Bee preference for floral resources
To determine whether bees demonstrate preferences for floral resources or foraged randomly, we compared the abundance of plant morphotypes in bee-collected pollen to the availability of plant resources in the environment. If bees forage randomly, we expect the representation of plants in pollen pellets to reflect plant resource availability. These analyses were done at the level of morphotypes, and to be included in the analyses a morphotype had to be present in both the pollen pellets and plant resource surveys at any point during the summer, but not necessarily in all periods. We examined the preference of each bee species over the entirety of the flowering season. We used period as a replicate and combined pollen collection periods 2 and 3 for these analyses to have equivalent periods for plant resource and pollen pellet collections (Table 1). The proportion of a morphotype in pollen pellets and resource surveys was calculated for each bee species during each comparable period, across all sites (Table 1). The data represent the mean proportion of a pollen type over all pollen pellets or over all resource survey quadrats for that period and bee species. Preference tests were performed separately for each bee species. We used a multivariate analysis of variance (MANOVA in proc GLM, SAS v.9.4) to determine if there were any overall preferences, followed by individual ANOVAs to test preference for individual morphotypes. A greater proportion of a morphotype in the pollen pellets relative to the resource surveys suggested a preference for that plant morphotype, while a lesser proportion was indicative of avoidance. The lack of a difference suggested that bees foraged randomly (i.e. based on the availability of plant resources).