Potential for detection of marker–trait associations in durum wheat using unbalanced, historical phenotypic datasets

Abstract

Identification of marker–trait associations in germplasm relevant to a breeding program can be an effective way to identify quantitative trait loci (QTL) useful for selection and is critical to the success of genome-wide selection strategies. This approach is most cost-effective if phenotypic data routinely collected by breeding programs is used, necessitating only addition of genotypic data. The objective of this work was to evaluate such an approach using unbalanced phenotypic data from durum wheat (Triticum turgidum L. var. durum) registration trials genotyped with diversity arrays technology (DArT) markers. Plant height, grain cadmium concentration and yellow colour loss during pasta manufacture were chosen as example traits because all are influenced by major genes associated with known QTL. A further evaluation was performed on semolina yellow pigment concentration, a more complexly-inherited trait, but with numerous QTL identified. In total, 870 informative DArT markers were used to detect marker–trait associations. The genome coverage of markers was uneven, with low coverage of chromosomes 4B and 5A. The DArT coverage of chromosome 4B was too sparse to identify markers strongly associated with the semidwarf height locus Rht-B1 and the lipoxygenase locus Lpx-B1, both known to reside on 4B. The 20 DArT markers associated with pigment concentration localized to chromosomes 1B, 2A, 5B, 6A, 7A and 7B, linked to the trait in other studies. One DArT clone showed sequence identity to a single wheat expressed sequence tag that maps to the same deletion bin as Psy1-A1, a gene previously associated with yellow pigment concentration in durum wheat. Three markers were associated with grain cadmium and explained similar proportions of the phenotypic variance as the Xusw14 marker known to be physically linked to Cdu-B1, a major locus on 5B regulating cadmium accumulation. The sequences of these three DArT markers were 98 % identical, and were used to identify a single gene in rice that is physically linked to other rice genes that co-localize with Cdu-B1 in durum wheat. The results suggest that this historical phenotypic dataset is useful for QTL discovery and would potentially be a ‘training population’ for genomic selection when a high-density, low-cost marker platform becomes available.