COS 161-6 - Hierarchical habitat utilization: a case study in eastern copperheads (Agkistrodon contortrix) reveals a snake in the grass for wildlife resource utilization models

Abstract: Ecologists recognize that studies accounting for the hierarchical (i.e., multi-level or multi-scale) nature of wildlife-resource relationships yield more accurate conclusions about how animals select habitat. Even so, these relationships are often only analyzed at the individual (e.g., selection of home ranges, component microhabitats, or both) or population scale (e.g., associating animal density with habitat attributes). In species with periodic use of or congregation in shared habitats, patterns of habitat use modeled at only the individual or population level could vary dramatically leading to potentially inaccurate recommendations for appropriate management. Individual-level models may fail to detect the importance of resources needed only occasionally or briefly by any one individual, even when those resources are used by a large proportion of the population (e.g., watering holes, communal hibernacula, seasonal breeding wetlands, etc.). Likewise, population-level models may successfully characterize conditions of shared resources but fail to detect important resources found in individual home ranges (e.g., foraging habitats). We used data from four years of radio-tracking eastern copperheads (Agkistrodon contortrix), a seasonally gregarious North American pit viper, to test for the presence of utilization hierarchies across individual and population levels. We modeled individual- and population-level habitat utilization using approaches analogous to many single-level studies in the spatial ecological literature. We then compared models to test for the presence of hierarchical utilization. As predicted, population-level models amplified characteristics of communal hibernacula and spring basking sites (forested habitats with ample rock), while individual-level models amplified characteristics of individual summer sites (open habitats without rock). Our multi-level comparison showed the potential for single-level utilization models to produce opposing inferences, which could limit the efficacy of conservation strategies based on wildlife activity modeled at a single level. Recognizing that wildlife can exhibit resource utilization hierarchies, and that level dependencies can change or even reverse model outputs, presents both a challenge and an additional tool for wildlife ecologists.