Molecular Plant Breeding 2024, Vol.15, No.6, 371-378 http://genbreedpublisher.com/index.php/mpb 375 4.3 Statistical analysis, GWAS model, and identification and validation of high-yield QTLs Various GWAS models were employed to identify yield-related QTLs. The Fixed and Random Model Circulating Probability Unification (FarmCPU) model was used in several studies (Yoosefzadeh-Najafabadi et al., 2021; Priyanatha et al., 2022). Additionally, machine learning algorithms like support vector regression (SVR) and random forest (RF) were compared with conventional methods, showing that SVR-mediated GWAS outperformed others in identifying relevant QTLs (Yoosefzadeh-Najafabadi et al., 2023). These advanced statistical models enhance the power and accuracy of QTL detection. Several QTLs linked to high-yield traits were identified. For example, two stable seed yield QTLs on chromosomes 9 and 17 were consistently detected across multiple environments. Another study identified five QTL regions controlling seed yield and seed oil and protein content. These QTLs are crucial for understanding the genetic basis of yield and can be targeted in breeding programs (Happ et al., 2021). Validation experiments were conducted to confirm the identified QTLs. For instance, candidate gene analysis surrounding the seed yield QTL on chromosome 9 identified Glyma.09G048900 as a potential gene influencing yield. Similarly, functional annotation of candidate genes supported the relevance of identified QTLs (Wang et al., 2021). These validation steps are essential to ensure the reliability of the GWAS findings. 4.4 Implications for future breeding programs The identified QTLs have significant implications for future breeding programs. The integration of GWAS with machine learning algorithms provides a robust framework for marker-assisted selection, enabling the development of high-yield soybean varieties (Yoosefzadeh-Najafabadi et al., 2023). The insights gained from these studies can guide breeders in selecting superior genotypes, ultimately improving soybean yield and productivity. By leveraging advanced genotyping techniques, comprehensive phenotypic measurements, and sophisticated statistical models, these studies provide a detailed understanding of the genetic architecture underlying soybean yield. The identified QTLs and candidate genes offer valuable targets for breeding programs aimed at enhancing soybean production. 5 Integration of GWAS Results into Soybean Breeding Programs 5.1 Marker-assisted selection (MAS) for yield improvement Marker-assisted selection (MAS) leverages identified quantitative trait loci (QTL) to enhance specific traits in crops. In soybean, several studies have identified QTLs associated with yield and other agronomic traits. For instance, a study identified 19 single-nucleotide polymorphisms (SNPs) significantly associated with seed yield, with stable QTLs on chromosomes 9 and 17 (Ayalew et al., 2022). These QTLs can be used to develop markers for MAS, facilitating the introgression of desirable traits into elite soybean lines. Another study highlighted the potential of MAS in improving seed protein and oil content, which are crucial for soybean's nutritional profile (Fields et al., 2023; Wang, 2024). The integration of these markers into breeding programs can significantly enhance yield and quality traits in soybean. 5.2 Genomic selection (GS) and prediction models for yield traits Genomic selection (GS) uses genome-wide markers to predict the breeding value of individuals, enabling the selection of superior genotypes. Studies have demonstrated the effectiveness of GS in soybean breeding. For example, genomic prediction models using ridge regression best linear unbiased prediction (rrBLUP) have shown high accuracy for traits such as protein content, oil content, and yield (Ravelombola et al., 2021; Qin et al., 2022). Additionally, machine learning algorithms like support vector regression (SVR) have outperformed traditional methods in identifying relevant QTLs for yield components, suggesting their potential in GS (Yoosefzadeh-Najafabadi et al., 2023). These advanced prediction models can accelerate the breeding cycle and improve the efficiency of selecting high-yielding soybean varieties. 5.3 Use of identified QTLs in multi-environment trials The stability of QTLs across different environments is crucial for their effective use in breeding programs. Multi-environment trials help in identifying QTLs that are consistently associated with yield traits under varying conditions. A meta-GWAS study involving multiple environments identified 483 QTLs, with some loci showing
RkJQdWJsaXNoZXIy MjQ4ODYzMg==