Using unbalanced data from a barley breeding program to estimate gene effects: the Ha2, Ha4, and sdw1 genes.

Abstract

The effects of genes in self-pollinated crops are usually estimated from designed experiments where selection is minimised. In this study, we used a large, but unbalanced, dataset from a barley breeding program to estimate the effects of the Ha2, Ha4, and sdw1 genes on grain yield, grain weight, grain protein, malt extract, and diastatic power. The Ha2 and Ha4 genes for resistance to cereal cyst nematode were under intense selection pressure, whereas the sdw1 gene, which reduces plant height, was under mild selection pressure. From a mixed-model analysis of mainly F₅-derived lines over 5 years, resistance due to the Ha2 gene was found to increase grain yields at 2 sites where the nematode was expected to be present, but not at 3 other sites. There was no significant effect of Ha4 on grain yield. Because of selection, data from later stages of evaluation were not useable for Ha2 or Ha4. From analyses of both early stage and later stages of evaluation, the semi-dwarf allele of the sdw1 gene increased grain yields at high-yielding sites, but decreased yields at low-yielding sites. The semi-dwarf allele reduced grain weight. The effects of Ha2 or Ha4 on malt extract and diastatic power were not significant, but the semi-dwarf allele at sdw1 reduced grain protein. We concluded that plant breeding data can be used to successfully estimate the effects of important genes, with bias due to selection minimised by the use of data from appropriate stages of selection and the use of appropriate statistical models.