Since the reproductive success of many plants hinges upon animal pollen vectors, variation in floral visitation behavior by pollinators has important consequences for plant fitness. While work investigating pollinator quality has focused on differences among rather than within pollinator species, sex-associated variation has been reported to drive a number of large and discrete differences in the behavior and morphology of many pollinators. Whether these differences are large or prevalent enough to impact community-level or network-level processes, however, is unknown. To determine whether male and female pollinators differ in their network roles and whether these differences affect network architecture, we analyzed an extensive dataset of 309 networks collected over 10 years. Using a simulated null model approach, we compared networks with nodes split by sex to networks split randomly. 1) At the node level, we predicted that the sexes would occupy different network roles, with opportunistically foraging males visiting more species but being less central to the networks, and with males and females visiting different communities of partners. 2) At the network level, we predicted that differences in partner identity and foraging patterns would result in more specialized and vulnerable networks overall.
Results/Conclusions
We report 2 main findings. 1) Contrary to predictions, females tended to have higher connectedness (degree) than males; the magnitude of this difference was significantly higher than null expectations in 11% of observations. Females were also less central to the networks (closeness centrality), less specialized in their partners (d’), and visited different partners than males (Morisita-Horn dissimilarity), with observed differences being significantly larger than the null in 19%, 14%, and 17% of observations. 2) Networks with nodes split by sex tended to be less nested (NODF), more specialized (H2) and as a result had lower niche overlap between nodes, with 7%, 14%, and 12% of networks showing values significantly different from expectations. Together, these results suggest that sex differences in pollinator behavior are widespread and can impact the roles individuals play within plant-pollinator communities as well as overall network function. It is also clear, however, that many species show relatively small differences between the sexes despite differences in behavior; this may be because intrinsic specialization or a limited flower community constrains the visitation patterns of the sexes to be similar. We conclude that understanding sex-associated variation is important in its broad generality and cannot be treated simply as random variation.