Document Type

Dissertation - Open Access

Award Date


Degree Name

Doctor of Philosophy (PhD)

Department / School

Agronomy, Horticulture, and Plant Science

First Advisor

Melanie Caffe


Oats are important cereal crops worldwide, primarily valued for their nutritional and health benefits. The primary goal of oat breeding programs is to accelerate genetic gain to develop oat varieties with the desired product profiles, with the aim of meeting the needs and expectations of growers, processors, and consumers. Achieving this goal involves conducting extensive and resource-intensive field trials, particularly for traits such as crop yield and grain quality, which are strongly influenced by environmental factors. The availability of cost-effective molecular markers and tools like genome-wide association studies (GWAS) and genomic selection have the potential to accelerate the oat breeding process, yet both methods remain unexplored for numerous oat traits. The first study conducted resistance screening and GWAS for Dreschelera avenae leaf spot using 313 US elite spring oats and 12,453 GBS SNP markers. The study identified makers linked to 10 QTL regions on chromosomes 3A, 4A, 4C, 4D, 5A, 6C, 7A, and 7C. The annotation of these QTL regions revealed the presence of genes like TIR-NBS-LRR, cysteine-rich RLK, and 35 other plant defense-related protein families. Additionally, we identified 19 lines with multiple resistant QTL alleles, offering potential for oat breeding against D. avenae leaf spot. The second study was conducted to identify markers associated with twelve oat yield, milling, and nutritional quality-related traits. We conducted a GWAS using a total of 648 oat PYT breeding lines grown in four locations over the span of 2015–2017, with a total of 23,306 high-quality GBS SNP markers. We identified 150 unique marker-trait associations that were distributed across 17 chromosomes, encompassing all three oat sub-genomes. A total of 18 SNP markers for plant height and 19 for days to heading were found on chromosomes 1D, 5A, 5D, 6C, 7A, 7C, and 7D. Oat yield and its components associated with SNPs were found on chromosomes 1C, 7C, and 7D, with a notable marker on 7D explaining 7% of the phenotypic variation (PEV). Thousand kernel weight (TKW) and test weight (TW) were associated with SNPs on multiple chromosomes (1A, 2C, 2D, 3A, 4A, 4C, 4D, 5A, 5C, 6A, 6C, and 7D). Remarkably, a TKW SNP marker on chromosome 7D accounts for 12.2% of the variation. Kernel sizes, ranging from thin to plump, were linked to makers on chromosomes 1A, 2A, 2C, 2D, 4C, 5A, 5C, 5D, 6A, 6C, 7A, and 7D. Oil content is influenced by regions on chromosomes 1A, 1D, 2D, 3D, 6A, and 7C, while both betaglucan and protein content have associations on chromosomes 1D, 2D, 4C, 5A, 6C, and 7D. These identified SNPs in the study provide valuable information for oat breeding programs aiming at developing new varieties with high yields and improved quality. The third study explored the variations in prediction accuracy of genomic selection models for multiple traits within oat breeding lines from the PYT evaluated across three years (2015–2017) and in multiple environments. A total of 14 different genomic prediction models were evaluated for a range of oat traits, including yield, agronomic, milling, and quality traits. The results revealed that no single model excelled in predicting all traits, with varying levels of accuracy observed across different models. Overall, looking at three-year averages, models BA, BB, BC, BL, BRR, GAU, GB, and RRB consistently had high average prediction accuracy across multiple traits; however, EN, LA, and RF were best for oil content and HD traits, making them ideal choices when those traits are a priority. The fourth study examined single-trait and multi-trait genomic selections using the same PYT data, specifically evaluating the Bayesian Single-Trait Multi-Environment (BSTME) and Multi-Trait Multi-Environment Model (BMTME) for twelve oat traits under various cross-validation (CV) and sparse testing strategies. The results emphasized the high predictive accuracy of both the MT-ME and ST-ME models under the sparse testing strategy. The CV2 cross-validation method yielded higher accuracy for all traits, suggesting that increasing environmental and genetic connections enhances prediction accuracy. Further, the study indicated that reducing the test size through sparse testing from 20% to 50% only led to a marginal decrease in accuracy, potentially reducing breeding costs, particularly during the early stages of line development, and facilitating accelerated breeding programs. The results of this study have significant implications for enhancing oat breeding. By identifying genomic regions linked to key oat traits like resistance to D. avenae, nutritional qualities, milling characteristics, and agronomic features, breeders can now employ marker-assisted selection, marker-assisted backcrossing, and gene stacking. Additionally, the findings shed light on the use of genomic selection and prediction models, especially for complex traits influenced by both genetics and the environment, providing a promising approach to more efficient and accurate oat breeding. These advancements can ultimately lead to the accelerated development of oat varieties with improved resistance and other desirable traits, benefiting both farmers and consumers.

Library of Congress Subject Headings

Oats -- Breeding.
Oats -- Genetics.


South Dakota State University

Available for download on Sunday, December 15, 2024



Rights Statement

In Copyright