Accounting for rank uncertainty in decision making for plant breeding
Date
2022-08
Authors
Bijari, Reyhaneh
Major Professor
Advisor
Olafsson, Sigurdur
Vardeman, Stephen
Min, Kyung
Nordman, Daniel
Li, Qing
Committee Member
Journal Title
Journal ISSN
Volume Title
Publisher
Altmetrics
Abstract
This dissertation is devoted to helping solve real-world plant breeding problems using innovative data science. There have been lots of efforts in the area of plant breeding to improve the quality of decisions made in such programs. While the use of new techniques has increased in this area, there exist a lot of limitations in these programs that tie to unavoidable uncertainties that need to be taken into account for proper analysis. This work addresses a plant breeding decision-making challenge that stems from having a very limited number of environments observed for each plant breeding trial. We propose new methods that plant breeders can utilize when analyzing the genotypes’ performance. Specifically, to capture the inherent uncertainty due to the specific set of environments observed, we propose a bootstrapping approach to estimating the distribution of rank and constructing confidence intervals around it. We also a new approach to comparing genotypes probabilistically and offer a new ranking method based on pairwise probabilistic comparisons of genotypes. The methods are provided in an R package for analysis of plant breeding experiments for all users. We believe plant breeding would benefit from the body of this work as it tries to fill the gap in the analysis of multi-environment trails’ data.
Series Number
Journal Issue
Is Version Of
Versions
Series
Academic or Administrative Unit
Type
dissertation