Roughly two-thirds of major crops require pollination by bees and other animals, an ecosystem service worth tens of billions of dollars annually. Crop pollination is also a rich system in which to understand the links between management, biodiversity, and ecosystem services. Pollinators typically compose diverse guilds within communities that include crops and other plants, and they respond strongly to local habitat features and management. Over the last 15 years, pollination ecologists have built a strong evidence base for the spatial dynamics of crop pollination. Dozens of field studies have illuminated the importance of both farm-scale and landscape factors on pollinators, and regular meta-analyses have synthesized these studies into overall findings. Based on these studies, we developed a general model to link landscape features, pollinator communities, and crop pollination services. Here we use this model within agricultural landscapes in Costa Rica and the U.S. to ask: Where would deforestation most affect pollination services to nearby farmers? What are optimal locations and compositions of restored pollinator habitats? Where are areas of likely mismatch between supply and demand for pollination services?
In Costa Rica, we find that the effects of deforestation on pollinators and crop yields vary more than 100-fold across the landscape. Forests with highest marginal value for pollination services are concentrated in relatively few parcels, whose loss would reduce yields significantly in surrounding farms. In the U.S., our model predicts that restoring bee habitats will have highly varied impacts on pollinators and crop production. The effectiveness of restored wildflowers or hedgerows depends on habitat quality of the surrounding landscape, foraging behavior of bees, and the pollination biology of the crop itself. Applying the model across the entire U.S., we identify 139 counties with substantial mismatches between trends in pollinator supply (wild bee abundance) and demand (cultivated area). These counties comprise 39% of the pollinator-dependent crop area in the U.S. Together, these modeling efforts help to generalize the findings of ongoing field studies and predict impacts of changing land use and management on pollinators and crop production.