Donate Sign up for e-network
CENTER for BIOLOGICAL DIVERSITY Because life is good

Content on this page requires a newer version of Adobe Flash Player.

Get Adobe Flash player


An Occasional Review of Goshawk Research and Conservation Issues

#5: Summary of Finn (2000) by Noah Greenwald

Finn, S.P. 2000. Multi-scale habitat influences on northern goshawk occupancy and reproduction on Washington’s Olympic Peninsula. MS thesis, Boise State University.

Finn (2000) studied the influence of habitat at multiple scales from the nest tree to the home range on the occupancy and reproductive success of 30 historic goshawk nest sites (located between 1975 and 1996) on the Olympic Peninsula, Washington. Forests used by goshawks on the Olympic Peninsula are dominated by sitka spruce and western hemlock in wetter areas and Douglas fir in drier areas. Goshawks on the Olympic Peninsula have darker plumage and may exhibit different foraging behavior than in other parts of North America, making them an interesting subject of study. [Abstractor’s note: The Olympic Peninsula may be within the range of the laingi (i.e. Queen Charlotte) subspecies].

From 1996 to 1998, Finn surveyed all reputably located historic nest sites on the peninsula for occupancy (located on territory in at least one of three years of survey) and reproductive success (number of fledglings produced). Not able to survey all 30 sites in all three years, Finn surveyed 20 of the territories in at least one year and 10 sites in all three years. This sampling regime was deemed adequate based on relative consistency in occupancy and reproduction across the years in the 10 territories.

Occupancy and reproductive success were related to 175 habitat attributes at six spatial scales, including the nest tree (.003 hectares), nest vicinity (.04 hectares), nest stand (9-146 hectares), nest area (38.5 hectares), post-fledging family area (PFA; 176.7 hectares), and home range (1885.5 hectares) using stepwise logistic regression for occupancy (binary) and stepwise multiple regression for reproductive success (continuous). At each scale, biologically and statistically relevant variables were chosen for further modeling by comparing box and whisker plots for occupied and unoccupied nest sites with the goal of choosing five to nine variables. Akaike’s Information Criterion and Mallow's Cp statistics were used to select the best model for variation in reproduction.

Twelve of the historic nest sites were occupied, and 18 were unoccupied. Occupancy was most significantly correlated with habitat attributes at coarse-grained scales, including the nest stand, PFA, and home range. Attributes of the nest tree, nest vicinity, and nest area were not useful predictors of occupancy. Goshawks were more likely to occupy nest sites with low percent shrub cover (mean = 19.0 percent, SE = 4.2) and a deep overstory canopy depth (mean = 28.7 meters, SE = 1.8; canopy depth is created by multiple canopy layers and was positively correlated with high canopy closure) in the nest stand and less non-forest cover (90 percent of non-forest cover was harvest areas) and heterogeneity in the home range. A model including an intercept, stand shrub cover in the nest stand, and an interaction term of non-forest habitat and a contrast index, which measures the heterogeneity of different habitat types in a specified area, correctly predicted goshawk occupancy at 82.9 percent of the nest sites (Wald X2 = 6.85, P = .033, Hosmer and Lemeshow’s X2 = 5.5, P = .70, df = 8). Goshawks were unlikely to occupy a nest site if non-forest habitat exceeded 20 percent in the home range and 15 percent of the PFA. If contrast was held constant, the likelihood of goshawk occupancy decreased by 20 percent for every 2 percent increase of non-forest cover in the home range. Similarly, if non-forest cover was held constant at 15 percent, likelihood of occupancy decreased by 25 percent for every 0.1 increase in the contrast index. Late seral forest was consistently > 40 percent of the landscape surrounding nest sites at all scales and would have contributed to a significant model of goshawk occupancy at the home range scale. Finn, however, believed non-forest cover, which was highly correlated with late seral forest cover, was a more “useful” explanatory variable because it provided “management targets and considered natural disturbance and permanent non-forest cover classes.”

Eight of the occupied nest sites produced at least one fledgling over the three years of study, with a total of 24 fledglings documented. Reproductive success was most significantly correlated to habitat attributes in the nest vicinity and nest stand. Reduced percent shrub cover (mean = 15.6 percent, SE = 8.2) in the nest vicinity and increasing forest decadence in the nest stand, measured by the number of snags (mean = 40 snags >= 15.2 dbh/hectare, SE = 6.7) and length of coarse woody debris (mean = 11.5 meters, SE = 1.3), were correlated with increased reproductive success (F(2,9) = 38.8, P = .0002, Adjusted R2 = .89).

These findings indicate that both goshawk occupancy and reproductive success are positively related to habitat attributes associated with late seral forests, such as forest decadence (snags and coarse woody debris), overstory canopy depth, and amount of late seral forest, and negatively associated with attributes associated with timber harvest. Goshawk occupancy was negatively correlated with the amount of non-forest cover and contrast — both primarily created by logging — in the PFA and home range. Finn concludes:

“Forest managers on the Olympic Peninsula can promote goshawk occupancy and reproduction by limiting the amount of non-forest cover (<20 percent) and heterogeneity (contrast index <1.0) in the landscape and by maintaining potential nest stands (>=38.5 hectares) having deep overstory canopies (>=25 meters), and reduced shrub cover (preferably <=20 percent).”

Photo © Robin Silver