Modeling avian habitat selection at multiple spatial scales.
Kroll, Andrew*,1, 2, Haufler, Jonathan 2, 1 Wildlife Biology Program, Missoula, MT, USA2 Ecosystem Management Research Institute, Seeley Lake, MT, USA
ABSTRACT- Natural selection should favor individuals who choose habitat features that enhance survival and reproductive success, leading to non-random patterns of habitat selection. We modeled hypotheses for habitat selection by the Dusky Flycatcher (Empidonax oberholseri) at three spatial scales (territory, nest patch, nest site). Territories (n=107) and nests (n=312) were monitored during 2002 and 2003 at 8 grand fir (Abies grandis) sites in central Idaho. We used multiple logistic regression to model potential mechanisms influencing habitat selection in open-cup nesting passerines and results from previous studies of habitat selection by the Dusky Flycatcher. We used information-theoretic (AIC) and k-fold cross-validation to identify the best models and to determine if the choice of a specific model selection criterion changed inferences about the mechanisms influencing habitat selection. Habitat features were selected at each scale, but we detected no differences between successful and unsuccessful territories. Models representing the predation reduction hypothesis (i.e., habitat features selected to reduce nest predation) had the most support at the nest site and patch scales (6 of the 10 best models). However, hypotheses concerning habitat selection are difficult to distinguish because the same variables may be significant in models for competing hypotheses at the same scale. Also, model selection criteria did not support the same models at the same scales, a crucial result given the increasing use of model selection procedures in ecological studies. Research that fails to examine the hierarchical nature of habitat selection may misinterpret how different mechanisms influence selection decisions and affect reproductive success and survival.
Key words: habitat selection, scale, reproductive success, modeling
All materials copyright The Ecological Society of America (ESA), and may not be used without written permission.