Date of Award

Fall 9-5-2025

Degree Type

Thesis

Degree Name

Master of Science (MS)

Department

Bioinformatics & Computational Biology

First Advisor

Yoel Stuart

Abstract

This thesis explores two distinct yet complementary aspects of evolutionary biology. The first chapter tests predictions from Fisher's geometric model of adaptation using a finely resolved fossil sequence of Gasterosteus doryssus spanning approximately 21,250 years. I analyze phenotypic changes in three armor traits during adaptation to a new environment and find that while step sizes follow the predicted exponential distribution—with many small steps and few large ones—they do not adhere to the model's temporal prediction that large-effect mutations should fix early in the adaptive walk. This discrepancy likely reflects adaptation from standing genetic variation rather than de novo mutation. The second chapter evaluates methodological approaches to missing data in quantitative genetics, comparing Complete Case Analysis (CCA) and Multiple Imputation (MI) across simulated datasets derived from threespine stickleback fish (Gasterosteus aculeatus) populations. I develop a simulation framework that uses Random Skewers as a metric to quantify how effectively each method preserves the original variance-covariance structure under different missingness scenarios. Results demonstrate that MI consistently outperforms CCA as missingness increases, with particularly pronounced advantages at higher levels of missing data. These findings have significant implications for paleontological studies and other contexts where incomplete datasets are unavoidable. Together, these chapters advance our understanding of both theoretical models of adaptive evolution and methodological approaches to evolutionary data analysis.

Download

Share

COinS