1 Introduction

Soybean (Glycine max (L.) Merr.) is an important crop used for both grain and oil production. It provides plant protein and oil for human food, animal feed, and industrial uses, and holds a highly important position in global agricultural production (Bhat and Yu, 2021; Xue et al., 2025). In recent years, with population growth, changes in dietary structure, and the expansion of animal husbandry and bio-based industries, global demand for soybean yield and quality has continued to increase (Vargas-Almendra et al., 2024). Climate change, emerging diseases, and limited arable land resources have also placed greater pressure on soybean production. Therefore, more efficient breeding methods are needed to develop new varieties with higher yield, better quality, and stronger stress resistance as soon as possible.

Traditional soybean breeding mainly relies on phenotypic selection and long-term field evaluation, and has achieved many results in genetic improvement. However, the overall breeding cycle is long, the labor demand is high, and the process is easily affected by the environment. These limitations are particularly evident in the improvement of complex quantitative traits such as yield, quality, and stress resistance (Ma et al., 2016; Bhat and Yu, 2021). With the development of molecular biology and genomics, introducing molecular genetic information into the breeding process has become an important approach to improving efficiency and shortening the breeding cycle.

Marker-assisted selection (MAS) has developed under this background. It uses DNA markers closely linked to target genes or quantitative trait loci (QTLs) to conduct indirect selection in breeding populations (Bhat and Yu, 2021). Compared with traditional phenotypic selection, MAS can complete genotypic identification at the seed or seedling stage, reduce repeated phenotypic evaluations, and improve screening efficiency (Lin et al., 2022). Commonly used markers currently include SSR, SNP, InDel, and KASP, which have been widely applied in early-generation selection, marker-assisted backcrossing (MABC), and the pyramiding of favorable alleles (Li et al., 2024; Wang et al., 2024).

With the completion of soybean genome sequencing and the application of high-throughput genotyping platforms, such as SoySNP50K and BARCSoySNP6K, a large number of genes and QTLs controlling important agronomic traits, disease resistance, and seed quality have been gradually identified, laying a foundation for the application of MAS in soybean breeding (Ravelombola et al., 2021; Lin et al., 2022). For example, MAS for maturity E gene loci (E1-E4, etc.) has been successfully applied to the improvement of early-maturing adaptability in high-latitude regions (Yerzhebayeva et al., 2023); the prediction accuracy of SNP markers for pod shattering resistance, such as KSS-SNP5, can exceed 90% (Kim et al., 2020); and various SSR and KASP markers have also been developed to improve quality traits such as seed protein, oil content, and Kunitz trypsin inhibitor (Ri̇az et al., 2023; Li et al., 2024). Studies have shown that MAS has become an important conventional breeding strategy for introducing key genes into the genetic backgrounds of elite varieties.

Although MAS has made progress in soybean breeding, its application still faces certain limitations. Many important agronomic traits, especially yield-related traits, are usually controlled jointly by multiple loci with small effects and are influenced by gene–environment interactions. This limits the predictive ability of MAS based on a small number of marker loci (Zhang et al., 2016; Ravelombola et al., 2021). Recent studies have shown that, in the improvement of such complex traits, genomic selection (GS) based on whole-genome marker information has higher prediction accuracy. Therefore, modern breeding strategies are increasingly inclined to combine MAS with GS in order to fully exploit the complementary advantages of both approaches (Bhat and Yu, 2021; Miller et al., 2023).

This study systematically analyzes the research progress and practical application effects of MAS in the improvement of important soybean traits, with a focus on its application achievements in disease resistance, stress resistance, agronomic traits, and quality improvement. It also discusses the limitations of MAS in the improvement of complex quantitative traits and its potential integration with emerging technologies such as genomic selection, with the aim of providing a reference for optimizing future molecular breeding strategies in soybean.

2 Application of MAS in Major-Effect Traits

2.1 Disease and pest resistance

Disease and pest resistance is one of the categories of major-effect traits in soybean that is most suitable for marker-assisted selection (MAS). Soybean cyst nematode (SCN) is one of the most destructive diseases and pests worldwide, and breeding resistant varieties is considered the most economical and environmentally friendly control strategy. At present, rhg1 and Rhg4 have been successfully identified and have become core loci for SCN resistance improvement. SSR, SNP, CAPS, and KASP markers developed around these loci have been widely used for early-generation screening of breeding populations and pyramiding of resistance genes, thereby improving resistance durability and reducing dependence on a single resistance source (Lin et al., 2022; Qu et al., 2025).

In addition to SCN, Phytophthora root rot and soybean rust are also among the most mature disease types for MAS application. Phytophthora root rot, caused by Phytophthora sojae, is mainly controlled by the Rps gene series. Map-based cloning of Rps11 has shown that major-effect disease-resistance genes can provide direct targets for the deployment of broad-spectrum resistance. Therefore, foreground selection and marker-assisted backcrossing using markers closely linked to Rps1, Rps3, Rps6, Rps11, and other genes have become effective strategies for introducing disease-resistance genes into elite genetic backgrounds (Lin et al., 2022). For soybean rust, pyramiding of the Rpp gene series has also been applied in breeding. However, because the pathogen Phakopsora pachyrhizi evolves rapidly and has complex pathotypes, relying only on a single major-effect gene makes it difficult to maintain long-term stable resistance. Therefore, a combined strategy of “major-effect genes + minor-effect QTLs” is more suitable.

In terms of insect pests, Rag series genes associated with traits such as resistance to soybean aphid have also been mapped and used for molecular detection, indicating that MAS is applicable not only to disease resistance, but also to the rapid introgression and pyramiding of major pest-resistance loci.

2.2 Abiotic stress tolerance

Abiotic stresses such as drought, salinity-alkalinity, high temperature, low temperature, and waterlogging continue to limit stable and increased soybean production. Under the background of climate change, both their frequency and damage intensity are increasing. Therefore, stress-resistance breeding has become an important direction in soybean improvement (Manghwar et al., 2022). Although most stress-resistance traits have quantitative genetic characteristics, recent GWAS and QTL mapping studies have identified a group of loci with relatively large effects in soybean, enabling MAS to conduct effective selection for certain key tolerance traits. Taking drought resistance as an example, related QTLs are mostly associated with root architecture, osmotic regulation, and the maintenance of redox homeostasis, and are closely related to yield and yield stability under drought conditions (Amangeldiyeva et al., 2025; Gai et al., 2025) (Figure 1). In addition, the practice of introducing drought-resistance QTLs through MAS in rice, which achieved yield improvement under stress conditions without an obvious yield penalty under normal environments, also provides methodological reference for drought-resistance breeding in soybean (Ramayya et al., 2021; Hassan et al., 2023).

