Biodiversity positively affects ecosystem function and becomes even more important when we want to maintain function at many places and times. Two knowledge gaps prevent a better understanding of how biodiversity drives ecosystem function at real-world spatial scales. The first is the role of dominance. We know that all ecological communities have high numerical dominance, but less is known about whether dominance is also high when species’ contributions to function are considered, rather than their abundances. The second is the degree to which changes in species identity across space increase the number of species required for sufficient functioning at a larger spatial scale. Here, we sampled wild bee pollinators off three wildflower species (Polemonium reptans, Phacelia tanacetifolia, and Monarda fistulosa) at 25 sites in New Jersey, and also measured how many pollen grains each type of bee deposits per flower visit. We asked two questions: (1) How many dominant pollinators (those accounting for ≥ 5% of all pollination) does each plant species have when dominance is defined at the within-site vs. across-site (i.e. when species’ contributions to function were summed across sites) scales? (2) Do dominant bee species turn over in identity more than expected across sites?
Results/Conclusions:
We collected 8,840 individuals of 124 bee species. We found that each plant species had few dominant pollinators at the across-site scale (P. reptans: 5 bee species; P. tanacetifolia: 6, M. fistulosa: 5). However, each plant species had many more dominant pollinators when dominance was defined at the within-site scale (i.e., when a pollinator is dominant if it accounts for 5% of function within at least one site; P. reptans: 34 bee species; P. tanacetifolia: 24; M. fistulosa: 12). To determine whether the identity of dominant pollinators changes more than expected, we used a null model that redistributed individual bees randomly across sites, while maintaining bee abundance at each site. Thus, the null maintains functional dominance at the across-site scale and the distribution of bee abundance across sites, while removing variation in bee species identity across sites (i.e., each species is equally likely to be found at each site). For all plant species, the observed number of dominant species at the within-site scale exceeded the null model CI. We conclude rapid changes in the identity of dominant species across sites leads to a higher-than-expected number of species needed to maintain ecosystem function at multiple sites simultaneously.