Effects of landscape and local habitat features on bird communities : a study of an urban gradient in Greater Vancouver

Bird species diversity and abundance trends in urban areas can provide evidence to predict the relative importance of local bird habitat under different landscape contexts. I examined the hypothesis that selected species and nesting guilds should be more closely associated with landscape level features, such as proximity to large forested areas (< 100 ha), than with local scale habitat measures (< 1 ha). I collected avian community data during surveys completed over a two year period at 285 point count stations along four linear road transects located in Vancouver and Burnaby, British Columbia. Stations were located along transects bisecting three large parks (>324 ha) and proceeding away from these parks along residential streets into highly urban and suburban areas. A total of 49 breeding bird species were observed including 36 common species and 13 species that were sighted only once. Canonical correspondence analysis was used to view the main associations between measured habitat variables and species distributions. Species richness declined with increasing urbanization and the gradient (CCA axis one site scores) was dominated by landscape level habitat measures. Park area-by-distance metrics, developed using G.I.S., had the highest correlation with CCA axis one indicating the importance of park area in the vicinity for many species of birds breeding in marginal residential areas. Different land use zones did not neatly separate the urbanization gradient into simple bird habitat categories. Habitat models were created for five nesting guilds and three selected species (Song Sparrow, Melospiza melodia, Spotted Towhee, Pipilo maculatus, and American Robin, Turdus migratorius) with sequential block addition of landscape and local variables. Local variables significantly improved predictability of landscape variable only models, but the difference was slight. Landscape variables alone were often good predictors of presence or absence of most groups of species (guilds), but were less sensitive than local variables at predicting individual species presence. Incidence (percent stations occupied) of several bird species increased with park area in the vicinity as an inverse function of distance. The results of this study suggest that matrix areas surrounding parks and reserves should be integrated into urban planning and development designs.Forestry, Faculty ofGraduat