Salinity-alkalinity stress is another key factor limiting regionalized soybean production. In addition to classical salt-tolerance loci, recent studies have further identified key genes such as GmPM30, whose favorable haplotypes can significantly improve survival and yield performance under salt stress by maintaining ion homeostasis and alleviating oxidative damage. The related haplotype-specific markers already have potential for application in MAS pipelines (Gai et al., 2025; Huang et al., 2025). In terms of waterlogging and heat tolerance, although their genetic bases are more complex, several major-effect QTLs or candidate regulatory genes associated with flooding survival rate, pollen viability, and flowering stability have been reported, such as GmDREB, GmNAC, GmWRKY, and GmHSP, indicating that MAS can still be used for targeted improvement around stable major-effect loci (Manghwar et al., 2022; Gai et al., 2025).

2.3 Agronomic trait improvement

Agronomic traits such as plant height, branching patterns, and growth habits directly affect light interception efficiency, lodging resistance, adaptability to mechanized harvesting, and final yield potential in soybean populations. Therefore, they are important targets for MAS of major-effect loci. GWAS and linkage analyses have identified multiple genomic regions associated with plant height and architecture in soybean, some of which overlap with determinate/indeterminate growth habit genes. The related SSR and SNP markers have been validated in breeding populations (Ravelombola et al., 2021; Podzorova et al., 2022). For example, SSR markers Satt244, Satt288, and Satt371 are significantly associated with plant height and related yield traits and can be used for early-generation selection of ideal plant-type materials, reducing lodging risk and improving population uniformity (Podzorova et al., 2022).

Flowering and maturity time are core traits determining regional adaptability of soybean, mainly regulated by photoperiod response genes such as E1, E2, E3, and E4. SNP markers closely linked to these loci have been used in breeding for adaptation to different ecological zones, allowing breeders to rapidly select genotypes with appropriate growth periods according to target latitude and cropping system (Ravelombola et al., 2021; Gai et al., 2025). For yield-related traits, although total yield is a typical polygenic trait, component traits such as seed weight, pod number, and per-plant yield still contain QTLs with moderate to large effects. For example, an approximately 11.5 Mb region on chromosome 10 is significantly associated with both seed weight and yield, showing high breeding value (Ravelombola et al., 2021). Therefore, MAS can be preferentially applied for precise manipulation of major-effect agronomic traits such as plant architecture and growth period, and can be combined with genomic selection (GS) to simultaneously capture the contributions of both major- and minor-effect loci to yield potential (Ravelombola et al., 2021; Gai et al., 2025).

3 Application of MAS in Quality Improvement

3.1 Protein and oil content

Soybean seed protein and oil contents are core traits determining nutritional quality and processing value. Both traits have relatively high heritability, but they are usually negatively correlated with each other and show certain trade-offs with yield, making their coordinated improvement difficult (Lee et al., 2019). Existing GWAS, linkage analyses, and multi-environment trials have shown that although protein and oil contents are mainly controlled by multiple genes, relatively stable large-effect QTLs exist in regions such as chromosomes 15 and 20. In addition, several regions on chromosomes 2, 8, and 14 can also explain a relatively high proportion of phenotypic variation, and therefore have become priority targets for MAS utilization (Jin et al., 2023). These results indicate that, in quality breeding, MAS is particularly suitable for early-generation screening targeting stable major-effect loci, thereby improving the efficiency of identifying high-protein or high-oil materials.

In terms of oil improvement, high-resolution GWAS has identified multiple SNP-enriched regions associated with oil content, some of which contain candidate genes related to lipid metabolism and transport. Meta-QTLs formed through meta-analysis of previous QTLs further improve the stability and transferability of markers (Kumar et al., 2022). Among them, GmSWEET39 on chromosome 15 has been identified as an important causal gene simultaneously affecting the protein-oil balance. Its different haplotypes correspond to “high-oil and low-protein” or “relatively high-protein” phenotypes, providing a functional marker basis for the targeted design of quality combinations (Figure 2) (Zhang et al., 2020). Therefore, by jointly tracking major-effect QTLs and key genes through functional markers such as KASP, MAS can, to a certain extent, alleviate the negative correlation between protein and oil and achieve targeted optimization of seed composition (Patel et al., 2025).

3.2 Specialized metabolites

In addition to protein and oil, soybean seeds are also rich in various specialized metabolites with nutritional and functional value, among which isoflavones and specific fatty acid compositions are the most representative. Isoflavones, such as genistein and daidzein, are synthesized through the phenylpropanoid-flavonoid pathway, and their natural variation is closely related to differences in the expression of key enzyme genes such as CHS, CHI, and IFS, as well as upstream regulatory factors (Kim et al., 2021; Zhao et al., 2025). Transcriptomic and multi-omics studies have shown that related biosynthetic genes are continuously highly expressed at specific developmental stages in high-isoflavone lines, and isoflavone accumulation is also coordinated with changes in amino acid metabolism, lipid metabolism, and antioxidant defense networks. Therefore, molecular markers linked to or functionally associated with these loci can provide an important basis for the targeted improvement of isoflavone content through MAS.

Fatty acid composition directly determines the nutritional structure, oxidative stability, and processing suitability of soybean oil. High-resolution mapping studies have identified multiple stable QTLs controlling the contents of palmitic acid, stearic acid, oleic acid, linoleic acid, and linolenic acid, including candidate genes such as GmFabG, GmACP, GmFAD8, and fatty acid desaturase-related regions. These QTLs and their tightly linked SNPs can be directly used to screen high-quality oil-type materials, such as those with high oleic acid and low linolenic acid contents (Li et al., 2017; Kumar et al., 2022; Wang et al., 2025).

3.3 Seed nutritional and processing quality

Seed nutritional and processing quality directly determines the application value of soybean in soybean products, plant protein, and feed processing. Its improvement goals have expanded from simply increasing total protein content to optimizing amino acid composition, reducing antinutritional factors, and improving processing suitability. GWAS studies have shown that the contents of essential amino acids such as cysteine, methionine, lysine, and threonine are regulated by multiple QTLs, and some loci are relatively independent of total protein content loci. This means that amino acid balance can be improved through MAS without significantly changing the overall protein level (Lee et al., 2019). In addition, in populations derived from cultivated soybean and wild soybean, some donor alleles that increase protein content have been mapped and shown to function stably with relatively small effects on oil content and agronomic traits, providing traceable targets for backcross introgression and early-generation screening (Huang et al., 2020).

Trypsin inhibitors, lipoxygenases, and certain storage protein variants are important targets for MAS because they affect protein digestibility, beany flavor, oil oxidative stability, and functional properties in food processing, respectively (Kumar et al., 2022; Yao et al., 2022). Molecular markers closely linked to these loci can be used to screen materials with low antinutritional factors, reduced off-flavor, and better processing performance, thereby achieving coordinated improvement of nutritional quality and processing quality. As QTLs related to protein, oil, amino acids, and processing traits are gradually integrated with genomic prediction models, breeders have been able to design ideal genotypes around the comprehensive goals of “high nutritional value-excellent processing characteristics–acceptable agronomic performance” (Sun et al., 2022; Patel et al., 2025).

