Jane A. Fitzgerald

Assessing Ecoregional‐Scale Habitat Suitability Index Models for Priority Landbirds

Abstract Emerging methods in habitat and wildlife population modeling promise new horizons in conservation but only if these methods provide robust population–habitat linkages. We used Breeding Bird Survey (BBS) data to verify and validate newly developed habitat suitability index (HSI) models for 40 priority landbird species in the Central Hardwoods and West Gulf Coastal Plain/Ouachitas Bird Conservation Regions. We considered a species’ HSI model verified if there was a significant rank correlation between mean predicted HSI score and mean observed BBS abundance across the 88 ecological subsections within these Bird Conservation Regions. When we included all subsections, correlations verified 37 models. Models for 3 species were unverified. Rank correlations for an additional 5 species were not significant when analyses included only subsections with BBS abundance >0. To validate models, we developed generalized linear models with mean observed BBS abundance as the response variable and mean HSI score and Bird Conservation Region as predictor variables. We considered verified models validated if the overall model was an improvement over an intercept-only null model and the coefficient on the HSI variable in the model was >0. Validation provided a more rigorous assessment of model performance than verification, and models for 12 species that we verified failed validation. Species whose models failed validation were either poorly sampled by BBS protocols or associated with woodland and shrubland habitats embedded within predominantly open landscapes. We validated models for 25 species. Habitat specialists and species reaching their highest densities in predominantly forested landscapes were more likely to have validated models. In their current form, validated models are useful for conservation planning of priority landbirds and offer both insight into limiting factors at ecoregional scales and a framework for monitoring priority landbird populations from readily available national data sets.

Application of Models to Conservation Planning for Terrestrial Birds in North America

Partners in Flight (PIF), a public–private coalition for the conservation of land birds, has developed one of four international bird conservation plans recognized under the auspices of the North American Bird Conservation Initiative (NABCI). Partners in Flight prioritized species most in need of conservation attention and set range-wide population goals for 448 species of terrestrial birds. Partnerships are now tasked with developing spatially explicit estimates of the distribution, and abundance of priority species across large ecoregions and identifying habitat acreages needed to support populations at prescribed levels. The PIF Five Elements process of conservation design identifies five steps needed to implement all bird conservation at the ecoregional scale. Habitat assessment and landscape characterization describe the current amounts of different habitat types and summarize patch characteristics, and landscape configurations that define the ability of a landscape to sustain healthy bird populations and are a valuable first step to describing the planning area before pursuing more complex species-specific models. Spatially linked database models, landscape-scale habitat suitability models, and statistical models are viable alternatives for predicting habitat suitability or bird abundance across large planning areas to help assess conservation opportunities, design landscapes to meet population objectives, and monitor change in habitat suitability or bird numbers over time.

Landscape-level forest cover is a predictor of cerulean warbler abundance

Abstract We examined support for the hypothesis that abundance of Cerulean Warblers (Setophaga cerulea) increases with percentage of bottomland and upland forest, and decreases with percentage of developed land at a local-habitat scale (within a 250-m buffer) and increases with percentage of all forest at a landscape scale (within a 10-km buffer). We conducted surveys along 16 rivers in Missouri and Arkansas from 1999 to 2006 and related habitat and landscape factors to counts of Cerulean Warblers in 123 5-km segments on these rivers. We detected 576 singing male Cerulean Warblers and found support for both local and landscape effects on Cerulean Warbler abundance. Model fit was good with an average correlation of 0.841 between predicted and observed values based on an eight-fold cross-validation procedure. The abundance of Cerulean Warblers increased 390.7, 8.7, and 4.1 times across the observed range of forest within 10 km, bottomland forest within 250 m, and upland forest within 250 m, respectively. Conservation and research need to address large-scale forest patterns in addition to local habitat for Cerulean Warblers. Further research is needed on abundance patterns across riparian and upland forests and demographic rates in this part of their range.