Molecular Breeding

, Volume 30, Issue 2, pp 951–966

Population structure revealed by different marker types (SSR or DArT) has an impact on the results of genome-wide association mapping in European barley cultivars

  • Inge E. Matthies
  • Theo van Hintum
  • Stephan Weise
  • Marion S. Röder
Article

DOI: 10.1007/s11032-011-9678-3

Cite this article as:
Matthies, I.E., van Hintum, T., Weise, S. et al. Mol Breeding (2012) 30: 951. doi:10.1007/s11032-011-9678-3

Abstract

Diversity arrays technology (DArT) and simple sequence repeat (SSR) markers were applied to investigate population structure, extent of linkage disequilibrium and genetic diversity (kinship) on a genome-wide level in European barley (Hordeum vulgare L.) cultivars. A set of 183 varieties could be clearly distinguished into spring and winter types and was classified into five subgroups based on 253 DArT or 22 SSR markers. Despite the fact, that the same number of groups was revealed by both marker types, it could be shown that this grouping was more distinct for the SSRs than the DArTs, when assigned to a Q-matrix by STRUCTURE. This was supported by the findings from principal coordinate analysis, where the SSRs showed a better resolution according to seasonal habit and row number than the DArTs. A considerable influence on the rate of significant associations with malting and kernel quality parameters was revealed by different marker types in this genome-wide association study using general and mixed linear models considering population structure. Fewer spurious associations were observed when population structure was based on SSR rather than on DArT markers. We therefore conclude that it is advisable to use independent marker datasets for calculating population structure and for performing the association analysis.

Keywords

BarleyPopulation structureGenome-wide association studiesDArTSSR

Supplementary material

11032_2011_9678_MOESM1_ESM.doc (40 kb)
Electronic Supplementary Material Table 1Number of mapped DArT and SSR markers used for the analysis of population structure, linkage disequilibrium (LD), kinship and association studies in a set of 183 barley cultivars and their average PIC values. The number of DArT-markers which were used in association mapping studies after removal of 5% minor allele frequency (MAF) is also shown. (DOC 40 kb)
11032_2011_9678_MOESM2_ESM.xls (120 kb)
Electronic Supplementary Material Table 2Population structure and assignment (>50% probability) of each cultivar to the five groups (Q1 to Q5) revealed by STRUCTURE analysis with both marker types (sheet a). Population structure and assignment (in % probability) of every cultivar to one of the five groups (Q1 to Q5) revealed by STRUCTURE analysis with the 22 SSRs (sheet b). Population structure and assignment (in % probability) of every cultivar to one of the five groups (Q1 to Q5) revealed by STRUCTURE analysis with the 253 DArTs (sheet c). Origin of cultivars and breeders (sheet d). (XLS 120 kb)
11032_2011_9678_MOESM3_ESM.xlsx (217 kb)
Electronic Supplementary Material Table 3GWAS with the kernel quality parameter glume fineness (sheet a), GWAS with the kernel quality parameter TGW (sheet b), GWAS with the malting quality parameter extract (sheet c), GWAS with the malting quality parameter friability (sheet d) considering the GLM. (XLSX 218 kb)
11032_2011_9678_MOESM4_ESM.xlsx (593 kb)
Electronic Supplementary Material Table 4Results of GWAS considering the MLM with four different combinations of marker types used for the estimation the Q-matrix and kinship (MLM_1 = Q5-DArT + K-DArT, MLM_2 = Q5-SSR + K-DArT, MLM_3 = Q5-DArT + K-SSR, MLM_4 = Q5-SSR + K-SSR). Two kernel quality parameters, such as glume fineness (sheet a), and TGW (sheet b), and two malting quality parameters like extract (sheet c), and friability (sheet d) were assessed. (XLSX 593 kb)
11032_2011_9678_MOESM5_ESM.ppt (153 kb)
Electronic Supplementary Material Fig. 1Fig. 1 Estimated probability of number of subgroups k (goodness of fit computed as lnPr (X|K) for the investigated set of 183 cultivars studied with two marker types (a) 22 SSRs, and (b) 253 DArTs. The ln likelihood L(K) mean values determined by STRUCTURE are plotted against the assumed number of subgroups (k1 to k20). (PPT 153 kb)
11032_2011_9678_MOESM6_ESM.ppt (147 kb)
Electronic Supplementary Material Fig. 2Scatterplot showing the distribution of the intrachromosomal LD-decay parameter r2 in 183 European barley cultivars and plotted against the genetic distance in cM. The horizontal line indicates the 95% percentile of the distribution of unlinked r2, which gives the critical value of r2. Second degree LOESS curve fitted to the plot (black bottom line). (PPT 147 kb)
11032_2011_9678_MOESM7_ESM.pdf (9 kb)
Electronic Supplementary Material Fig. 3Proportion of marker pairwise r2 intrachromosomal measurements above and below background linkage disequilibrium with a critical r2 of 0.21 plotted as a logarithmic function of the genetic distance (in classes) of the entire set of 183 European barley cultivars investigated with 862 mapped DArT markers considering 5 % MAF. (PDF 10 kb)
11032_2011_9678_MOESM8_ESM.pptx (189 kb)
Electronic Supplementary Material Fig. 4Cumulative distribution of the observed p values assessed for different variantsof the MLM_QK with respect to the marker type used for assessing population structure and kinshipinformation (a) MLM_1 = Q5-DArT + K-DArT, (b) MLM_2 = Q5-SSR + K-DArT, (c) MLM_3 = Q5-DArT + K-SSR, (d) MLM_4 = Q5-SSR + K-SSR). Following traits were considered: Glume fineness,thousand grain weight (TGW), extract, and friability. (PPTX 190 kb)
11032_2011_9678_MOESM9_ESM.pptx (319 kb)
Electronic Supplementary Material Fig. 5Genomewide association studies of 183 barley cultivars considering the MLM_QK for four traits (a) extract, and (b) thousand grain weight. Population structure (Q) and kinship (K) was taking into account by using the Q5 matrix calculated either with SSR- or DArT markers. This information was used in different combinations on the MLM in order to assess the effect of the marker type on the rate of signifianct association results (MLM_1 = Q5-DArT + K-DArT, MLM_2 = Q5-SSR + K-DArT, MLM_3 = Q5-DArT + K-SSR, MLM_4 = Q5-SSR + K-SSR). The calculated p-values were converted into –log10(p). The significance thresholds p < 0.05, and p < 0.001 are indicated by dashed lines. (PPTX 320 kb)

Copyright information

© Springer Science+Business Media B.V. 2011

Authors and Affiliations

  • Inge E. Matthies
    • 1
  • Theo van Hintum
    • 2
  • Stephan Weise
    • 1
  • Marion S. Röder
    • 1
  1. 1.Leibniz Institute of Plant Genetics and Crop Plant Research (IPK)GaterslebenGermany
  2. 2.Centre for Genetic Resources, The NetherlandsWageningen University and Research CentreWageningenThe Netherlands