Abstract Anthropogenic drivers of global change threaten bee diversity and the pollination services they provide. Despite the importance of bees, their conservation is complicated by limited and often heavily biased occurrence data. A recent state‐wide survey of insect pollinators across New York, United States generated a large spatial dataset of bee species occurrence records from community scientists, historical collections, and recent surveys. To predict the distributions of most of the state's bee species, we combined the state survey records with occurrence data from across the contiguous United States before applying an ensemble modeling approach consisting of balanced random forest and small bivariate generalized linear models. We predicted the spatial distribution of bee species richness using a stacked species distribution model with climate, land cover, and soil covariates. To inform bee diversity conservation, we predicted spatial variation for each species and groups of species sharing similar life history traits. We also estimated the statewide distribution of range‐size rarity, ecological uniqueness, and climate exposure. We found that the richness of modeled species is high across the state, with the greatest richness in regions with low soil clay content and intermediate forest cover. The fine spatial scale and extent of our gridded data layers match the scale of conservation action in the state, providing an opportunity to incorporate wild bee diversity into broader statewide conservation planning. Conserving bee pollinators is not straightforward, and decisions should be based on broader conservation priorities that incorporate bee biodiversity indicators into decision‐making. Here, we present a roadmap for the inclusion of these vital pollinators in conservation decisions by leveraging the best available data and methods robust to small sample sizes to provide spatially explicit data products representing the distribution of bee diversity.
Buckner et al. (Tue,) studied this question.