A Soybean Haplotype Map and the Initial Application of Association Analysis in Soybean Germplasm.

The entire USDA Soybean Germplasm Collection consisting of 18,682 accessions of cultivated soybean [Glycine max (L.) Merr.], including landraces and some cultivars, as well as, 1,116 accessions of the wild soybean [G. soja (Sieb. et Zucc.)] were analyzed with a 60,800 bead type Illumina Infinium array.  A total of 47,337 SNPs spread across the 20 soybean chromosomes produced successful allele calls and were polymorphic.  The data were useful for developing a core collection of landrace and wild soybean genotypes that spanned the diversity of their respective groups.  An important objective of the genetic analysis was the development of a haplotype map that will permit the analysis of the germplasm collection via the application of genome wide association (GWA) analysis.  Levels of linkage disequilibrium were highly variable in the landraces.  In many genome regions, the density of SNPs will have to be increased to permit the successful identification of loci controlling phenotypic traits.  However, in other regions sufficient tag SNPs are available to allow successful GWA in the landrace germplasm.  LD was much less extensive in the wild soybean germplasm.  Clearly a much greater SNP density is required for effective association analysis in the wild soybean germplasm.  To examine GWA, a set of 160 landrace genotypes were selected with seed protein levels exceeding 44% and an otherwise identical set of 160 genotypes, in terms of Asian country of origin, maturity, and other traits were selected that had “normal” seed protein content (38-40%).  The 320 genotypes were grown in replicated field trials at Beltsville, MD and Lincoln, NE and seed protein and oil content were determined.  Using the SNP data on these genotypes, analyses was applied to adjust for population structure and identify genome regions associated with seed protein and oil content.  A number of genome regions that had previously been associated with seed protein concentration via QTL analysis were identified using GWA.