Evidence for grassland and agriculture in the landscape suppressing soybean aphid (Aphis glycines) abundance and improving field-scale soybean yields

While there is some evidence that landscape properties affect crop yields through cascading effects on agricultural insect communities, previous studies have extrapolated field-scale yields from findings in small plots or cage experiments. Understanding how landscape context impacts yield is critical for growers and policymakers, as it is the scale at which growers manage their land and make pest insect management decisions. Thus, our objective was to assess the contribution of landscape composition to crop yield at field scale (bushels per acre) through impacts on crop pests and natural enemies.We predicted that soybean aphid abundance would be lower, biological control potential higher, and crop yields higher in fields with more grassland in the surrounding landscape, due to grassland habitat providing prey resources early in the growing season for generalist aphid predators. 

Our approach was two-tiered. The first, our "intensive" sampling, consisted of weekly sampling at 28 working soy fields in 2014, representing a wide range of landscape and climatic conditions.  At these sites, we measured pest densities with weekly field observations, biological control potential using clip-cage assays, and yields using standardized quadrats, as well as field-scale yields.  The second, our "extensive" sampling, consisted of analyzing grower-reported field-scale yields from 74 working soy fields between 2011 and 2014.  Using linear mixed effect models, we tested how yield varied with landscape composition, growing degree days, pest densities, and biological control potential.

Our results supported our hypothesis that increasing grassland in the landscape is strongly associated with higher field-scale yields. Our finding was consistent in both our "intensive" (p<0.01) and "extensive" (p<0.01) site analysis.  Yet we did not find evidence of a trophic cascade, as these relationships were not consistent with patterns in pest densities and biological control potential. We also did not find any relationships between landscape composition and quadrat-scale yields. Our results thus demonstrate the importance of using field-scale yields to understand patterns at the scale most relevant to growers and that of semi-natural habitat in landscapes to support improved crop yields, potentially through conservation of generalist predators.