Climate change is altering the ranges and phenology (timing of life history events) of many species. One particularly important and understudied element of these spatial and temporal shifts is the effect that they will have on species interactions, which in turn has important implications for ecosystem functioning. Understanding how climate change will impact interactions requires information not just on the spatial distribution of participating species––will species occupy in the same location––but also on the temporal dimension of co-occurrence––will species be active at the same times. Here, we examine some novel applications of the MaxEnt species distribution modeling framework with which we predict the spatial and temporal distribution of plant-pollinator interactions. We present case studies of this modeling framework using museum specimens and community science data for several plant-pollinator pairings and determine the extent to which spatial and phenological patterns can predict the occurrence of interactions between these groups.
Results/Conclusion
While analysis for this project is still ongoing, preliminary results suggest that spatial overlap alone does not adequately predict plant-pollinator interaction occurrence. While relying upon spatial co-occurrence of species appears superficially effective at predicting where interactions are unlikely to occur, tests of niche equivalency show that the predictions by these co-occurrence models differ substantially from models built using empirical data from observed plant-pollinator interactions. We expect that phenological overlap represents an important overlooked component of interaction occurrence and are continuing to develop a modeling framework that reflects both the spatial and temporal element of plant-pollinator interaction occurrence. With these results, we offer some insights into how distribution models can be used to predict climate-driven shifts in species interactions and highlight the importance of widespread pollinator monitoring to provide robust data for these types of analyses.
