Accelerating genetic gain in plant breeding demands increased selection intensity, enhanced selection accuracy, broader genetic diversity, and shortened breeding cycle. As phenomic platforms and genomic resources continue to evolve, integrating these complementary datasets offers opportunity to improve breeding pipelines. This dissertation explores multiple approaches to potentially incorporating phenomic information with genomic data, aiming to increase prediction accuracy and selection intensity, and identify genomic regions for economically important traits in soft winter wheat. Infrared thermal imaging enabled high-resolution differentiation of Fusarium head blight (FHB)-resistant and -susceptible genotypes at the single-spike level. However, field-scale implementation requires careful consideration due to uncontrolled factors under field conditions, where no direct relationship between plot-level infrared thermal readings and FHB-related traits was established. Hyperspectral imaging demonstrated superior predictive ability and prediction accuracy over genomic prediction alone for deoxynivalenol (DON) content. Integrating phenomic and genomic predictions by model blending enhanced prediction accuracy and allowed clustering-based selection on predicted DON content in F4:5 breeding lines. Additionally, multiple strategies for integrating UAV-derived vegetation indices (VIs) to improve genomic prediction accuracy were evaluated, with varying degrees of success depending on how UAV-derived information were used as fixed effect, as well as training set composition. Beyond enhancing prediction, phenomic data facilitated the identification of key genomic regions associated with DON content and grain yield, underscoring the potential of phenomic information as a phenotypic input in genome-wide association studies. Collectively, these findings support the potential application phenomics-genomics integration in potentially improving wheat breeding. By enhancing selection accuracy, identifying informative genomic regions, and potentially reducing selection intervals, this work lays the foundation for a more efficient breeding pipeline. However, careful consideration is essential when implementing combined phenomic-genomic approaches to ensure robust, field-applicable results.
Read
