Linking species-level network metrics to flower traits and plant fitness

Theoretical models indicate that the structure of plant-pollinator networks has important implications for the reproduction and survival of species. However, despite the growing information on the mechanisms underlying such a structure, it is still difficult to predict the functional consequences of species' structural positions in these networks. From the plant perspective, species position and roles in pollination networks might be related to traits describing flower attractiveness, availability and dependence on pollinators. In turn, both network metrics and plant traits might influence plant species fitness. During two field seasons, we collected data from the 23 most abundant plant species in a rich coastal community, to evaluate the association between population and floral traits (floral abundance at population level and flowers per individual, flower shape and size, flowering length, nectar volume, pollinator dependence), species-level network metrics (linkage level, specialization -d '-, weighted closeness centrality, network roles related to modularity) and plant fitness (seeds/flower, seed weight). Flowering length, flower size, flower abundance and pollinator dependence were positively related to greater generalization, as measured by various indices. More abundant species and those with larger flowers showed higher linkage levels (i.e. higher number of pollinator species), whereas longer flowering periods were negatively related to d ' and positively related to closeness centrality and important roles in the network. Likewise, plants species more dependent on pollinators occupied more central positions in the network. Furthermore, centrality in the networks was significantly associated with plant fitness. Specifically, central species in the network produced more and heavier seeds than the others. However, other plant traits, such as flower size and pollinator dependence had additional direct effects on seed production. Synthesis. Our study highlights how population and floral traits define the positions and roles of species structuring the pollination communities. Moreover, the relationships between network metrics and plant reproduction indicate, for the first time, the functional implications of these structural positions at the interspecific level of community assembly.