Neglecting rice milling yield and quality underestimates economic losses from high-temperature stress

Abstract

Future increases in global surface temperature threaten those worldwide who depend on rice production for their livelihoods and food security. Past analyses of high-temperature stress on rice production have focused on paddy yield and have failed to account for the detrimental impact of high temperatures on milling quality outcomes, which ultimately determine edible (marketable) rice yield and market value. Using genotype specific rice yield and milling quality data on six common rice varieties from Arkansas, USA, combined with on-site, half-hourly and daily temperature observations, we show a nonlinear effect of high-temperature stress exposure on yield and milling quality. A 1 °C increase in average growing season temperature reduces paddy yield by 6.2%, total milled rice yield by 7.1% to 8.0%, head rice yield by 9.0% to 13.8%, and total milling revenue by 8.1% to 11.0%, across genotypes. Our results indicate that failure to account for changes in milling quality leads to understatement of the impacts of high temperatures on rice production outcomes. These dramatic losses result from reduced paddy yield and increased percentages of chalky and broken kernels, which together decrease the quantity and market value of milled rice. Recently published estimates show paddy yield reductions of up to 10% across the major rice-producing regions of South and Southeast Asia due to rising temperatures. The results of our study suggest that the often-cited 10% figure underestimates the economic implications of climate change for rice producers, thus potentially threatening future food security for global rice producers and consumers.

Figure 1. Rice (Oryza sativa L.) developmental stages.

This figure describes the developmental stages relevant to this study. a, Vegetative, reproductive and ripening growth stages are defined relative to the observed 50% heading and emergence dates at each station. The vegetative, reproductive, and ripening stages are defined as the intervals [emergence, H –30 ), [H –30, H +5 ], (H +5, harvest ], respectively, where H denotes 50% heading. The ripening stage is divided into early and late grain filling denoted by W2 and W3 to account for the differential effects of temperature on the physiological processes occurring during these periods. b, Harvest dates are not available for the paddy yield data, so we approximate harvest as 40 days after 50% heading. Harvest dates are available for the milled rice yield and quality data, but some plots were not harvested at maturity to allow harvest moisture content (HMC) to decrease. To avoid inclusion of temperatures beyond maturity, H +40, harvest is used as the harvest date for milling yield and quality trials.

Figure 2. High temperature effects on paddy rice yield.

Represented above are the marginal effects and total effects of high temperature on paddy yield. a, Marginal effect of a 1°C increase in average daily temperature during the vegetative stage on paddy yield. b, Marginal effects of an additional unit of high temperature exposure, TDN, during W1, W2, and W3. c, Response of mean paddy yield to 1°C, 2°C, and 4°C increases in average growing season temperature. Error bars indicate 95% confidence intervals calculated using heteroskedasticity robust standard errors.

Figure 3. Nonlinear response of mean milled rice outcomes of tested genotypes given 1°C, 2°C, and 4°C increases in mean growing-season temperature.

Average milled non-chalky head rice (red), chalky head rice (blue), and broken rice (green) per hectare were derived from estimated changes in paddy and milling quality outcomes, by scenario, using sample mean values of paddy and milling quality outcomes as baselines. Estimated changes in mean paddy and milling quality outcomes illustrated in Figures 2 and 3 were used to calculate the milled mass per hectare values illustrated above. Estimated changes in paddy yield from the baseline were calculated using the pooled sample of paddy yield trials and do not vary across cultivar. These changes are presented in Figure 2. Total milling revenue ($1,000 ha⁻¹) is presented at the end of each bar. We used the sample means of milled medium-grain (Bengal and Jupiter) and long-grain (XL723, Wells, LaGrue, and Cypress) U.S. rice prices from 2007–2010, the sample period, for the price of non-chalky head rice, and the sample mean of Arkansas brewers’ milled rice prices over the 2007–2010 period for the price of broken and chalky-head rice kernels.

Figure 4. Changes in paddy and milling yield, and milling revenue across 1°C, 2°C, and 4°C increases in average growing-season temperature.

The consequences of stopping the analysis at the paddy yield level are depicted above. Milling revenue lossses are greater for long-grain cultivars than medium-grain cultivars because high-temperature exposure causes much larger declines in HRY among long-grain cultivars. Medium grains are also less susceptible to chalk formation under heat-stressed conditions, but this disparity contributes relatively less to the disproportional response of milling revenue.