Aspects of mutualistic network structure touted as important drivers of community resilience may exacerbate or reduce community vulnerability in unique ways depending on the type of external perturbation. Studies have identified diverse mechanistic factors that contribute to plant and pollinator losses including reduction of nesting habitat and food resources, parasites and pathogens, and phenological challenges impelled by climate change. By modeling the linkage and dependencies between plants and pollinators as interaction networks under various external perturbations, we may derive a better understanding of the unique risks associated with the anthropogenically driven declines based on network structure and biogeographic context. We utilize a five year dataset of plant and pollinator interactions to simulate plant and pollinator declines across 18 different communities, and we measure the contribution of network structure and biogeographic context to community resilience. The four modelled decline scenarios included frost damage to early blooming plants, drought, habitat loss, and social bee decline.
Results/Conclusions
Network structural metrics were partially predicted by the biophysical properties of habitat, and in turn, predicted 50-69% of the variation in network resilience. Meadow area partially explained pollinator diversity and linkage, while meadow soil moisture retention partially explained a higher partner diversity of plants and lower partner diversity of pollinators, but this result was not significant. The frost damage scenario was most detrimental to plant-pollinator networks with low moisture retention, and the resilience of networks to such perturbations were best predicted by species diversity, links per species, network nestedness and partner richness of plants. Drought, habitat loss, and social bee decline scenarios were most impactful in meadows with higher soil moisture retention, and the resilience of networks under these scenarios was best predicted by the number of compartments, interaction evenness, and other factors depending on the scenario. The resilience of mutualistic networks has rarely been investigated with regard to present-day externally driven threats, and these results underscore the utility of investigating network structure with regard to biogeographical context as well as threat-specific vulnerabilities.