Boreal forest landbirds in relation to forest composition, structure, and landscape: implications for forest management

Abstract

We examined a landbird community and its relationship to environmental variables within the boreal forest in north–central Ontario to evaluate its potential usefulness as an indicator of sustainable forest management. Our study had two components. First, we compared bird assemblages in mature forest stands inside Pukaskwa National Park (n = 17) with similar forested stands in a logged landscape (n = 18) over 3 years. We found significant separation of sites in the two treatments based on an ANOSIM (analogue of the standard univariate one-way ANOVA test) of the bird communities (R = 0.238, p < 0.001). We identified four significant indicators of the park landscape (bay-breasted warbler ( Dendroica castanea (Wilson, 1810)), black-throated green warbler ( Dendroica virens (J.F. Gmelin, 1789)), ovenbird ( Seiurus aurocapillus (L., 1766)), and red-eyed vireo ( Vireo olivaceus (L., 1766))) and five indicators of the logged landscape (black-backed woodpecker ( Picoides arcticus (Swainson, 1832)), brown creeper ( Certhia americana Bonaparte, 1838), winter wren ( Troglodytes troglodytes (L., 1758)), white-throated sparrow ( Zonotrichia albicollis (J.F. Gmelin, 1789)), and yellow-bellied sapsucker ( Sphyrapicus varius (L., 1776))). Some relationships were attributable to differences in vegetation, whereas other differences were attributable to the landscape context. Second, we used generalized additive models to examine the relationship of individual species with four sets of environmental data (understorey floristics, forest structure, overstorey composition, and landscape context) using the 35 sites noted above and 18 additional mature forest sites in the logged landscape (n = 53). We found that all four types of variables were frequently included in the best model based on Akaike's information criterion (AIC) (structure in 23 models, landscape in 20 models, overstorey in 19 models, and understorey in 15 models). We discuss our results in terms of their implications to forest management and note that our ability to map habitat for forest birds is substantially compromised by the lack of good spatial estimates of environmental variables that describe bird habitat.