Riparian bird communities along an urban gradient: Effects of local vegetation, landscape biophysical heterogeneity, and spatial scale

Urbanization is an important driver of ecosystem change that can have deleterious effects on regional native biodiversity. Yet we know little about the potential value of urbanizing areas for maintaining local and regional species diversity. Few studies have explicitly examined how the spatial arrangement and composition of biophysical elements within a metropolitan area contribute to the structure and composition of urban biodiversity. This thesis focuses on avian community responses and uses observational studies that consider local vegetation and landscape factors in order to further understanding of the ecological and conservation implications of urbanization across multiple temporal and spatial scales.;A synthesis of three urban systems demonstrates the usefulness of gradient analysis approach for understanding fine-and coarse-scale processes influencing urban bird distributions. Results illustrate differences between two urban-to-rural gradient paradigms and the importance of conducting investigations at multiple spatial, temporal and biological scales. Recommendations are provided to improve our understanding of urban bird communities using gradient analyses and emphasize the future need to derive a common framework that incorporates the biophysical and social heterogeneity of urban systems.;An urban gradient study of riparian bird communities within metropolitan Cincinnati, Ohio during spring migration found that bird species responses varied based on migratory strategies and across spatial scales. Long-distance Neotropical migrant species density, richness, and evenness responded most strongly to landscape and vegetation measures and were positively correlated with areas of wide riparian forests and less development within 250 m. Resident Neotropical migrants density, richness, and evenness increased with wider riparian forests (> 500 m) without buildings, while en-route migrants utilized areas having a wide buffer of tree cover (250 to 500 m) regardless of buildings; both resident and en-route landbirds were positively associated with native vegetation composition and mature trees.;To better understand the relative importance of proximate versus landscape features and the influential spatial scales of these landscape features, I focused on breeding riparian bird species and the influences of two biophysical features of the urban environment--vegetation and built elements--within 1 km for the same riparian study area. At the proximate scale, native tree and understory stem frequency were the most important vegetation variables; native tree frequency had a positive influence on 35 species and a negative influence on 13 of the 48 species and native understory frequency had a positive influence on 27 species and a negative influence on 21 species. At the landscape scale, the vegetative features (both tree cover and grass cover) were most important variables included in competitive models across all species; tree cover positively influenced 15 species and negatively influenced 5 species and grass cover positively influenced 22 species and negatively influenced 5 species. Building density was an important variable for 13 species, and positively influenced 6 species and negatively influenced 7 species. In a comparison of multiple scales, models with only landscape variables were adequate for some species, but models combining local vegetation and landscape information were best or competitive for 42 of the 44 species. Local-vegetation-only models were rarely competitive. Combined models at small spatial scales (≤ 500 m) were best for 36 species of the 44 species and these models commonly included tree cover and building density. Only eight species had best models at larger scales (> 500 m); grass cover was most the important variable at larger scales.;In conclusion, understanding the processes that create repeatable patterns in urban bird distributions is a challenge that requires investigation at multiple spatial, temporal, and biological scales. These findings provide managers and land-use planners with species-specific information and emphasize the importance of acknowledging both proximate and landscape influences in habitat modeling.