predator should select prey that offers maximum energetic benefits in terms of size, with minimum energetic costs and risks incurred during prey capture (Pyke et al. 1977). Risk imposed on the predator during prey capture is likely influenced not only by relative prey size but also by other prey and predator life history attributes, such as prey defense morphology (e.g., the presence of horns, spines, or plates) and behavior (e.g., grouping) that may repel predator attacks (Packer 1983; Caro 2005), and predator social structure (e.g., cooperative hunting) that may improve hunting success rates (Kruuk 1975; Fanshawe and FitzGibbon 1993). Stankowich and Caro (2009) predict that larger prey species, by being better able to defend themselves against predators as well as by being more exposed, should benefit more from defensive weaponry (e.g., horns) than smaller species. This suggests that predation risk is a function of the interaction between prey size and weaponry. Furthermore, predators that hunt cooperatively, at least to the extent that all group members hunt simultaneously, can kill prey that is larger relative to their own size than solitary predators (Kruuk 1975; Fanshawe and FitzGibbon 919 920 JOURNAL OF MAMMALOGY 1993). Males of sexually dimorphic predators also kill larger prey than females (Radloff and du Toit 2004). Prey size is an important determinant of prey preference of the large African predators: cheetah Acinonyx jubatus, leopard Panthera pardus, lion Panthera leo, spotted hyaena Crocuta crocuta, and African wild dog Lycaon pictus (Hayward and Kerley 2005; Hayward 2006; Hayward et al. 2006a, 2006b, 2006c; Clements et al. 2014). Risk presented by prey species is a further determinant of prey preference in cheetah, leopard, and wild dog (Hayward et al. 2006a, 2006b, 2006c). Hayward and colleagues classed these risks a priori according to documented predator–prey interactions, on a scale from 0 (no threat) to 2 (severe threat; known deaths of predators caused by the species). The mechanism behind this threat level is unknown, with interspecies analyses preventing a finer assessment of the influence of weaponry on prey preference, since weaponry can differ between age and sex classes within species. The implications of differences in preference between prey age and sex classes on prey population dynamics (FitzGibbon 1990a, 1990b; Gervasi et al. 2015) are therefore overlooked. Furthermore, by pooling predator sexes and hunting group sizes, these broadscale studies do not account for differences in size and hunting behavior between and within predator sexes, despite the fact that several single-site studies have shown predator hunting group size to be an important determinant of prey size for lion, spotted hyaena, and wild dog (Mills 1990; Scheel and Packer 1991; Creel and Creel 2002). The influence of the interaction between prey morphology and predator social structure on the prey preferences of large predators has yet to be accounted for in multisite predator prey preference reviews. The cheetah is a suitable study species with which to assess the relationship between prey morphology and predator social structure in determining prey preference. It is a highly specialized predator evolved for rapid pursuit hunting (Mills and Harvey 2001), which is believed to impose morphological limitations on the size of prey that can be successfully captured without injury (Hayward et al. 2006b). The high energy expenditure during such hunts, penalizing unsuccessful attempts, is reflected by the high hunting success rate (Bertram 1979). These findings suggest that the cheetah is a highly selective predator, targeting prey that offer the greatest likelihood of successful capture without injury. Following optimal foraging theory (Pyke et al. 1977), we predict that the risk imposed by horns will lower the upper body size threshold of horned prey compared to non-horned prey, which is preferred by cheetah. Furthermore, the cheetah displays a social structure varying both between and within the sexes (Caro 1994). Female cheetah are solitary except when rearing cubs, while male cheetah are either solitary or form stable coalitions (Caro 1994). Male cheetah, being 25% larger than females, have been shown to hunt more large prey than females (Radloff and du Toit 2004) and male coalitions hunt more large prey than solitary male cheetah (Caro 1994). Consistent with these patterns described at single sites, when assessed across multiple sites and measured in terms of prey preference, we predict that (1) the size dimorphism between cheetah sexes will result in solitary male cheetah preferring larger prey and a broader weight range of prey than female cheetah and (2) a cooperative hunting advantage will result in male coalition cheetah preferring larger prey and a broader weight range of prey than solitary cheetah of either sex. In intact African ecosystems, cheetah coexist with other competitively dominant predators, such as lion and spotted hyaena, which kleptoparasitize and kill them (Caro 1994). Intraguild competition has been predicted to promote cheetah preference for medium-sized prey, which can be consumed before kleptoparasites arrive (Radloff and du Toit 2004; Hayward et al. 2006b). Where the risk of kleptoparasitism is low, cheetah preferred prey weight range has been shown to expand (Bissett and Bernard 2007), though cheetah in denser habitats that afford refuge from kleptoparasitism do not prefer larger prey than those in more open areas (Hayward et al. 2006b). Given these conflicting findings, it is necessary to account for the influence of dominant predators in the system when assessing cheetah prey preference. Using 1,290 cheetah kills from multiple sites throughout South Africa, we test our predictions by (1) assessing the influence of prey size, prey weaponry, cheetah social structure, and the density of dominant predators on cheetah prey preference, and (2) using a novel approach to quantify and compare the degree of preference displayed by cheetah for different prey sizes. This study is unique in its integrated assessment of the influence of prey morphology (size and horns), predator social structure (sex and solitary/coalition hunting), and dominant predator density on cheetah prey preferences across multiple sites in South Africa. Materials and Methods Data sourcing.