The chimney swift is a small long-distance migrant whose range expanded dramatically with European settlement and the increase in artificial nest structures (i.e., chimneys) that followed (Cink and Collins 2002). Prior to European settlement, the species was probably thinly distributed and relied on tree cavities for nesting. Nesting in trees is now rare (Graves 2004), and most nests and roosts are concentrated in urban areas (Cink and Collins 2002). The species is weakly territorial (typically 1 nest per cavity), and population declines may be due to loss of nest sites as large open chimneys become scarce. Home range sizes are largely unknown.
For a bird that occurs in such close association with humans, extraordinarily little data is available on chimney swift habitat preferences. We assumed habitat suitability for this species was mostly affected by the availability of nest and roost sites within the proper landscape context (i.e., open chimneys near foraging areas). To identify these locations, we estimated the proportion of foraging habitats in a 1 km buffer around each pixel of developed landcover. We assumed birds could travel 1 km from nesting-roosting areas to foraging habitats (defined as water, grassland, pasture-hay, recreational grasses, or forest landcover classes) and these habitats needed to be >1 ha to accommodate the aerial foraging maneuvers of this species. Because chimney swifts are semi-colonial, we also assumed that as foraging habitat increased in the 1 km buffer, developed pixels were increasingly isolated and would be of lower habitat suitability
. We used a quadratic curve
to quantify the relationship between landscape composition and chimney swift habitat suitability.
Chimney swifts occurred in all 88 subsections of the CH and WGCP. Spearman rank correlation identified a significant (P ≤ 0.001) positive relationship (r = 0.50) between average HSI score and mean BBS route abundance across subsections. Negative binomial regressions between these variables at the subsection and route scales indicated a positive relationship between HSI scores and abundance and an improvement of the HSI model over the null model for predicting chimney swift abundance.