Describing habitat suitability of bobcats (Lynx rufus) using several sources of information obtained at multiple spatial scales

We investigated habitat use of bobcats (Lynx rufus) across spatial scales in New Hampshire, USA, by integrating independent sources of information into one statewide model of habitat suitability. First, statewide distribution of bobcats (second-order selection) was estimated using observations of bobcats solicited from the public. Probability functions indicated that areas where bobcats were not detected (no sightings) were characterized by high elevations and deep snow during winter. That pattern was corroborated by camera traps that were distributed among high and low-elevation landscapes. Second, bobcats equipped with GPS collars in two study areas were used to determine habitat use within established home ranges (third-order selection). Resource-selection probability functions developed from locations of marked bobcats indicated they selected forests, shrub/scrub, and wetlands, and avoided developed areas, agricultural areas, and open water relative to availability. Bobcats also avoided areas with high road densities and selected areas close to forest edges, and preferred rugged south-facing slopes. Finally, these measures of bobcat distribution were combined to obtain fine-scale estimates of bobcat habitat suitability across New Hampshire. A comparison of models of habitat suitability to an index of bobcat abundance (effort-corrected detections of bobcats by hunter) indicated that the hierarchical model (second-order multiplied by third-order of selection) provided a better description of statewide bobcat habitat than either single-scale analyses. As a result, we recommend caution when extrapolating information on the distribution of a species obtained at a limited spatial scale.