Abstract
Warmer temperatures significantly influence crop yields, which are a critical determinant of food supply and human well-being. In this study, a probabilistic approach based on bivariate copula models was used to investigate the dependence (described by joint distribution) between crop yield and growing season temperature (T(GS)) in the major producing provinces of China for three staple crops (i.e., rice, wheat, and maize). Based on the outputs of 12 models from the Coupled Model Intercomparison Project Phase 6 (CMIP6) under Shared Socioeconomic Pathway 5-8.5, the probability of yield reduction under 1.5 °C and 2 °C global warming was estimated, which has great implications for agricultural risk management. Results showed that yield response to T(GS) varied with crop and region, with the most vulnerable being rice in Sichuan, wheat in Sichuan and Gansu, and maize in Shandong, Liaoning, Jilin, Nei Mongol, Shanxi, and Hebei. Among the selected five copulas, Archimedean/elliptical copulas were more suitable to describe the joint distribution between T(GS) and yield in most rice-/maize-producing provinces. The probability of yield reduction was greater in vulnerable provinces than in non-vulnerable provinces, with maize facing a higher risk of warming-driven yield loss than rice and wheat. Compared to the 1.5 °C global warming, an additional 0.5 °C warming would increase the yield loss risk in vulnerable provinces by 2-17%, 1-16%, and 3-17% for rice, wheat, and maize, respectively. The copula-based model proved to be an effective tool to provide probabilistic estimates of yield reduction due to warming and can be applied to other crops and regions. The results of this study demonstrated the importance of keeping global warming within 1.5 °C to mitigate the yield loss risk and optimize agricultural decision-making in vulnerable regions.