4 Gene Pyramiding and MAS

4.1 Gene pyramiding strategies

Gene pyramiding refers to a breeding strategy that integrates multiple independent genes or QTLs with favorable effects into the same genetic background, aiming to simultaneously express multiple desirable traits in a single variety, thereby achieving broader, more stable, or more durable improvement effects than individual genes (Dormatey et al., 2020; Haque et al., 2021). In soybean, this strategy is particularly suitable for improving disease and pest resistance, stress tolerance, and certain quality traits, which are determined by multiple functionally complementary major- or moderate-effect loci. Unlike traditional cumulative selection based on phenotypes, MAS can directly track multiple target loci in segregating generations, improving pyramiding efficiency and reducing misselection caused by environmental interference (Das et al., 2017). In disease-resistance breeding, pyramiding multiple R genes or major-effect QTLs, or combining major genes with minor-effect resistance loci, has been shown to enhance resistance spectrum and durability. This principle also applies to complex adaptive traits such as drought, salt, and waterlogging tolerance (Dormatey et al., 2020; Haque et al., 2021).

Marker-assisted backcrossing (MABC) is the core technical pathway for achieving gene pyramiding. Its basic framework includes foreground selection, recombinant selection, and background selection, aiming to restore the recurrent parent genome as quickly as possible while introducing target genes (Haque et al., 2021). In multiple crops, MABC can restore more than 95% of the recurrent parent genome within 3-4 backcrosses and successfully pyramid 4-10 resistance or tolerance loci, which is significantly faster than traditional backcrossing methods (Das et al., 2018; Pandit et al., 2021). With the introduction of high-density SNP chips, KASP, multiplex detection systems, and genomic selection models, breeders can now simultaneously track multiple loci, optimize population size, and improve pyramiding success rates, providing an actionable approach for modular improvement of complex target traits in soybean (Jarallah et al., 2025).

4.2 Breeding efficiency and limiting factors

Gene pyramiding through MAS improves traditional breeding efficiency. Its core advantage lies in the ability to directly identify target gene combinations at the seedling or early-generation stage, thereby reducing the number of materials carried into subsequent generations and saving field space, phenotyping costs, and breeding cycle time (Haque et al., 2021). In multiple crops, marker-assisted gene pyramiding can essentially restore elite genetic backgrounds within 2–4 backcrosses, whereas traditional backcrossing usually requires about six generations. It can also pyramid multiple resistance or tolerance loci, enhancing resistance levels and yield stability (Kumar et al., 2018; Haque et al., 2021). This means that breeders can not only simultaneously track multiple target loci through multi-marker detection, but also focus resources on the most promising recombinants using foreground and background selection, improving the accumulation efficiency of favorable genes in populations (Ramalingam et al., 2020).

However, multi-gene pyramiding is not without cost. As the number of target genes increases, the required population size rapidly expands; at the same time, gene linkage, QTL×QTL interactions, and donor fragment linkage drag increase screening difficulty and may weaken the expression of target genes in new genetic backgrounds (Das et al., 2017). Even if molecular detection confirms the target genotypes, rigorous background selection and multi-environment field trials are still needed to validate the combined performance of resistance, stress tolerance, yield, and adaptability.

4.3 Application of gene pyramiding in soybean breeding

In soybean breeding, MAS-based gene pyramiding has gradually shifted from theoretical strategy to practical application, with the most mature field still being disease-resistance breeding (Figure 3). Studies have used marker-assisted backcrossing to simultaneously introduce the Phytophthora root rot resistance gene Rps2, powdery mildew resistance gene Rmd-c, and effective nodulation-related genes into high-yielding variety backgrounds, resulting in new materials with multiple resistances and good agronomic performance. This demonstrates the feasibility of simultaneous multi-gene introgression in soybean (Ramalingam et al., 2020). The same concept also applies to the pyramiding of soybean cyst nematode resistance genes Rhg1/Rhg4, different Rps genes, and rust resistance genes, aiming to build broader-spectrum and more durable resistance systems within a single variety (Haque et al., 2021).

In terms of abiotic stress, although mature pyramiding cases reported in soybean are still fewer than those in crops such as rice, the relevant technical pathway has become relatively clear. As major-effect QTLs and candidate genes related to drought, salt, waterlogging, and heat tolerance continue to be validated, breeders can use MAS to integrate loci corresponding to different stress-resistance mechanisms—such as genes related to osmotic regulation, maintenance of ion homeostasis, and optimization of root architecture—into the same elite genetic background, thereby improving varietal stability and adaptability under complex environments (Dormatey et al., 2020; Pandit et al., 2021). In quality improvement, gene pyramiding also has clear potential. For example, joint selection of protein- and oil-related QTLs, fatty acid composition loci, and low-antinutritional-factor genes is expected to simultaneously improve nutritional value, processing quality, and marketability (Das et al., 2017; Zampieri et al., 2023).

5 Limitations of MAS

5.1 Minor-effect QTLs

Although MAS has achieved significant results in the utilization of major genes and large-effect QTLs, its application is often clearly limited by the complex genetic architecture of minor-effect QTLs for typical quantitative traits such as yield, stress resistance, and certain quality traits. Such traits are usually jointly controlled by a large number of small-effect loci, with each individual locus explaining only a very low proportion of phenotypic variation. The remaining variation is also jointly influenced by undetected loci, environmental noise, and epistatic interactions. Therefore, relying only on a small number of markers makes it difficult to fully capture the complete genetic basis (Yáñez et al., 2023; Oh et al., 2025). When the target phenotype is jointly formed by multiple minor-effect QTLs through additive or interactive effects, the advantage of traditional MAS in “targeted selection of a few loci” is significantly weakened. This is also one of the important reasons why MAS is often inferior to genomic selection in the improvement of complex quantitative traits (Yáñez et al., 2023; Li and Lin, 2024).

Another key limitation lies in the poor environmental stability and transferability of minor-effect QTLs. Many minor-effect loci detected in specific populations and environments are often difficult to repeatedly validate under different genetic backgrounds, climatic conditions, or cultivation management practices, reflecting significant genotype × environment interactions and background-dependent marker–trait associations (Chang-Brahim et al., 2024; Li and Lin, 2024). Precise mapping of such loci usually requires large populations, high-density markers, and repeated validation across multiple environments in order to effectively distinguish true minor-effect QTLs from statistical false positives and reduce the risk of recombination between markers and causal variants (Song et al., 2023). In practical breeding, minor-effect QTLs are often difficult to stably convert into routine MAS tools in the same way as major genes.

5.2 Cost and operational challenges

Although MAS can theoretically significantly improve breeding efficiency, its routine promotion is still constrained by both cost and operational conditions, especially in resource-limited breeding programs. First, high-throughput genotyping, SNP chips, and sequencing-based detection platforms still involve relatively high costs when applied to large-scale populations, and often require dedicated instruments, reagents, and data processing software. As a result, although the testing cost per sample may decrease, the initial infrastructure investment and continuous operating expenses remain high (Song et al., 2023; Chang-Brahim et al., 2024). Even when low-cost strategies such as pooled DNA, crude DNA extraction, or multiplex PCR are used to reduce the unit sample cost, laboratory platform construction, quality control, and professional personnel allocation still constitute important barriers (Ru et al., 2015).

