Effect of climate change-induced water-deficit stress on long-term rice yield

(mm day-1), PERC is the amount of percolation and seepage losses (mm day-1), and WL is the amount of water required to establish a water layer (mm month-1). For analysis, the period of each growth stage was suggested as being consistent with the period of one single month dur- ing the long-term experimental period. To determine the total ETC and PERC values through- out the entire growth stage, their daily values were multiplied by the number of days in each growth stage. The PERC was suggested to be 5 mm day-1 in the analysis based on well-satu- rated clay soils at the research farm [36]. The WL was assumed to be 100 mm throughout the growth stage [36]. The value of the CWS represents the water status of crop growth under weather conditions. A negative CWS value indicates that the crop is water deficient, and a smaller value represents a lower water level (greater water-deficit stress) in crop growth. In cases where all the water needed for optimal crop growth is provided by rainfall, irrigation is not required, and the CWS equals zero. The CWS determined in this study was inspired by the formula for irrigation water need (IN) [36], IN = ETC + PERC + WL—PE. The value of the CWS equals the negative number of the value of IN. A simple linear regression equation was used to represent the linear time trend of the CWS. The underlying relationship between x and t can be described by x ¼ b0 þ b1t þ ε ð5Þ where x is water stress, t is the year corresponding to x, β0 is the intercept, β1 is the regression

coefficient representing the rate of change in the CWS per year, and ε is the model error. A multiple linear regression model was used to describe the relationship between rice yield and water stress, using the value of rice yield as the response variable and the values of CWS during the initial, crop development, reproductive, and maturity stages (CWSI, CWSD, CWSR, and CWSM) as explanatory variables for each rice cultivar in each cropping season. Y ¼ b0 þ bCWSI CWSI þ bCWSD CWSD þ bCWSR CWSR þ bCWSM CWSM þ ε ð6Þ where β0 represents the model intercept, β is the regression coefficient for water stress in each growth stage, and ε is the model error. The total rice yield change (kg ha-1) related to water stress under each growth stage was computed using the regression coefficients for water stress (βCWS) and the estimated change in crop water status (ΔCWS) throughout each cultivation period. Total yield change ¼ bCWS � DCWSTotal years ð7Þ Results Long-term temporal variations in rice yield In the cool cropping season, the yield of the four rice cultivars for the period 1925–1944 ranged between 1,530 and 8,055 kg ha-1, with an average ± standard deviation (SD) of 4,236 ± 1,114 kg ha-1. The yield of the four rice cultivars for the period 1945–1983 ranged between 2,550 and 8,215 kg ha-1, with an average ± SD of 4,708 ± 774 kg ha-1. The yield of the four rice cultivars during 1996–2019 ranged between 4,181 and 9,268 kg ha-1, with an average ± SD of 6,523 ± 1,080 kg ha-1 (Fig 1a–1d). In the warm cropping season, the yield of the four rice culti-

vars for the period 1925–1944 ranged between 2,181 and 6,493 kg ha-1, with an average ± standard deviation (SD) of 3,880 ± 873 kg ha-1, the yield of the four rice cultivars for the period 1945–1983 ranged between 2,180 and 6,384 kg ha-1, with an average ± SD of 4,149 ± 796 kg ha-1, and the yield of the four rice cultivars for the period 1996–2019 ranged between 2,861 and 7,269 kg ha-1, with an average ± SD of 4,704 ± 971 kg ha-1 (Fig 1e–1h). PLOS ONE Effect of climate change-induced water-deficit stress on long-term rice yield PLOS ONE | https://doi.org/10.1371/journal.pone.0284290 April 17, 2023 5 / 14