Journal of Scientific Research and Reports

Breeding programmes are expected to deliver cultivars that combine high yield potential with reliable performance under warmer, drier and more variable growing conditions. Two classical instruments underpin that expectation: the genetic diversity assembled in breeding populations, and the combining ability of candidate parents estimated through structured mating designs. This review examines how far those instruments, as currently applied, support the development of high-yielding and climate-resilient varieties, and where the supporting evidence is weaker than the surrounding literature implies. Evidence was drawn from peer-reviewed quantitative genetics, genomics, agronomy and breeding-systems literature identified through scholarly databases and citation searching, with emphasis on cereals and other field crops for which multi-environment data are available. Four unresolved problems emerge. First, the widely repeated claim that modern breeding has progressively eroded diversity is only partly supported; measured trends depend strongly on the germplasm sampled and the marker system used, and diversity within elite pools behaves differently from diversity across genebank collections. Second, the mating-design literature reports a broadly consistent predominance of general over specific combining ability for yield under abiotic stress, yet many individual studies are too small, too tester-dependent and too sparsely replicated across environments to sustain the inferences drawn from them. Third, general combining ability interacts with environment often enough that parental rankings obtained under one managed stress regime transfer imperfectly to others, which weakens the practice of selecting parents from a single stress trial. Fourth, genomic and enviromic prediction now estimate combining ability more cheaply than exhaustive crossing, but gains in predictive accuracy have not been matched by comparable evidence of gains in realised, on-farm genetic progress. The available evidence supports a reorientation from estimating combining ability in isolation towards evaluating parental value jointly with environmental characterisation, cross-variance prediction and deployment speed. Confidence in this reorientation is limited by uneven crop and geographical coverage.