Native bees are an ecologically diverse group of pollinators in global decline due at least in part to invasive species, pesticides, and habitat loss. Although guidelines exist for land managers to restore pollinator habitat, these "best management practices" (BMPs) include other pollinator taxa that may have different requirements than bees, do not give particular attention to rare bee species, or describe practices that are impractical for land managers. Using co-production science, our team of land managers and researchers sampled bee communities in 100 wildlife openings on six National Forests (NF) within the Great Lakes Basin of the United States during 2017-2019. We found that bee communities responded to site factors and management practices, including prescribed fire, mechanical methods (e.g., felling, brushhogging, mowing), herbicides, and pollinator plantings. Bee abundance, diversity, and rarity were strongly related to soil properties, landscape context, and the plant community, including small-statured woody species, which collectively informed our BMPs. For instance, mechanical treatments were most beneficial for openings with clayey or organic soils while prescribed fire was most effective in openings with well-drained soils. Our BMPs highlight effects of treatment combinations, including negative effects on rare species when herbicides were combined with plantings and positive effects on abundance and rare species when prescribed fire was combined with mechanical treatments. Since our BMPs were generated in collaboration with land managers, they better conform to their needs and constraints, contributing to more effective native bee conservation. Our manuscript describes a 3-year co-production experiment evaluating how ongoing habitat restoration practices within 100 forest openings affected the inhabitant native bee communities across a large sampling region of the United States: the Great Lakes Basin. Due to the geographically expansive nature of the project and the independently operating entities performing the sampling, we employed a series of sophisticated analyses to infer relationships among the plant community, local and landscape factors, and management practices (fire, herbicide, mechanical methods, pollinator plantings, and their combinations) in the past 20 years on bee community composition as well as bee community metrics of abundance, diversity, and rarity. Since the goal of the project was to create best management practices (BMPs) for conserving forest-opening bee communities, we produced a scoring system that placed the three bee community metrics on the same scale, then derived BMPs based on the results that were tailored to both placement of the proposed management area within an opening and its surrounding landscape as well as the greater need of conservation intervention for any of the three modeled aspects of the bee community. image
