Field evaluation of a habitat-relation model for the American marten

Monitoring programs aimed at indicator species like the American marten (Martes americana) may index ecosystem conditions throughout the animal's range. Habitat-relation models may be useful to monitor indicator species and their habitats temporally. We constructed and field-tested a habitat-relation model for American martens to determine current distribution of the species in the Black Hills, South Dakota and establish baseline data for the population. We conducted a track-plate-box survey, selecting survey quadrats using the model, field reconnaissance, and area used by radiocollared martens. We performed stepwise logistic regression to test the model and predict the presence of martens in random quadrats throughout the study area and in quadrats surveyed by the United States Department of Agriculture (USDA) Forest Service. Of 46 10.2-km 2 quadrats surveyed, 14 (30%) had boxes visited by martens. High-quality habitat was positively and low-quality habitat negatively correlated to martens' presence. Logistic model coefficients predicted that martens would be detected in 6 of 68 (9%) random quadrats. When field-tested (post-analysis), the model correctly predicted outcomes for 3 (60%) of 5 quadrats surveyed.