Genomic analysis and predictive modeling in the Northern Uniform Soybean Tests

Abstract

The Northern Uniform Soybean Tests (NUST) are a regional field trial network coordinated by the United States Department of Agriculture to evaluate experimental soybean (Glycine max L.) strains developed by public institutions. Historical data from the NUST compiled, curated, and reported herein comprise a valuable multi-environment trial dataset including relevant elite soybean germplasm from maturity groups 00 to IV evaluated over 28 years in 199 locations, totaling 1652 environments. Our aim was to characterize the genetic structure of the NUST experimental strains, perform genome-wide association studies using historical phenotypic data, and assess the usefulness of these historical data for genomic prediction model training. Molecular marker information was collected on 2544 unique NUST experimental strains using the BARCSoySNP6K assay. High fixation index values between early and later maturity groups were observed in a region on chromosome 10 near the known soybean maturity gene E2. We failed to find strong genetic divergence between strains from different breeding programs, reflecting the germplasm sharing common among public programs. Genome-wide association analyses on important agronomic traits identified marker-trait associations, many of which overlap with quantitative trait loci previously reported in the literature. Genomic prediction models trained using the historical NUST data produced moderate to high predictive abilities in most cases, suggesting these data could make useful contributions to training sets. We have made these data publicly available as a data resource for others to study genotype–phenotype relationships within elite public soybean germplasm and develop predictive models for advancement and implementation of genomics-assisted breeding.