—The literature was reviewed for publications and dissertations detailing cheetah kill, prey abundance (census) and, if possible, prey demographic (sex and age ratio) data from sites in Africa. Unpublished data were sourced from additional sites. Kill data were derived from incidental observations, observations from tracking cheetah using VHF collars and continuous follows (Hayward et al. 2006b). Continuous follows are widely regarded as the superior method of ascertaining the diet of a predator, minimizing the undercounting of small prey and the inclusion of scavenged prey (Bertram 1979; Mills 1992). However, because cheetah rarely scavenge and spend a relatively brief period of time on kills (regardless of size), there is effectively no difference between the 2 methods in measuring cheetah diet (Mills 1992). Published data were augmented with those from the gray literature because few published studies included data in the detail required by this study. All unpublished and dissertation data were derived from standard, widely used methods (incidental observations and continuous follows—Mills 1992). Data from 9 South African sites, which varied in rainfall from 315 mm·yr−1 to 875 mm·yr−1, were obtained (Clements 2013; Supporting Information S1). Limited availability/access to complete datasets from elsewhere in Africa prevented a broader geographical scope. Kill data were separated into 3 cheetah social class categories: female, solitary CLEMENTS ET AL.—CHEETAH PREY PREFERENCES921 male, and male coalition cheetah according to the sex and number of cheetah observed either making the kill or feeding from it (Supporting Information S1). As a result of insufficient data classification at sites with both solitary female cheetah and those rearing cubs, female cheetah kill data were pooled. At 3 sites, distinct changes in prey abundance (see Bissett 2007; Clements 2013) necessitated the division of data into 2 time periods (Supporting Information S1). Temporal partitioning of kill data is common practice in studies of carnivore feeding ecology, and is not believed to result in autocorrelation since a fundamental determinant of whether a prey item is captured is the probability of the predator encountering that prey item, which varies with temporal changes in prey abundance (Creel and Creel 2002; Hayward and Kerley 2005; Hayward et al. 2006b). Prey abundance data were obtained from aerial census data or road count data. Standard visibility correction factors, used previously in thicket and savanna ecosystems, were applied to aerial counts (Rapson and Bernard 2007; Owen-Smith and Mills 2008). No visibility corrections were applied to road counts (performed at just 1 site), since information on the density of habitats in which prey were counted was not available. For 3 sites, prey demographic data (the ratio of adult males to adult females to juveniles) could not be obtained and ratios were inferred from directly adjacent sites in 2 instances and from a site 58 km away in the 3rd instance. We ensured sites were directly comparable both in terms of vegetation type and in terms of management history in order to minimize the influence of these factors on prey demographic ratios (Supporting Information S2). Allocating prey masses and determining prey abundances.— For each listed potential prey species, adult males, adult females, and juveniles (hereafter referred to as species-demographic-classes) were allocated a standard “species-classmass”, assuming juvenile mass to be 30% of the adult female mass (Skinner and Chimimba 2005; Supporting Information S3). Juveniles were considered to be individuals less than a year old. For warthogs Phacochoerus africanus, adult kills were rarely sexed, and so all adults were pooled, calculating species-class-mass as an average between adult male and female body mass (Supporting Information S3). Within each species, males and females were categorized as horned or non-horned (Skinner and Chimimba 2005), with all juveniles categorized as non-horned. Warthogs of both sexes carry tusks and hence adults were included in the “horned” category. Abundance of each prey species-demographic-class at each site was calculated using species abundance and demographic data. The total abundance of censused prey at each site was calculated by summing the census data for all prey species at that site. Prey weighing more than 1,200 kg were omitted, since giraffes are the largest prey species recorded to have been killed (Hayward et al. 2006b). Carnivores occasionally appear in the kill records but are rarely consumed (Radloff and du Toit 2004) and were therefore omitted. Calculating preferences for prey species-demographicclasses.