UC Davis

The primary goal of the research presented in this dissertation is to assess the prospects of utilizing modern plant breeding methodology to address the lack of biomass yield improvement in alfalfa. Biomass yield improvement in alfalfa has remained stagnant for the last ~30 years and cultivated areas are in steady decline. Significant advancements have been made in genomics and phenomics in recent years with applications in plant breeding. Chapter 1 reviews relevant literature including the role alfalfa plays in western agriculture, history, biology, breeding, and an introduction to genomic selection and high throughput phenotyping. Chapter 2 seeks to empirically test genomic selection for forage dry matter yield in alfalfa. Genomic selection has proven successful for increasing the rate of genetic gain in other crops, and models have been developed for a variety of traits in alfalfa. However, no practical testing has occurred to date. Chapter 3 builds on the findings from the previous chapter by addressing some limitations in the predictive model. Namely, building a model based on family bulks and half-sib family performance, rather than on an individual plant basis, to better reflect the commercial environment alfalfa is grown in. The results from chapter 2 and 3 provide encouraging prospects with prediction accuracies around 0.3. However, one of the greatest limitations for a successful genomic selection breeding program is the size of the training population, particularly in crops with limited resources such as alfalfa. Chapter 4 explores the potential to greatly increase the size of breeding trials without a comparable increase in costs. We utilized drone based multispectral imagery to estimate biomass of alfalfa and a range of forage grasses across a range of plot types typically used in forage breeding trials. Remote sensing and high throughput phenotyping massively reduce the labor component of data collection while simultaneously providing accurate estimates of forage biomass.