MAS is not simply “molecular detection replacing phenotypic evaluation”. Instead, it requires the highly integrated implementation of marker development, DNA extraction, genotyping, data management, and conventional hybridization and selection procedures, which places high demands on a team’s capabilities in molecular biology, statistical analysis, and bioinformatics (Chang-Brahim et al., 2024). In practical operation, differences among laboratories in DNA quality, marker platforms, interpretation standards, and data management may all affect result consistency and technical reproducibility. Meanwhile, some breeding teams lack a clear cost–benefit evaluation framework and are uncertain about the breeding stage at which MAS should be introduced, making it difficult for molecular markers to be truly embedded into routine breeding decision-making processes (Ru et al., 2015). In developing countries or emerging breeding systems, insufficient funding, limited testing services, and pressure from short-term production goals further intensify this problem, resulting in an obviously uneven pattern of global MAS application.

5.3 Limitations of MAS in complex traits

The limitations of MAS are most prominent in the improvement of complex traits, especially in polygenic traits such as yield, adaptability, and comprehensive stress resistance. Such traits are usually determined by a large number of minor-effect loci and are simultaneously influenced by gene–environment interactions (G×E) and epistatic effects. Therefore, the phenotypic variation explained by a single or a few QTLs is very limited, making it difficult to accurately predict final performance (Yáñez et al., 2023). Taking soybean yield as an example, it is essentially the combined result of multiple component traits such as pod number, seed weight, branch number, and growth period. If only a few molecular markers are tracked, it is insufficient to capture the complete genetic basis, and the improvement magnitude may also be difficult to justify the cost and operational complexity of MAS implementation (Yáñez et al., 2023; Oh et al., 2025).

Marker–trait associations in complex traits often show clear environmental dependence: a QTL that is significant in one population or environment may have a weakened effect or even disappear under another genetic background, management condition, or climatic context, thereby reducing the stability and transferability of MAS results (Chang-Brahim et al., 2024; Li and Lin, 2024). Increasing numbers of studies advocate limiting MAS to major genes and large-effect QTLs, while assigning the improvement of highly polygenic traits more to genomic selection (GS), because GS can use whole-genome marker information to integrate a large number of minor-effect loci and their interaction effects (Yáñez et al., 2023; Kumar et al., 2025).

6 MAS in the Genomic Era

6.1 High-throughput genotyping technologies

With the rapid development of genomics, high-throughput genotyping technologies have significantly expanded the marker resources and application boundaries of MAS. Due to their wide distribution, high stability, and ease of automated detection, SNPs have become the core marker type in modern molecular breeding. Commercial SNP chips can typically detect tens of thousands to hundreds of thousands of loci simultaneously and have been widely used for QTL mapping, routine MAS, genomic selection (GS), and breeding quality control (Kumar et al., 2024). In soybean, high-density platforms such as SoySNP50K have been used for constructing genetic maps and screening key loci, providing a basis for high-precision MAS (Bhat et al., 2016).

The popularization of NGS has further promoted the development of sequencing-based genotyping (GBS). GBS can simultaneously discover and genotype thousands of SNPs at a relatively low cost through reduced-representation genome sequencing, combining high throughput with considerable flexibility. Even at lower sequencing depths, it can support genetic analysis of complex traits and prediction of breeding values (Sinha et al., 2016; Sinha et al., 2023). New platforms such as GBTS and mSNP liquid-phase chips provide layered schemes with approximately 1K to over 40K markers, allowing the same system to simultaneously serve major-effect gene MAS and whole-genome prediction (Guo et al., 2019; 2021). Therefore, high-throughput genotyping technologies have enabled MAS to transition from “few-locus detection” to a new stage of “high-density, customizable, and integrable” applications.

6.2 Integration with genomic selection

Methodologically, the core difference between MAS and genomic selection (GS) lies in how molecular marker information is utilized. MAS primarily relies on a few discovered and validated markers that are tightly linked to major genes or large-effect QTLs, making it suitable for targeted selection of Mendelian or oligogenic traits such as disease resistance, maturity, and specific quality loci. In contrast, GS assumes that all genome-wide markers may be linked to causal variants and predicts genomic estimated breeding values (GEBVs) by simultaneously estimating all marker effects, making it more suitable for complex quantitative traits controlled by a large number of minor-effect loci (Bhat et al., 2016; Nanthini et al., 2025). In crops such as soybean, with the development of high-throughput SNP chips and GBS platforms, the predictive potential of GS for complex traits such as yield, stress tolerance, and comprehensive quality has continuously improved, while MAS still maintains an advantage in managing major-effect loci.

For highly polygenic complex traits, GS generally offers higher prediction accuracy and genetic gain per unit time than MAS, because it can integrate a large number of minor-effect loci across the genome and their combined effects without requiring individual markers to reach significance thresholds (Budhlakoti et al., 2022). However, when major-effect loci that explain a substantial proportion of phenotypic variation exist for the target trait, MAS still retains strong competitiveness. Therefore, the currently more practical strategy is not to replace MAS with GS, but to integrate the two: use MAS to introgress or fix key major-effect genes, such as disease-resistance genes, maturity genes, or important quality loci; simultaneously utilize GS to optimize the genetic background composed of numerous minor-effect loci, thus improving both Mendelian and quantitative components (Sinha et al., 2023; Kumar et al., 2024; Jarallah et al., 2025). This complementary approach of “precise control of major-effect loci + whole-genome background prediction” is becoming an important direction in soybean molecular breeding in the genomic era.

7 Future Prospects of MAS in Soybean Breeding

7.1 Improving accuracy and validation of molecular markers

The key to further enhancing the application value of MAS in soybean lies in continuously improving the accuracy, stability, and cross-population transferability of molecular markers. Currently, many markers are still based on “tight linkage with target genes or QTLs” rather than directly corresponding to causal variants, and their predictive ability may be reduced in different genetic backgrounds or ecological environments due to recombination or haplotype differences. This is particularly evident for traits regulated by multiple loci, such as seed weight, sucrose content, and seed quality (Bhat and Yu, 2021; Hasan et al., 2021). In the future, fine mapping, candidate gene analysis, and haplotype analysis should be more extensively employed to gradually upgrade traditional linked markers to functional markers or high-resolution targeted SNP panels, enabling markers to more directly reflect causal variants and thereby improving stability and accuracy in routine breeding (Yang et al., 2023).

The practical breeding value of molecular markers does not depend on “significance” but on “sufficient validation”. Future validation systems should emphasize cross-validation, independent population testing, and multi-environment trials to systematically evaluate marker stability across different germplasms and stress conditions. For example, SNPs associated with soybean pod shattering resistance can maintain over 90% prediction accuracy across different breeding stages and environments, indicating that rigorous validation is a prerequisite for markers to enter routine MAS pipelines (Kim et al., 2020). Only sufficiently validated markers have large-scale application value for critical decisions such as parental selection, hybrid design, and release of gene-edited materials. For complex traits, validated key markers can further be integrated with GS models to improve prediction efficiency and decision reliability (Xue et al., 2025).

