Phenologically Explicit Robustness Metric Reveals Increased Vulnerabilities in Temporal Plant-Pollinator Networks
Robustness is a useful metric in evaluating the ability of an ecosystem to resist further extinction after species loss. This measure is straightforward to calculate from empirical interaction data, however doing so typically requires a static network of time-aggregated interactions, ignoring the finer-scale variation imposed by species phenologies. Although the analysis of time-aggregated ecological networks has been successful in capturing their overall structure, it often leads to the loss of information of the within-season variation that ecosystems typically exhibit. We demonstrate how this approach can also dramatically overestimate robustness by assuming all interactions are concurrent. Here, we develop a new measure of temporal robustness that incorporates the time-varying nature of species interactions. We apply our measure to plant-pollinator systems using both simulated and empirical networks, and obtain analytical predictions for the temporal robustness of randomly generated networks. In each case, our analysis reveals a substantial loss of information in the established static-network robustness measure, which obscures the crucial role of phenology in determining the vulnerability of pollinators to species loss. We find that pollinators active over long time periods appear more robust after time-aggregation, but typically have a much lower temporal robustness, indicating that long-lived pollinators may be more vulnerable to plant loss than previously thought.