—To determine how and why a predator selects its prey, cognizance of prey availability is imperative. If a predator is presented with 2 (or more) prey items in equal abundance and selects the 1 more frequently than expected by chance, it is referred to as a preferred prey item (Johnson 1980). While in many field studies measuring preference for equally available food items is not possible, the term preferred prey has been used to refer to a prey item that is selected (killed) by a predator more frequently than expected based on the prey item’s relative availability (Hayward and Kerley 2005). If the predator selects proportionally fewer prey than expected based on the prey item’s availability, then it is referred to as an avoided prey item. Prey preference was calculated following previous prey preference studies (Hayward and Kerley 2005; Hayward et al. 2006b), using the Jacobs’ Index (JI—Jacobs 1974): JI i = (ri − pi ) (ri + pi − 2ri pi ) JI standardizes the relationship between prey relative abundance pi (i.e., the proportion p that prey item i makes up of the total abundance of censused prey at a site) and the proportion of cheetah kills that prey item i comprises ri, to a value ranging between +1 and −1, where +1 indicates complete preference and −1 indicates complete avoidance. For each cheetah social class, JI values were calculated for each prey species-demographic-class at each site. Assessing drivers of prey preference.—We used generalized linear mixed models (GLMMs) to assess whether prey size, cheetah social class, prey weaponry, and the density of dominant predators influenced cheetah preference for each prey species-demographic-class. JI values are bounded, non-normal, and overdispersed; we therefore fitted the GLMMs using a negative binomial distribution (Zuur et al. 2009). Models based on this distribution require an integer predictor variable, and so we allocated JI values into 0.1 interval bins and allocated each bin an integer value representing degree of preference, commencing at 1 for maximum prey avoidance (bin −1 ≤ JI < −0.9) and terminating with 20 for maximum prey preference (bin 0.9 ≤ JI < 1). We combined data from all sites and timeperiods; including site, and sample-period nested within site, as random effects. We developed a global model that included prey mass, prey weaponry, cheetah social class, and the combined density of lion and spotted hyaena at the site as fixed effects. Since we predicted prey weaponry, cheetah social class and the density of other large predators would interact with prey mass to influence prey preference, these 2-way interactions were included in the global model. Given the potential biases arising from applying visibility correction factors to aerial but not road census data, we further included census method as a methodological predictor. We generated 12 nested models by removing combinations of fixed effects and interactions from the global model, while keeping the biological relevance of the model in mind and always including the methodological predictor. Akaike’s Information Criterion (AIC) was used for model selection, where models with a change in AIC greater than 2 (when compared to the best model) were considered to be less informative than the best model (Akaike 1974; Burnham and Anderson 2002). The best model was compared 922 JOURNAL OF MAMMALOGY to the null (intercept only) model using a likelihood ratio test in order to assess whether the predictor variables significantly increased model fit. Prior to model fitting, we examined the fixed effects for multicollinearity to avoid correlation among covariates. Plots of fitted and observed values and residuals were examined for deviations from the assumptions regarding homogeneity and normality (Zuur et al. 2009), and as a result prey mass was log-transformed. Models were run using the R package “lme4” (Bates et al. 2014). Quantifying preferred prey weight ranges.—Prey weaponry and predator sociality were significant predictors of cheetah prey preference (see “Results”). We therefore assessed cheetah prey weight preferences by distinguishing between horned and non-horned prey and solitary and coalition cheetah (pooling solitary male and female cheetah kills). As the density of dominant predators influenced cheetah prey preference (see “Results”), we excluded sites in our subsequent analyses where dominant predators were absent (resulting in 879 kill records for the subsequent analyses). A mean JI value across sites was calculated for each prey species-demographic-class with 2 or more site-specific JI values. These mean JI values (Supporting Information S3) were standardized (+1 to ensure non-negative values) and used as an index of prey preference. To determine solitary and coalition cheetah prey size preference for horned and non-horned prey, segmented generalized linear models were used (Hudson 1966; Muggeo 2008) as they provide a statistically robust means of detecting changes in prey preference as a function of prey mass. Following Clements et al. (2014), prey species-class-masses were ranked from the lightest to the heaviest and allocated corresponding incremental integer values (mass-ranks; Supporting Information S3) in order to control for uneven size distribution of prey species. For each prey species-class-mass, a corresponding cumulative JI+1 value was calculated, commencing at the prey speciesclass-mass with a mass-rank of 1. Mass-ranks were plotted against cumulative JI+1 values and break-points were estimated using segmented models (Muggeo 2008). Starting with a linear model, we incrementally increased the number of break-points in each segmented model, and selected the optimum number of break-points (if any) using AIC (Akaike 1974). Mass-ranks, at which the best-fit segmented model detected break-points in the relationship between mass and preference, were converted back into corresponding prey species-class-masses in order