Cotton Genomics and Genetics 2025, Vol.16, No.3, 148-162 http://cropscipublisher.com/index.php/cgg 156 Figure 2 (A) Susceptible (Sicot 714B3F recurrent parent) and (B) resistant two-spotted spider mite cotton germplasm from the CSIRO cotton breeding program (Photos: Lucy Egan). (C) Progress of breeding mite resistant germplasm showing that as backcross (BC) generation number increases mite resistance scores have remained lower than the susceptible recurrent parent, Sicot 714B3F, and relatively stable. Data from C. Trapero, used with permission (Adopted from Conaty et al., 2022) CSIRO has also developed a new parent selection strategy in combination with GS. For example, in response to the negative correlation between yield and quality that has been troubled in the past, they used the prediction model to select hybrids with better fiber quality without significantly reducing yield. Today, CSIRO's cotton varieties enjoy a reputation in the international market for their excellent quality. This successful case shows that integrating AI-driven genomic prediction into the traditional breeding process can achieve a "win-win" in breeding efficiency and breeding effect. CSIRO's experience also provides a reference for other crop breeding, that is, it is necessary to establish a high-quality phenotype-genotype database, continuously optimize the prediction model, and gradually expand the application scope of GS in actual decision-making. It can be expected that CSIRO will further try to integrate environmental data, phenotypic images and other information into predictions in the future, and build a more intelligent breeding decision support system to maintain its international leading position in cotton breeding. 5.2 Comparison of genomic selection methods in U.S. public cotton breeding programs In the United States, since commercial cotton breeding is mainly dominated by private enterprises, public breeding units are particularly active in exploring new technologies. In recent years, the United States Department of Agriculture (USDA) has conducted feasibility studies on cotton genomic selection in collaboration with several universities. The goal is to evaluate the effects of GS on different traits and provide decision-making basis for public breeding programs. Billings et al. (2022) collected a large amount of phenotypic data from regional trials of cotton in the United States, including yield, quality and disease resistance traits, and used existing cotton high-density SNP chips to perform typing analysis on these varieties. They used multiple statistical models for
RkJQdWJsaXNoZXIy MjQ4ODYzNA==