Degree

Doctor of Philosophy (PhD)

Department

Plant, Environmental Management and Soil Sciences

Document Type

Dissertation

Abstract

Plant breeding is a crucial tool in addressing global challenges such as climate change, population growth, and the need for more resilient crops. Rice serves as a staple food for a significant portion of the global population and has served as a model crop for genetic and genomic studies. Genomic selection (GS) has become valuable tool in plant breeding, allowing breeders to predict breeding lines’ performance based on their genotypes. The training (TS) set is a central component of genomic selection and the optimization of the training set is crucial for efficient implementation of genomic selection. The first two objectives of this project were to 1) investigate the effect of haplotype sizes on genomic selection accuracy and epistasis, 2) study the effect of Clearfield® herbicide carryover on rice yield trials and genomic prediction training set accuracy. Additionally, a third objective of this project was the map and characterize a spontaneous mutation that arose in the LSU rice breeding program and causes a short grain phenotype.

The investigation of haplotype sizes on genomic selection accuracy and epistasis revealed that increased recombination decreases haplotype sizes and increases resolution. Smaller haplotype sizes in the TS were associated with increased prediction accuracy, particularly for highly quantitative traits. Additionally, the inclusion of epistatic effects in the model increased overall genetic variance but did not significantly improve prediction accuracy.

Clearfield® herbicide carryover has significant implications for rice yields, with Clearfield (CL) genotypes generally exhibiting higher yields compared to Conventional (CN) materials in the same trial. When predicting CL genotypes, using a Clearfield-only training set appears optimal, leveraging the shared genetic background and similar environmental responses. In contrast, a combined Clearfield and Conventional training set is optimal for predicting CN lines. This highlights the importance of training set optimization in genomic selection for accurate prediction of different rice genotypes.

Finally, the discovery of a spontaneous mutation within an advanced experimental line was observed, characterized, and fine-mapped. Sequence analysis identified a single, functional mutation that distinguishes the mutant from all wild-type lines. The recessive phenotype is caused by a disruption in the srs3 gene on chromosome 5. DNA marker (PACE) assays were developed for the functional mutation and validated across a panel of US breeding germplasm, showing a perfect association with the mutant allele.

Date

4-11-2024

Committee Chair

Famoso, Adam

Available for download on Tuesday, April 01, 2025

Share

COinS