Bumble bees (Bombus sp.) are vital pollinators with several species experiencing declines in North America in part due to agricultural intensification. With the reduction of suitable natural habitat and the rise of urbanized environments, there is growing interest in understanding bee populations within human-modified landscapes. Some evidence suggests that urban areas may serve as potential refugia for pollinators. For example, the endangered rusty patch bumble bee (B. affinis) is commonly found near metropolitan areas in the Upper Midwest. However, higher occupancy in urban areas might be due to observational bias or variability in species detection. Our goal was to assess bumble bee occupancy using a systematic field survey design conducted during the summer of 2019 within the metropolitan area of Madison, WI, and the surrounding rural landscape. Based on repeated transect surveys, we estimated detection (p) and occupancy (psi) probabilities using hierarchical occupancy models. We quantified detection probabilities of nine bee species, including declining (e.g., B. affinis and B. fervidus) and abundant species (e.g., B. impatiens). Finally, we assessed how multi-scale variability in abiotic and landscape factors influence occupancy across an urban-rural landscape matrix using spatial occupancy predictions for common bumble bee species (B. bimaculatus, B. impatiens and B. vagans).
Results/Conclusions
Many of the rare and declining bumble bee species had relatively low detection probabilities (p<0.4; psi<0.1), but common species were detected consistently across the entire urban-rural gradient (p>0.5; psi>0.5). We detected B. affinis in sites characterized by agriculture and low-intensity development, but this species was not detected in highly urbanized (i.e., high impervious area) landscapes. In contrast, B. fervidus was detected in sites that contained at most 60% of habitat classified as highly developed. Our spatial predictions suggest higher occupancy of common bumble bees in urban habitat relative to agricultural areas. Higher occupancy in urban areas could be due to urban heat island effects or high availability of floral resources. We conclude that urban areas may benefit some, but not all bumble bee species. Continuous monitoring of bee populations within urban-rural landscape gradients is essential to our understanding of bee performance in human-modified landscapes and to inform pollinator conservation efforts. We expect that the integration of hierarchical occupancy models, which disentangle the variation in detection and occupancy probabilities, will prove useful in estimating unbiased population estimates of insect pollinators and in addressing questions relating to the impacts of urbanization on bumble bees and other pollinator communities.