Large, mobile organisms can move swiftly through their environment and probably select resources, from those available, at a relatively coarse grain. The northern goshawk (Accipiter gentilis), a large mobile raptor, is known as a forest generalist throughout its holarctic range, yet individual populations seem to prefer specific habitats in a given region. To determine goshawk habitat relationships and the spatial scales at which they operate I surveyed all known historical nest sites on the Olympic Peninsula, Washington (N = 30) and compared goshawk occupancy and reproductive success at these sites to existing habitat conditions at a range of spatial scales that might be important to a breeding pair. A historical nest site was defined as a location that contained a large stick nest and was occupied by an adult goshawk at least one year between 1975-1995. 1 estimated goshawk occupancy one year at each historical nest site, from 1996-1998, using standardized aural broadcast surveys within a 170-314 ha circular plot surrounding the most recently built nest. I assessed among-year variation in occupancy by surveying a subset of 10 historical nest sites each of the three years. Occupancy remained consistent among years at these 10 sites (Fisher’s Exact Tests, P < 0.07) so I characterized all 30 historical sites as ‘occupied’ or ‘unoccupied’ based on one year surveys. Reproductive success at occupied historical sites was estimated during two visits to the site during the post-fledging period. The maximum number of fledged young observed was used to index the relative success of each breeding pair.
I characterized occupancy and reproductive success at historical sites in terms of existing habitat condition by comparing them to 175 individual habitat attributes at six spatial scales – the nest tree (0.003 ha), nest vicinity (0.04 ha), nest stand (9 – 146 ha), nest area (38.5 ha), Post-fledging Family Area (PFA; 176.7 ha), and home range (1885.5 ha) – surrounding the historical nest sites. I measured attributes at the first three scales using ground-based silvicultural descriptions of vegetation and topographic structure. Conditions at the latter three scales describe habitat-patch dynamics within the broader landscape using a Geographic Information System. These six scales encompass the spatial range of resources available to a breeding pair of goshawks as determined by my field observations or by published estimates of goshawk space-use. I compared each discrete spatial scale to goshawk occupancy and reproductive success individually and then examined cross-scale habitat influences on goshawks using stepwise logistic regression for the binary response variable (occupancy) and linear multiple regression for the continuous response variable (reproductive success). Habitat attributes at the coarse-grained scales (i.e., the nest stand, PFA, and home range) provided the most significant correlates of goshawk occupancy although reproductive success was more strongly associated with finer-scale conditions. Habitats attributes in the nest tree (0.003 ha), nest vicinity (0.04 ha), and nest area (38.5 ha) were not useful predictors of goshawk occupancy. Goshawks were more likely to occupy historical nest sites with a deep (x = 28.7 m, SE = 1. 8, N = 12) overstory canopy depth and low percent shrub cover (x = 19.0%, SE = 4.2) in the nest stand (9 – 146 ha) and less non-forest cover (x = 11.2%, SE = 3.0) and heterogeneity (contrast index, x = 0.9, SE = 0.4) in the home range (1885.5 ha). Reproductive success at occupied sites was most strongly correlated with reduced percent shrub cover (x = 15.6%, SE = 8.2, N = 8) in the nest vicinity and increasing forest decadence (x = 40.0 snags (³ 15.2 dbh)/ha, SE = 6.7; x = 11.5-m long coarse woody debris, SE = 1.3) in the nest stand. A selection of these variables combined to provide significant models of goshawk occupancy (logit [occupancy] = 3.78 – 0.061(stand shrub cover) – 0.116(home range non-forest cover home range contrast index); Wald X2 = 6.85, P = 0.033, df = 2) and reproductive success (success = 1.28 + 0.015(stand snag density) – 0.027(vicinity shrub cover); F(2,9) = 38.8, P = 0.0002, Adj. R2 = 0. 89).
Forest managers on the Olympic Peninsula can promote goshawk occupancy and reproduction by limiting the amount of non-forest cover (< 20%) and heterogeneity (contrast index < 1.0) in the landscape and by maintaining potential nest stands (³ 38.5 ha) having deep overstory canopies (³ 25 in), and reduced shrub cover (preferably < 20%). The landscape targets can be met by considering pre-existing landscape conditions when planning timber harvest prescriptions. Micro-scale overstory canopy and percent shrub cover targets may be easily achieved by specific planting and tree-thinning prescriptions, which can be incorporated into current forest management activities. Because habitat conditions influence goshawk historical site occupancy and reproductive success differently, both descriptors of goshawk demography must be monitored when testing the utility of these models.