Professor Rutgers University, New Jersey, United States
Abstract: Background and Methods
Despite concern that habitat loss is threatening wild bees, surprisingly little is known about the habitat needs of any particular bee species. This is because most studies of bee habitat focus on bee communities in aggregate, rather than specific species. The spatial scale at which particular bee species respond to habitat in their landscape is also largely unknown. While it is often assumed that larger bees respond at larger scales, empirical evidence for this relationship is limited. To address these issues, we assessed the habitat associations and scales of response of 89 bee species in New Jersey, New York, and Pennsylvania, using 33,391 specimens collected at 198 sites between 2003 and 2020. For each bee species, we used GLMMs and AICc comparisons to assess their response to each of 11 habitat types, while also determining the best spatial scale at which to measure those habitats (from 100 m to 8 km). Given the many comparisons made, we used a bootstrapped null model to adjust our AICc threshold for “significance” and reduce expected type I error. Finally, we tested whether these scales of response could be explained by body size or other functional traits.
Results and Discussion
We found 215 significant habitat responses among 63 (of 89) species. The number of significant responses per species ranged from 0 to 8. No habitats were universally good or bad, but a few stood out. In particular, deciduous forest and pasture elicited mostly positive responses, while wetlands and high-intensity development elicited mostly negative responses. These results are pertinent to bee conservation, which typically focuses on only a single habitat type – open meadows. Our results suggest that wild bee conservation should expand to include diverse habitat types, especially deciduous forests. Scales of response were strongly right skewed, with a strong mode at 100-300 m, but many responses even at the max of 8 km (median = 1.2 km, mean = 2 km). Scales of response were also unpredictable, varying both within and among species, and being unrelated to bees’ body size, sociality, resource specialization or nest substrate. These results suggest that the spatial scale at which bees respond to their environment are not fixed or easily predicted by coarse functional traits. Instead, scales of response may be determined by idiosyncratic species-level traits (e.g. dispersal behavior), or aspects of the landscape itself (e.g. patch size and arrangement).
