Fig 1.
Percent visitation of bee groups.
Percent visitation of bee groups were collected at 19 urban agricultural sites using pan traps (black bars) and visually observed (gray bars) visiting cucumber flowers during 2011.
Fig 2.
Bee community composition in relation to local and landscape variables.
NMDS ordination depicts the relationship between study sites (black points) and landscape variables (blue vectors) in 2-dimensional space. Impervious cover was the only landscape variable significantly negatively correlated with axis 2 (R2 = 0.48, P = 0.009). No variables were significantly correlated with axis 1. Open circles represent bee species with abundances greater than 35 individuals, which highlights the most common species collected.
Table 1.
Model selection results for bee abundance by body size and nesting habit.
Table 2.
Variable weights calculated for each explanatory variable in the full models for abundance.
Table 3.
Model selection results for bee richness (rarefied), community composition, and fruit set.
Table 4.
Variable weights calculated for each explanatory variable in the full model for bee richness, community composition, and fruit set.
Fig 3.
Linear regression showing the relationship between NMDS axis 2 and percent impervious cover.
A significant negative correlation was found between bee community composition (NMDS axis 2) and impervious cover measured in the 500 m surrounding study sites (R2 = 0.34; P = 0.0078).
Fig 4.
Linear regression showing the relationship between fruit set and percent hardscape.
A significant negative correlation was found between fruit set measured using sentinel cucumber plants at each study site and percent hardscape (R2 = 0.23; P = 0.038).