Fluctuating Selection and its Effects on Neutral Genetic Variation

Abstract

Natural selection has become a pervasive theory of evolution since its first proposal over 160 years ago. Some forms of selection have been intensively studied such that we now have a robust understanding of their dynamics, can predict their effects, and have wide-ranging evidence in empirical data from natural populations. Fluctuating selection, when selection shifts in strength or direction over time, is a form of natural selection that has been studied sporadically over the last century. As a consequence, we have a number of models of fluctuating selection but a comprehensive understanding of its effect on genomic variation has remained elusive. Fluctuating selection was originally studied in the form of phenotype observations, for example, Fisher and Ford studied wing patterns in a population of scarlet tiger moth and noticed patterns varied in proportion over time. More recently, fluctuating selection has been identified at the molecular level using allele frequencies. With recent advances in sequencing and decreases in associated costs, the last decade has seen a multitude of evidence of fluctuating selection in natural and experimental populations. A large quantity of this has stemmed from investigations of cosmopolitan Drosophila melanogaster populations over seasonal time scales but is also observed in other species such as Arabidopsis thaliana, non-biting midge, and stickleback. This thesis contains a methodical investigation of fluctuating selection, with an emphasis on its effects on neutral genetic variation. In Chapter 1, I perform an updated review of the field of fluctuating selection, compiling the abundance of recent evidence of fluctuating selection in natural and experimental populations as well as analytical studies of the dynamics of loci under this form of selection. Additionally, I highlight the gaps that remain in our current knowledge of fluctuating selection, with the hope of filling some with the subsequent work in this thesis. In Chapter 2, I benchmark population genetic simulation frameworks to identify an efficient and robust way to simulate non-standard forms of natural selection using fluctuating selection as an example. I compare four simulation frameworks using classical and updated datarecording methods to identify the workflow that confers the most efficient computational resource usage for use in the following chapters. Chapter 3 comprises a population genetic analysis of the effects of seasonally fluctuating selection on linked neutral genetic variation. A single-locus model of fluctuating selection is used to characterise its effect on linked neutral genetic variation using a range of population genetic statistics. I then compare the signatures of fluctuating selection to common selection types (i.e. positive and balancing selection) to determine if fluctuating selection can be differentiated from these forms. In Chapter 4, I simulate multilocus fluctuating selection to examine its impact on genomewide effective population size. A recent theoretical study found that seasonally fluctuating selection can decrease diversity genome-wide. I find that under realistic model parameters estimated from Drosophila data, fluctuating selection indeed leads to a strong reduction in effective population size, and thus reduces genome-wide genetic diversity. I discuss the results in the context of Lewontin’s Paradox, the observation that the range of genetic diversity across species is much smaller than the extent of variation in population size. Together, this thesis synthesises and expands our knowledge of fluctuating selection and its implications on population genetic measures and approaches.