The word “epigenetics” once meant simply “development”—that is, the way the genome worked itself into an organism through the production and regulation of proteins and absorption of food and materials from the outside. Now, however, the term means roughly “forms of inheritance that rest on modification of the DNA sequence,” and by “DNA sequence” I mean the sequence of four bases (A, G, C, and T) that constitutes the DNA code.
We now realize, though, that those four DNA can be modified, and in an inherited way, in a manner that can affect the development, behavior, or structure of an organism. Such modification often takes place via DNA methylation, in which some of those four bases acquire methyl groups, thereby changing how the DNA functions.
Such methylation, as you’ll see by reading the Wikipedia link above, is important in organismal development—something we’ve realized only in recent decades. For example, there is differential “imprinting” of DNA via differential methylation in male versus female parents, and this results in the DNA in the zygote doing different things depending on whether it came from mother or father (organisms have paired chromosomes, getting one from each parent). This has led to speculations about the evolution of differential imprinting resulting from different interests of mother and father in how and which zygotes develop.
Methylation is also important in silencing the X chromosomes of female mammals that have two X chromosomes. This silencing equalizes the gene dosage between XX females and XY males (Y chromosomes barely have any genes), so normal development usually means keeping the dose of X-linked genes (and hence the amount of proteins they make) the same in both sexes.
Anyway, that kind of epigenetics is itself based on the DNA code. That is, the A, C, T and G sequences, and the environment in they find themselves, are “programmed” by natural selection to add or remove methyl groups from other parts of the DNA. Such adaptive epigenetic programming must perforce rest on the sequence of DNA bases, because methylation of the DNA is not inherited in a stable way. In imprinting, for instance, those bases that are methylated differently between the sexes are “reset” in the offspring: the methylation vanishes and is re-constituted before reproduction by whatever sex the embryo happens to be. That reconstitution is programmed in the DNA code itself. Methylated bases in DNA don’t usually get passed on from one generation to the next. There are two important points to add.
First, as I said, if methylation is itself an adaptation produced by natural selection, it ultimately must rest on changes in the sequence of unmethylated A, C, T, and G bases in the DNA. Only unmethylated bases are stably inherited, and evolution demands stable inheritance. There must be something about the DNA sequence that controls its own methylation. (Note that some methylations can last for several generations, though that’s not common.) For a population to change over time and acquire adaptations (or features that evolve through nonadaptive processes like genetic drift), whatever types of replicators that are inherited must remain unchanged, with of course the exception of mutations in the DNA code. But if the DNA code changed unpredictably back and forth each generation, natural selection and evolution wouldn’t work.
Second, there are also epigenetic changes that are induced not by the DNA sequence but by the environment. Temperature, starvation, and other environmental factors can cause methylation of the DNA as well. The thing is, though, that such changes, because they’re rarely passed on to future generations, cannot serve as the basis of evolutionary change. Such changes constitute true Lamarckian inheritance, i.e., the inheritance of acquired characteristics.
And lots of studies show us that Lamarckian inheritance doesn’t operate. Changes that are induced by the environment, or the organism’s “striving,” can’t somehow get incorported into the DNA. An athlete, for example, doesn’t produce kids with bigger muscles. And after millennia of circumcision, Jewish boys are still obstinately born with foreskins, giving my readers plenty to argue about.