(Internal) Improving predictions of crop plant genetic performance

Advisor: Lewis Lukens, Plant Agriculture

Proposed co-advisor (UoG): TBC

In plants, genotype and environment interact in complex ways to generate variation in quantitative traits such as grain yield. As a result, a genotype’s value for yield in one environment may be a poor predictor of the genotype’s yield in another environment. This situation is problematical because growers choose genotypes to grow based on yields in past environments, and these could substantially differ from current environments. To address this issue, this research project seeks to: 1) Define regions in Ontario in which specific crop genotypes have the most consistent performances, and 2) Explore if predictions of genotype performance in an environment can be improved by utilizing genomic information. The work will use machine learning and statistical methods to generate models that estimate location-specific genotypic values from environmental data and genome-wide SNP data. Models trained on prior years will be evaluated based on the accuracies of their subsequent years’ predictions. A successful project would help increase grain production without increasing fertilizer or pesticide inputs.

This is a one-semester project.