Checking billy barr’s snow log…
Abstract Species interactions, such as those between plants and pollinators, are known to be shaped by both evolutionary history and ecological factors. However, little is known about how multiple factors (e.g. phylogeny, phenology, abundance and functional traits) interactively affect interaction patterns. Using a plant–bumblebee network comprising 2,428 interactions between 29 plant species and 12 bumblebee species in the Himalaya–Hengduan Mountains, we tested for phylogenetic signal and whether phylogenetic pattern was explained by abundance, phenology and 13 plant and bumblebee functional traits. We also tested whether trait matching in two pairs of plant–bumblebee traits explained interaction frequencies at both species and individual levels. The network showed significant phylogenetic signal; closely related bumblebees tended to visit similar sets of plants, but not vice versa. Among all the measured factors, nectar volume and sugar concentration, rather than other phylogenetically constrained factors, were most important for explaining phylogenetic pattern (64% and 54% respectively). Although long–tongued bumblebee species tended to interact with long–tubed flowers, trait matching did not predict short–tongued bee interaction frequencies. Despite this, trait matching was apparent at the level of individual bees, reflecting intraspecific variation in tongue length and body size. Different selection pressures may exist within this bumblebee community, resulting in specialized, co–evolved traits in long–tongued species and adaptive generalization in short–tongued species to gain access to a variety of flowers. Together, these findings contribute to our understanding of phylogenetic trait–based structure in species interaction networks. In particular, this study provides new evidence for the importance of nectar rewards in structuring interaction patterns in pollination mutualisms. The results also demonstrate that trait matching may occur at the individual level, despite not being detectable at the species level, and underline the necessity of taking intraspecific trait variation into account in studies of community structure. A free Plain Language Summary can be found within the Supporting Information of this article.