Prey specialization by cougars on feral horses in a desert environment.
Natural controls on the distribution, abundance, or growth rates of exotic species are a desirable mode of intervention because of lower costs compared to anthropogenic controls and greater social acceptance. In the Great Basin, cougars (Puma concolor) are the most widely distributed carnivore capable of killing large ungulate prey. Populations of feral horses (Equus ferus) are widely distributed throughout the Great Basin and can grow at rates up to 20%/year. Although cougars exhibit distributional overlap with horses, it has been assumed that predation is minimal because of differences in habitat use and body-size limitations. To evaluate this hypothesis, we monitored the diets of 21 global positioning system (GPS)-collared cougars in the western Great Basin (5 males, 8 females) and eastern Sierra Nevada (2 males, 6 females) from 2009-2012. We investigated 1,310 potential kill sites and located prey remains of 820 predation events. We compared prey composition and kill rates of cougars inhabiting the Sierra Nevada and Great Basin, and among male and female cougars across seasons. We used generalized linear mixed models (GLMMs) to examine the effects of prey availability and habitat characteristics on the probability of predation on horses by cougars. Mule deer (Odocoileus hemionus) comprised 91% of prey items killed on the Sierra Nevada reference site but only comprised 29% of prey items in the Great Basin study area. Average annual kill rates for deer differed between the Sierra Nevada (x = 0.85 deer/week, range = 0.44-1.3) and Great Basin (x = 0.21 deer/week, range = 0.00-0.43). Diets of cougars in the Great Basin were composed predominantly of horses (59.6%, n = 460 prey items; 13 individuals). Ten cougars regularly consumed horses, and horses were the most abundant prey in the diet of 8 additional individuals in the Great Basin. Cougars on average killed 0.38 horses/week in the Great Basin (range=0.00-0.94 horses/week). Differences in predation on horses between the sexes of cougars were striking; Great Basin females incorporated more horses across all age classes year-round, whereas male cougars tended to exploit neonatal young during spring and summer before switching to deer during winter. Within GLMM models, the probability of predation on horses compared to other prey species increased with elevation, horse density, and decreasing density of mule deer on the landscape, and was more likely to occur in sagebrush (Artemesia spp.) than in pinyon (Pinus monophylla)-juniper (Juniperus osteosperma) forests. Behavior of individual cougars accounted for more than a third of the variation explained by our top models predicting predation on horses in the Great Basin. At landscape scales, cougar predation is unlikely to limit the growth of feral horse populations. In the Great Basin ecosystem, however, cougars of both sexes successfully preyed on horses of all age classes. Moreover, some reproductive, female cougars were almost entirely dependent on feral horses year-round. Taken together, our data suggest that cougars may be an effective predator of feral horses, and that some of our previous assumptions about this relationship should be reevaluated and integrated into management and planning.