Bees and moths are globally important pollinators. Xeric barrens in the largely mesic northeastern USA support high levels of pollinator diversity, including rare bees and moths. We investigated the response of bee vs moth communities to abiotic and vegetation drivers in barrens across the region. We sampled environment, vegetation, bees, and moths for 2-4 years in 20 preserves. Employing random forest analysis, we tested the role of 26 abiotic and vegetation predictors of bee vs moth abundance, species richness, Shannon-Wiener Diversity Index, evenness, and species composition. Variables related to climate, canopy cover, and soils were the most important predictors of abundance, diversity, and species composition for both bees and moths. Vegetation variables, such as species richness of shrubs and hostplants, were also important for bees. The direction of these relationships contrasted sharply between bees and moths: bees were more abundant and species rich in more open, sandy sites and moths the opposite. Surprisingly, bee-moth contrasts in diversity did not hold for Shannon-Wiener Diversity. Habitat preferences for a subset of moth xeric specialists were much more similar to bees than to other moths, with a preference for open, sandy conditions. Contrasts between bees and moths in habitat preferences likely stemmed from differences in life history: bees rely on flowers for feeding and porous substrates for nesting, whereas most moth adults rely on flowers but many moth caterpillars use woody plants as hosts. The contrast between bees and moths for species richness vs. Shannon-Wiener Diversity raises important general questions about the conservation value of these two metrics. Our results suggest that, because of differences in habitat preferences among pollinators, barrens management for spatial and temporal heterogeneity is likely to promote the highest abundance and diversity of resident pollinator communities.