In nature, plants and pollinators interact in complex networks, but studies of how pollinator diversity affects pollination-function have largely been limited to monocultural crops. In this context, numerically dominant pollinators are most important for function. However, in native plant communities, we might expect more pollinator species to become important due to stochastic effects (e.g., annual fluctuations in the bee community) and/or biological effects (e.g., species-specific phenologies or floral preferences). Using 3 years of observations from an experimental garden of 12 native plant species, we ask, (i) how does the number of bee species that are important to pollination (defined as those providing >5% of visits to at least one plant in one year) increase with increasing plant diversity? and (ii) to what extent is this effect driven by sampling and annual fluctuations in the bee community (stochastic effects), vs. bee phenology and preferences (biological effects)? To answer (i), we resampled our data to simulate plant communities of varying richness and composition, then measured accumulation of important bee species across 1 to 3 years of sampling. To answer (ii), we compared these results with those of three randomization-based null models that successively narrowed the timeframe over which observations were randomized.
Results/Conclusions
We collected 5452 individual bees of 69 species visiting the 12 plant-species in the three years of our experiment. The number of bee species important to pollination increased strongly across plant species and time, from 3.9 for single plant species in a single year to 31 bee species for 12 plants after 3 years. This accumulation followed a saturating pattern but did not asymptote. Our null models show some accumulation of important bee species could be expected due to sampling effects and annual fluctuations in the bee community, but much more is expected to result from differences in the phenology and preferences of bee species. Notably, 17 of the 31 bee-species important to pollination were rare, constituting < 1% of total individuals in our data set. Our results contrast with many previous biodiversity-ecosystem function studies based on monoculture crops, which show little contribution from rare bees. We show that niche differences (i.e. differences in phenology and preference) among pollinator species create a greater need for pollinator diversity in native plant communities. Therefore, to better understand real-world biodiversity-pollination function relationships, we need to move beyond monocultures to consider pollination-function in the context of complex mutualistic networks.