7.2 Integrating MAS with multi-omics data

Future soybean MAS will no longer be limited to DNA-level linked markers, but will rely more on the coordinated integration of genomic, transcriptomic, proteomic, metabolomic, and high-throughput phenotypic data to systematically reveal the causal chains between “genotype–regulatory network–phenotype” (Hasan et al., 2021; Vargas-Almendra et al., 2024). For example, combining SNP information obtained from GWAS with transcriptomic expression patterns can more accurately distinguish truly functional candidate genes from markers that are merely in linkage. Further overlaying metabolomic data helps to pinpoint key pathways directly involved in seed component formation, stress responses, or developmental regulation, thereby improving the precision of functional marker development (Budhlakoti et al., 2022).

The value of multi-omics integration lies not only in improving marker development accuracy but also in enhancing the biological interpretability of MAS and its ability to improve complex traits. Studies have shown that incorporating a small number of metabolite markers obtained from metabolome association analysis into prediction models can increase the prediction accuracy of heterosis or complex traits by 4%-14%, sometimes even exceeding that of full genome–metabolome models (Xu et al., 2025). Applying similar strategies to soybean allows breeders to prioritize markers from transcripts, proteins, or metabolites with clear functional links to sucrose, isoflavones, protein, oil, or stress responses, making MAS not just “tracking loci” but “fixing favorable alleles within regulatory networks” (Ri̇az et al., 2023; Xue et al., 2025). In the future, multi-omics-driven MAS can also synergize with GS and gene editing to provide more interpretable and precise molecular breeding strategies for complex traits.

7.3 Developing climate-adapted soybean varieties

Climate change continues to intensify the combined stresses of high temperature, drought, flooding, and emerging pests and diseases on soybean production. Therefore, developing climate-adapted varieties with high yield, stable yield, stress resistance, and superior quality has become one of the core goals of future molecular breeding in soybean (Budhlakoti et al., 2022; Vargas-Almendra et al., 2024). In this process, MAS still plays an irreplaceable role, particularly suitable for tracking and pyramiding major disease-resistance genes, maturity-related genes, and certain stable stress-resistance loci. For highly polygenic traits such as drought tolerance, heat tolerance, and broad adaptability, MAS is best combined with GS to simultaneously capture the contributions of numerous minor-effect loci to complex stress responses (Budhlakoti et al., 2022). Therefore, future climate-adapted soybean breeding should not focus solely on single-stress tolerance but should prioritize fixing key adaptive alleles through MAS, and then optimize the whole-genome background using GS to maintain yield and quality stability under complex environments.

The future climate-adapted soybean breeding system is likely to adopt a comprehensive “MAS + GS + gene editing + rapid breeding” model: MAS is used to identify and fix key alleles, GS predicts combined breeding values under different environments, CRISPR/Cas is used to precisely introduce superior allelic variants when natural variation is insufficient, and rapid breeding shortens generation time to accelerate fixation of target genotypes (Fang et al., 2021; Vargas-Almendra et al., 2024). Studies have shown that optimizing hybrid combinations using genomic prediction can significantly improve the efficiency of soybean genetic improvement. This strategy can also be extended to prioritize the selection of hybrid combinations that combine stress resistance, disease resistance, suitable maturity, and superior quality (Miller et al., 2023). Therefore, MAS in the future will not be applied in isolation but will serve as a “key-locus anchoring tool” in climate-adapted breeding, embedded within a more systematic molecular design breeding framework to support global food and feed security.

8 Concluding Remarks

Marker-assisted selection (MAS) has become an important technical foundation of modern soybean molecular breeding and has profoundly changed the way genetic variation is discovered, tracked, and utilized. With the development of molecular marker systems from RFLP and SSR to high-density SNP chips and NGS-based detection platforms, breeders are able to more efficiently associate genomic regions with disease resistance, adaptability, quality, and some yield-related traits. In soybean, MAS is particularly effective for traits controlled by single genes or a small number of major-effect genes. It has achieved stable results in resistance to soybean cyst nematode, resistance to Phytophthora root rot, regulation of maturity, and improvement of certain seed quality traits, and has significantly shortened the breeding cycle through marker-assisted backcrossing and gene pyramiding.

However, the advantages of MAS are mainly concentrated in the management of major-effect loci, and clear limitations remain in the improvement of complex quantitative traits. Traits such as yield stability, climate adaptability, and multiple-stress tolerance are usually jointly affected by numerous minor-effect QTLs, gene–environment interactions, and genetic background effects. A small number of markers are difficult to fully explain their phenotypic variation, and therefore their predictive ability is usually weaker than that of genomic selection (GS). MAS is also constrained by the accuracy of phenotypic evaluation, marker stability across populations, environmental dependence of QTLs, and continuous genotyping costs. These problems are particularly prominent in resource-limited breeding systems.

In the future, MAS will not be replaced, but will be repositioned within an integrated genomic breeding framework. A more efficient strategy is to use MAS to precisely track key disease-resistance genes, major adaptive loci, and functional markers, while using GS to optimize polygenic backgrounds. At the same time, multi-omics analysis, high-throughput phenotyping, and artificial intelligence prediction should be integrated to improve the efficiency of candidate locus discovery and cross-environment prediction ability. On this basis, gene-editing technologies such as CRISPR/Cas will enable breeding to gradually move from “indirect selection based on linked markers” toward “direct modification or regulation of causal alleles.”

The development trajectory of MAS in soybean shows that it has evolved from a single molecular tool into a key component of modern precision breeding systems. As long as molecular markers are fully validated and applied synergistically with GS, gene editing, and multi-omics technologies, MAS will remain an important bridge connecting key genes with breeding practice, and will continue to play a central role in developing new soybean varieties with high yield, superior quality, stress resistance, and adaptation to climate change.

Acknowledgments

The authors thank Mr. Zhang for his support and assistance in material compilation. The authors also thank the two anonymous reviewers for their careful review of the manuscript.

Conflict of Interest Disclosure

The authors affirm that this research was conducted without any commercial or financial relationships that could be construed as a potential conflict of interest.

