Patterns of Regional Yield Stability in Association with Regional Environmental Characteristics

Regional‐level recurring spatial patterns of yield variability are important for commercial activities, strategic agricultural planning, and public policy, but little is known about the factors contributing to their formation. An important step to improve our understanding is recognizing regional spatial patterns of yield variability in association with regional environmental characteristics. We examined the spatial distribution of county‐level mean yields and CVs of mean yields of four functionally different crops—corn ( Zea mays L.), soybean [ Glycine max (L.) Merr.], alfalfa ( Medicago sativa ), and oat ( Avena sativa L.)—in Iowa using Moran's Index of spatial autocorrelation. Patterns of association with 12 county‐level climatic, edaphic, and topographic environmental characteristics were examined using partial least squares regression. Two distinct geographic provinces of yield stability were identified: one in the northern two‐thirds of the state characterized by high mean yields and high yield constancy, and one in the southern third of the state characterized by low mean yields and low yield constancy. Among eight partial least squares regression models, which explained 50 to 81% of variation of mean yields and yield CVs, mean organic matter and mean depth to seasonally high water table had greatest relative importance to mean yields of grass crops and legume crops, respectively. Among the CV models, variables describing water availability were of greatest relative importance, with less distinct differences between grass and legume crops. Partial least squares regression is a potentially powerful tool for understanding regional yield variability.