Variation in habitat thresholds: An analysis of minimum habitat requirements of North American breeding birds

Many species show dramatic changes in population extinction or persistence probability at particular habitat amounts. These 'extinction thresholds' could be translated to conservation targets, under the condition that we can derive generalities. I investigated the level of variation in landscape-level habitat thresholds for a suite of North American, forest-associated, breeding birds. Records from Breeding Bird Atlases and the availability of remotely-sensed land cover data allowed me to compare habitat thresholds for 25 species across the states of Massachusetts, Michigan, New York, Ohio, Pennsylvania, and Vermont. I show that variation in thresholds is considerable (Chapter II, III), as thresholds range from 7 to 90% forest cover between species, within regions, and even from 12 and 90% forest cover within species across regions (results for White-throated Sparrow (Zonotrichia albicollis)). I found no universal trend in this variation, although a few species showed a significant increase in threshold amounts with increasing forest cover in the landscape. In Chapter IV, I show that it is possible to assess vertical habitat structure with light detection and ranging (lidar) data. The availability of detailed habitat metrics, such as maximum canopy height and canopy heterogeneity, allowed me to detect detailed extinction thresholds for five species of cavity breeding birds and the Cerulean Warbler (Setophaga cerulea), a species of great conservation concern. Models also showed that some species persist at low forest cover, even though they demonstrate a peak in extinction probability at intermediate levels of forest cover (Chapter V). These peaks in extinction probability correspond with a peak in change in amount of forest cover over time, indicating that change in habitat might be predictive of extinction probability. Estimation of species-specific thresholds, I propose, provides information that can potentially be used to set management targets (Chapter VI). I conclude that we should be wary of extrapolation of thresholds and emphasize that estimation of thresholds should be considered a tool for understanding the process of habitat loss, not a goal in itself.