COS 76-6 - A global assessment of pollinator limitation in agriculture systems: Prevalence, risks, and the relative importance of wild insects versus honey bees

Background/Question/Methods

Declining pollinator populations could threaten global food security. Three quarters of the world’s food crops are animal pollinated, and pollination limitation (i.e., reduced crop yields due to lack of pollination) has been detected in several studies. However, pollinator declines are less likely to affect yield stability for those crops that are not pollen limited. Here, we conduct the first comprehensive analysis of the extent of pollination limitation in agricultural systems worldwide. We ask 1) what is the global prevalence of pollen limitation?, 2) are there risk factors, such as crop type or region, that predict where pollen limitation is more pervasive?, and (3) what is the relative importance of honey bees versus wild insects as crop visitors? We answer these questions using the CropPol database, a new open-access compilation of 137 studies of 49 crop species. First, we use breakpoint regression and AICc model comparison to assess whether individual crop systems best fit a “pollination limited”; “somewhat limited”; or “not limited” model for crop yield. Second, we used logistic regression to evaluate whether crop system characteristics can predict the occurrence of limitation. Lastly, we compared whether honey bee flower visits, wild insect visits, or insect richness best predicted yield values.

Results/Conclusions

We found that 30% of global crop systems exhibited some pollinator limitation. Most of the factors we tested, including crop species identity, geographic region, bloom season, climate, and management practices did not predict pollinator limitation. However, preliminary results identified that crop fields that were closer to semi-natural habitat were pollinator limited less often. Finally, for most crop species, yields were best predicted by a model that included both honey bee and wild insect visitation, contradicting previous studies using smaller datasets which concluded that honey bee visitation does not increase crop yield. Our findings are promising because they suggest that farmers could ameliorate the effects of observed crop limitation with pollinator management, such as additional honey bee hives. We estimate that for those crops which were limited, increasing pollinator visitations could close yield gaps between high (90^th percentile) and low (10^th percentile) yielding farms by a median of 48%. In addition, while most crop systems were not pollinator limited, we can predict that crops cultivated in landscapes lacking semi-natural land are likely at higher risk of pollinator limitation. Thus, targeted efforts to restore and enhance wild bee habitat in simplified landscapes could simultaneously address global food security and bee declines.