Assessing the status of the bee fauna of the northeastern United States across a century of global change

Background/Question/Methods

Pollination is an important ecosystem function because eighty-seven percent of angiosperm species worldwide are pollinated by animals. However, despite concern about possible pollinator declines, long-term data on the status of pollinator populations is scarce. We present the first long-term study of relative rates of change for species on multiple genera of North American bees (Hymenoptera: Apoidea: Anthophila). We databased > 30,000 bee specimens from the collections of leading North American museums, with specimen collection dates spanning a 140 year period from 1870 to 2010. In order to control for unknown sampling effort, which is a limitation of museum data, we binned the data in 6 time periods of 25 years each and standardized the number of specimens in each bin by using resampling methods. In addition, we used logistic regression to analyze changes over time for the 132 native species with more than 50 specimens each. We also assessed which life history traits characterize declining species by using phylogenetically controlled general linear models.

Results/Conclusions

The number of bee species present within equal-size samples of specimens decreased over time, with the most recent time period bin showing the fewest species. In addition, the number of non-native bee species increased since 1900.  When analyzing the individual responses, 32 species present a significant positive response, while 33 present a significant negative response, and 67 showed no significant trend over time. At the genus level, Andrena, Melissodes and Osmia contain many declining species. Species of Bombus, Halictus and Megachile present mixed responses and Colletes, Lasioglossum and Sphecodes contain many species that show increasing patterns. Traits associated with decline include oligolecty (dietary specialization; p-value = 0.002) and having a short flying period (p-value = 0.04). Southern species are increasing in the study area (p-value < 0.001), which we speculate may be due to climate warming.