New methods and hierarchical models for estimating intensity of diurnal habitat use by merriam’s turkeys in managed forests of northern Arizona

The response of native wildlife species to habitat change and the increased pace of large-scale forest restoration and fuels reduction treatments in managed forests of the southwestern United States is not well studied. Thus, there is a need to develop contemporary statistical models for evaluating the effect of these activities on animal populations prior to extensive and rapid habitat modification while accounting for small sample sizes typical of many field efforts. We used a maximum likelihood-based modeling framework to estimate seasonal intensity of diurnal habitat use by a population of Merriam’s turkeys (Meleagris gallopavo merriami) occupying a ponderosa pine (Pinus ponderosa) forest in northern Arizona. Using an information-theoretic approach and multimodel inference, we fitted linear mixed models that were capable of accommodating high-resolution global positioning system data on turkeys. We accounted for spatial and temporal structuring in these locational data and incorporated multiple random effects. We considered a modest number (12) of key habitat variables (i.e., fixed effects) in these models that were specified a priori and included remote sensing-derived, high-resolution data on forest structure. On average, fixed kernel estimates of summer home range and core area use were larger than winter estimates. Relationships between the intensity of use\and habitat variables were different in each season, and digitally derived estimates of forest structure and vegetation often were strong predictors. In both seasons turkeys tended to use areas closer to small canopy gaps but not use areas with high densities of forest edge or that were proximate to tertiary roads. Low and intermediate values of the normalized difference vegetation index, derived as a surrogate measure of the amount of deciduous or herbaceous understory vegetation on a site, were also important predictors of use in the summer and winter respectively. Individuals showed greater habitat use in canyon bottoms and steeper terrain at higher elevations, as well as areas closer to water in the summer. During the winter season, we detected a strong positive relationship with areas that were thinned or burned in previous years. Our results provide insight into which forest or habitat parameters are contributing to the intensity of habitat use by Merriam’s turkeys and will help to mitigate the adverse timing or targeting of large-scale forest restoration and fuels reduction treatments. Our modeling approach is highly transferable to the management and conservation of other species and ecosystems.