Plant-pollinator communities can be dynamic systems that gain and lose species as the flowering season progress. Even for species that remain active across the flowering season, shifts in plant or pollinator species relative abundance and in pollinator preferences can further alter interaction patterns. Consequently, plant-pollinator interaction networks should be dynamic over time as well, even within a season. Despite this, interaction networks used to describe the plant-pollinator community are typically characterized as static snapshots, constructed from observations collected within a short time period or pooled across weeks, months, or an entire flowering season. This approach would obscure any intra-annual changes in plant-pollinator interaction patterns as well as in the very structure of the network, with implications for understanding the stability of the community. In this study, we investigate how plant-pollinator community interactions and network structure change over short time scales throughout a flowering season. Specifically, we evaluate the degree of interaction turnover and determine the relative contribution of two underlying processes, species turnover and interaction rewiring. Additionally, we use interaction network metrics to quantify how the structure of community-level interactions changes across bi-weekly sampling periods and relative to the entire-season network.
Results/Conclusions
We constructed 9 interaction networks from >1600 hours of pollinator observations collected bi-weekly across a flowering season in a temperate grassland. We find interaction turnover between consecutive sampling periods and near complete species turnover from early to late in the season for both plant and pollinator species. However, our preliminary analysis indicates that a greater proportion of interaction turnover between consecutive sampling period is attributable to species rewiring rather than species turnover. Metrics of network structure varied across the season and were different than values calculated from the entire pooled network. For example, while the overall interaction network based on data pooled across the season was significantly nested, networks constructed from single sampling periods were not necessarily so, indicating that network structure itself can change over the season nested. Our work sheds light on the intra-annual variation of plant-pollinator interactions and the importance of considering interaction networks as dynamic systems.