Fig. 3
From: Development and selective grain make plasticity 'take the lead' in adaptive evolution
Evolving a single map creates similar phenotypic distributions in the other maps. A population whose individuals initially exhibit a random phenotypic distribution in t = 0 (A, small panels) is evolved to fit a target phenotypic distribution (ST = 1) using as an input just one kind of phenotypic determinant (i.e., genetic, environmental, or parental variation). Other targets (ST = −1) give similar results (see Additional file 1: Fig. S6). In each generation, one individual is exposed to 10 different input values (0 < x < 1) of a single phenotypic determinant (the colour of each dot in A–B represents value of this input). This parametric variation produces a set of ten potential phenotypes whose slope is compared to the target to evaluate the individual’s fitness (see “Methods”). After 105 generations in a mutation-selection-drift scenario (where other sources of phenotypic variation are frozen), the population has a narrow phenotypic distribution in the evolved map (A, large panels). In (B) we uncover variation in the other maps by introducing parametric variation (0 < x < 1) in the phenotypic determinants that were kept fixed during the evolutionary trial. Results reveal that selection on a single map creates significant side-effect phenotypic distributions in the other maps that are not the target of selection. C Correlations in the side-effect maps are significant across all parameter values at which the parameter of the evolved map is frozen. p = 64 individuals; n = 30 replicates, GRN + Multilinear model