Body mass, habitat generality, and avian community composition in forest remnants

Aim Many theories of biodiversity and biogeography assume that species respond equally to variability in habitat area and isolation. This assumption does not allow for differential responses due to interspecific competition or other mechanisms, and therefore does not allow community composition to be predicted. As body size is relevant to area requirements and interspecific dominance, a natural experiment was conducted to quantify the differential responses of avian species abundance to variability in remnant area, isolation and forest cover based on average species body mass. Location Deciduous forest remnants of varying area and isolation throughout the State of Delaware, USA. Methods Forest remnants within stratified area and isolation classes were randomly selected for breeding bird surveys; total forest cover (ha) within 2 km of each survey point was subsequently determined as a covariate. Surveys were conducted within 100-150 m from the edge of each remnant and detected bird species were divided into five classes based on a log(2)-transformation of body mass (very small, small, medium, large and very large). Assuming a negative binomial distribution, the abundance of detected individuals in each mass class was analysed using generalized linear models with remnant area, isolation, local forest cover and two-way interactions specified as independent variables. The same analyses were conducted for individual species where sample size allowed. Results Very small, small and very large bodied species decreased in abundance with decreasing local forest cover and remnant area and with increasing remnant isolation, while large species increased in abundance. Medium-sized species decreased in abundance with increasing forest cover, did not respond to remnant isolation and showed a concave, curvilinear response to increasing remnant area. Large and medium-sized species were the most abundant birds in small, isolated remnants despite occurring in the largest remnants with the more abundant very small and small species, suggesting that communities are not randomly organized. Main conclusions Regardless of presumed habitat associations, large and medium-sized species are of the appropriate size to be dominant competitors when forest resources are limiting, and thus may be considered 'generalists'. The smallest species may be excluded entirely from small, isolated remnants even though such remnants meet their ecological needs; the needs of very large species are not met in small remnants. The applicability of biodiversity theories to community composition, species abundance and, by extension, to conservation, can be improved by incorporating differential responses based on body mass into their assumptions.