Flower constancy
An individual bee was categorized as flower constant if the pollen pellet brought back to the hive contained pollen grains from a single morphotype or family. We calculated the proportion of foragers that were flower constant each day for a bee species, time period, and site. Linear mixed models (proc mixed in SAS v. 9.4) were used to determine the impact of bee species, time period, site, and their two-way interactions on the proportion of foragers that were flower constant. Day was the replicate in the model and bee species, period, site, and the two-way interactions were fixed effects. The three-way interaction term, site*period*species, was included as a random effect in the model, and provided the error term against which each fixed effect was tested. For each statistically significant factor, multiple means comparisons were performed to determine differences among levels of each factor.