References 

Amangeldiyeva A., Yerzhebayeva R., Mazkirat S., Didorenko S., Bastaubayeva S., Maikotov B., Kassenov R., Jenisbayeva A., and Shavrukov Y., 2025, Yield and yield component trait analysis with DArT genotyping for GWAS in soybean grown in drought conditions of Kazakhstan, Frontiers in Plant Science, 16: 1674201.

https://doi.org/10.3389/fpls.2025.1674201

Bhat J.A., and Yu D., 2021, High-throughput NGS-based genotyping and phenotyping: role in genomics-assisted breeding for soybean improvement, Legume Science, 3(3): e81.

https://doi.org/10.1002/leg3.81

Bhat J.A., Ali S., Salgotra R.K., Mir Z.A., Dutta S., Jadon V., Tyagi A., Mushtaq M., Jain N., Singh P., Singh G., and Prabhu K., 2016, Genomic selection in the era of next generation sequencing for complex traits in plant breeding, Frontiers in Genetics, 7: 221.

https://doi.org/10.3389/fgene.2016.00221

Budhlakoti N., Kushwaha A.K., Rai A., Chaturvedi K.K., Kumar A., Pradhan A.K., Kumar U., Kumar R., Juliana P., Mishra D., and Kumar S., 2022, Genomic selection: a tool for accelerating the efficiency of molecular breeding for development of climate-resilient crops, Frontiers in Genetics, 13: 832153.

https://doi.org/10.3389/fgene.2022.832153

Chang-Brahim I., Koppensteiner L.J., Beltrame L., Bodner G., Saranti A., Salzinger J., Fanta-Jende P., Sulzbachner C., Bruckmüller F., Trognitz F., Samad-Zamini M., Zechner E., Holzinger A., and Molin E.M., 2024, Reviewing the essential roles of remote phenotyping, GWAS and explainable AI in practical marker-assisted selection for drought-tolerant winter wheat breeding, Frontiers in Plant Science, 15: 1319938.

https://doi.org/10.3389/fpls.2024.1319938

Das G., Rao G.J., Varier M., Prakash A., and Prasad D., 2018, Improved Tapaswini having four BB resistance genes pyramided with six genes/QTLs, resistance/tolerance to biotic and abiotic stresses in rice, Scientific Reports, 8(1): 2413.

https://doi.org/10.1038/s41598-018-20495-x

Das G., Patra J.K., and Baek K.H., 2017, Insight into MAS: a molecular tool for development of stress resistant and quality of rice through gene stacking, Frontiers in Plant Science, 8: 985.

https://doi.org/10.3389/fpls.2017.00985

Dormatey R., Sun C., Ali K., Coulter J.A., Bi Z., and Bai J., 2020, Gene pyramiding for sustainable crop improvement against biotic and abiotic stresses, Agronomy, 10(9): 1255.

https://doi.org/10.20944/preprints202008.0088.v1

Fang Y., Wang L., Sapey E., Fu S., Wu T., Zeng H., Sun X., Qian S., Khan M., Yuan S., Wu C., Hou W., Sun S., and Han T., 2021, Speed-breeding system in soybean: integrating off-site generation advancement, fresh seeding, and marker-assisted selection, Frontiers in Plant Science, 12: 717077.

https://doi.org/10.3389/fpls.2021.717077

Gai Y., Liu S., Zhang Z., Wei J., Wang H., Liu L., Bai Q., Qin Q., Zhao C., Zhang S., Xiang N., and Zhang X., 2025, Integrative approaches to soybean resilience, productivity, and utility: a review of genomics, computational modeling, and economic viability, Plants, 14(5): 671.

https://doi.org/10.3390/plants14050671

Guo Z., Wang H., Tao J., Ren Y., Xu C., Wu K., Zou C., Zhang J., and Xu Y., 2019, Development of multiple SNP marker panels affordable to breeders through genotyping by target sequencing (GBTS) in maize, Molecular Breeding, 39(3): 37.

https://doi.org/10.1007/s11032-019-0940-4

Guo Z., Yang Q., Huang F., Zheng H., Sang Z., Xu Y., Zhang C., Wu K., Tao J., Prasanna B., Olsen M., Wang Y., Zhang J., and Xu Y., 2021, Development of high-resolution multiple-SNP arrays for genetic analyses and molecular breeding through genotyping by target sequencing and liquid chip, Plant Communications, 2(6): 100230.

https://doi.org/10.1016/j.xplc.2021.100230

Haque M.A., Rafii M.Y., Yusoff M.M., Ali N.S., Yusuff O., Datta D.R., Anisuzzaman M., and Ikbal M.F., 2021, Recent advances in rice varietal development for durable resistance to biotic and abiotic stresses through marker-assisted gene pyramiding, Sustainability, 13(19): 10806.

https://doi.org/10.3390/su131910806

Hasan N., Choudhary S., Naaz N., Sharma N., and Laskar R.A., 2021, Recent advancements in molecular marker-assisted selection and applications in plant breeding programmes, Journal of Genetic Engineering and Biotechnology, 19(1): 128.

https://doi.org/10.1186/s43141-021-00231-1

Hassan M.A., Dahu N., Hongning T., Qian Z., Yueming Y., Yiru L., and Shimei W., 2023, Drought stress in rice: morpho-physiological and molecular responses and marker-assisted breeding, Frontiers in Plant Science, 14: 1215371.

https://doi.org/10.3389/fpls.2023.1215371

Huang J., Q., Cai Z., Xia Q., Li S., Jia J., Chu L., Lian T., Nian H., and Cheng Y., 2020, Identification and mapping of stable QTLs for seed oil and protein content in soybean [Glycine max (L.) Merr.], Journal of Agricultural and Food Chemistry, 68(23): 6448-6460.

https://doi.org/10.1021/acs.jafc.0c01271

Huang S., Xia Y., Yang J., Si Y., Chen X., Zhang H., Liu T., Zheng W., Chen X., Zhao Z., Zheng X., Lu Q., Li S., and Xiang F., 2025, A coding SNP in GmPM30 enhances soybean salinity tolerance and yield through the GmLEA1-GmPM30-GmLEC1 module, Advanced Science, 12(44): e09391.

https://doi.org/10.1002/advs.202509391

Jarallah M.H., Mouhsan Z.M., and Alamry A.T., 2025, Integration of genetic tools: strategies for combining traditional and modern breeding to combat plant diseases, European Journal of Ecology, Biology and Agriculture, 2(4): 79-91.

https://doi.org/10.59324/ejeba.2025.2(4).07

Jin H., Yang X., Zhao H., Song X., Tsvetkov Y., Wu Y., Gao Q., Zhang R., and Zhang J., 2023, Genetic analysis of protein content and oil content in soybean by genome-wide association study, Frontiers in Plant Science, 14: 1182771.

https://doi.org/10.3389/fpls.2023.1182771

Kim D.G., Lyu J.I., Lim Y.J., Kim J.M., Hung N.N., Eom S.H., Kim S., Kim J., Bae C., and Kwon S.J., 2021, Differential gene expression associated with altered isoflavone and fatty acid contents in soybean mutant diversity pool, Plants, 10(6): 1037.

https://doi.org/10.3390/plants10061037

Kim J.M., Kim K.H., Jung J., Kang B.K., Lee J., Ha B.K., and Kang S., 2020, Validation of marker-assisted selection in soybean breeding program for pod shattering resistance, Euphytica, 216(11): 166.

https://doi.org/10.1007/s10681-020-02703-w

Kumar A., Sandhu N., Dixit S., Yadav S., Swamy B.P.M., and Shamsudin N.A.A., 2018, Marker-assisted selection strategy to pyramid two or more QTLs for quantitative trait-grain yield under drought, Rice, 11(1): 35.