to identify prey weight ranges of differing preference. Prey weight ranges of differing preference were thereafter assessed for prey preference/avoidance by calculating the JI of each prey weight range at each site listed in Supporting Information S1 (Clements et al. 2014). The mean JI value of each prey weight range across sites was tested for significant preference or avoidance using a single sample t-test against a mean of zero where data conformed to the assumptions of normality, and a Wilcoxon signed-rank test where data did not. Prey weight ranges found to be either preferred (mean JI significantly greater than 0) or killed relative to their abundance (mean JI not significantly different from 0) are together referred to as “accessible” weight ranges. A mean JI value significantly less than zero indicated an avoided prey weight range. Segmented models were performed using the R package “segmented” (Muggeo 2008). Female size and the presence of horns in South African antelope species.—We generated a list of the antelope species (family Bovidae) indigenous to South Africa and recorded the mass and presence/absence of horns for the adult female of each species (Supporting Information S4; Skinner and Chimimba 2005). For each mass, we determined, according to our preference results, whether a prey individual with such a mass would be avoided by cheetah if it (1) had horns and (2) did not have horns. We then allocated antelope into 1 of 2 categories: (1) antelope with adult females of a mass where only non-horned prey are accessible to solitary or coalition cheetah and (2) antelope with females of a mass where both horned and non-horned prey are accessible to solitary and coalition cheetah. The number of antelope species with horned and non-horned adult females in each category was summed. We excluded antelope with horned adult females weighing more than 313 kg, since no non-horned prey weigh more than this (Supporting Information S3), thus preventing comparison. Difference in the number of species with horned versus non-horned females in each category was assessed using a chisquare test. Statistical analyses were performed in the statistical programme R (R Development Core Team 2013), at a significance level of 0.05. Results The best model predicting cheetah prey preference (see Supporting Information S5 for the top 4 candidate models) had a significantly improved fit when compared with the null model (χ2 = 215.04, d.f. = 10, P < 0.001). The best model included a significant negative relationship between prey mass and prey preference (Table 1). This relationship was stronger when prey had horns, depicted by a significant negative interaction between prey mass and the presence (compared with absence) of horns (Table 1). The relationship between prey mass and prey preference was weaker for male coalition cheetah, compared with female cheetah, depicted by a significant positive interaction between prey mass and male coalition (compared with female) cheetah (Table 1). There was no difference in the relationship between prey mass and prey preference for female and solitary male cheetah, evidenced by the nonsignificant interaction between prey mass and solitary male (compared with female) cheetah (Table 1). Finally, there was a significant negative relationship between the density of dominant predators and prey preference, and census method did not influence prey preference (Table 1). Segmented models revealed changes in the relationship between prey weight and degree of cheetah prey preference at sites where dominant predators were present (Fig. 1). For solitary and male coalition cheetah, preferred weight ranges of prey were only evident in non-horned prey, whereas horned prey were killed relative to their abundance or avoided (Fig. 2; CLEMENTS ET AL.—CHEETAH PREY PREFERENCES923 Table 1.—Best generalized linear mixed model (AIC = 3602.9) of cheetah prey preference, as a function of prey mass, prey weaponry, cheetah social class, and density of lion and spotted hyaena. (: represents an interaction between 2 fixed effects.) Results include coefficient estimate β, standard error SE(β), associated Wald’s z-score β/SE(β), and significance level P for all predictors. AIC = Akaike’s Information Criterion. β SE(β) z P 4.146 −1.228 1.310 −0.602 −1.524 −1.956 −0.804 0.319 1.023 0.051 0.328 0.169 0.418 0.461 0.503 0.573 0.213 0.242 0.262 0.144 12.642 −7.252 3.126 −1.305 −3.029 −3.411 −3.765 1.318 3.915 0.355 < 0.001* < 0.001* 0.002* 0.19 0.002* < 0.001* < 0.001* 0.19 < 0.001* 0.723 Fixed effect Intercept log(prey_mass) prey_horned versus prey_not_horned solitary_male_cheetah versus female_cheetah coalition_cheetah versus female_cheetah density_dominant_predators log(prey_mass):prey_horned versus log(prey_mass):prey_not_horned log(prey_mass): solitary_male_cheetah versus log(prey_mass): female_cheetah log(prey_mass): coalition_cheetah versus log(prey_mass): female_cheetah census_method_road versus census_method_aerial * Significant. Cumulative Jacobs Index + 1 Prey mass (kg) 10 9 8 7 6 5 4 3 2 1 0 14 12 10 8 6 4 2 0 11 62 0 5 16 150 Solitary cheetah - horned prey AIC = 13, n = 26 15 20 25 30 10 42 1192 179 365 1192 Coalition cheetah - horned prey AIC = 10, n = 21 0 5 10 15 20 25 20 18 16 14 12 10 8 6 4 2 0 20 18 16 14 12 10 8 6 4 2 0 4 8 0 145 313 Solitary cheetah - non-horned prey AIC = 19, n = 28 10 15 20 25 30 5 0 48 31 18 5 62 119 313 Coalition cheetah - non-horned prey AIC = 10, n = 22 10 15 20 25 Prey mass rank Fig. 1.