University of Illinois Urbana-Champaign

The role of genomic selection and advanced statistical modeling in domesticating perennial sorghum

Widener, Sarah Joy Rhodes RIchards

This item's files can only be accessed by the System Administrators group.

Permalink

https://hdl.handle.net/2142/132795

Description

Title
The role of genomic selection and advanced statistical modeling in domesticating perennial sorghum
Author(s)
Widener, Sarah Joy Rhodes RIchards
Issue Date
2025-12-05
Director of Research (if dissertation) or Advisor (if thesis)
Lipka, Alexander E
Doctoral Committee Chair(s)
Lipka, Alexander E
Committee Member(s)
  • Jarquin, Diego
  • Sacks, Erik
  • Jamann, Tiffany M
Department of Study
Crop Sciences
Discipline
Crop Sciences
Degree Granting Institution
University of Illinois Urbana-Champaign
Degree Name
Ph.D.
Degree Level
Dissertation
Keyword(s)
  • Perennial Sorghum
  • Sorghum
  • Genomic Prediction
  • Genotype by Environment Interactions
  • Genomic Best Linear Unbiased Prediction
  • MultiBLUP
  • Environmental Covariates
  • Food Security
  • Sustainable Agriculture
  • GenSelect
Abstract
The perennialization of agronomically valuable grain crops, such as sorghum (Sorghum bicolor (L.)), has the potential to significantly enhance sustainable agriculture and food security, particularly in regions experiencing soil degradation, water scarcity, and resource limitations. Genomic prediction (GP) is a crucial tool for accelerating the selection process in breeding perennial sorghum for key domestication and perenniality traits. The goal of the research presented in this dissertation is to assess and implement the most effective GP model for accelerating breeding cycles. This work focused on two key questions. First, are there benefits of incorporating genetic architecture (using multi-kernel) into GP predictive ability? Second, can accounting for Genotype × Environment (GxE) interactions, including the use of environmental covariates (ECs), improve predictive ability? We found that partitioning single nucleotide polymorphisms (SNPs) based on a priori genetic features did not consistently boost predictive ability when compared to standard genomic best linear unbiased prediction (GBLUP) for these populations. Therefore, we inferred that the added complexity of MultiBLUP is not worth the negligible gains. In our multi-environment studies, we found that simple traits such as flowering time showed high and stable predictive ability. In contrast, perenniality traits were more complex to predict, and including ECs had little effect on improving their predictive ability. We developed GenSelect, an accessible R Shiny application that implements advanced GP models (GBLUP, MultiBLUP, GxE with and without ECs), allowing practitioners to easily upload their own data and identify the optimal model for their unique accessions. This application makes these models usable and broadens the potential impact of this research. This dissertation aims to identify the most effective and practical genomic prediction strategies for accelerating the breeding of perennial sorghum. These findings and the GenSelect tool facilitate the development of a practical GP pipeline for perennialized sorghum, guiding future efforts to create environmentally sustainable crops and enhance food security. Future research should focus on obtaining below-ground environmental covariates and collecting data in more diverse environments to boost multi-environment predictive abilities for perenniality traits.
Graduation Semester
2025-12
Type of Resource
Thesis
Handle URL
https://hdl.handle.net/2142/132795
Copyright and License Information
Copyright 2025 Sarah Widener

Owning Collections

Graduate Dissertations and Theses at Illinois PRIMARY
Graduate Theses and Dissertations at Illinois