Determination of critical habitat for the endangered Nelson's bighorn sheep in southern California

The United States Fish and Wildlife Service's (USFWS) designation of critical habitat for the endangered Nelson's bighorn sheep (Ovis canadensis nelsoni) in the Peninsular Ranges of southern California has been controversial because of an absence of a quantitative, repeatable scientific approach to the designation of critical habitat. We used 12,411 locations of Nelson's bighorn sheep collected from 1984-1998 to evaluate habitat use within 398 km2 of the USFWS-designated critical habitat in the northern Santa Rosa Mountains, Riverside County, California. We developed a multiple logistic regression model to evaluate and predict the probability of bighorn use versus non-use of native landscapes. Habitat predictor variables included elevation, slope, ruggedness, slope aspect, proximity to water, and distance from minimum expanses of escape habitat. We used Earth Resources Data Analysis System Geographic Information System (ERDAS-GIS) software to view, retrieve, and format predictor values for input to the Statistical Analysis Systems (SAS) software. To adequately account for habitat landscape diversity, we carried out an unsupervised classification at the outset of data inquiry using a maximum-likelihood clustering scheme implemented in ERDAS. We used the strata resulting from the unsupervised classification in a stratified random sampling scheme to minimize data loads required for model development. Based on 5 predictor variables, the habitat model correctly classified >96% of observed bighorn sheep locations. Proximity to perennial water was the best predictor variable. Ninety-seven percent of the observations were within 3 km of perennial water. Exercising the model over the northern Santa Rosa Mountain study area provided probabilities of bighorn use at a 30 x 30-m2 pixel level. Within the 398 km 2 of USFWS-designated critical habitat, only 34% had a graded probability of bighorn use to non-use ranging from ???1:1 to 6,044:1. The remaining 66% of the study area had odds of having bighorn use <1:1 or it was more likely not to be used by bighorn sheep. The USFWS designation of critical habitat included areas (45 km2) of importance (2.5 to ???40 observations per km2 per year) to Nelson's bighorn sheep and large landscapes (353 km2) that do not appear to be used (<1 observation per km2 per year).