—Segmented relationship between the mass-rank of each prey species-demographic-class and predator preference, for solitary and male coalition cheetah. Preference is indicated by cumulative Jacobs’ Index +1 value. Actual prey masses that correspond to the lowest, break-point, and highest prey mass-ranks are indicated above the figures. Table 2). For solitary cheetah, the presence of horns decreased the weight threshold above which prey are avoided from 145 kg for non-horned prey (JI>145 kg = −0.91 ± 0.05, P < 0.001) to 62 kg for horned prey (JI62–150 kg = −0.85 ± 0.10, P = 0.02; JI>150 kg = −0.96 ± 0.02, P = 0.02; Fig. 2; Table 2). Male coalition cheetah did not exhibit a weight threshold above which non-horned prey are avoided (Fig. 2). Adult male zebra, weighing 313 kg, were the largest non-horned prey present at any of the study sites and male coalition cheetah killed prey up to this weight relative to prey abundance (JI119–313 kg = −0.22 ± 0.23, P = 0.41; Fig. 2; Table 2). This translated into male coalitions killing non-horned prey relative to their abundance at weights up to 168 kg greater than solitary cheetah (Fig. 2; Table 2). Male coalition cheetah avoided horned prey above a weight threshold of 179 kg (JI179–365 kg = −0.85 ± 0.09, P < 0.001; JI>365 kg = −1 ± 0; Fig. 2; Table 2), thereby killing horned prey relative to their abundance at weights up to 117 kg greater than solitary cheetah (Fig. 2; Table 2). There were significantly more antelope species with horned adult females in the weight range where horns result in avoidance by cheetah, than there were antelope species with horned adult females in the weight range where the females are accessible to cheetah irrespective of horn presence/absence (χ2= 8.27, d.f. = 1, P < 0.01). Discussion This study highlights the influence of prey morphological traits (size and weaponry), predator behavioral traits (sociality), as well as the density of dominant predators on a large predator’s prey preference. Site-specific prey preferences may be further influenced by prey species composition and 924 JOURNAL OF MAMMALOGY abundance, ecology (e.g., prey scarcity), behavior, and its relation to sex (e.g., group defense; males tending to be more solitary with reduced group vigilance), and how the prey individual responds to the landscape of fear, which can also depend on available habitat (Laundré et al. 2001; Hayward 2011). Similarly, measures of prey preference can be influenced by detectability of both kills and prey communities during census, which may vary as a function of sampling technique, herd size, and habitat (which varies seasonally). As this study is based on a multisite, multispecies analysis, for a prey weight range to be found to be preferred or avoided, it must be so across a diversity of prey communities and habitats. This study therefore presents patterns of the influence of prey morphology and predator social structure on predator prey preference that are robust across a multitude of habitats, climates, and prey communities. These trends should serve as a useful departure point from which hypotheses regarding predator–prey interactions at a finer scale can be formulated. Preferred 320 Relative Avoided Prey mass (kg) 280 240 200 160 120 80 40 0 Horned Non-horned Solitary cheetah Horned Non-horned Coalition cheetah Fig. 2.—The weight ranges of horned and non-horned prey that are preferred, killed relative to their abundance, and avoided by solitary and male coalition cheetah. (Weight ranges preferred and killed relative to their abundance are together referred to as “accessible prey weight ranges.”) The influence of prey morphology on cheetah prey preference.—Upper prey size thresholds were identified above which large body size serves to provide protection to prey from cheetah predation. This supports previous studies on predator–prey size relationships (Sinclair et al. 2003). Notably, this study further shows that the influence of prey size and the presence of horns interact, lowering the mass threshold above which prey is avoided. Our findings therefore provide quantitative evidence that horns serve as a successful antipredatory mechanism, particularly in prey that are on the cusp of being too large to be predated. This further provides support for the prediction that larger species, by being better able to defend themselves against predators as well as by being more exposed, should benefit more from having defensive weaponry than smaller species (Packer 1983; Stankowich and Caro 2009). In all antelope species, males possess horns that are widely thought to have evolved for territorial disputes or mate acquisition (Geist 1966; Janis 1982). In contrast, the presence of horns in females is highly variable across antelope species and its evolution has been attributed to an antipredatory mechanism, the defense of territories against other females, and as a buffer against aggression toward male offspring by dominant males (Packer 1983; Estes 1991; Stankowich and Caro 2009). The occurrence of horned females is significantly greater in antelope species where females are large enough to escape into the avoided weight range by having horns, supporting the prediction that predation is an evolutionary driver of horns in large female antelope. Small prey also benefit from horns. While not influencing the accessibility of smaller prey to cheetah, horns do influence preference within this accessible prey weight range. Small prey size, independent of horn presence/absence, limits absolute protection from cheetah predation, but the presence of horns reduces predation below levels experienced by non-horned prey of similar size. Furthermore, horns in small-sized prey may serve to reduce predation by predators smaller than cheetah, such as black-backed jackal Canis mesomelas. The influence of prey size and horns on cheetah prey preference shows that, Table 2.