An approach to identify and levels of nesting habitat selection: a cross-scale analysis of Goshawk preferences

Abstract

Nest site selection may result from diverse agents that work on different spatial scales, and amultilevel analysis can offer a more realistic vision of the search image (proximate factors) and of the scale of nesting habitat selection by a species. Cross-scale analyses of nesting habitat preferences are scarce : in this respect, the Goshawk (Accipiter gentilis) is one of the most intensely investigated species, but most studies have focused only on a single habitat feature. This study, conducted in Burgundy (Eastern France), describes Goshawk nest site preferences by using a multilevel spatial scale analysis (nest tree level, stand level, landscape level), and identifies the habitat level that represents the supposed proximate factor for this species. We identified 57 active Goshawk nest sites (6 .7 active nests/ 100 km2) . The stepwise logistic regression showed that 4 variables of the nest stand structure (high tree dbhs, high crown volumes, high flight space and short distance to trails) and 2 variables of the landscape surrounding nest trees (low avian prey richness for both 100-500 and 501-2000 g classes) were significant predictors of Goshawk nest site selection (98% ofnests correctly reclassified). We hypothesised that Goshawk species choose nest sites on the basis of the overall structural features of the stand, and then focus on a particular nest tree .