Incorporating preferential prey selection and stochastic predation into population viability analysis for rare prey species

There is increasing evidence that predation can cause the decline and extinction of small populations of prey, and that stochastic predation resulting from variation in prey selection by individual predators can have significant consequences for population persistence. Modelling approaches that ignore variation in prey selection exhibited by individual predators may inaccurately predict the effect of predation on prey populations, especially over longer time scales. We assess the impacts of variation in prey selection by building PVA models for endangered huemul deer Hippocamelus bisulcus that sequentially include and exclude observed stochasticity in predation among individual pumas Puma concolor. Our results indicated that huemul are at risk of extinction in all scenarios modelled, although the immediacy of this risk differed based on model structure and time period considered. Specifically, modelling predation as a random effect based on an interrupted Poisson process rather than as a directional and continuous change in survival rates, resulted in significantly longer estimates of time to extinction independent of the assumed intensity of predation. Our results highlight the importance of determining whether specialist predators are driving predation on rare prey, and when they are, incorporating said stochastic predation when attempting to predict persistence probabilities of rare prey using PVA models. Since results of PVA models are commonly used to develop conservation strategies, we advocate for the inclusion of stochastic predation in future PVA models where warranted to more accurately inform strategies for the conservation of rare prey and their predators.