https://doi.org/10.1186/s12284-018-0227-0

Kumar B., Yankanchi S., Singh R., P., Sarkar D., Kumar P., Kumar K., Choudhary M., Jat B., and Jat H.S., 2025, Dissecting the genetic architecture of polygenic nutritional traits in maize through meta-QTL analysis, Food Chemistry: Molecular Sciences, 10: 100256.

https://doi.org/10.1016/j.fochms.2025.100256

Kumar R., Das S.P., Choudhury B.U., Kumar A., Prakash N.R., Verma R., Chakraborti M., Devi A., Bhattacharjee B., Das R., Das B., Devi H., Das B., Rawat S., and Mishra V.K., 2024, Advances in genomic tools for plant breeding: harnessing DNA molecular markers, genomic selection, and genome editing, Biological Research, 57(1): 80.

https://doi.org/10.1186/s40659-024-00562-6

Kumar V., Goyal V., Mandlik R., Kumawat S., Sudhakaran S., Padalkar G., Rana N., Deshmukh R., Roy J., Sharma T., and Sonah H., 2022, Pinpointing genomic regions and candidate genes associated with seed oil and protein content in soybean through an integrative transcriptomic and QTL meta-analysis, Cells, 12(1): 97.

https://doi.org/10.3390/cells12010097

Lee S., Van K., Sung M., Nelson R., LaMantia J., McHale L.K., and Mian M.R., 2019, Genome-wide association study of seed protein, oil and amino acid contents in soybean from maturity groups I to IV, Theoretical and Applied Genetics, 132(6): 1639-1659.

https://doi.org/10.1007/s00122-019-03304-5

Li B., Fan S., Yu F., Chen Y., Zhang S., Han F., Yan S., Wang L., and Sun J., 2017, High-resolution mapping of QTL for fatty acid composition in soybean using specific-locus amplified fragment sequencing, Theoretical and Applied Genetics, 130(7): 1467-1479.

https://doi.org/10.1007/s00122-017-2902-8

Li J., and Lin X., 2024, Analyzing the impact of marker-assisted selection on livestock productivity and genetic diversity, Animal Molecular Breeding, 14.

https://doi.org/10.5376/amb.2024.14.0014

Li S., Guo C., Feng X., Wang J., Pan W., Xu C., Wei S., Han X., Yang M., Chen Q., Wang J., Hu L., and Qi Z., 2024, Development and validation of kompetitive allele-specific polymerase chain reaction markers for seed protein content in soybean, Plants, 13(24): 3485.

https://doi.org/10.3390/plants13243485

Lin F., Chhapekar S.S., Vieira C.C., Da Silva M.P., Rojas A., Lee D., Liu N., Pardo E., Lee Y., Dong Z., Pinheiro J., Ploper L., Rupe J., Chen P., Wang D., and Nguyen H., 2022, Breeding for disease resistance in soybean: a global perspective, Theoretical and Applied Genetics, 135(11): 3773-3872.

https://doi.org/10.1007/s00122-022-04101-3

Ma Y., Reif J., Jiang Y., Wen Z., Wang D., Liu Z., Guo Y., Wei S., Wang S., Yang C., Wang H., Yang C., Lu W., Xu R., Zhou R., Wang R., Sun Z., Chen H., Zhang W., Wu J., Hu G., Liu C., Luan X., Fu Y., Guo T., Han T., Zhang M., Sun B., Zhang L., Chen W., Wu C., Sun S., Yuan B., Zhou X., Han D., Yan H., Li W., and Qiu L., 2016, Potential of marker selection to increase prediction accuracy of genomic selection in soybean (Glycine max L.), Molecular Breeding, 36(8): 113.

https://doi.org/10.1007/s11032-016-0504-9

Manghwar H., Zaman W., Chugh V., Kaur D., Purwar S., Kaushik P., Sharma V., Kumar H., Rai A., Singh C., and Dubey R., 2022, Applications of molecular markers for developing abiotic-stress-resilient oilseed crops, Life, 13(1): 88.

https://doi.org/10.3390/life13010088

Miller M.J., Song Q., and Li Z., 2023, Genomic selection of soybean (Glycine max) for genetic improvement of yield and seed composition in a breeding context, The Plant Genome, 16(4): e20384.

https://doi.org/10.1002/tpg2.20384

Nanthini B., Venkatachalam S.R., Arutchenthil P., Natarajan S.K., and Jaya Prabhavathi S., 2025, Review on genomic selection in plant breeding, Plant Science Today, 12(3): 2348-1900.

https://doi.org/10.14719/pst.8556

Oh H., Mengist M., G., Giongo L., Pottorff M., Spencer J., Perkins-Veazie P., and Iorizzo M., 2025, Unraveling the genetic architecture of blueberry fruit quality traits: major loci control organic acid content while more complex genetic mechanisms control texture and sugar content, BMC Plant Biology, 25(1): 36.

https://doi.org/10.1186/s12870-025-06061-4

Pandit E., Pawar S., Barik S.R., Mohanty S.P., Meher J., and Pradhan S.K., 2021, Marker-assisted backcross breeding for improvement of submergence tolerance and grain yield in the popular rice variety ‘Maudamani’, Agronomy, 11(7): 1263.

https://doi.org/10.3390/agronomy11071263

Patel J., Patel S., Cook L., Fallen B.D., and Koebernick J., 2025, Soybean genome-wide association study of seed weight, protein, and oil content in the southeastern USA, Molecular Genetics and Genomics, 300(1): 43.

https://doi.org/10.1007/s00438-025-02228-8

Podzorova T., Zatybekov A., Didorenko S., and Abugalieva S., 2022, Genetic variation of soybean collection based on microsatellite DNA markers related to plant height, Eurasian Journal of Applied Biotechnology, (2): 3-12.

https://doi.org/10.11134/btp.2.2022.1

Qu S., Hu S., Song G., Sun H., Liu F., Zhang M., Zhan Y., Li Y., Teng W., Li H., Zhao X., and Han Y., 2025, Genome-wide association study and molecular marker development for resistance to soybean cyst nematode in soybean, The Plant Genome, 18(3): e70083.

https://doi.org/10.1002/tpg2.70083

Ramalingam J., Alagarasan G., Savitha P., Lydia K., Pothiraj G., Vijayakumar E., Sudhagar R., Singh A., Vedna K., and Vanniarajan C., 2020, Improved host-plant resistance to Phytophthora rot and powdery mildew in soybean (Glycine max (L.) Merr.), Scientific Reports, 10(1): 13928.

https://doi.org/10.1038/s41598-020-70702-x

Ramayya P., Vinukonda V., Singh U., Alam S., Venkateshwarlu C., Vipparla A., Dixit S., Yadav S., Abbai R., Badri J., T R., Padmakumari A., Singh V., and Kumar A., 2021, Marker-assisted forward and backcross breeding for improvement of elite Indian rice variety Naveen for multiple biotic and abiotic stress tolerance, PLoS One, 16(9): e0256721.

https://doi.org/10.1371/journal.pone.0256721

Ravelombola W., Qin J., Shi A., Song Q., Yuan J., Wang F., Chen P., Yan L., Feng Y., Zhao T., Meng Y., Guan K., Yang C., and Zhang M., 2021, Genome-wide association study and genomic selection for yield and related traits in soybean, PLoS One, 16(8): e0255761.

