Spatial processes decouple management from objectives in a heterogeneous landscape: predator control as a case study.

ABSTRACT

Predator control is often implemented with the intent of disrupting top-down regulation in sensitive prey populations. However, ambiguity surrounding the efficacy of predator management, as well as the strength of top-down effects of predators in general, is often exacerbated by the spatially implicit analytical approaches used in assessing data with explicit spatial structure. Here, we highlight the importance of considering spatial context in the case of a predator control study in south-central Utah. We assessed the spatial match between aerial removal risk in coyotes (Canis latrans) and mule deer (Odocoileus hemionus) resource selection during parturition using a spatially explicit, multi-level Bayesian model. With our model, we were able to evaluate spatial congruence between management action (i.e., coyote removal) and objective (i.e., parturient deer site selection) at two distinct scales the level of the management unit and the individual coyote removal. In the case of the former, our results indicated substantial spatial heterogeneity in expected congruence between removal risk and parturient deer site selection across large areas, and is a reflection of logistical constraints acting on the management strategy and differences in space use between the two species. At the level of the individual removal, we demonstrated that the potential management benefits of a removed coyote were highly variable across all individuals removed and in many cases, spatially distinct from parturient deer resource selection. Our methods and results provide a means of evaluating where we might anticipate an impact of predator control, while emphasizing the need to weight individual removals based on spatial proximity to management objectives in any assessment of large-scale predator control. Although we highlight the importance of spatial context in assessments of predator control strategy, we believe our methods are readily generalizable in any management or large-scale experimental framework where spatial context is likely an important driver of outcomes.