90x90 m is an optimal spatial scale to study the habitat use by hazel grouse

This study reports an analysis at three different spatial scales with the aim at identifying the most explicative scale to investigate the role of management, stand structure and ground layer on the habitat of hazel grouse (Bonasa bonasia L.). Three locations with increasing density of nesting sites were selected in the Trudner Horn Park, Bozen Province (Northern Italy). In each site a lattice regular grid of 100 30 730m cells was sampled, providing that it contained at least one nesting site. Within each cell we drew two orthogonal transects. On each 1m segment we recorded herbaceous and canopy cover and we counted hazel grouse presence indices. On each 5m segment we recorded stand structure. In the whole cell we measured woody species cover and top height, evidence of cuttings, Rufa ant nests and other management variables. The effects of management, stand structure and ground layer sets of explanatory variables on presence/absence and on the count of presence indices were analyzed using generalized linear mixed models (GLMM). Separate GLMM analyses were performed within each of the three sets of variables and for each spatial scale (1 71, 2 72 and 3 73 grid cells). The significant variables were then further analyzed by variation partitioning (VP) to determine the unique and joint fractions of variation explained by the variables for each spatial scales. At the largest spatial scale both the presence/absence and the count models showed higher explained variation (88-43% respectively) than at small (19-10%) and intermediate (63-40%) spatial scale. While the pure effect of single sets of explanatory variables could be higher at smaller spatial scale, the joint variation shared between the three groups was the highest in the 3 73 grid cells (90 790m) scale, where the largest part of the explained variation in the presence/absence model was related to joint variation between ground layer and stand structure (30%) and between the three groups (29%)