Consistent negative response of US crops to high temperatures in observations and crop models

Abstract

High temperatures are detrimental to crop yields and could lead to global warming-driven reductions in agricultural productivity. To assess future threats, the majority of studies used process-based crop models, but their ability to represent effects of high temperature has been questioned. Here we show that an ensemble of nine crop models reproduces the observed average temperature responses of US maize, soybean and wheat yields. Each day >30 °C diminishes maize and soybean yields by up to 6% under rainfed conditions. Declines observed in irrigated areas, or simulated assuming full irrigation, are weak. This supports the hypothesis that water stress induced by high temperatures causes the decline. For wheat a negative response to high temperature is neither observed nor simulated under historical conditions, since critical temperatures are rarely exceeded during the growing season. In the future, yields are modelled to decline for all three crops at temperatures >30 °C. Elevated CO₂ can only weakly reduce these yield losses, in contrast to irrigation.

Figure 1. Comparison of statistically estimated effects of temperatures on observed and simulated US yields in rainfed counties.

Columns are maize (a,d,g), soybean (b,e,h) and wheat (c,f,i). a–c show regression coefficients and d–f show the histogram of times spent in individual temperature bins as the sum of times derived for each grid point across the growing seasons. g–i show rainfed counties (black outlines) with their per cent land-use share (colours) of the respective crop (for wheat only counties with predominantly winter wheat). Black lines in a–c: coefficients γ_h derived from log-transformed observed yields (Methods; equation (1)). Green/blue lines: coefficients of the ensemble median rainfed/irrigated simulated yields. Estimates are derived by a panel regression of US county data, where the considered crop is grown under predominantly (>90%) rainfed conditions. Shaded areas represent 95% confidence intervals. Simulated coefficients are marked by coloured dots if they are significantly different from the observed coefficients (confidence intervals do not overlap).

Figure 2. Comparison of statistically estimated effects of temperatures on observed and simulated US yields in irrigated counties.

Unconstrained irrigation is assumed on the irrigated areas specified by MIRCA2000 (ref. 24). Columns are maize (a,d), soybean (b,e) and wheat (c,f). a–c show regression coefficients and d–f show irrigated counties (black outlines) with their per cent land-use share (colours) of the respective crop. Counties are considered as irrigated if >75% of the crop-specific-harvested area is irrigated. Black and blue lines in a–c represent coefficients γ_h for observed and simulated yields, respectively. Shaded areas are 95% confidence intervals. Results for individual models are shown in Supplementary Fig. 33.

Figure 3. Simulated yield responses to temperature under future climate change in rainfed counties.

Columns are maize (a,d), soybean (b,e) and wheat (c,f). a–c show regression coefficients and d–f display temperature histograms for the historic (dashed grey) and future (solid red) periods; future climate is evaluated over 2071–2099 based on RCP8.5. Green tone lines in a–c are ensemble yield responses to temperature under rainfed conditions. Blue tone lines are ensemble yield responses under irrigation. Solid lines are derived with fixed present-day [CO₂], while dotted lines include elevated [CO₂] according to RCP8.5. Shaded areas are 95% confidence intervals. Rainfed counties are defined in Fig. 1.