Figure legends

Figure 1. Map of collection sites in Adams County, Pennsylvania, USA. Each yellow marker is the location of a single Blue Vane trap which was left out to capture bees from April to October for 6 years. The four shapes show the collection sites that were closer than 900 m and were lumped together for data analysis. The town of Biglerville is seen at the bottom right.
Figure 2. A) Species accumulation curve shows how the average number of species detected increases with the total number of bees collected. The flattening of the curve suggests that most, but not all, of bee biodiversity is represented in our collections. B) Rank abundance curve shows the number of individuals collected for each species ranked from highest to lowest, note the log y-axis. In our dataset of over 26,000 bees, only 10 species had over 1000 individuals while over half of the species had 5 or fewer individuals.
Figure 3. Patterns of bee biodiversity across months, and changes across years. All model relationships were highly significant (P<0.002). Abundance (A, E) is the average number of bees collected per site. Richness (B, F) is the number of bee species per site. Diversity (C, G) is the inverse Simpson’s diversity index. Phylogenetic structure (D, H) is standardized effect size of mean pairwise distance (higher values are more even, and lower values are more clustered). Line fits and adjusted R2 values are from general additive models and the shading represents 95% confidence intervals of the models.
Figure 4. Effects of months (A) and years (B) on bee community composition. Both months and years have significant effects on bee composition but differences among months are larger than among years. Data are visualized using non-metric multidimensional scaling on a Bray-Curtis dissimilarity matrix which includes species abundances. The R2 values are the variance explained from perMANOVA models.
Figure 5. Species-level phenological patterns and changes in abundance over time for 40 species with at least 30 individuals collected. The colored heatmap shows the percentage of individuals captured for each species, therefore a value of 100% would mean all individuals of that species were captured in that one month. The black and gray points represent the positive or negative change in abundance over time. The size of the points are from coefficients from linear models (i.e. slope of the relationship between year and abundance using standardized data). The phylogeny has our focal species amended (see methods) to a genus-level tree by Hedtke et al (2013).
Figure 6. The distributions of both seasonality and phenological breadth among bee genera and families. Seasonality is the median Julian date in which each genus and species was captured across 6 years of continuous sampling. Phenological breadth is a measure of the length of time in which bees are active. Error bars show the highest and lowest values for species in each genus. Red dotted lines illustrate the conceptual idea of “phenological syndromes”. The bottom left quadrant are early emerging species with narrow phenological breadth. Species in the bottom right quadrant emerge in summer but still have narrow breadth. Species in the top right have wide phenological breadth and are most abundant in summer.