Avian species richness and abundance in a rapidly urbanizing landscape in the Southern Piedmont region of Georgia, United States of America

Species richness of Neotropical Migrant birds and the densities of thirteen species of birds were modeled across a gradient of urbanization in a part of the Southern Piedmont in central west Georgia. Species richness and densities of each species were quantified at 650 points across Meriweather, Muscogee, and Harris counties in Georgia in 2002 and 2003.; To infer landscape attributes, I used a 2002 Landsat thematic image with seven land-use classifications. Buffers of 100, 200, and 1000 m were generated around each point. Within each of the buffers, I calculated the proportion of each land-use and in the largest buffers I also calculated configuration statistics such as the amount of edge (unweighted and weighted to account for structural differences), patch density, average patch size, and others.; I used Poisson regression to model species richness at a point count station with landscape attributes. Akaike Information Criterion was used to rank models by relative fit and the numbers of parameters in the model. In 2002 the top ranking model had percent urban cover (1000 m), percent transitional cover (200-m) and mixed hardwoods (200-m). In 2003 the top ranking model had weighted edge density (1000-m), percent transitional cover (1000-m) and mixed hardwoods (200-m). An analysis of habitat association demonstrated that some birds, such as Black-and-white Warbler, and Wormeating Warbler were associated with extremely low levels of urbanization---less than ten percent in the 1000-m radius buffer around a point. The Neotropical migrants that were most tolerant of urbanization were Gray Catbird and Great Crested Flycatcher.; For the analysis of densities I selected a number of disturbance-dependent species that have been suggested to have been declining regionally or globally as well as Brown-headed Cowbirds, Molothrus ater, a species that is a brood parasite and whose densities may influence the reproductive success of other birds. Most of these species were absent from a majority of sites, so the distribution was marked by a preponderance of zeros. To account for the extra zeros, I used zero-inflated Poisson regression. Most of the absences were driven by the proportion of urban cover in the landscape. Where birds were present, densities were driven by land-uses associated with that species' ecology.