Regional variability in extinction thresholds for forest birds in the north-eastern United States: An examination of potential drivers using long-term breeding bird atlas datasets

Abstract

Aim: Demand for quantitative conservation targets has yielded a search for generalities in habitat thresholds, particular amounts of habitat at which extinction probabilities change strongly. These thresholds are thought to vary across regions, but investigation of this variability has been limited. We tested whether thresholds (of forest separating extinction from persistence) increased as either average forest cover in landscapes decreased or the degree of fragmentation increased. Location: Massachusetts, Michigan, New York, Ohio, Pennsylvania and Vermont. Methods: We used segmented logistic regressions to estimate thresholds in the relationship between extinction probability and forest cover for 25 forest-breeding birds, comparing estimated thresholds across states. We also selected landscapes from our entire study area in which landscape-level forest cover and degree of fragmentation varied independently and compared thresholds. Results: We found that thresholds in extinction probability varied widely among species (7-90% forest cover) and within species across states [e.g. 12-90% for white-throated sparrow (Zonotrichia albicollis)]. Additional analyses showed no indications that thresholds correlated with the degree of fragmentation or forest cover across the landscape; we found considerable variability in thresholds across landscapes, species and even landscapes in which (average) fragmentation and forest cover were similar. Main conclusions: Extinction threshold estimates varied tremendously across species and landscapes. Thus, habitat thresholds are difficult to generalize as they depend on many factors beyond landscape fragmentation and habitat availability (e.g. landscape characteristics such as matrix quality). Our findings highlight the need to avoid oversimplification and generalization of habitat thresholds, especially as they might prove counterproductive to conservation efforts. Instead, we propose that we evaluate thresholds for individual species - preferably using species-centred habitat definitions in threshold modelling - to derive generalities for ecological and conservation applications.