| Association mapping is a promising approach to identifying marker-trait associations within broader germplasm than traditional bi-parental quantitative trait loci (QTL) mapping approaches. Linkage disequilibrium (LD) between observable markers and unobservable QTL allow for QTL detection, as such patterns of LD impact the ability to detect QTL. Forces like selection, genetic drift and admixture create LD, while recombination decays LD. First, we compare the pattern of LD between two subpopulations within a barley breeding program, one that is essentially closed and the other with recent introduction of parents for disease resistance ('reopened'). Over 1,500 single nucleotide polymorphism (SNP) were genotyped; within the closed subpopulation 261 (31.4%) were included in LD analysis and 877 (57.1%) in the reopened subpopulation. Patterns of LD were highly variable across the genome, the closed subpopulation had 39 LD blocks of an average 3.24 cM while the reopened subpopulation had 108 blocks averaging 1.16 cM. We estimate genome coverage for association mapping purposes to be 25.1% in the closed subpopulation and 44.1% in the reopened subpopulation. Second, in related germplasm consisting of 182 breeding lines from a single breeding program, we conduct association mapping for beta-glucan related traits and then malt quality traits. More than 3,000 SNPs were genotyped, 756 were utilized in the final analysis. Genome coverage was estimated to be 36.5%. More than 100 QTL related to beta-glucan and 94 for the additional malting quality traits were detected on all seven chromosomes. A major previously described QTL for multiple malting quality traits on 4H was successfully detected within the single breeding program. Comparison of haplotypes across this region indicate that inclusion of breeding lines with more recent exotic parentage was critical for this detection. We conclude that association mapping within a single breeding program can detect both novel and previously described QTL, but that greater marker density is needed for full genome coverage. |
