Cotton Genomics and Genetics 2025, Vol.16, No.3, 148-162 http://cropscipublisher.com/index.php/cgg 153 that affect boll number and boll weight at the same time. These findings provide gene targets for directly improving yield through gene editing and other means. In summary, the research on genomic prediction breeding in cotton yield improvement has made initial progress. In the future, with more perfect models and more training data, it is expected to achieve high-precision prediction and efficient selection of yield traits. 4.2 Prediction studies on stress resistance traits Common adversities in cotton production include diseases (such as Verticillium wilt, Fusarium wilt), drought, high salt, etc. Breeding new stress-resistant varieties is the key to ensuring stable yield. Traditional stress-resistant breeding is often time-consuming and labor-intensive, and genomic prediction is expected to accelerate this process. In terms of disease-resistant breeding, genomic selection technology has shown feasibility (Li, 2024). Zhang et al. (2025) used multi-environment Verticillium wilt resistance identification data of 1 152 upland cotton germplasms to construct a GS model for disease resistance traits. Based on the multi-year stable QTL information training model, they predicted the disease resistance of an F2:3 segregating population, and the correlation coefficient was above 0.5, which was significantly better than phenotypic screening alone. More importantly, the model predicted that the selected materials showed higher disease resistance in the field, proving that GS is practical in cotton disease resistance breeding. This study also located 10 major disease resistance QTLs in combination with GWAS and aggregated them in breeding materials, reflecting the power of association analysis combined with GS (Figure 1). For drought resistance, salt and alkali resistance and other stresses, molecular biological methods are more often used at home and abroad to clone functional genes or create transgenic materials. There are few reports on the application of GS in drought-resistant breeding. The reasons are that drought resistance phenotypes are difficult to obtain and that environmental interactions are strong, making predictions complicated. However, some indirect traits such as drought-related physiological indicators can be used as alternative phenotypes to apply GS models. At present, studies have summarized the physiological and molecular regulatory mechanisms of cotton drought and salt tolerance, providing candidate markers and genes for subsequent genome predictions (Ma et al., 2021). For example, the cloned GhCBL1-GhCIPK signaling pathway genes are involved in the regulation of cotton drought resistance. If corresponding markers can be developed, they can be incorporated into the GS model to improve the prediction accuracy of drought resistance. Epigenetic information has also been shown to be related to stress resistance. Zhao et al. (2024) found that 36% of the expression variations of resistance-related genes were associated with DNA methylation variations, but not with conventional genetic variations. These epigenetic markers independent of DNA sequences can also be used to assist prediction models, thereby improving the ability to capture stress resistance. It can be expected that with the development of multi-omics technology, stress-resistant breeding will shift from single gene engineering to whole genome integration optimization. Combining genomic selection with high-pressure screening (such as artificial inoculation of pathogens and simulated drought stress) is expected to quickly select stress-resistant superior plants from a large number of offspring. In general, the research on genomic prediction of cotton stress resistance traits has just started but has broad prospects. By constructing an intelligent model that comprehensively considers genomic, epigenetic and environmental factors, we can expect to achieve early prediction of disease resistance and stress resistance potential and accelerate the breeding process of new stress-resistant cotton varieties. 4.3 Advances in predicting fiber quality traits Fiber quality (including fiber length, strength, fineness, etc.) is the core indicator for measuring the value of cotton, and is also a trait that has a trade-off with yield in breeding. Genomic prediction is of special significance in improving fiber quality, because improving quality in traditional breeding often comes at the expense of yield. Through GS, it is expected to discover gene combinations that increase yield without reducing quality, and achieve synergistic improvement of the two. At present, cotton fiber quality is one of the most significant areas of GS research. CSIRO's experiments have shown that the accuracy of GS prediction of fiber quality is much higher than that of yield: in its study of 1 385 materials, the prediction accuracy of the average length of the upper half of the fiber and the specific breaking strength reached 0.76 and 0.65 respectively. This means that long fibers and high-strength materials can be reliably distinguished based on genotype alone, which provides the possibility for quality-oriented selection. The reason for this phenomenon may be that the genetic control of fiber quality is
RkJQdWJsaXNoZXIy MjQ4ODYzNA==