Do ecosystem service-providers and rare bees prefer the same plant species? A three-year experimental field study

Background/Question/Methods

In plant-pollinator networks, the specialist species with few interaction partners tend to interact with highly linked generalist partners, which themselves interact with other generalists. We predicted that this nested structure of plant-pollinator networks could result in convergence of the traditional biodiversity approach and the ecosystem services-based approach to pollinator conservation/restoration. Specifically, we predicted that the rare specialist pollinators in need of protection, and the common generalists that provide ecosystem services to crops, would use the same generalist plants. We tested this prediction using a field experiment consisting of 102 monospecific plots of 17 plant species from which we collected pollinators. All plant species were present at the same abundance, thus removing any sampling effects caused by a skewed species-abundance distribution for plants, and allowing us to directly measure pollinator preference. To define rare and ecosystem service-providing (ES) bee species we used independent databases on the bees of our region. We used negative binomial mixed models to determine whether plant species varied in the abundance of rare or ES bees. We then used a rank correlation to test whether plant species that attract a high (or low) abundance of rare bees also attract high (or low) abundance of ES bees.

Results/Conclusions

In our 3-year field experiment, we collected 7702 bee specimens of 112 species. Using independent data sets collected by our lab group, we defined the rare bee species to be the 87 species found at only one or two sites, out of 13,738 specimens of 262 species collected at 192 sites in our study region. We defined the ES bees to be the 12 bee species accounting for 62% of all crop flower visits, using 11,800 specimens of 152 bee species collected from 5 crops in our study region. Both rare and ES bees had strong, significant preferences (P<0.01 in all cases) for certain plant species in all bloom periods, and consistently so across years. Furthermore, the rare and ES bees preferred the same plants: plant species ranks according to abundance of rare and ES bees were positively, significantly correlated (P<0.001). These results suggest that pollination networks can inform the design of pollinator restorations to to simultaneously achieve conservation of rare bee species and ecosystem service-providing species.