The relationship between robustness and evolution

Developmental robustness (canalization) is a common attribute of traits in multi-cellular organisms. High robustness ensures the reproducibility of phenotypes in the face of environmental and developmental noise, but it also dampens the expression of genetic mutation, the fuel for adaptive evolution. A reduction in robustness may therefore be adaptive under certain evolutionary scenarios. To better understand how robustness influences phenotypic evolution, and to decipher conditions under which canalization itself evolves, a genetic model was constructed in which phenotype is explicitly represented as a collection of traits, calculated from genotype, and the degree of robustness can be explicitly controlled. The genes were sub jected to mutation, altering phenotype and fitness. We then simulated the dynamics of a population evolving under two classes of initial conditions, one in which the population is at a fitness optimum and one in which it is far away. The model is formulated with two robustness parameters in the genotype to phenotype map, controlling robustness over a tight (γ) or a broad (α) range of values. Within the robustness range determined by γ, high robustness results in a equilibrium population fitness closer to the optimal fitness value than low robustness. High robustness should be favored, therefore, under a constant optimal environment. This situation reverses when populations are challenged to evolve to a new phenotype optimum. In this situation, low robustness populations adapt faster than high robustness populations and reach higher equilibrium mean fitness. A larger set of phenotypes are accessable by mutation when robustness is low, in part explaining why low robustness is favored under this condition. A larger range of robustness could be sampled by varying α, revealing a complex relationship between robustness and both the initial rate of phenotypic adaptation as well as the final equilibrium population mean fitness. Intermediate values of α produced a bifurcation in evolutionary trajectories, with some populations remaining at low population mean fitness, and others escaping to achieve high population mean fitness. We then allowed robustness itself to be encoded by a mutable genetic locus that could co-evolve along with the phenotype under selection. Low robustness genotypes are initially favored when adapting to a new optimal phenotype. A high robustness genotype then replaces it, well before maximum fitness is achieved, and moreover appears to prevent further invasion into the population of a low-robustness genotype. This phenomenon was dependent on having tight linkage (and sufficiently low mutation rate) between the robustness locus and the loci encoding phenotype.