https://doi.org/10.1371/journal.pone.0255761

Riaz A., Raza Q., Kumar A., Dean D., Chiwina K., Phiri T., Thomas J., and Shi A., 2023, GWAS and genomic selection for marker-assisted development of sucrose enriched soybean cultivars, Euphytica, 219(9): 97.

https://doi.org/10.1007/s10681-023-03224-y

Ru S., Main D., Evans K., and Peace C., 2015, Current applications, challenges, and perspectives of marker-assisted seedling selection in Rosaceae tree fruit breeding, Tree Genetics and Genomes, 11(1): 8.

https://doi.org/10.1007/s11295-015-0834-5

Shi A., Gepts P., Song Q., Xiong H., Michaels T.E., and Chen S., 2021, Genome-wide association study and genomic prediction for soybean cyst nematode resistance in USDA common bean (Phaseolus vulgaris) core collection, Frontiers in Plant Science, 12: 624156.

https://doi.org/10.3389/fpls.2021.624156

Sinha D., Maurya A.K., Abdi G., Majeed M., Agarwal R., Mukherjee R., Ganguly S., Aziz R., Bhatia M., Majgaonkar A., Seal S., Das M., Banerjee S., Chowdhury S., Adeyemi S., and Chen J.T., 2023, Integrated genomic selection for accelerating breeding programs of climate-smart cereals, Genes, 14(7): 1484.

https://doi.org/10.3390/genes14071484

Song L., Wang R., Yang X., Zhang A., and Liu D., 2023, Molecular markers and their applications in marker-assisted selection (MAS) in bread wheat (Triticum aestivum L.), Agriculture, 13(3): 642.

https://doi.org/10.3390/agriculture13030642

Sun B., Guo R., Liu Z., Shi X., Yang Q., Shi J., Zhang M., Yang C., Zhao S., Zhang J., He J., Zhang J., Su J., Song Q., and Yan L., 2022, Genetic variation and marker-trait association affect the genomic selection prediction accuracy of soybean protein and oil content, Frontiers in Plant Science, 13: 1064623.

https://doi.org/10.3389/fpls.2022.1064623

Vargas-Almendra A., Ruiz-Medrano R., Núñez-Muñoz L.A., Ramírez-Pool J.A., Calderón-Pérez B., and Xoconostle-Cázares B., 2024, Advances in soybean genetic improvement, Plants, 13(21): 3073.

https://doi.org/10.3390/plants13213073

Wang X., Qi Y., Sun G., Zhang S., Li W., and Wang Y., 2024, Improving soybean breeding efficiency using marker-assisted selection, Molecular Plant Breeding, 15: 1-9.

https://doi.org/10.5376/mpb.2024.15.0025

Wang X., Zhang M., Li F., Liu X., Zhang C., Zhang F., Zhao K., Yuan R., Lamlom S., Ren H., Qiu H., and Zhang B., 2025, Genome-wide association study reveals key genetic loci controlling oil content in soybean seeds, Agronomy, 15(8): 1889.

https://doi.org/10.3390/agronomy15081889

Xu Y., Yang W., Qiu J., Zhou K., Yu G., Zhang Y., Wang X., Jiao Y., Wang X., Hu S., Zhang X., Li P., Lu Y., Chen R., Tao T., Yang Z., Xu Y., and Xu C., 2025, Metabolic marker-assisted genomic prediction improves hybrid breeding, Plant Communications, 6(3): 101199.

https://doi.org/10.1016/j.xplc.2024.101199

Xu Y., Zhang X., Li H., Zheng H., Zhang J., Olsen M., Varshney R., Prasanna B., and Qian Q., 2022, Smart breeding driven by big data, artificial intelligence, and integrated genomic-enviromic prediction, Molecular Plant, 15(11): 1664-1695.

https://doi.org/10.1016/j.molp.2022.09.001

Xue Y., Tang X., Zhu X., Zhang R., Yao Y., Cao D., He W., Liu Q., Luan X., Shu Y., and Liu X., 2025, Leveraging GWAS-identified markers in combination with Bayesian and machine learning models to improve genomic selection in soybean, International Journal of Molecular Sciences, 26(19): 9586.

https://doi.org/10.3390/ijms26199586

Yang Q., Zhang J., Shi X., Chen L., Qin J., Zhang M., Yang C., Song Q., and Yan L., 2023, Development of SNP marker panels for genotyping by target sequencing (GBTS) and its application in soybean, Molecular Breeding, 43(4): 26.

https://doi.org/10.1007/s11032-023-01372-6

Yao D., Zhao Q., Li T., Wang J., Wang L., Liu Y., Hao W., and Liu H., 2022, Genetic analysis and quantitative trait locus mapping using the major gene plus polygene model for soybean [Glycine max (L.) Merr.] main quality trait, Legume Research - An International Journal, 45(1): 18-24.

https://doi.org/10.18805/lrf-705

Yerzhebayeva R., Didorenko S., Amangeldiyeva A., Daniyarova A., Mazkirat S., Zinchenko A., and Shavrukov Y., 2023, Marker-assisted selection for early maturing E loci in soybean yielded prospective breeding lines for high latitudes of Northern Kazakhstan, Biomolecules, 13(7): 1146.

https://doi.org/10.3390/biom13071146

Yáñez J.M., Barría A., López M.E., Moen T., Garcia B.F., Yoshida G.M., and Xu P., 2023, Genome-wide association and genomic selection in aquaculture, Reviews in Aquaculture, 15(2): 645-675.

https://doi.org/10.1111/raq.12750

Zampieri E., Volante A., Marè C., Orasen G., Desiderio F., Biselli C., Canella M., Carmagnola L., Milazzo J., Adreit H., Tharreau D., Poncelet N., Vaccino P., and Valè G., 2023, Marker-assisted pyramiding of blast-resistance genes in a japonica elite rice cultivar through forward and background selection, Plants, 12(4): 757.

https://doi.org/10.3390/plants12040757

Zhang H., Goettel W., Song Q., Jiang H., Hu Z., Wang M.L., and An Y.Q.C., 2020, Selection of GmSWEET39 for oil and protein improvement in soybean, PLoS Genetics, 16(11): e1009114.

https://doi.org/10.1371/journal.pgen.1009114

Zhang J., Song Q., Cregan P.B., and Jiang G.L., 2016, Genome-wide association study, genomic prediction and marker-assisted selection for seed weight in soybean (Glycine max), Theoretical and Applied Genetics, 129(1): 117-130.

https://doi.org/10.1007/s00122-015-2614-x

Zhao K., Hong H., Liu X., Wang X., Zhang C., Zhang F., Yuan R., Lamlom S., Ren H., and Zhang B., 2025, Multi-omics integration reveals heavy ion-induced enhancement of soybean isoflavone biosynthesis, Physiologia Plantarum, 177(5): e70508.

https://doi.org/10.1111/ppl.70508