Cotton Genomics and Genetics 2025, Vol.16, No.3, 117-125 http://cropscipublisher.com/index.php/cgg 119 2.3 Limitations of QTL mapping, including environmental interactions and resolution Although QTL mapping has many advantages, it also has some disadvantages. For example, its resolution may not be high, which is related to the number of materials and population types you use. If there are not enough samples, the QTL region found may be relatively large, containing many genes, and it is difficult to say which one is working (Zhang et al., 2016). In addition, the environment has a great influence on QTL. Some QTLs are obvious in one place, but the effect is different in another place or year (Shen et al., 2007). Only a small number of QTLs can be stably expressed in various environments, so it is difficult to find stable QTLs that can be used in breeding. In addition, traits such as yield and fiber quality are often affected by multiple genes together, and their effects will interfere with each other, which makes the application of QTL in actual breeding more complicated (Wu et al., 2022). 3 Genomic Selection: Principles and Implementation 3.1 Theoretical basis and advantages over traditional selection methods In the matter of breeding, everyone hopes to select good varieties faster, but the traditional method is slow and relies on trait performance, which is observed from generation to generation. Later, genetic marker-assisted selection was introduced, but it only focuses on a few "key genes". But the problem is that not all traits are determined by a few genes. For complex traits such as yield and stress resistance, there are many genes behind them that play a role at the same time, and even each gene does not contribute much. At this time, genomic selection (GS) becomes more appropriate. It does not "pick the key points" like the traditional method, but takes into account the markers of the entire genome, which is more comprehensive. Moreover, it can predict which individuals have potential without measuring phenotypes (Goddard and Hayes, 2007). Using GS for selection is more efficient and accurate - especially when facing traits that do not have a single "dominant gene", traditional methods are often unable to cope with it (Meuwissen et al., 2016). 3.2 Models and statistical approaches in genomic prediction To use GS, you have to build a model. But you can't just build it casually, you have to be able to handle massive amounts of genomic data. Several common models now, such as GBLUP, Bayesian models, and machine learning methods, actually have the same purpose - to predict the breeding value of an individual based on the marker information of the whole genome (Wang et al., 2018). However, the actual operation may be more complicated than you think. A common problem is that there are too many markers, but not enough phenotypic data. This imbalance will affect the accuracy of modeling. To compensate, some models will also consider non-additive effects, interactions between genes and the environment, or multiple traits together (Larkin et al., 2019). Of course, whether the effect is good or not depends on whether the model you choose is appropriate, the population structure, the density of markers, and the strength of the trait heritability itself (Robertsen et al., 2019). It doesn't mean that everything will be fine if GS is used, but the details must be coordinated. 3.3 Integration into breeding pipelines for rapid genetic gain Integrating GS methods into the breeding process allows us to select good lines earlier and more accurately, so that the time and cost of breeding excellent varieties will be much less. GS can be used at every stage of breeding, whether it is early selection or later verification, and is especially suitable for those traits that are expensive or difficult to measure (Crossa et al., 2017). If a dynamically updated training population is used, the prediction model is optimized regularly, and other new technologies, such as high-throughput phenotyping technology, are combined, the entire breeding efficiency will become higher and the genetic gain will be greater (Bassi et al., 2016). Therefore, GS is changing the way modern breeding is done and providing new means for the changing needs of agriculture (Voss-Fels et al., 2018). 4 Combining QTL Mapping and Genomic Selection 4.1 Marker-assisted selection guided by major QTLs Marker-assisted selection (MAS) is a method of guiding breeding based on the located major effect QTL. It can help select genes that are beneficial to yield and fiber quality into target varieties (Cao et al., 2024). Breeders can find plants with ideal traits by selecting molecular markers that are closely linked to these important QTLs. This
RkJQdWJsaXNoZXIy MjQ4ODYzNA==