Discussion
Our study is the first to demonstrate that bees occupy the aerosphere immediately above the forest canopy; furthermore, the community above the canopy was compositionally distinct with similar abundances compared to lower strata (understory, midstory, canopy). These findings expand our understanding of forest bee communities and build on earlier research that revealed differences between understory and canopy bees (Cunningham-Minnick & Crist, 2020; Milam et al., 2022; Ulyshen et al., 2010; Urban-Mead et al., 2021). However, when attempting to characterize the forest bee community, the importance of sampling the canopy-aerosphere interface hinges upon the question of whether these bees should be considered as part of the forest community, or if they are transients moving among resources. The fact that the above-canopy assemblage was generally characterized by many species that were not observed at lower strata and were also associated with non-forested habitats (e.g., Agapostemon texanus Cresson,Halictus parallelus Say, Peponapis pruinosa(Say); Harrison et al., 2018) suggests that while some bees may forage on floral resources available at tree crowns in the spring, many others may be moving over the forest to access other habitat patches or resources throughout the season, as reported in other insect taxa (Wainwright et al., 2017). Nevertheless, the presence of an abundant and species rich bee assemblage at the canopy-aerosphere interface, which had not previously been considered, suggests that more studies are needed to address the extent to which these bees should be considered a subset of the forest bee community.
This study also demonstrated how the vertical stratification of forest bee communities changed throughout the flight season. Our observations confirm the findings of previous studies that have documented a positive relationship between day of year and bee abundance in the canopy (i.e., Cunningham-Minnick & Crist, 2020; Ulyshen et al., 2010; Urban-Mead et al., 2021). However, while this pattern may reflect a response to the depletion of floral resource availability near the ground in temperate forests (Proesman et al., 2019; Ulyshen et al., 2010), it remains unclear why bees moved to the canopy. It is possible that the canopy provides alternative foraging resources (Campbell et al., 2018; Ulyshen et al., 2010), nesting opportunities (Cunningham-Minnick & Crist, 2020), or favorable conditions associated with tree leaf phenology (Urban-Mead et al., 2021). Our results add a layer of complexity to the issue by demonstrating that this temporal pattern extends vertically beyond the forest canopy and involves a compositionally distinct subset of the bee community that may be responding to a mix of environmental cues. For example, the highest bee abundance across sampling points in the spring was at the canopy-aerosphere interface during the peak bloom of A. rubrum , suggesting that this stratum may provide access to floral resources of the forest canopy. However, it seems unlikely that summer bees above the canopy were foraging or nesting since forest floral resources were depleted and most were soil-nesting species. Vegetation height has been negatively associated with bee abundance and diversity (Roberts et al., 2017); therefore, bees may instead use the canopy-aerosphere interface for movement or dispersal since this space lacks the obstacles created by the vegetation structure of forest interiors. Alternatively, bees may be physiologically driven to take advantage of the greater light intensities and warmer temperatures above the canopy compared to other strata to forage earlier or later in the day (Fig. A5; Kebler et al., 2019; Roubik, 1993). It is also possible that some species were seeking mates above the canopy. For instance, groups of male Apis mellifera L. mate with females 10 – 40 m above the ground (Ruttner, 1966); similarly, male groups of someBombus species will fly to higher elevations to mate with emerging females, a behavior known as ‘hill-topping’ (Goulson et al., 2011). Though A. mellifera and Bombus spp. comprised 56% of the overall abundance of bees above the canopy, these behaviors are unlikely to explain our findings because only three individuals of these species were males. Alternatively, males of two solitary soil-nesting species, Andrena imitatrix Cresson and A. mandibularisRobertson, comprised 57% of bee abundance above the canopy in the spring, though we are not aware of any studies addressing hill-topping behavior in these species or the Andrena genus. Thus, it appears that many of the species above the canopy were flying over the forest to unknown ends.
There were notable differences in bee assemblages among the other strata that may be best explained through life-history traits. For instance, bees in our study that nest in moist, decayed wood (e.g.,Augochlora pura (Say), Lasioglossum coeruleum(Robertson), L. cressoni (Robertson), L. subviridatum(Cockerell)) or pithy twigs (e.g., Hylaeus spp., Ceratinaspp.) were nearly absent above the canopy (< 1%), while 77% were found in the canopy and midstory, and only 22% of bees from this guild were sampled from the understory. Our findings are consistent with other studies that demonstrated a high abundance of wood-nesting bees within the canopy (e.g., Urban-Mead et al., 2021; Cunningham-Minnick & Crist, 2020; Campbell et al., 2018; Ulyshen et al., 2010) and suggest that bees that nest in wood, including species that nest in moist decayed wood, or ‘soft’ wood, exhibit a preference for canopy strata within forests likely due to the availability of wood nesting substrates. Nesting substrate in the canopy has yet to be tested as a mechanism in explaining high abundances of wood-nesting bees within the higher strata of forests due to the known availability of dead and rotting wood on the forest floor, despite a lack of correlation between coarse woody debris on the ground and the abundance of this guild in the canopy (Urban-Mead, 2021; Campbell et al., 2018; Ulyshen et al., 2010). Therefore, studies that quantify potential nesting substrates for wood-nesting bees within the canopy, including those that nest in ‘soft’ wood, are clearly needed to resolve these discrepancies (Harmon-Threatt, 2020).
Milam et al. (2022) found that the inclusion of canopy sampling in addition to understory sampling did not influence their ability to characterize the forest bee community. Our study supports their conclusion when only considering bees below the maximum height of the canopy (i.e., understory, midstory, and canopy strata), but further demonstrates that the bee community above the canopy is distinct from lower strata. The existence of bees above the forest canopy is highly relevant to understanding pollinator ecology and may have additional implications for their conservation vis a vis our understanding of the effects of habitat fragmentation and isolation on bee movements and related population processes (Proesman et al., 2019; Roberts et al., 2017; Winfree et al., 2009). Though our study was limited in sampling intensity, it clearly demonstrates the complexities of spatiotemporal bee dynamics within forests, suggests a new perspective on the role of forests in the surrounding landscape, and emphasizes caution when drawing conclusions about forest bee communities that were sampled with vertically or temporally restricted designs. Thus, our study supports the growing body of literature that asserts the need for additional baseline research of forest bee communities along the full vertical gradient to inform forest management and bee conservation.