Improving Spring Wheat Breeding Efficiency through Genomic, Phenomic, and Multi-Environment Modeling Strategies

Peter Schmuker

Bread wheat (Triticum aestivum) is a cornerstone of agriculture in Eastern Washington. Wheat produced by Washington grain growers is exported to East Asian countries including Japan, South Korea, Indonesia, and Tawain. Millers, bakers, and consumers in these countries have stringent market class expectations associated with soft wheat quality including high milling yields, low flour protein content, and desirable cookie and cake quality. Phenotyping wheat end use quality traits is expensive, slow, and destructive. The second, third, and fourth chapters of this dissertation focus on evaluating cost-effective methods to screen and predict wheat end use quality traits. Chapter two evaluates the potential to predict total flour yield and break flour yield of the club wheat (Triticum compactum), a specialty market class of soft wheat, through spectroscopy. Using a comprehensive multi-year dataset, the results for this chapter found that non-linear calibration methods could predict milling yields more accurately than linear methods and the correlation between total flour yield and break flour yield. Model updating for independent growing seasons by including a subset of samples as additional testing observations was shown to improve validation accuracy, suggesting that targeted phenotyping of testing populations could improve model robustness. In chapter three, the sugar snap cookie diameter phenotype was modeled through phenomics and secondary traits. Secondary end use quality traits like solvent retention capacity tests can be used to assess end use quality during early generation stages in a breeding program when seed amounts are limited and baking tests are not feasible. The results from this chapter demonstrated that phenomic prediction based on spectral data acquired on intact grain could predict cookie diameter as accurately as individual secondary traits. The prediction accuracies of spectral calibrations were substantially more accurate than regression from flour protein content which was phenotyped through an appropriate calibration. The ten-times cost discrepancy between acquiring spectral data and evaluating multiple secondary traits make phenomics a powerful tool to improve soft wheat end use quality. The fourth chapter compares phenomics, genomic selection, phenomics assisted genomic selection, and marker assisted selection to predict quality traits in soft and hard spring wheat breeding populations. Phenomics assisted genomic selection through a multi-trait approach gave promising improvements over univariate genomic selection for multiple hard and soft wheat quality traits. Marker assisted selection was evaluated through diagnostic trait linked markers currently used by the Spring Wheat Breeding Program to screen crossing block parents, which could only effectively predict flour swelling volume in soft wheat genotypes. Through the manipulation of selection intensity, phenomic prediction alone could be used to achieve comparable genetic gain to genomic selection for milling yield and consumer product traits when selection intensity with genomic selection is lax. The fifth chapter focuses on optimizing variety testing practices for spring and winter wheat testing networks. The current number of locations evaluated per year, testing cycle length, and number of plots evaluated per year are sufficient for most rainfall zones. The lowest rainfall and intermediate spring wheat rainfall zones could benefit from additional locations per year to improve network precision. Current mega-environment groupings could be altered from four to two, leading to an improvement in network precision in a cost neutral manner. Combining mega-environments would also minimize the loss of network precision if trialing practices must be scaled back for budgetary reasons.

Improving Spring Wheat Breeding Efficiency through Genomic, Phenomic, and Multi-Environment Modeling Strategies

Files and links (1)

Abstract

Metrics

Details