Climate influences broadly, landscape influences narrowly: Implications for agricultural beneficial insects

Insects provide critical ecosystem services, like pollination, in both natural and agricultural ecosystems. Delivery of these services depends on their ability to develop, survive, and move through their environment. Whether they can do this depends on the weather, climate, and landscape; but a changing climate means these systems are potentially vulnerable to disruption. Short-term fluctuations in weather can disrupt development, impede movement, and affect survival, while long-term climate norms influence environmental niches and influence species distribution. Landscape composition also influences beneficial insect distribution and has the potential to reduce the impacts of climate change. Here we use a database of >97,000 bee occurrence records, collected from 320 sampling sites across a 90,000+ km² area in the North American Prairies to generate models of species occurrence for 50 species, sampling in and around crop fields. We use a tree-based machine learning method with extreme gradient boosting to create predictive classification models. These models are then used to analyze the relative importance of weather, climate, and landscape variables. The variables with the highest mean absolute importance are cumulative degree days, cumulative precipitation, and percent tree cover. When we analyzed individual species models, bee taxonomic groups responded most strongly to weather, and the direction of response corresponded to trait-grouping. The responses to landscape were weak and species-specific. The results indicate that pollination service supply is largely determined by heat and moisture, and that cavity nesters and ground-nesters have opposite responses to rising temperature, which could impact taxonomic and functional diversity.