What I find interesting is that methylation of cytosine, one common epigenetic modification, increases local mutation rates. Methylation typically reduces expression levels of nearby genes. A concomitant increase in local mutation rate may further disrupt transcription factor binding sites, also reducing expression. Thus, multi-generation methylation at a locus could increase the odds of genetically encoding the epigenetic information (assuming that the new selection pressure remains constant).
I've glossed over some details (e.g., somatic vs germ-line methylation patterns).
I've glossed over some details (e.g., somatic vs germ-line methylation patterns).