Genomic Selection Predicts Wheat Baking Quality Years in Advance

A new study from Montana State University has found that genomic selection can accurately predict key baking quality traits in wheat years before they are traditionally measured. Published in Theoretical and Applied Genetics, the research demonstrates how breeders can use these models to make better-informed decisions early in the breeding process, potentially saving time and costs.

The study analyzed twelve years of data from the Montana State University Spring Wheat Breeding Program, examining 842 wheat lines across 12 quality traits, including flour yield, dough properties, and loaf volume. The team used genotyping-by-sequencing to profile the lines, resulting in over 21,000 genetic markers. High broad-sense heritability values were observed for nearly all traits, making them well-suited for genomic selection.

The researchers compared two analytical frameworks: a conventional two-step approach and a single-step reaction norm model. The single-step model, which accounts for genotype-by-environment interactions, showed higher predictive ability, particularly for traits like bake time, flour yield, and mix time. The study also highlighted the risks of data leakage, where information from validation sets can artificially inflate model performance.

The findings suggest that breeders can now select for end-use quality traits earlier in the pipeline, accelerating genetic gain and ensuring that wheat meets market demands. As genomic datasets grow and reference genomes improve, these methods are expected to become standard tools in wheat breeding.