Development and validation of scale-dependent habitat models for cavity-nesting birds in Fishlake National Forest, Utah

I examined the relative ability of landscape-scale, stand-scale, nest-tree scale, and multi-scale habitat models to predict nest presence for cavity-nesting birds on Fishlake National Forest, Utah. Using data from 2001, I built stepwise logistic regression models for cavity nesting birds and two cavity-nesting sub-guilds. I validated the habitat models using both internal validation and external validation. Cavity-nesting birds were associated with habitat at multiple spatial scales. Multi-scale and micro-scale models were the best predictors of nest-site presence. Model accuracy decreased as spatial extent increased, and model accuracy was similar for both internal and external validation. The results highlight the importance of the nest tree for cavity-nesting birds. Macro-scale models, which are inexpensive and spatially explicit, cannot currently be used to accurately predict nest presence for cavity nesting birds in aspen forests of southern Utah.