Spatial-explicit responses of maize yield to temperature and precipitation across the United States

Abstract:

Spatial-explicit responses of maize yield to temperature and precipitation across the United States

Ru Xu, Yan Li^*

*Email: yanli.geo@gmail.com

The United States produces >30% maize production in the world. How maize yield responds to climate fluctuation is pivotal for accurate projection of climate change impact on crop productivity. Previous studies reported nonlinear temperature response of maize yield with a fixed optimal threshold (around 29 ℃). However, the actual crop yield response to climate, especially at large areas, is complicated and varies under different natural and human management conditions. Here, we spatially quantify the response types of maize yield to precipitation and maximum temperature using statistical county-level yield data from 1980 to 2010. Despite a large fraction of unresponsive type, the leading yield responses to precipitation are linear increase (29.47%), inverted-U-shaped (28.33%) which gives the optimal threshold, while the leading responses to max temperature are linear decrease (49.43%) and inverted-U-shaped (16.91%). The optimal yield thresholds for both maximum temperature and precipitation exhibit considerable spatial variations with higher thresholds appearing under warmer and wetter conditions, respectively. The precipitation threshold also increases with higher temperature, suggesting a more beneficial role of precipitation under hot conditions. Irrigation tends to lower precipitation threshold and raises temperature threshold relative to rainfed maize. When comparing yield responses derived from statistical data with ensemble simulations from AgMIP (Agricultural Model Intercomparison and Improvement Project), crop models show poor performance in representing yield response to precipitation. This misrepresentation, that is higher percent of increase type (62 %. vs 20%) and higher optimal threshold to precipitation (>90 mm) compared to statistical data, leads to a strong yield overestimation bias under extreme rainfall conditions. Our study highlights the importance spatial variations of maize yield response to climate which need to be better characterized in crop models.