Item

Abstract

The phenotypic variation generated by mutation is not random, in that all variants are not equally likely to be generated. Rather, variation is biased in direction and extent by the structure of the G-P map and the particular mutational processes that reconfigure it. I use experimental evolution in a bacterial model organism to explore how such biases in the production of variation can interact with natural selection. The thesis consists of two main projects. The first investigates the role of bias in shaping the phenotypic outcome of adaptation. I uncover examples in our experimental populations where natural selection fails to 'find' the fittest available phenotype. The underlying cause is traced to differences in the relative likelihood of the mutational paths required to reach each phenotype. Thus, bias may at times channel evolution toward sub-optimal adaptive outcomes. The next project examined how bias can itself be shaped by natural selection so as to facilitate further adaptation – the ‘evolution of evolvability’. To this end, I conducted an experiment where bacterial lineages were challenged to repeatedly activate and then inactivate a focal phenotype that was in turn beneficial then deleterious across time. Failure to reach the target phenotypic state resulted in extinction of that lineage and replacement by a contemporary extant lineage from the population. This lineage-level birth-death process allowed selection to operate on the emergent property of evolvability, leading to lineage adaptation in the form of a 'contingency locus' that facilitated rapid switching between the target states. Moreover, the increased speed with which the target phenotypes were generated enabled additional adaptive changes to take place in these lineages – a possibility that was stalled in lineages not possessing the rapid-switching ability.