—The prey mass-ranks at which segmented models revealed break-points in cheetah prey preference, with corresponding prey weight ranges and preference statistics. JI = Jacobs’ Index. Cheetah Prey weaponry Preference Mass-rank range Corresponding weight range (kg) JI Significance Solitary Horned Relative Avoided Avoided Preferred Relative Relative Avoided Relative Relative Avoided Avoided Relative Relative Preferred Relative ≤ 8.3 8.3–15.1 > 15.1 ≤ 11.7 11.7–17.2 17.2–23.5 > 23.5 ≤ 3.4 3.4–11.5 11.5–17.5 > 17.5 ≤ 5.9 5.9–12.2 12.2–16.6 > 16.6 ≤ 62 62–150 > 150 ≤ 31 31–48 48–145 > 145 ≤ 42 42–179 179–365 > 365 ≤ 18 18–62 62–119 > 119 −0.19 ± 0.14 −0.85 ± 0.10 −0.96 ± 0.02 0.35 ± 0.14 0.12 ± 0.22 0.21 ± 0.12 −0.91 ± 0.05 −0.17 ± 0.42 0.03 ± 0.14 −0.85 ± 0.09 −1 ± 0 0.00 ± 0.27 −0.15 ± 0.12 0.87 ± 0.01 −0.22 ± 0.23 t = 1.32, d.f. = 6, P = 0.24 W = 0, n = 7, P = 0.02 W = 0, n = 7, P = 0.02 t = 2.46, d.f. = 6, P = 0.049 t = 0.54, d.f. = 6, P = 0.61 t = 1.69, d.f. = 6, P = 0.14 t = −17.65, d.f. = 6, P < 0.001 t = −0.41, d.f. = 4, P = 0.72 t = 0.21, d.f. = 4, P = 0.85 t = −9.25, d.f. = 4, P < 0.001 n = 6 t = 0.00, d.f. = 4, P = 0.99 t = −1.26, d.f. = 4, P = 0.28 t = 162.03, d.f. = 4, P < 0.001 t = −0.92, d.f. = 4, P = 0.41 Non-horned Male coalition Horned Non-horned CLEMENTS ET AL.—CHEETAH PREY PREFERENCES925 in the face of predation by cheetah, it is better for a prey individual to be big and horned. In multipredator systems, size- and horn-derived protection from smaller predators such as cheetah would not necessarily provide protection from predation by larger predators that hunt larger prey (e.g., lion—Radloff and du Toit 2004; Owen-Smith and Mills 2008). However, while a large body size and the presence of horns may not eliminate the risk of predation for most prey species, it will determine the suite of predators that the species is vulnerable to, with a bigger size and the presence of horns reducing total predation pressure. Therefore, predation pressure, even in multipredator ecosystems, could have contributed to the selection for large body size and the presence of horns in prey species. The influence of cheetah sex and social structure on cheetah prey preference.—Cheetah preference for prey of a given mass did not differ between female and solitary male cheetah, suggesting that sexual dimorphism in cheetah is not sufficient to facilitate a clear hunting advantage to the slightly larger males when the latter are solitary. While this is contrary to the finding of Radloff and du Toit (2004), the majority of male cheetah data used in their study were from coalitions of cheetah, not solitary males (F. Radloff, Cape Peninsula University of Technology, pers. comm.). Therefore, while Radloff and du Toit (2004) only compared cheetah sexes and not group sizes, their results may have been biased by group hunting in males, since we show that male coalition cheetah have a larger weight range of prey accessible to them than solitary cheetah. Drivers of sociality within the order Carnivora have been debated, including the benefits of strength in numbers for defense of cubs, kills, and territories and the ability to hunt and kill larger prey cooperatively (Macdonald 1983; Packer et al. 1990; Caro 1994). For cooperative hunting to be a driver of sociality, groups members must accrue greater per capita forage returns by hunting as a group than alone, and food intake must be a key factor affecting fitness (Caro 1994). Ultimately, the costs of sociality, such as food and mate sharing (Kruuk 1975; Caro and Collins 1986; Packer et al. 1990; Caro 1994), need to be outweighed by the benefits. Costs and benefits will vary, depending for example on the availability and density of food and the risks of accessing, capturing, and defending it (Macdonald 1983; Radloff and du Toit 2004). While acknowledging that sociality in predators is therefore likely to be the result of several interacting variables, this study shows that coalition males derive increased resource access as a direct benefit of their sociality. Female cheetah with cubs will require more food to meet increased energetic demands, and may accomplish this by increasing their kill rate, as seen in leopard and puma Puma concolor (Bothma and Coertze 2004; Laundré 2008), or hunting more prey in the upper prey size limits imposed on solitary hunters (Laurenson 1995). We did not investigate this, but with the accumulation of datasets distinguishing between kills of solitary and cub-rearing females, these predictions can be tested. The influence of dominant predators on cheetah prey preference.—It has been predicted that a reduction in kleptoparasitism may result in an expansion of a cheetah’s prey weight preferences (Radloff and du Toit 2004; Bissett and Bernard 2007). Contrary to this prediction, we show that the density of lion and spotted hyaena does not influence prey weight preferences of cheetah. The density of dominant predators does, however, influence the number of prey speciesdemographic-classes that are preferred. The reduction in prey preference with increasing dominant predator density, independent of prey mass, suggests that at higher densities of dominant predators, cheetah diet becomes more specialized. This functional response may be as a result of cheetah avoiding certain habitats where dominant predators more commonly occur, or within which cheetah are more visible to dominant predators (Durant 1998; Vanak et al. 2013). This would represent a spatially focused mechanism for mesopredator release in the absence of lion and spotted hyaena (Prugh et al. 2009). We did not quantify the cheetah’s