Recognizing the spatial distribution of populations is a primary step in conservation since it provides basic information necessary for designing further studies. However, the lack of sufficient occurrence records for rare populations is a major obstacle in precisely defining their distributions. The southern Black Hills (BH) of South Dakota are the northeastern limit of the breeding range of Virginia’s Warbler (Oreothlypis virginiae, VW) and are separated by over 200 km from the nearest breeding population in Wyoming. Although VW is a species of conservation concern in South Dakota, few studies have focused on precisely identifying its distribution and none have examined landscape-level habitat associations for this population. Hence, the primary goal of this study was to identify the potential spatial distribution of VW at the northeastern extent of its breeding range. We applied a suite of standard species distribution modeling techniques to recent survey data for VW in the southern BH to examine landscape-level habitat characteristics that might predict VW occurrence. We averaged predictions of the distribution models to create an ensemble model weighted by AUC values to predict VW distribution in the southern BH and surrounding region and compared individual distribution models with the ensemble model.
Results/Conclusions
The ensemble model performed better than any single modeling technique for predicting VW occurrence. The abundance of shrub (mountain mahogany and skunkbush sumac) land cover, terrain ruggedness index, NDVI and land cover type were the most important variables associated with VW occurrence in most of the models. These results are consistent with previous studies at the local habitat scale, which documented that high shrub cover, steep slopes and scattered pine trees were habitat elements favored by this population. The ensemble prediction map revealed that additional scattered locations in the central, northern and northeastern BH, northwestern Nebraska and southeastern Wyoming might serve as potentially suitable breeding habitats for VW. These results can guide future surveys with the aim of more precisely defining the distribution and preferred habitats of VW populations at the northeastern extent of their breeding range. In addition, this study highlights landscape-level habitat associations for VW in the southern BH, which could help clarify conservation priorities, facilitate studies of source-sink population dynamics, and answer other important conservation questions for this disjunct population.