Ancient museum collection and modern research intersect in century-old study of bees

In a quiet nature reserve in southern Michigan, an Agricultural Research Service (ARS) scientist and her collaborators connected ancient collections of wild bee samples with modern technology to better decode the pollinators’ ecological traits and habits, links that are critical to environmental stability.

Kelsey Graham, a scientist in the ARS Insect Pollinator Research Unit, co-led the collaborative, intensive study of wild bees at the University of Michigan’s E.S. George Preserve with a sampling period spanning 1921 to 2018, which, in tandem with advanced computational analyses, revealed long-term trends in the bee population that could hold the keys to new and improved conservation approaches.

“These studies highlight clear indicators of an urgent need for diligent and consistent conservation efforts to protect bee diversity, which is crucial to the health of our ecosystem, human health and agricultural productivity,” Graham said.

In a recent publication of Proceedings of the Royal Society BGraham’s research paper, “A Century of Wild Bee Sampling: Historical Data and Neural Network Analysis Reveal Ecological Traits Associated with Species Loss,” explains how the study hit turning points along the way, finding alarming declines in species richness, evenness, and overall diversity of the bee community. The researchers also found that 64 percent of the most common bee species showed abundance declines of more than 30 percent.

“In 1972 and 1973, the late zoologist Francis C. Evans detected 135 species of bees, while our recent surveys in 2017 and 2018 recorded only 90, with only 58 species present during both sampling periods,” Graham noted. “These samplings indicate a substantial shift in the composition of the bee community.”

To better understand why some species have disappeared from the reserve, the ARS team and its partners used neural networks, which determined that certain types of bees were more likely to disappear. Specifically, the researchers found that ground-nesting oligolectic bees (that is, bees that collect pollen from a few types of plants and nest in the ground) and kleptoparasitic bees (that steal food from other bees) are most vulnerable.

In comparison, the study found that polylectic cavity-nesting bees (or bees that collect pollen from a variety of plants and nest in cavities) were more likely to remain in the reserve.

Similarly, the results demonstrated that bees that are active for long periods each year have a better chance of remaining in the community if they collect pollen from a variety of plants.

In short, bees with certain characteristics, such as being picky about food, will continue to struggle compared to their more flexible counterparts.

The scientists also noted the importance of the climate response, as bee species from the contemporary sampling period had a more southerly overall distribution compared to the historical community, indicating that communities are shifting in response to warming temperatures.

This study, Graham explained, demonstrates the utility and importance of publicly available, long-term historical data for deciphering complex indicators of bee population trajectories, findings that might otherwise have been obscured in a smaller scope and time frame.

“Combining traditional analysis techniques with neural networks helped us reveal changes in geographic ranges and declines in bee abundance and diversity based on species traits,” Graham said. “Such analyses help us understand trends in bee populations to inform the science and practice of bee conservation.”