Distribution of priority grassland bird habitats in the Prairie Pothole Region of Canada / Bradley C. Fedy, James H. Devries, David W. Howerter, and Jeffery R. Row.

Includes bibliographical references.

Grassland ecosystems and the species that rely on them are one of the most urgent habitat conservation concerns in
North America. Fundamental to any landscape conservation efforts is the identification of priority habitats to help target management
efforts. Many avian species associated with prairie ecosystems have experienced population declines along with continued loss of prairie
habitats. Additionally, given the long history of research in avian systems and the close grassland associations of some species, birds
are excellent candidate taxa for the identification of priority habitats and can provide an informed starting point for multispecies
assessments. We used data from the North American Breeding Bird Survey (1997-2014) to develop species distribution models for 15
grassland bird species across the Prairie Pothole Region of Canada. Model performance varied widely across species. Ten species
demonstrated good model performance (average Boyce Index > 0.64 across 5-fold cross validation). We used these 10 species to assess
the influence of habitat covariates on the relative probability of occurrence, to compare the spatial scales of selection, and to generate
multispecies habitat priority maps. Of the nine habitat covariates considered, most species predictably demonstrated positive associations
with grassland habitats and avoidance of areas of high tree and shrub cover. Two covariates representing wetland abundance were also
frequently included in the top models. The area covered by wetlands (w.area) was present in the top model for 5 of 10 species with a
consistently estimated negative coefficient. However, a covariate, which represented the number of wetland basins (w.basins), was
present in the top model for 8 of 10 species with an estimated positive coefficient for all but 1 species, representing a preference for
more heterogeneous wetland landscapes. The larger spatial scales we considered tended to have greater explanatory power than smaller
spatial scales and were thus more prevalent in the top models. The multispecies priority habitat maps that we produced can be used for
future assessments of potential habitat management actions. Our work provides a critical foundation for the incorporation of grassland
bird conservation goals into on-going landscape-planning initiatives in the Prairie Pothole Region of Canada.