Using a novel land cover classification to identify multi-scale drivers of site occupancy for shrubland habitat conservation

Context Understanding habitat selection at multiple scales is important for guiding conservation efforts. This is difficult for species that occupy early-successional habitats (ESH) due to a lack of accurate representation of shrub/sapling cover in land cover data. Objectives We used a novel, LIDAR-derived land cover classification that accurately identifies shrub/sapling cover to investigate how habitat composition and configuration influence site occupancy for two declining songbird species. Methods The Golden-winged Warbler ( Vermivora chrysoptera ) is a threatened species of conservation concern and a flagship for ESH conservation; it hybridizes and competes with the closely related Blue-winged Warbler ( V. cyanoptera ). We developed occupancy models using 1 and 30 m resolution land cover data that accurately represents shrub/sapling cover and with 30 m resolution National Land cover Data (NLCD) which does not. To account for potential spatial autocorrelation, we ran spatial occupancy model versions of the top performing models and compared the outputs. Results Spatial occupancy models performed better than non-spatial models; therefore, we present results and parameter estimates from these models. We confirm that elevation, forest and shrub/sapling cover are important habitat features for Vermivora and demonstrate specific extents and optimum amounts that these cover types should be maintained when managing for these species. Aggregating land cover to 30 m resolution and using NLCD identified the same cover types and extents as important; however, NLCD model predictions were unrealistic for shrub/sapling cover because it does not accurately represent this cover type. Conclusions Our results demonstrate that habitat models for shrubland-dependent species would be improved by using land cover maps that accurately represent shrub/sapling cover, specifically those that use LIDAR data. For species that require a complex mosaic of cover types, such models would lead to improved communication and implementation of conservation efforts.