MUTATIONAL CONTRIBUTIONS TO GENETIC VARIANCE‐COVARIANCE MATRICES: AN EXPERIMENTAL APPROACH USING INDUCED MUTATIONS IN ARABIDOPSIS THALIANA

Genetic potential for evolutionary change and covariational constraints are typically summarized as the genetic variance‐covariance matrix G , and there is currently debate over the extent to which G remains effectively constant during the course of adaptive evolution. However, G provides only a temporally restricted view of constraints that ignores possible biases in how new mutations affect multivariate phenotypes. We used chemical mutagenesis to study the effect of mutations as summarized by the mutational covariance matrix, M , in Arabidopsis thaliana . By introducing mutations into three isogenic strains of A. thaliana , we were able to quantify M directly as the genetic variance‐covariance matrix of mutagenized lines. Induced mutations generally did not alter the means of the six morphology and life‐history traits we measured, but they did affect the levels of available genetic variation and the covariances among traits. However, these effects were not consistent among the three isogenic lines; that is, there were significant differences among the lines in both the number of mutations produced by ethyl‐methane‐sulfonate treatment and the M matrices they induced. The evolutionary implications of the dependence of M on the number of mutations, the particular genetic background, and the mutagenic sampling of loci in the genome are discussed in light of commonly applied models of multivariate evolution and the potential for the genetic architecture itself to change in ways that facilitate the coordinated evolution of complex phenotypes.