Population Structure and Linkage Disequilibrium in U.S. Barley Germplasm: Implications for Association Mapping

Abstract

Previous studies have shown that there is considerable population structure in cultivated barley (Hordeum vulgare L.), with the strongest structure corresponding to differences in row number and growth habit. U.S. barley breeding programs include six‐row and two‐row types and winter and spring types in all combinations. To facilitate mapping of complex traits in breeding germplasm, 1816 barley lines from 10 U.S. breeding programs were scored with 1536 single nucleotide polymorphism (SNP) genotyping assays. The number of SNPs segregating within breeding programs varied from 854 to 1398. Model‐based analysis of population structure showed the expected clustering by row type and growth habit; however, there was additional structure, some of which corresponded to the breeding programs. The model that fit the data best had seven populations: three two‐row spring, two six‐row spring, and two six‐row winter. Average linkage disequilibrium (LD) within populations decayed over a distance of 20 to 30 cM, but some populations showed long‐range LD suggestive of admixture. Genetic distance (allele‐sharing) between populations varied from 0.11 (six‐row spring vs. six‐row spring) to 0.45 (two‐row spring vs. six‐row spring). Analyses of pairwise LD revealed that the phase of allelic associations was not well correlated between populations, particularly when their allele‐sharing distance was >0.2. These results suggest that pooling divergent barley populations for purposes of association mapping may be inadvisable.

Population structure and linkage disequilibrium in elite barley breeding germplasm from the United States

“…Nine subpopulations (sp1-sp9) were identified by the program STRUCTURE and subsequently confirmed by principle component analysis (PCA). Out of the nine subpopulations, seven were very similar to the respective subpopulations identified by Hamblin et al (2010) which were based on half of the germplasm and half of the SNP markers, but two subpopulations were found to be new. One subpopulation was dominated by six-rowed spring lines from Utah State University (UT) and the other was composed of six-rowed spring lines from multiple breeding programs (USDA-ARS Aberdeen (AB), Busch Agricultural Resources Inc. (BA), UT, and Washington State University (WA)).…”

Development of genome-wide SNP assays for rice

“…1) (Ashikari and Matsuoka 2006, Fukuoka et al 2009, Tanksley and McCouch 1997, Yara et al 2010. SNPs are also extensively used to determine population substructure , Flint-Garcia et al 2005, Hamblin et al 2010, Hyten et al 2007, Nordborg and Weigel 2008, understand the history of domestication (Kovach et al 2009, Shomura et al 2008, Sweeney et al 2007, Takano-Kai et al 2009, and to explore the ancestry of specific alleles, regions of chromosomes or populations (He et al 2006.…”

“…The pronounced subpopulation structure of O. sativa , Garris 2005, Glaszmann 1987) is apparent at all levels of analysis. Genome-wide SNP frequencies are characteristic of each subpopulation and/or breeding program, and make it possible to trace the ancestry of accessions (Hamblin et al 2010. For this reason, a comprehensive SNP discovery initiative requires thoughtful selection of materials for re-sequencing to ensure that patterns of variation are adequately sampled.…”

Applications of Graphical Clustering Algorithms in Genome Wide Association Mapping

“…The output from the algorithm returns a subset of markers which is not necessarily the largest subset of independent markers, nontheless the number of markers returned could still be large enough to get an accurate handle on population stratification. We now briefly discuss the application of of this to real data, more details are available in (Hamblin, 2010).…”