Coordinated evaporative demand and precipitation maximize rainfed maize and soybean crop yields in the USA.

To understand how climate change affects crop yields, we need to identify the climatic indices that best predict yields. Grain yields are most often predicted using precipitation and temperature in statistical models, assuming linear dependences. However, soil water availability is more influential for plant growth than precipitation and temperature, and there is ecophysiological evidence of intermediate yield maximizing conditions. Using rainfed maize and soybean yields for 1970–2010 across the USA, we tested whether the aridity index, that is, the ratio of precipitation and potential evapotranspiration seasonal totals and a proxy of soil water availability, better predicts yield than growing season precipitation total, average temperature and their interaction. We also tested for non‐monotonic responses allowing for intermediate yield‐maximizing conditions. The aridity index alone explained 77% and 72% of maize and soybean yield variability, compared with 78% and 73% explained by temperature, precipitation and their interaction. Yield responses were non‐monotonic, with yields maximized at intermediate precipitation and temperature as well as at intermediate aridity index of 0.79 for maize and 0.98 for soybean. The yield maximizing precipitation also increased with growing season average temperature, faster in maize than soybean. The intermediate yield maximizing conditions show that rainfed maize and soybean yields could both increase and decrease depending on whether climatic conditions come closer to or deviate from the yield maximizing conditions in the future. In most counties, during 1970–2010, the precipitation and aridity index were lower and temperature higher compared with those maximizing yields, suggesting that climate change will reduce yields. [ABSTRACT FROM AUTHOR]