Bumble bees (Bombus spp.) rely on pollen as the single protein source for larval development, but relatively little is known about the role of pollen protein as a factor in bumble bee foraging decisions. In this study we asked whether the composition of pollen loads carried by foraging bees was predicted by floral resource abundance and total protein content in floral rewards. To address this, we collected 100 pollen-collecting bumble bees from a 20-ha early successional field in Virginia in both June and July 2015. Bumble bees were identified to species, and pollen loads were removed, mounted, and stained using glycerin and acid fuchsin. We sampled 300 pollen grains per load and identified them by comparison to a reference collection made from the same field. Each month we also estimated floral resource abundance within the field using 400 4m x 1m belt transects. We collected flowers from the most commonly represented plant species in the pollen loads. Using a particle counter, we determined the number of pollen grains within a flower, and using the Bradford assay, we determined the percent protein content within the pollen. Using these data we calculated the total protein reward per flower.
Results/Conclusions
We collected six species of Bombus, but B. griseocollis (n = 79) and B. impatiens (n = 107) accounted for 93% of all bees, so we restricted analyses to these two species. Four plant species (Carduus acanthoides, Securigera varia, Solanum carolinensis, Verbascum thapsis) accounted for over 90% of the pollen grains carried by these bees. Although bees differed significantly in their pollen collection, Securigera was the most common pollen in the loads of both species in June, and Solanum was the most common in both species in July. The total protein reward per flower in Carduus was 5-6 times greater than in Solanum and Securigera and was twice the protein in Verbascum. For B. griseocollis, pollen representation in the pollen load increased with the relative abundance of the host plant but decreased with increasing pollen protein. For B. impatiens, pollen representation increased with protein content but decreased with host plant abundance. These two bees appear to have different strategies for the exploitation of floral resources, with one responding positively to resource abundance and the other responding positively to resource quality. We recommend that both resource quality and quantity be addressed when evaluating the relative importance of mechanisms in pollinator decline.