Biophysical constraints determine the selection of phenotypic fluctuations during directed evolution

Phenotypes of individuals in a population of organisms are not fixed. Phenotypic fluctuations, which describe temporal variation of the phenotype of an individual or individual-to-individual variation across a population, are present in populations from microbes to higher animals. Phenotypic fluctuations can provide a basis for adaptation and be the target of selection. Here we present a theoretical and experimental investigation of the fate of phenotypic fluctuations in directed evolution experiments where phenotypes are subject to constraints. We show that selecting bacterial populations for fast migration through a porous environment drives a reduction in cell-to-cell variation across the population. Using sequencing and genetic engineering we reveal the genetic basis for this reduction in phenotypic fluctuations. We offer one interpretation for this reduction by developing a simple, abstracted, simulation model of the evolution of phenotypic fluctuations subject to constraints. We find that directed evolution applied to constrained phenotypes results in a decrease (increase) in phenotypic fluctuations when selection is weak (strong) without explicitly specifying a mechanistic basis for phenotypic fluctuations. This result is a generic property of selection acting on phenotypic fluctuations near a bound on the selected phenotype. We explore the mechanism of the observed reduction of phenotypic fluctuations in our experimental system, discuss the relevance of our abstract model to the experiment and explore its broader implications for evolution. PACS numbers: 87.23.Kg,87.17.Jj