Interannual variation in quantitative plant-pollinator networks

Background/Question/Methods

Understanding patterns of temporal variation in plant-pollinator networks and their underlying causes has important implications for mechanisms of evolution and resilience of communities and ecosystem services to environmental change. Plant-pollinator interactions are among the most studied ecological interactions. In the last decade, the use of networks for describing community wide plant-pollinator systems has captured the attention of ecologists and resulted in an upsurge of studies. These studies have yielded generalities about network structure. Still, temporal variation in these networks has been largely ignored. Here we use quantitative plant-pollinator networks spanning five years with two main objectives: 1) to quantify inter-annual variability in species composition, occurrence of pairwise interactions and overall network structure; 2) to explore two possible mechanisms responsible for this variability, namely phenological shifts and “rewiring” of interactions.

Results/Conclusions

We observed 791 different interactions among 55 species of plants and 160 species of animals. Of these interactions, a majority were observed only in a single year (63%) and very few were consistently observed across all years (2%). We found that community composition and the occurrence of interactions varied considerably across years (Bray-Curtis Distance: plants = 0.33; pollinators = 0.47; interactions = 0.79). To disentangle interaction turnover from species turnover, we calculated the number of possible interactions (those observed in at least one year) that were realized given that the interacting partners were present in the same year. We found that only around half of these possible interactions were observed. For variance in network structure, we observed significantly less variance across years in connectance, average number of links and evenness than expected from null models. Lastly, we found that Bray-Curtis distance in interactions did not correlate with distance in flowering phenology (i.e., phenological shifts) for pairwise combinations of years. In summary, this study shows a high turnover in interactions regardless of temporal overlap of interaction partners. The flexible nature of these interactions suggests that networks may be more robust than previously assumed. It also highlights the need for multiple years of data to capture the variability of plant-pollinator interactions.