RESUMO
Plasticity is found in all domains of life and is particularly relevant when populations experience variable environmental conditions. Traditionally, evolutionary models of plasticity are non-mechanistic: they typically view reactions norms as the target of selection, without considering the underlying genetics explicitly. Consequently, there have been difficulties in understanding the emergence of plasticity, and in explaining its limits and costs. In this paper, we offer a novel mechanistic approximation for the emergence and evolution of plasticity. We simulate random "epigenetic mutations" in the genotype-phenotype mapping, of the kind enabled by DNA-methylations/demethylations. The frequency of epigenetic mutations at loci affecting the phenotype is sensitive to organism stress (trait-environment mismatch), but is also genetically determined and evolvable. Thus, the "random motion" of epigenetic markers enables developmental learning-like behaviors that can improve adaptation within the limits imposed by the genotypes. However, with random motion being "goal-less," this mechanism is also vulnerable to developmental noise leading to maladaptation. Our individual-based simulations show that epigenetic mutations can hide alleles that are temporarily unfavorable, thus enabling cryptic genetic variation. These alleles can be advantageous at later times, under regimes of environmental change, in spite of the accumulation of genetic loads. Simulations also demonstrate that plasticity is favored by natural selection in constant environments, but more under periodic environmental change. Plasticity also evolves under directional environmental change as long as the pace of change is not too fast and costs are low.