preferred and accessible prey weight ranges at sites devoid of dominant predators, since no data at such sites were available for male coalition cheetah. However, as the density of dominant predators does not influence cheetah prey mass preference or avoidance, the preferred and accessible prey weight ranges of cheetah, quantified in this paper, should be reliable independent of the presence or absence of dominant predators. Our findings provide support for the influence of selective predation on the evolution of horns and body size in prey species, and the influence of expanded resource access on the evolution of sociality in predator species. With the diverse range of social systems found in large predator species, and the varied size and weaponry array found in prey species, these findings should allow for hypotheses to be developed and tested regarding predator prey preferences, and thereby predator–prey interactions and their evolutionary drivers, in other predator–prey communities. Acknowledgments We thank T. Caro, N. Owen-Smith, M. Hayward, and 2 anonymous reviewers for their useful comments on an earlier version of this paper. We thank C. Bissett, T. Burke, M. Grover, E. Larson, J. O’Brien, S. Razzaq, and R. Slater for providing data. HSC was funded by Samara Private Game Reserve, the National Research Foundation (NRF), and Nelson Mandela Metropolitan University. CJT was funded by the NRF and a Claude Leon Postdoctoral Fellowship. Supporting Information The Supporting Information documents are linked to this manuscript and are available at Journal of Mammalogy online (jmammal.oxfordjournals.org). The materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supporting data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author. 926 JOURNAL OF MAMMALOGY Supporting Information S1.—Sites and sources of data used in this study. Supporting Information S2.—Habitat and management details for sites for which prey demographic data were inferred. Supporting Information S3.—Prey species-class-masses, mass-ranks, and mean Jacob’s Index (JI) values of speciesdemographic-classes used in segmented model analyses, for solitary and coalition cheetah, separated into horned and non-horned prey. Supporting Information S4.—South African antelope (Bovidae) species and adult female masses. Supporting Information S5.—Formulas and model comparison statistics for the 4 best-fit generalized linear mixed models predicting the degree of cheetah prey preference for a given prey species-demographic-class (Pr = degree of prey preference; M = log(prey mass); H = prey weaponry; C = cheetah social group; D_LH = density of lion and spotted hyaena; X = census method; : = interaction between 2 fixed effects). Literature Cited Akaike, H. 1974. A new look at the statistical model identification. IEEE Transactions on Automatic Control 19:716–723. Bates, D., M. Machler, B. Bolker, and S. Walker. 2014. Fitting linear mixed-effects models using lme4. Cornell University, Ithaca, New York. http://arxiv.org/abs/1406.5823. Accessed 8 February 2016. Bertram, B. C. B. 1979. Serengeti predators and their social systems. Pp. 221–285 in Serengeti: dynamics of an ecosystem (A. R. E. Sinclair and M. Norton-Griffiths, eds.). University of Chicago Press, Chicago, Illinois. Bissett, C. 2007. The feeding and spatial ecologies of the large carnivore guild on Kwandwe Private Game Reserve. Ph.D. dissertation, Department of Zoology, Rhodes University, Grahamstown, South Africa. Bissett, C., and R. T. F. Bernard. 2007. Habitat selection and feeding ecology of the cheetah (Acinonyx jubatus) in thicket vegetation: is the cheetah a savanna specialist? Journal of Zoology (London) 271:310–317. Bothma, J. P., and R. J. Coertze. 2004. Motherhood increases hunting success in southern Kalahari leopards. Journal of Mammalogy 85:756–760. Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach. 2nd ed. Springer, New York. Carbone, C., and J. L. Gittleman. 2002. A common rule for the scaling of carnivore density. Science 295:2273–2276. Carbone, C., G. M. Mace, S. C. Roberts, and D. W. MacDonald. 1999. Energetic constraints on the diet of terrestrial carnivores. Nature 402:286–288. Caro, T. M. 1994. Cheetah of the Serengeti Plains: group living in an asocial species. University of Chicago Press, Chicago, Illinois. Caro, T. M. 2005. Antipredator defenses in birds and mammals. University of Chicago Press, Chicago, Illinois. Caro, T. M., and D. A. Collins. 1986. Male cheetahs of the Serengeti. National Geographic Research 2:75–86. Clements, H. S. 2013. Incorporating prey demographics and predator social structure into prey preference and carrying capacity estimates for cheetah. M.S. thesis, Nelson Mandela Metropolitan University, Port Elizabeth, South Africa. Clements, H. S., C. J. Tambling, M. W. Hayward, and G. I. H. Kerley. 2014. An objective approach to determining the weight ranges of prey preferred by and accessible to the five large African carnivores. PloS One 9:e101054. Creel, S., and N. M. Creel. 2002. The African wild dog: behavior, ecology, and conservation. Princeton University Press, Princeton, New Jersey. Durant, S. M. 1998. Competition refuges and coexistence: an example from Serengeti carnivores. Journal of Animal Ecology 67:370–386. Estes, R. D. 1991. The significance of horns and other male secondary sexual characters in female bovids. Applied Animal Behaviour Science 29:403–451. Fanshawe, J. H., and C. D. FitzGibbon. 