Soybean (Glycine max L.) is a global staple crop valued for its seeds, which contain about 40% protein in total weight, making them a rich source of plant-based protein. Enhancing soybean seed protein content (SPC) has long been a central focus of breeding research. In this study, we used a recombinant inbred line (RIL) population derived from the cross Dongnong L13 × Henong 60 (RIL6013). We performed quantitative trait locus (QTL) mapping for SPC using a high-density genetic linkage map and IciMapping v4.2 software to identify and predict candidate genes associated with SPC. We employed the mrMLM method to detect significant quantitative trait nucleotides (QTNs) linked to SPC, followed by genomic selection (GS) based on these QTNs. We conducted simulation breeding, using genotypic data from significant QTNs and GS outcomes, to identify hybrid combinations for selecting high-protein soybean varieties. This analysis revealed 16 QTL associated with SPC, including a region containing the gene Glyma.02G250200 on chromosome 2. Genome-wide association study (GWAS) identified 37 significant QTNs, which we used as a single-nucleotide polymorphism (SNP) set for GS, together with the best linear unbiased prediction (BLUP) values of phenotypic data. We employed the five conventional statistical models BayesA, BayesB, BayesC, BayesLASSO, and GBLUP to perform genomic prediction. The prediction accuracy for all five GS models exceeded 0.65. Based on the five GS outcomes we devised five breeding schemes, followed by simulation breeding. The average genotypic values of the virtual progeny generated through these simulations were significantly higher than those of the parental populations. Simulation breeding identified 22 hybrid combinations optimal for high-protein selection. Of these, the line HN138 from the RIL6013 population was the parental line for multiple high-potential hybrid combinations and may thus represent the most suitable genetic background within RIL6013 for the breeding of cultivars with high protein content.

Pods are unevenly distributed on soybean (Glycine max L. Merr.) plants, with significantly fewer pods in the lower regions, which limits overall yield. Elucidating the genetic basis of pod formation in the lower part of the plant holds substantial theoretical and practical significance for breeding efforts aimed at increasing seed yield. In this study, we evaluated a four-way recombinant inbred line (FW-RIL) population and a germplasm population (GP) of soybean across seven and five environments, respectively, to assess pod number in the lower part (PNL) of the soybean plant. We identified quantitative trait loci (QTL) and quantitative trait nucleotides (QTN) associated with PNL. We systematically screened candidate genes potentially involved in regulating PNL within linkage disequilibrium (LD) blocks of QTN that colocalized with QTL. Finally, we developed a molecular-assisted selection (MAS) model based on QTN derived from the GP and identified the optimal breeding schemes using the B4L (breeding for pure lines) ISB (in silico breeding) model. We identified 25 QTL in the FW-RIL population and 93 QTN in the GP, including 5 QTN that colocalized with the QTL. In LD blocks surrounding the QTN AX-90477863, we identified Glyma.09G040000 as a candidate gene associated with PNL. Using a MAS model, the 93 QTN accounted for 52.5% of the standing phenotypic variation in PNL in the GP. Using this model, we selected 16 hybrid combinations with PNL genotypic values exceeding the breeding target of 16. Our findings enhance our understanding of the genetic basis of soybean pod number and provide technical support for the molecular breeding of high-yielding soybean varieties.

Soybean seeds contain approximately 40% protein, making soybeans an important source of plant-based protein. Research on QTN mapping, molecular design breeding and mining of genes related to seed protein formation provides guiding significance for the analysis of the underlying genetic mechanisms of seed protein formation and the selection of high-protein varieties. The seed protein contents (SPCs) of 144 lines of a soybean four-way recombinant inbred line (FW-RIL) population were determined in 8 environments. A three-variance component multisite random effects mixed linear model (3VmrMLM) was used to conduct a genome-wide association study on protein content. A single detected QTN explained 0.53%–3.37% of the phenotypic variation. A molecular-assisted selection breeding model containing the 18 QTNs explained 51.97% of the phenotypic variation in protein content. Eight biparental and five tri-parental crosses that produced excellent lines with the greatest protein content-related genotype values that could be generated by phenotypic and molecular-assisted selection were screened. An LD block of 17 QTNs (QEIs) was identified, and one key candidate gene related to protein formation was predicted by haplotype analysis. The proportion of Hap 1 varieties in the spring-sowing soybean region in North China was lower than those in the Huang-Huai-Hai soybean region in Central China and the multiripe soybean region in South China. The proportion of Hap 1 varieties among the wild varieties and landraces was greater than that among the improved varieties. The results of this study provide important insights into the genetic basis of soybean protein content and information to aid in molecular design breeding methods to improve protein content.