Model Bee-havior: modifications and a formal sensitivity analysis for adapting BEE-STEWARD to simulate population dynamics of the endangered bumble bee Bombus affinis

Native bee species in the United States provide key pollination services that support both natural ecosystems and agriculture. However, pesticide use, habitat loss, pathogens, competition, and climate change have contributed to the decline of native bee populations. Bombus affinis (rusty patched bumble bee) is an endangered species that has a historic range throughout the Midwest and Northeastern United States and southern Canada. After its recent dramatic decline, it inhabits small pockets located primarily in the Midwest. The U.S. Environmental Protection Agency recognizes deleterious effects of pesticide exposure on ecosystems and wildlife; simulations from population models could predict the impacts of exposure and other stressors to B. affinis. Therefore, we adapted a published agent-based bumble bee model, BEE-STEWARD, to simulate B. affinis foraging patterns, life history and population dynamics in a critical habitat zone near Madison, Wisconsin, where recent numbers of B. affinis have been reported. Weather-based daily foraging hours, redefined habitats for the Upper Midwest, and modifications allowing high throughput analyses were implemented in the model. We performed a formal sensitivity analysis to explore which parameters (food availability, phenology, etc.) significantly affected the model outputs (number of foragers, colony density, and number of overwintering gynes) in scenarios without pesticide exposure. The results of this sensitivity analysis will inform future parameterization efforts and calibration of the model to empirical B. affinis forager and colony density datasets within critical habitat zones. Subsequent research will integrate pesticide exposure and effects modules in the model for simulating agricultural and other high exposure scenarios.