IMPACT OF CLIMATE CHANGE ON WHEAT YIELD STABILITY IN SEMI‑ARID AGRICULTURAL REGIONS

Abstract

The climate change aggravates the amount of extreme weather and the distribution of the precipitation, which puts the stability of the wheat production in semi-arid regions at risk. With a multi-model framework, comprising of crop modeling (CERES-Wheat), statistical analysis, machine learning and CMIP6 climate projections, the paper quantifies the impacts of climatic variability, heat stress, and drought on wheat production over 40 years at 12 semi-arid sites. Findings demonstrate that the heat stress in combination with drought stress enhances the yield variability 341 times more and the coefficient of variation increases by 18.1 to 62.9. Critical nonlinear temperature response: The yield damages will decrease by 3-4 percent or one day with temperatures above 33C during the grain-filling of five or more days in sequence. The SPEI drought index based on RDIst (1-week lead time, r = 0.76) was the initial sign of the loss of yield. In models, benchmarking indicated that XGBoost was the most accurate in predicting extreme events and deep learning-based genomic selection was able to increase genetic gain by 27 percent in predicting height tolerance in response to heat compared to more traditional approaches. By 2050, the return period of compound heat-drought events under SSP5-8.5 is decreased to 1.9 years (compared to 26 years (under SSP5-8.5)) and 95 percent of the yield loss is caused by anthropogenic forcing. Adaptations to systems to include heat tolerant genotypes, optimum sowing and deficit irrigation improved resilience by 76%. We find that the semi-arid wheat systems are becoming destabilized at a faster rate than ever unless multi-stress-tolerant cultivars, advanced seasonal prediction with machine learning, and agronomic information to prevent regional food crises are adopted at a frenzy.