ABSTRACT Cropland abandonment has become a widespread phenomenon in China, posing a potential threat to national food security. Despite its growing prevalence, a comprehensive understanding of abandonment risk and the associated potential grain losses remains limited, and effective early warning mechanisms are notably lacking. This study analyzes the spatiotemporal evolution of abandoned croplands, simulates cropland abandonment risk using machine learning methods, and further assesses the associated potential grain losses. Results indicate that from 1991 to 2021, the cumulative area of abandoned cropland in China reached 60.18 Mha, with a maximum abandonment frequency of five times, and 15.98% of the cropland experiencing repeated abandonment. Newly abandoned croplands during 2021–2023 were predominantly located in the hilly regions of Southwest China, where most areas underwent abandonment for the first time. In 2023, 79.35% of cropland was classified as having low risk or very low risk of abandonment, while 6.83% remained under high risk or very high risk, mainly located in Inner Mongolia and Xinjiang. Cropland abandonment poses a substantial threat to food security, especially in high‐risk and very high‐risk areas. Under the current cropping structure, croplands categorized as medium to very high risk may result in a potential grain loss of 41.53 Mt., with high‐risk and above areas contributing 71.23% of the total loss. This study provides valuable insights for optimizing cropland allocation, developing early warning systems for cropland abandonment, and strengthening food security governance in China.
Li et al. (Tue,) studied this question.