1993. Factors influencing the hunting success of an African wild dog pack. Animal Behaviour 45:479–490. FitzGibbon, C. D. 1990a. Why do hunting cheetahs prefer male gazelles? Animal Behaviour 40:837–845. FitzGibbon, C. D. 1990b. Mixed-species grouping in Thomson’s and Grant’s gazelles: the antipredator benefits. Animal Behaviour 39:1116–1126. Geist, V. 1966. The evolution of horn-like organs. Behaviour 27:175–214. Gervasi, V., E. B. Nilsen, and J. D. C. Linnell. 2015. Body mass relationships affect the age structure of predation across carnivore-ungulate systems: a review and synthesis. Mammal Review 45:253–266. Gittleman, J. L. 1985. Carnivore body size: ecological and taxonomic correlates. Oecologia 67:540–554. Hayward, M. W. 2006. Prey preferences of the spotted hyaena (Crocuta crocuta) and degree of dietary overlap with the lion (Panthera leo). Journal of Zoology (London) 270:606–614. Hayward, M. W. 2011. Scarcity in the prey community yields antipredator benefits. Acta Oecologica 37:314–320. Hayward, M. W., P. Henschel, J. O’Brien, M. Hofmeyr, G. Balme, and G. I. H. Kerley. 2006a. Prey preferences of the leopard (Panthera pardus). Journal of Zoology (London) 270:298–313. Hayward, M. W., M. Hofmeyr, J. O’Brien, G. I. H. Kerley, P. Henschel, and G. Balme. 2006b. Prey preferences of the cheetah (Acinonyx jubatus) (Felidae: Carnivora): morphological limitations or the need to capture rapidly consumable prey before kleptoparasites arrive? Journal of Zoology (London) 270:615–627. Hayward, M. W., and G. I. H. Kerley. 2005. Prey preferences of the lion (Panthera leo). Journal of Zoology (London) 267:309–322. Hayward, M. W., J. O’Brien, M. Hofmeyr, and G. I. H. Kerley. 2006c. Prey preference of the African wild dog Lycaon pictus (Canidae: Carnivora): Ecological requirements for conservation. Journal of Mammalogy 87:1122–1131. Hudson, D. J. 1966. Fitting segmented curves whose join points have to be estimated. Journal of the American Statistical Association 61:1097–1129. Jacobs, J. 1974. Quantitative measurements of food selection: a modification of the Forage Ratio and Ivlev’s Electivity Index. Oecologia 14:413–417. Janis, C. 1982. Evolution of horns in ungulates: ecology and paleoecology. Biological Reviews 57:261–318. Johnson, D. H. 1980. The comparison of usage and availability measurements for evaluating resource preference. Ecology 61:65–71. Kruuk, H. 1975. Functional aspects of social hunting in carnivores. Pp. 119–141 in Function and evolution in behavior (G. Baerends, CLEMENTS ET AL.—CHEETAH PREY PREFERENCES927 C. Beer, and A. Manning, eds.). Oxford University Press, Oxford, United Kingdom. Laundré, J. W. 2008. Summer predation rates on ungulate prey by a large keystone predator: how many ungulates does a large predator kill? Journal of Zoology (London) 275:341–348. Laundré, J. W., L. Hernandez, and K. B. Altendorf. 2001. Wolves, elk, and bison: reestablishing the “landscape of fear” in Yellowstone National Park, USA. Canadian Journal of Zoology 79:1401–1409. Laurenson, M. K. 1995. Behavioral costs and constraints of lactation in free-living cheetahs. Animal Behaviour 50:815–826. Macdonald, D. W. 1983. The ecology of carnivore social behaviour. Nature 301:379–384. Mills, M. G. L. 1990. Kalahari hyaenas. Unwin Hyman, London, United Kingdom. Mills, M. G. L. 1992. A comparison of methods used to study food habits of large African carnivores. Pp. 1112–1124 in Wildlife 2001: populations (D. R. McCullough and R. H. Barrett, eds.). Elsevier Science Publishers LTD, London, United Kingdom. Mills, M. G. L., and M. Harvey. 2001. African predators. Struik Publishers, Cape Town, South Africa. Muggeo, V. 2008. Segmented: an R package to fit regression models with broken-line relationships. R News 8:20–25. Owen-Smith, N., and M. G. L. Mills. 2008. Predator-prey size relationships in an African large-mammal food web. Journal of Animal Ecology 77:173–183. Packer, C. 1983. Sexual dimorphism: the horns of African antelopes. Science 221:1191–1193. Packer, C., D. Scheel, and A. E. Pusey. 1990. Why lions form groups: food is not enough. The American Naturalist 136:1–19. Prugh, L. R., C. J. Stoner, C. W. Epps, W. T. Bean, W. J. Ripple, A. S. Laliberte, and J. S. Brashares. 2009. The rise of the mesopredator. Bioscience 59:779–791. Pyke, G. H., H. R. Pulliam, and E. L. Charnov. 1977. Optimal foraging: a selective review of theory and tests. The Quarterly Review of Biology 52:137–154. R Development Core Team. 2013. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Radloff, F. G. T., and J. T. du Toit. 2004. Large predators and their prey in a southern African savanna: a predator’s size determines its prey size range. Journal of Animal Ecology 73:410–423. Rapson, J. A., and R. T. F. Bernard. 2007. Interpreting the diet of lions (Panthera leo); a comparison of various methods of analysis. South African Journal of Wildlife Research 37:179–187. Scheel, D., and C. Packer. 1991. Group hunting behaviour of lions: a search for cooperation. Animal Behaviour 41:467–709. Sinclair, A. R. E., S. Mduma, and J. S. Brashares. 2003. Patterns of predation in a diverse predator-prey system. Nature 425:288–290. Skinner, J. D., and C. T. Chimimba. 2005. The mammals of the southern African subregion. Cambridge University Press, Cambridge, United Kingdom. Stankowich, T., and T. Caro. 2009. Evolution of weaponry in female bovids. Proceedings of the Royal Society of London, B. Biological Sciences 276:4329–4334. Vanak, A., D. Fortin, M. Thaker, and M. Ogden. 2013. Moving to stay in place: behavioral mechanisms for coexistence of African large carnivores. Ecology 94:2619–2631. Zuur, A. F., E. N. Ieno, N. J. Walker, A. A. Saveliev, and G. M. Smith. 2009. Mixed effects models and extensions in ecology with R. Springer Science and Business Media, New York. Submitted 1 June 2015. Accepted 26 January 2016. Associate Editor was I. Suzanne Prange.