COS 115-10
More pollinator species are required for pollination function at larger spatial scales, but high regional dominance can suppress this effect

Thursday, August 14, 2014: 4:40 PM
Regency Blrm D, Hyatt Regency Hotel
James R. Reilly, Department of Ecology, Evolution, & Natural Resources, Rutgers University, New Brunswick, NJ
Ignasi Bartomeus, Estación Biológica de Doñana (EBD-CSIC), Sevilla, Spain
Daniel P. Cariveau, University of Minnesota, MN
Faye Benjamin, Ecology & Evolution Graduate Program, Rutgers University, New Brunswick, NJ
Rachael Winfree, Department of Ecology, Evolution, and Natural Resources, Rutgers University, New Brunswick, NJ

Ecological understanding of the biodiversity-ecosystem functioning relationship is based largely on experiments, which by necessity are done at small scales. Here we use natural pollinator communities and the ecosystem services they provide studied across a 5000 km2area to extend previous results to larger scales. Specifically, we investigated how the number of bee species required to pollinate three crop plants changes with increasing spatial scale.  Pollinator abundance and identity were measured by collecting bees visiting flowers of watermelon, cranberry, and blueberry (16 sites each) in New Jersey and Pennsylvania. Per-species contribution to pollination function was measured as conspecific pollen grains deposited per flower visit, and total function per site was estimated by multiplying pollinator abundance by pollen deposition per bee.  We explored two questions: 1) how does the number of bee species required to achieve a target level of pollination at all sites change with increasing spatial extent of the study, and 2) how important is beta-diversity among the dominant species in driving this effect?  We used a minimum set approach to determine the number of bee species, chosen from the complete list of species observed at a site, necessary to achieve a target level of pollination at each site.


We collected a total of 3883 individual bees of 81 species visiting flowers of the three crops.  Following previous studies, we used 50% of observed pollination level as our target.  As spatial extent increased, the number of species required increased by a factor of 5.5 (watermelon), 3.7 (cranberry), and 6.8 (blueberry).  Despite this large increase, the absolute number of species required to reach the target at all 16 sites was only 18% of total species in the system for watermelon and 15% for cranberry.  For blueberry, 44% of total species were required, consistent with the higher species turnover and lower dominance observed in that system (Evar evenness index for blueberry: 0.32, cranberry: 0.19, watermelon: 0.19). Under a null model that randomized species identity, the total number of species required increased across all sites by a factor of 2.3 (watermelon), 3.2 (cranberry), and 1.5 (blueberry), and asymptoted more slowly.  This suggests that high regional dominance (lack of beta-diversity among the dominant species) is suppressing the number of species needed for crop pollination, especially in watermelon and cranberry.  Our results highlight the importance of considering species abundance variation when analyzing ecosystem functions in real-world systems.