Applying genotypic principal component scores as latent phenotypes in genome-wide and epistatic analyses of soybean agronomic traits

Abstract Identification of marker trait associations (MTAs) for agronomic traits of soybean (Glycine max L. Merr.) can often be limited by confounding genotype by environment interactions. In this study, phenotypic data was derived from the calculation of genotypic principal component scores by GGEbiplot (gPCs) from a multiple year and location agronomic dataset to assess the validity and feasibility of using gPC scores in genome-wide association analysis (GWAS) in comparison with traditional phenotypes. Important Quantitative Trait Loci (QTL) were discovered for maturity, seed oil content, yield, and plant height that were not detected using the traditional phenotypes. MTAs were detected by GWAS analysis with PC1, PC2, and PC4 phenotypes. QTL for maturity associated with the E1 and E3 soybean maturity loci demonstrate the validity of this approach by detecting these well studied regions. Epistatic analysis revealed QTL controlling both oil and protein content but did not uncover significant interactions associated with other traits. This result further contributes to the understanding of complex gene networks controlling pleiotropic traits such as seed oil and seed protein content. QTL for the studied traits are reported across six Glycine max chromosomes with 15 genes and one gene cluster proposed as candidates controlling agronomic traits.