Note: Genetic assimilation is sometimes used to describe "eventual extinction of a natural species as massive pollen flow occurs from another related species and the older crop becomes more like the new crop." This usage is unrelated to the usage below.
Genetic assimilation is a process by which the effect of an environmental condition, such as exposure to a teratogen, is used in conjunction with artificial selection or natural selection to create a strain of organisms with similar changes in phenotype that are encoded genetically. Despite superficial appearances, this does not require the inheritance of acquired characters, although epigenetic inheritance could potentially influence the result. Genetic assimilation is merely a method of overcoming the barrier to selection imposed by genetic canalization of developmental pathways.
If there is no canalization of a developmental pathway, genetic variation of pathway components results in a continuous spectrum of phenotypes, often distributed in a bell curve. In these cases artificial selection can be done in a straightforward way, by choosing offspring from one end of the curve and using them to breed the next generation. However, when a pathway is strongly canalized, all of the individuals, except perhaps a few at the furthest extreme of the bell curve, physically look the same regardless of their genotype - under normal environmental circumstances. However, a given genetic make-up does not predestine the same outcome under all possible circumstances - instead, it determines a norm of reaction that varies with the environment (phenotypic plasticity). There may be a way to stress an organism so that canalization breaks down, and many aberrant individuals can be selected for further breeding - these are said to phenocopy the desired genetic trait. With several generations of artificial selection in this manner, perhaps aided by mutagenesis, the genetic variation can be reduced to that of the furthest extreme of the original population, until canalization is overwhelmed even under normal environmental conditions. At this point the environmentally induced abnormality has been duplicated genetically.
The classic example of genetic assimilation was a 1953 experiment by C. H. Waddington, in which Drosophila embryos were exposed to ether, producing a bithorax-like phenotype  (a homeotic change) Flies which developed halteres with wing-like characteristics were chosen for breeding for 20 generations, by which point the phenotype could be seen without ether treatment.
Genetic assimilation in natural selection
It has not been proven that genetic assimilation occurs in natural evolution, but it is difficult to rule it out from having at least a minor role, and research continues into the question. Mathematical modeling suggests that under certain circumstances, natural selection will favor the evolution of canalization that is designed to fail under extreme conditions . If the result of such a failure is favored by natural selection, genetic assimilation will occur. In the 1960s C. H. Waddington and J. M. Rendel argued for the importance of genetic assimilation in natural adaptation as a means of providing new and potentially beneficial variation to populations under stress. Their contemporary Williams argued that genetic assimilation proceeds at the cost of a loss of developmental plasticity, and should be a minor mechanism. If it occurs frequently, genetic assimilation could contribute to punctuated equilibrium in evolution, as organisms repeatedly evolve systems of canalization, then break out of them under adverse circumstances.
To illustrate the phenomenon, Waddington (1975) managed to induce an extreme environmental reaction in the developing embryos of Drosophila. In response to ether vapor a proportion of embryos developed a radical phenotypic change, a second thorax. At this point in the experiment we would say that bithorax isn't innate; it is a kind of chimera induced by an unusual environment. But then Waddington continually selected for Drosophila with the developmental capacity to respond to the environmental stress. After about 20 generations of selection, some Drosophila were obtained that developed bithorax without being exposed to ether treatment. What happened, according to Waddington, is that selection favored a particular pathway that led to the production of the optimal (in this case desired) effect. Eventually the pathway became canalized, hence the end-state, bithorax, appeared regardless of environmental conditions.
Genetic assimilation generally describes the production of phenotypes with altered or decreased responsiveness to environmental conditions; the phenotype produced under a stressful condition becomes the phenotype for every condition. Genetic accommodation can be used to refer more broadly to changes in gene frequency that result from environmentally induced phenotypes. When used by contrast with genetic assimilation, the term can be applied more specifically to refer to the outcome that may be obtained when selection under stressful conditions is used to obtain a phenotype with increased responsiveness to environmental conditions. For example, H.F. Nijhout found that a black mutant line of Manduca sexta caterpillars sometimes became green under heat-shock conditions; selection of green caterpillars for thirteen generations yielded a polyphenic line that reliably became green under heat-shock, but remained black at cool temperatures. Either genetic assimilation or (other) genetic accommodation can be produced by similar selection procedures, and it may not be possible to predict in advance which phenomenon will occur. The underlying biological basis of these phenomena can be quite similar - temperature sensitive mutations and mutations affecting the activity of a gene without temperature sensitivity can each be produced by a small change in the sequence of a protein.
Genetic compensation describes the situation that occurs when an environmental condition changes the phenotype, but the new phenotype is not favored by selection. The outcome is a genetic change that shifts the expressed phenotype back to its original state despite the altered environment. For example, in salmon, anadromous sockeye populations migrate into the ocean to develop, where they ingest high levels of carotenoids which they use to produce an intense red coloration. "Residuals", salmon which do not enter the ocean, do not receive this nutrition and are a green color. However, they are thought to be the progenitors of nonanadromous kokanee salmon, which despite remaining in freshwater lakes develop an intense red coloration. Similar situations can be described for the pigmentation of tanagers and guppies. Genetic compensation may play a role in speciation by creating genetic incompatibilities between phenotypically similar populations within a species.
Genetic assimilation experiments have been comparatively rare in modern studies, because most geneticists are more interested in relating the activity of a gene to that of other genes. Those relationships are pursued by the study of genetic interaction, which is similar in concept. In a genetic interaction study, the experimenter begins with a strain that has a weak phenotype due to a known mutant allele, and screens flies for second mutations that create a stronger phenotype. Genetic interaction studies are typically used to identify mutant alleles with relatively severe effects, at least in the genetic background of the known mutant allele, which can be readily localized by genetic mapping and further characterized. The objective of these studies is to work out which genes have related functions --- often genes paired in this manner are later shown to code proteins that physically interact within the cell or catalyze sequential steps of a chemical reaction.
Note: references are currently incorporated as external links in the text