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

#7: Summary of Stephens (2001) by Noah Greenwald

Stephens, R.M. 2001. Migration, habitat use, and diet of northern goshawks that winter in the Uinta Mountains, Utah. Thesis, University of Wyoming, Laramie.

Stephens studied the winter ecology of northern goshawks in the Uinta Mountains of Utah, including migration, habitat use at landscape and stand scales, core range size, prey species, and mortality. Eighteen goshawks were monitored using radio-telemetry during the winters of 1998/99 and 1999/2000.

Migration. Stephens classified goshawks as migratory if their winter range did not include the nest stand. Distance migrated was measured as the furthest distance a goshawk was located from its nest. Eleven of 15 goshawks migrated, including two of five males and nine of 10 females. Goshawks migrated an average of 55 km, with males averaging 15 km and females averaging 68 km. Nine of the goshawks migrated to lower elevations, in several cases utilizing pinyon/juniper woodlands and in one case a cottonwood riparian corridor surrounded by salt-desert scrub. Fall migration occurred from late October to late December, with birds returning to their nesting territories from mid-January to late March.

Habitat use. At the landscape scale, Stephens compared mean patch density/km2, mean number of vegetation types/km2, mean length of forest edge (km/km2), and dominant vegetation between 12 goshawk core ranges and 12 random ranges of equal size. Mean patch density, mean number of vegetation types, and mean length of forest edge were all greater in goshawk ranges than random areas, but none showed significant differences. Lack of significant differences may be due to small sample size, error in the vegetation classifications, or lack of significant selection for these variables by northern goshawks. A power analysis indicated that all three tests had a 32 percent or less chance of detecting a significant difference.

Stephens identified four habitat types utilized by wintering goshawks, including mixed conifer forest comprised of lodgepole pine, subalpine fir, and Douglas fir; pinyon/juniper woodlands; a combination of mixed conifer and woodland; and lowland riparian. Four goshawks utilized mixed conifer, four utilized woodland, three utilized combined mixed conifer and woodland, and one utilized lowland riparian. Mixed conifer forests were present in "much higher" percentages in goshawk core ranges than in the study area. Although present in only a small number of core ranges, oak, ponderosa pine, and lowland riparian were also found in higher percentages within core ranges than in the study area. Pinyon/juniper woodlands and agricultural areas were found in equal proportions, and sagebrush, aspen, dry meadow, and salt-desert scrub were all found in lower proportions in goshawk core ranges compared to the landscape.

At the stand scale, Stephens compared canopy closure, tree density, DBH, PJ canopy (measures the size of individual tree crowns as a surrogate for DBH), tree height, and sapling and shrub density between goshawk and random locations. Using aerial and ground telemetry, 138 individual goshawk locations were identified, 134 of which were in forested environments. By comparison, only 103 of 138 random plots were in forested environments, suggesting overall selection for forests. For analysis of individual stand traits, non-forested plots were excluded. Goshawk locations had significantly greater canopy closure than random locations, using a significance level of .015. Stephens concluded that no other variables were significant. However, tree density was significantly greater at goshawk locations than random locations, at a significance level of .05 (p = .034). Stephens did not explain why he used the more conservative significance level. Levels of canopy closure or tree density at goshawk locations were not provided.

Similar to other studies (e.g. Beier and Drennan 1997, Good 1998), Stephens also identified locations where goshawks made a kill. This was accomplished by fitting goshawks with posture-sensitive switches that altered the signal pulse rate to allow the researchers to identify when goshawks were flying or perching. When goshawks were observed to be potentially feeding, the researchers walked to the bird's location and attempted to locate prey remains. Stand traits of kill and random locations did not significantly differ at a significance level of .015. Using a significance of .05, however, both canopy closure and tree DBH were significantly greater at kill sites than random sites.

Core Range Size. Using a Kernal estimator in ARCVIEW 3.2, Stephens calculated 50-percent polygons as an estimate of goshawk core ranges. Core ranges varied from 1.0 to 79.5 km2 with a mean of 25.8 km2 +/- 25.3 km2. Stephens notes that the large variance in core range size was in part due to three goshawks with large home ranges that were fragmented by agriculture. Stephens concludes: "These large winter ranges were likely a function of fragmentation and prey availability. Goshawks in agricultural landscapes probably had to cover larger amounts of land to make an adequate amount of kills."

Prey species. Using the methods described above, Stephens identified 23 prey individuals, the majority of which were pine squirrels (Tamiasciurus hudsonicus, n = 9) and cottontails (Syvilagus sp., n = 9). Goshawks primarily hunted pine squirrels in mixed conifer forests and cottontails in pinyon/juniper woodlands. During winter, pine squirrels are one of the few species that doesn't hibernate. Thus, goshawks either remained in mixed conifer forests and specialized on squirrels or migrated to lower elevations and specialized on cottontails. Other prey included two black-tailed jackrabbits (Lepus californicus), a ruffed grouse (Bonasa umbellus), a snowshoe hare (Lepus americanus), and a starling (Sturnus vulgaris).

Mortality. Stephens recovered six dead goshawks, including three males and three females, and attempted to determine cause of death. Hypothesized causes of mortality included predation in two instances, collision in one instance, old-age in one instance, and starvation in two instances. The two instances of predation were believed to be caused by a pine marten, based on tracks, and a golden eagle, based on its occurrence in the area.


Photo © Robin Silver