Refinement of the Arc-Habcap model to predict habitat effectiveness for elk

Abstract

Wildlife habitat modeling is increasingly important for managers who need to assess the effects of land management activities. We evaluated the performance of a spatially explicit deterministic habitat model (Arc-Habcap) that predicts habitat effectiveness for elk. We used five years of radio-telemetry locations of elk from Custer State Park (CSP), South Dakota, to test predicted habitat effectiveness by the model. Arc-Habcap forage and coverforage proximity components predicted elk distribution in CSP. However, the cover component failed to predict elk distribution in CSP. Habitat effectiveness calculated as the geometric mean of the model components failed to predict elk distribution and resulted in under-utilization of habitats predicted to be good and over-utilization of habitats predicted to be poor. We developed a new formula to calculate habitat effectiveness as an arithmetic average of the model components that weighted forage more than cover or cover-forage proximity. The new formula predicted actual elk distribution across categories of habitat effectiveness. Elk selected cover and forage areas ≤100 m from cover-forage edges. Arc-Habcap predicted that areas adjacent to roads were not usable by elk. Elk used areas adjacent to primary roads, but use was less than the proportional area comprised for primary roads, and about equal to proportional area adjacent to secondary roads and primitive roads. All sapling/pole and mature structural stages of ponderosa pine (Pinus ponderosa) were considered as both forage and cover by Arc-Habcap and consequently considered optimal in the cover-forage model component. We suggested revisions for both the cover-forage proximity component and areas adjacent to roads.