Rare and declining bee species are key to reliable pollination of wildflowers and crops across large scale

Ecosystem functions provided by insects, such as pollination, pest control, decomposition, are critical for human well-being. Hundreds of experiments have found species richness promotes ecosystem function, but this literature is dominated by controlled, small-scale studies of plant communities. We know much less about how the loss of insect biodiversity will affect ecosystem function at larger scales in non-experimental systems. We conducted landscape-scale studies of wild bee pollination of six plant species - three wildflowers and three crops, at 16-24 sites per plant species. Using independent datasets, we defined at-risk bee species as those that are currently rare or historically declining. We engaged with two knowledge gaps in biodiversity-function research. First, the amount of function provided by at-risk species is unknown for all taxa, including insects. Averaging across plant species, we found that 15 percent of pollination function was provided by at-risk bee species. Second, deciding how to analyze large-scale (i.e., multi-site) data is not trivial. Should data be aggregated across sites to reveal species broadly important to ecosystem function, or should we ask how many species are needed to maintain ecosystem function across all sites simultaneously? We found 12 times more at-risk bee species were needed to meet pollination thresholds across all sites simultaneously, compared to meeting one region-scale pollination threshold. By showing that at-risk species are needed to maintain pollination across many sites at once, our results show that changes in bee biodiversity could limit the reliable pollination of wildflowers and crops.