Avian community dynamics in a fragmented tropical landscape

As humans alter habitats worldwide, developing reliable methods of assessing biodiversity and community attributes of interest (e.g., species richness, turnover, and extinction rates) is important. Frequently, estimates of community-level attributes are biased because the estimators make assumptions of the data that are violated; many published studies assume equal detectability across species, sites, or time. The accuracy of estimators of species richness and community-level vital rates (e.g., extinction and colonization) can be increased by using probabilistic estimation methods, which do not assume that all species are detected, or that the data assume a particular statistical distribution. Using these estimation methods, we examined avian community dynamics in a fragmented tropical landscape using data from five years of a mark-release-recapture study. For the resident understory avifauna in each of five small (similar to 0.3-20 ha), isolated forest fragments in southern Costa Rica, we estimated species richness, rate of change in species richness, extinction and turnover rates of species, and the number of colonizing species over temporal scales of one month, one year, and two years. We expected that community dynamics would be higher in smaller fragments than in larger fragments, reflecting greater temporal variability of avian communities in relatively small habitat patches. Additionally, a selective logging operation was conducted at one of our sites during the midpoint of this study, which gave us the opportunity to examine how community-level vital rates may reflect the effects of that perturbation. Our results demonstrate that avian communities in the larger fragments were more stable than those in the smaller fragments, and that the selectively logged fragment was the most unstable of all. We found that extinction rates were more similar across our sites than were colonization rates, and that the higher instability of the small fragments was due primarily to higher levels of colonization. Although our sample size (n = 5) precludes strong inference, our findings are consistent with the prediction of higher local dynamics within small fragments and after logging. Taken together, these findings suggest that smaller fragments are more dynamic over time, and that ecological processes and multitrophic relationships at these dynamic sites may be in a constant state of flux.