The Structure of Behavioral Variation Within a Genotype

Abstract

Variation is a broad feature of animal and human behavior. Phenotypic variability is commonly explained as the sum of genetic and environmental variance. However, in at least some cases the amount of behavioral variability observed in clonal or isogenic animals matches or exceeds that of animals with higher genetic diversity. Here, we focused on characterizing the structure and dimensionality of individual behavior within a single inbred genotype of the fruit fly \textit{Drosophila melanogaster} and compared it to that of outbred flies with high genetic diversity. In the first part of this thesis, we developed MARGO, a software package for tracking and implementing closed-loop stimulus paradigms for hundreds of animals simultaneously. In the second part of the thesis, we conducted a high-throughput screen of individual behavior with both inbred and outbred flies. Using MARGO to implement a battery of behavioral experiments, we found that sparse correlations between pairs of behaviors was common but that individual behavior was largely high dimensional. We also characterized individual RNA expression and found that models could predict a subset of the individual behaviors when trained on individual gene expression data. We further used previously published datasets to conduct a broad analysis of behavioral dimensionality and found that it was lower following neuronal manipulation. Collectively, this work constitutes a substantial contribution to understanding the structure of behavioral variability.