Abstract. 1. Despite a growing interest in forest canopy biology, very few studies have examined the vertical distribution of forest bees. In this study, bees were sampled using 12 pairs of flight-intercept traps suspended in the canopy (≥15 m) and near the ground (0.5 m) in a bottomland hardwood forest in the southeastern United States.
2. In total, 6653 bees from 5 families and 71 species were captured. Augochlora pura (Say) (Halictidae), accounted for over 91% of all bees collected and was over 40 times more abundant on average in the canopy than near the ground.
3. Even after removing A. pura from the dataset, bee abundance, richness and Shannon’s diversity were considerably higher in the canopy than near the ground.
4. According to both non-metric multidimensional scaling and analysis of similarities, the bee community in the canopy was distinct from that near the ground. Based on indicator species analysis, A. pura, Lasioglossum imitatum (Smith), Lasioglossum versatum sensu Mitch., and Lasioglossum zephyrum (Smith) were significantly associated with the canopy whereas Andrena personata Robertson and Lasioglossum macoupinense (Robertson) were significantly associated with the ground.
5. Augochlora pura was consistently more abundant in the canopy than near the ground throughout the season, but was more so in mid-to-late summer (i.e., June–September), a period coinciding with low floral resource availability. As a group, the remaining bee community exhibited a similar pattern.
6. We suspect that bees frequent the canopy, particularly during times of low nectar and pollen availability, to acquire non-floral resources such as honeydew and sap.