Improving the accuracy of selection for late maturity α-amylase in wheat using multi-phase designs

The assessment of grain defect traits is assuming greater importance in wheat germplasm selection. Late maturity alpha-amylase is one such characteristic that renders wheat unsuitable for high value end products, even though the grain may appear sound. Phenotyping defect traits typically involves a multi-phase process, where genetic material for assay has been affected by non-genetic sources of variation in one or more previous linked stages of experimentation or preparation. The adoption of appropriate statistical design and analysis methods in these situations is, however, not widespread. Substantial sources of non-genetic variation were identified in the analysis of a designed experiment to measure late maturity alpha-amylase expression, indicating the potential for improved selection decisions. A simulation study based on these results suggests that significant gains over current methods in the accuracy of phenotyping this grain defect can be achieved with sound multi-phase statistical design and analysis techniques. Although restricted in scope, the simulation also indicates that a considerable increase in estimated heritability could be expected from the proposed methodology.