Abstract
Climate change imposes significant challenges on global agriculture through extreme temperatures, erratic precipitation, drought and salinity, threatening food security. Conventional breeding methods, while effective, are too slow to meet the urgent need for developing climate-resilient crops. Genomic prediction (GP) offers a transformative solution by enabling early and accurate selection of superior genotypes through genome-wide markers and machine learning (ML) algorithms. This chapter outlines the application of GP for breeding abiotic stress-tolerant crops, highlights the role of multi-environment trials (MET) and emphasizes the integration of environmental data (enviromics) to improve predictive accuracy. Recent advances in computational approaches, including deep learning and transformer-based models, are discussed for their potential to enhance GP efficiency under complex genotype-by-environment interactions. Despite challenges, such as limited adoption in low-resource breeding programs, prospects of integrating multi-omics, pan-genomics and gene editing promise to further optimize GP. These innovations position GP as a critical tool for developing climate-resilient crops and ensuring future food security.