Capturing cross-scalar variation of habitat selection with grid sampling: an example with hazel grouse (Tetrastes bonasia L.)

Abstract

The objective of this study was to investigate a grid-based sampling design to determine the cross-scalar selection of habitat by a territorial animal species: the hazel grouse (Tetrastes bonasia L.). In each of three sites with increasing hazel grouse nest site density, three lattice grids were used to measure both the habitat variables and the species occurrence in 100 30 × 30 m cells. We calculated the average values for habitat variables, as well as use versus non-use by the species, at three spatial scales: small (1 × 1 cell), intermediate (2 × 2 cells) and large (3 × 3 cells). Generalised linear mixed models were integrated into a method of variation and hierarchical partitioning and used to assess the relationship between the habitat variables and the species preferences at each scale. In all scales, species selection was associated with ground layer composition. Selection was also associated with the composition of the woody layer and negatively associated with dominance of tor grass (Brachypodium rupestre (Host) Roem. & Schult.) at the two larger scales. Both litter cover and thinning contributed positively to the habitat selection at the two smaller scales. The other variables were significant only at one scale or explained a relatively low proportion of the variation at multiple scales. Neither the management nor the stand structure variables played a significant independent role across scales when compared with ground layer variables. The total variation explained was highest (ca. 90 %) at the large scale. This finding indicates the possibility of obtaining cross-scalar hazel grouse preferences from grid-based sampling, provided that spatial autocorrelation in the data is handled appropriately.