Linking plant-pollinator networks with traits to understand interaction variation

Laura Burkle, Department of Ecology, Montana State University, Bozeman, MT, Evan Fricke, Rice University, Houston, TX and J. Simone Durney, Ecology, Montana State University, Bozeman, MT

Background/Question/Methods
Ecologists are intrigued by the causes and consequences of spatial and temporal variation in the occurrence of species interactions. Species traits and phenology are well-documented drivers of plant-pollinator interactions. Additionally, we increasingly appreciate spatiotemporal scale-dependence of interaction network structure. Yet we lack an understanding of the degree to which species traits and phenology contribute to plant-pollinator interactions at different spatial and temporal scales. Further, the role of intraspecific trait variation (ITV) in driving interspecific patterns of interactions is poorly understood.
First, we aim to understand how predictable plant-pollinator interactions are based on traits and phenology, and how this predictive capacity varies across spatial and temporal scales. Second, we test hypotheses about the role of ITV in plant-pollinator network structure.
We sampled flowering plants, bees and their interactions across 140 sites in Montana over four years. For bees, we measured intraspecific variation in intertegular distance (which is related to tongue length, body size, and foraging distances). For plants, we measured flower size and number, height, and specific leaf area. By considering three spatial scales and four temporal scales, we developed boosted regression tree models to investigate how well species identities, phenology, traits, or their combination predict plant-pollinator interactions at each scale.
Results/Conclusions
Given changing climate and shifts in species’ ranges, there is interest in trying to predict which species will interact at broad spatiotemporal scales, where the predictive capacity is strongest, and the most important factors to consider in these predictions. Plant-pollinator interactions were difficult to predict at small spatiotemporal scales, and predictions became better at increasingly large scales. Traits improved predictions at short temporal scales, and traits became less meaningful compared to phenology at long time scales and broad spatial scales.
While there was little evidence of relationships between ITV in bee body size or in plant traits with interaction richness (e.g., diet breadth) across species at local scales, it became important at landscape scales. Plant species with higher coefficients of variation in flower size interacted with fewer bee species (i.e., more specialized), but only at broad spatial scales. ITV in bee body size was positively related to plant-pollinator network structure, while interspecific variation was not. Both intraspecific and interspecific plant trait variation were important for network structure, and these relationships varied in magnitude and direction depending on the trait. Intraspecific trait variation has important implications for interaction network structure at a variety of spatiotemporal scales and deserves additional attention.