Genomic Selection for Yield and Seed Composition Stability in an Applied Soybean Breeding Program

Authors

Benjamin Harms, University of Nebraska - LincolnFollow

First Advisor

Dr. David Hyten

Date of this Version

5-2023

Comments

A THESIS Presented to the Faculty of The Graduate College at the University of Nebraska In Partial Fulfillment of Requirements For the Degree of Master of Science, Major: Agronomy, Under the Supervision of Professor David L. Hyten. Lincoln, Nebraska: May, 2023

Copyright © 2023 Benjamin Mark Harms

Abstract

Stability traits are of primary importance in plant breeding to ensure consistency in phenotype across a range of environments. However, selection efficiency and accuracy for stability traits can be hindered due to the requirement of obtaining phenotype data across multiple years and environments for proper stability analysis. Genomic selection is a method that allows prediction of a phenotype prior to observation in the field using genome-wide marker data and phenotype data from a training population. To assess prediction of stability traits, two elite-yielding soybean populations developed three years apart in the same breeding program were used. The individuals in each population were tested across three years and seven or more environments, allowing for calculation of observed stability and assessment of prediction accuracy. The primary goal of this research was to provide an overview of genomic selection for yield stability, protein content stability, and oil content stability in an applied soybean breeding program. Factors affecting prediction accuracy were assessed, including SNP density, SNP marker type, and stability measure. Briefly, predictive abilities were low across all stability traits and stability measures for prediction across populations, ranging from -0.01 to 0.37. During applied prediction of non-parametric measures for yield stability, we obtained rank coincidence of roughly 0.65. When individuals in the top half of predicted stability are selected, roughly 65% of those individuals are expected to be in the top half of observed stability. For prediction of protein and oil content stability for static environmental variance stability, we obtained rank coincidence of 0.59 and 0.58. While predictive abilities were too low for use in a breeding program, rank coincidence gave more promising results for applied genomic selection for stability traits. With improvement in methods such as prediction model, SNP type, and greater training and validation phenotype environments, there is potential for genomic selection to effectively improve stability in a breeding program by implementing selection at an early stage when phenotype data are insufficient to select for stability.

Advisor: David L. Hyten