Native bee community functional diversity explains sentinel plant pollination in an intensive agricultural landscape

Background/Question/Methods

More attention has been placed on measuring pollinator performance in agroecosystems at a community level. Previous studies have identified important pollinator functional traits and evaluated effectiveness, but not often in a community context. To study the agroecological bee community and its functional attributes as a whole, we conducted a manipulative floral removal experiment in agricultural landscapes of Southwestern Ohio. For the removal, we selected an understory shrub with a strong spring floral pulse that dominates woodlot edges adjacent to intensive corn and soybean agriculture. Next, to quantify community function we conducted a sentinel plant experiment, measuring seed and fruit set of an obligate out-crossing domesticated cucumber (Cucurbita sativa) placed at four distance classes (20m, 50m, 100m and 200m) from forest edges into the crop field. Bees were sampled during peak bloom of the shrub at each distance class using colored pan traps continuously for two weeks starting the end of May. All bees were identified to species, and distance between tegula, front wing length, and body length was measured. Additionally, life history functional traits were obtained from published literature. Generalized linear mixed models with site and sample date as random effects were used to analyze the data. Finally, we assigned graded numerical values to each functional trait so that one functional value could be assigned to each bee species for a community-level function analysis.

Results/Conclusions

We collected 1060 bees spanning 20 genera and 87 species. The species composition, species richness, and overall bee abundances were not different between treatments. However, intertegula to body length ratios and front wing length to intertegula ratios increased in removal plots (p<0.005). Although no difference between treatments was detected in abundance of eusocial species, parasitic bee abundance increased within floral removal plots (p=0.01). Of the 200 mature cucumbers analyzed, there was a significant decrease in seeds per cucumber within treatment plots (p<0.0001), indicating a direct reduction in pollination services of the bee community, despite a lack of overall differences in bee species composition, richness, and abundance. The observed variation in sentinel plant pollination may be explained by the shifts in functional traits within the bee community. The applicability of this model is limited, but as environmental disturbances continue and bee diversity declines, the need to evaluate and translate individual bee traits into a community function potential is of urgent importance.