Doctor of Philosophy (PhD)


Plant, Environmental Management & Soil Science

Document Type



Plant breeding dramatically improved crops performances during human history and will play a pivotal role in shaping the future of agriculture. However, this activity requires large amount of time and resources. The availability of multi-year datasets and abundant DNA information enables new analyses and breeding approaches that can increase the productivity and efficiency of a breeding program. In recent years, the LSU Rice Breeding Program has implemented marker assisted selection (MAS) and genomic selection (GS) to predict performance of rice genotypes before field testing. The goal of this project was to conduct analyses and test molecular approaches to increase the efficiency of the breeding program. The objectives were: 1) investigate the effect of planting date on rice analyzing multi-year planting date experiments; 2) develop and validate a marker set to implement GS; 3) evaluate GS across elite rice populations; 4) evaluate crosses between different rice market classes developed with MAS. The analysis of planting date experiments showed that planting date significantly affects rice key traits. Grain yield, maturity, and milling yield decrease with delayed plantings. Thus, planting date must be considered when developing new improved varieties. A marker set to implement GS for southern US rice germplasm was designed for outsourcing. The validation on experimental data from 2018 to 2020 produced good GS predictive ability across different traits and segments of the breeding program. Consequently, all experimental lines tested in the last five years were genotyped with it. Four bi-parental and two multi-parent elite populations representing southern US rice germplasm were used to test GS across different traits and different training populations. The comparison of genomic and phenotypic response to selection showed that GS has similar efficacy in predicting the future performance of lines for high priority traits like grain and milling yield. The analysis showed that large amount of data improves the accuracy of selection, and that training set optimization can maximize GS accuracy. Finally, MAS made possible to efficiently develop populations from between grain classes crosses. The resulting lines showed improvement only for few traits, while no great advantages were observed for most of the other agronomic traits.



Committee Chair

Famoso, Adam N.