Development of user-selectable diverse sets of cultivated and wild soybean germplasm for genetic and breeding applications.

After decades of intensive breeding, modern US soybean [Glycine max (L.) Merr.] varieties have achieved significant improvements in yield, quality, and stress tolerance, but these gains have come at the cost of severely reduced genetic diversity. To reduce vulnerability and promote efficient use of germplasm, diverse sets (DS) of varying sample sizes were defined for the entire USDAARS Soybean Germplasm Collection and 13 maturity groups using the SoySNP50K single-nucleotide polymorphism (SNP) profile. The average retained genetic diversity of the 50K SNPs was then compared between 10 DS and 10 random sets (RSs) at different sizes. DS consistently outperformed random sampling: in cultivated soybean, DS captured 94.9%-98.4% of SNP diversity compared with 73.1%-93.9% for RS; in wild soybean, DS captured 92.8%-97.9% compared with 83.4%-97.7% for RS. The performance of DS was further validated using whole-genome sequences from 1511 accessions, demonstrating that DS could retain the diversity predicted by the SNP subset across 1308 cultivated and 203 wild soybean genomes of different sample sizes. DS was also effective in capturing genetic diversity across different traits. To allow users to select DS, a "Soy-DS Selector" approach was proposed, and a table containing germplasm clusters across the USDA collection and different maturity groups was created. This resource enables researchers to tailor combinations based on maturity groups, accession and sample size preferences, and seed availability. The study provides both methodology and resources that can streamline germplasm evaluation, maximize resource utilization, and enhance future genetic improvement in soybean. Several DS have already been used by US soybean breeders in their programs.