Landscape effects on breeding songbird abundance in managed southern Appalachian forests

Abstract

Many studies have demonstrated adverse effects of forest fragmentation on breeding forest songbirds in North America, and the viability of regional populations is thought to depend on large, unfragmented forests. However, we know relatively little about the landscape scale consequences of management in the forested landscapes that are presumed to be important to maintaining songbird metapopulations. The southern Appalachians, a mostly forested region, contains the largest network of public lands in the eastern U.S. Most of these public lands are managed by the U.S. Forest Service. To begin to understand the landscape scale effects of forest management in the southern Appalachians, I examined the relationship between the relative abundance of different species of breeding songbirds and local and landscape scale habitat variables in two predominately mid- to late-successional National Forests in the southern Appalachian Mountains, USA: the French Broad Ranger District of Pisgah National Forest (North Carolina) and the Nolichucky Ranger District of Cherokee National Forest (Tennessee). As part of the study, I explored two statistical problems frequently encountered in species-environment analysis: count data and spatial autocorrelation. Results from classical normal-errors regression models were similar to results from Poisson and negative binomial models that explicitly model counts. Normal-errors regression models were then modified to account for spatial autocorrelation using a conditional gaussian autoregressive model. Most species, especially Neotropical migrants, were significantly correlated with at least one landscape variable. These correlations included both landscape composition (i.e., the proportion of different landcover types) and landscape pattern (i.e., the spatial arrangement of landcover types) variables at 500 m to 2 km landscape scales. However, these landscape effects explained only a small fraction of the variation in bird relative abundance, and most species appear to respond primarily to elevation and local habitat factors in my study area. My results are consistent with other studies that have reported only weak to moderate landscape effects on songbird abundance in large managed forests. These results should not be interpreted as being inconsistent with results from studies in highly fragmented forests that have reported strong effects of patch size, patch isolation, and landscape scale forest cover on breeding songbirds.