Population dynamics and community structure of granivorous forest rodents in a fragmented landscape

I studied the effects of forest fragmentation on population dynamics and community structure of small mammals within the Indian-Pine watershed of west-central Indiana. A comparison of survival, sex ratios, and other population attributes of white footed mice (Peromyscus leucopus) and eastern chipmunks (Tamias striatus) inhabiting four woodlot fragments and three continuous forests indicated that in woodlots white-footed mice exhibited greater population densities and body weights, and eastern chipmunks exhibited lower survival rates. Competitive release is the most likely mechanism producing the results for mice, whereas increased predation in woodlots could account for the results observed in chipmunks. I also developed logistic and multiple linear regression models to relate presence/absence and population density, respectively, to measured landscape and habitat attributes of 35 woodlots of various sizes and two continuous forest sites using spring livetrapping from 1992 through 1996. Species richness of forest-dwelling small mammals increased with area and was highest within the continuous forest sites. Nested subsets of the full complement of species were found in smaller woodlots. White-footed mice and eastern chipmunks were ubiquitous across the landscape, and their densities were inversely related to forested area. Fox squirrels (Sciurus niger) were found in the majority of study sites (31/37), but their densities were not related to landscape variables. Red squirrels (Tamiasciurus hudsonicus) were unevenly distributed across the landscape and found most often in woodlots with large core areas and simple shapes, possibly indicating sensitivity to edge. Southern flying squirrels (Glaucomys volans) and gray squirrels (S. carolinensis) were restricted to continuous forest sites and large $({\u3e}4.5$ ha) woodlots proximal to other wooded habitat. Addition of measures of local habitat structure and mast production to models incorporating only landscape-level variables improved the predictive ability of 11 of 14 models. Interspecific competition explained only a minor portion of the variation in densities of granivorous forest rodents. Thus, my results demonstrate that the effects of fragmentation vary among species and can be predicted largely on the basis of patch and landscape features. An increased understanding of organisms occupying fragmented landscapes is attained by examining processes at multiple spatial scales