The decline of wild bee populations poses a significant threat to pollination services in both natural and agricultural ecosystems. Wild bees contribute substantially to crop pollination worldwide, and their decline has substantial economic and ecological consequences. Several factors contribute to these declines, including climate change, habitat loss, disease, dietary stress, and pesticide use. Agricultural intensification plays a major role in exacerbating many of these factors, reducing floral and nesting resources for bees while increasing their exposure to pesticides. Neonicotinoids and pyrethroids are two pesticide classes of particular concern due to their widespread use and documented harm to bees. While laboratory and field studies have shown negative effects, large-scale, continent-wide assessments are limited. This study addresses this gap by analyzing a large dataset of bee occurrence records from across the contiguous United States, combined with land use and pesticide application data, to model species occupancy and assess the impacts of pesticide use.

Widespread bee declines have been reported in Europe and North America, but the status of most species is poorly known. Insect pollination, primarily from wild and managed bees, is crucial for a large percentage of crop and flowering plant species. Wild pollinators play a particularly important role, enhancing yields even in the presence of managed honeybees. Major drivers of wild bee decline include climate change, land-use change and habitat loss, disease, dietary stress, and pesticide use, often linked to agricultural intensification. Agricultural intensification reduces resource diversity and increases exposure to pesticides, particularly neonicotinoids and pyrethroids. Studies have shown these pesticides' harm to individual bees, but their large-scale effects remain less understood. While some studies linked neonicotinoid use to population-level extinction rates, a comprehensive understanding of pesticide impacts in relation to other drivers is critical for effective conservation strategies.

This study used one of the largest databases of bee records for the contiguous United States, aggregated from museum specimens, surveys, and community science observations, encompassing 178,589 unique observations from 1,081 bee species across six families. Multispecies occupancy models were employed to estimate the effects of pesticide use, animal-pollinated agriculture, and honeybee colonies on wild bee distributions. The analysis focused on neonicotinoids and pyrethroids, weighting pesticide use by their LD50 values for honeybees to account for toxicity differences. Data on pesticide use was obtained from the USGS Pesticide National Synthesis Project; honeybee colony data from the National Agricultural Statistics Service; and agricultural distribution data from the Crop Data Layer. The analysis used a multispecies framework, modeling occupancy across counties within species' geographic ranges over seven 3-year time periods (1995-2015). Structural causal model analysis with directed acyclic graphs (DAGs) was used to select predictors and control for confounding variables. Three types of multispecies occupancy models were used, each run independently on each bee family (or combined families where sample size was limited). Occupancy was modelled at county level, inferring species absences if another species within the same genus was observed at the same site and year. The model considered the effect of pesticide use, animal-pollinated agriculture, and honeybee colonies.

The analysis revealed a significant negative relationship between pesticide use (neonicotinoids and pyrethroids combined) and wild bee occupancy across all five bee families studied. The effect was strongest for Andrenidae and Apidae, with a 43.3% decrease in occupancy probability for Apidae with an increase in pesticide use from zero to the maximum observed value. The negative impact of pesticide use was consistent across different classifications of animal-pollinated agriculture and was robust to variations in the occupancy interval length. Genus-level analysis revealed varying responses to pesticide use, ranging from a 54% decline to a 62% increase in occupancy. Animal-pollinated agriculture showed a positive, though not always statistically significant, effect on occupancy for certain families, varying by genus. The effect of honeybee colonies on wild bee occupancy was not statistically significant across any groups.

The study's findings strongly support the hypothesis that pesticide use, specifically neonicotinoids and pyrethroids, negatively affects wild bee occupancy across the contiguous United States. The consistency of the negative effect across multiple bee families and the robustness to different model specifications underscore the significance of these findings. The identified negative impact on important crop pollinators, such as bumblebees and *Andrena* bees, emphasizes the ecological and economic implications. The results align with previous experimental and smaller-scale observational studies, demonstrating that the negative effects of pesticides observed at smaller scales can translate to large-scale patterns in species occupancy. While the study did not directly address habitat conversion, it highlighted that habitat loss was weakly correlated with pesticide use and animal-pollinated agriculture, suggesting that pesticide use may be a more dominant driver of occupancy trends. The limitations of the honey bee data used, which lacked detailed seasonal and spatial resolution, must be considered when interpreting the findings regarding honeybee effects on wild bee occupancy. The study’s findings underscore the need for integrated pest and pollinator management (IPPM) strategies, such as the use of hedgerows, to support pollination services and wild bee populations.

This large-scale study provides compelling evidence that pesticide use significantly impacts wild bee occupancy in the United States. The negative effect of neonicotinoids and pyrethroids is widespread and considerable, emphasizing the importance of reducing pesticide reliance through strategies like IPPM. Future research could investigate abundance patterns (currently unavailable), delve deeper into species-specific responses to pesticide combinations, and explore the combined effects of pesticides with other factors like habitat loss and climate change. Expanding this research methodology to other regions with appropriate data availability is crucial for global pollinator conservation efforts.

The study relied on presence-only data, necessitating assumptions about detection probabilities. While the chosen approach has shown robustness, uncertainties associated with detection and data collection methods should be considered. The coarse resolution of some datasets (e.g., honeybee colony data) may limit the detection of fine-scale impacts. The study did not explicitly account for some potentially important factors, such as habitat conversion, which may confound the relationships observed. Finally, the study focuses on occupancy rather than abundance, meaning that while the findings are concerning, they might not capture the full extent of population declines.