Reversibility in Evolution Considered From the Standpoint of Genetics1

Summary 1. In Drosophila the great majority of mutations are reversible in direction, and very commonly the “reverse mutation” appears to reconstitute precisely the original gene. The mutation from the normal to the abnormal type has been found usually to be a change from a more active to a less active condition of the gene, and is hence to be regarded as constituting, itself, a reversal of the original direction of evolution. The so‐called reverse mutation, in such a case, is really a mutation in the direction of past evolution. As the latter changes usually occur less readily than the former (except in the rare cases of highly mutable mutant genes), it is to be inferred that evolution proceeded contrary to the prevailing mutation pressure, and hence only by the aid of selection. Thus, with selection relaxed, a certain reverse evolution would tend to occur, so far as individual loci were concerned. 2. The reversibility of most mutations is significant in the theory of the gene and of evolution in showing, first, that most mutations pre not mere losses of genes. Secondly, as Timoféëff‐Ressovsky, Zimmer and Delbrück (1935) have pointed out, the fact of their fairly high reversibility indicates that most mutations involve canalized reactions in a unified gene structure. For, in the case of radiation mutations at any rate, it can be shown that activation of any one of a large number of atoms, and of gene parts, results in sensibly the same mutation; that this principle applies even to reverse mutations indicates that precisely the same gene‐part as was struck in making the original mutation need not be struck again to make the reversal. This adds to the already existing evidence (Muller, 1932) that the gene mutation process involves a chain of reactions of which the primary one may even have lain outside the gene. It is at present uncertain whether or not this process involves a breakage and linear rearrangement of the chromonema similar to that occurring in obvious gene rearrangements but much more minute. This conception, which might require a revision of older notions of distinctly delimited genes, seems however to meet with some difficulties in explaining the comparative readiness with which apparently exact reversions may be produced. 3. Considerations of mutation frequency show that a mere removal of selection for given genes would be attended by reversal of evolution, in the sense of loss of organs and of traits (e.g. pigmentation) dependent on organized reaction systems, in, geologically, a comparatively short time. In practice, there are difficulties in the way of such a stoppage of selection for given genes (or even for given alleles of them) inasmuch as there is a tendency for genes (and gene differences) to become increasingly pleiotropic in the course of evolution, through mutational transfer of functions. Stopping selection in respect to the major function of a gene, then, can only slowly, with a speed dependent on the recency with which the gene has acquired this function, lead to a genetic reversal, involving loss of this function. For such a process is now contingent upon the establishment of mutations in other genes, that accidentally happen to have the effect of taking over the secondary functions of the gene in question. Eventually, however, loss of any function must follow stoppage of selection for it, since an ever greater number of mutations must become established (both through selection for other functions, and through “drift”) that happen to disturb the organized reaction system whereby the given function is carried out. 4. There can be apparent reversal of evolution with respect to given characters brought about by selection of mutations as well as by the genetic disintegration attendant upon mere removal of selection. But in neither case will the final product be genically identical or even very similar to the archetype. For the mutations of many different genes have equivalent end‐effects, especially in the case of the “small mutations”, which are more numerous and less harmful, and hence more apt to furnish evolutionary material than the large ones. For this reason the determination of the exact mutational path of evolution involves a large element of accident and, considered from a genic point of view, this path can never really be retraced, nor paralleled, in a second evolutionary sequence, nor can the same complex genic system be twice arrived at. The probability of the phenotypic similarity being thorough‐going will depend, among other things, on the length and complexity of the path to be retraced (or paralleled), and on the extent to which the reverse (or parallel) selection applies to all features at once (as a departure in one respect will tend to influence the conditions for other features). 5. In the case of a longer, more complex path, there is an increasing role played by the complicating circumstance (mentioned in (3)) that some of the evolutionary steps have later acquired accessory functions that can no longer be dispensed with readily. At the same time their own genetic basis has spread so as to depend on an increasing number of genes, by a kind of genetic diffusion. These circumstances will often prevent even the appearance of retracement, so that an equivalent end‐result (e.g. adoption of fish‐like form by mammals) will obviously embody a quite different developmental mechanism or have a demonstrably different morphological or physiological basis. There will thus be a tendency for the old gene reactions of development and of physiology to persist in the basis of the life complex and only to be overlaid, as it were, by the newer acquirements, which would tend to develop later in ontogeny (recapitulation), and this principle would apply no matter whether these newer acquirements represented progressively different stages, or more or less phenotypic reversal to an earlier stage (a superficial reversion). Nevertheless, especially in the case of shorter paths (and more closely related organisms) there should often be the possibility both of parallel and reverse evolution involving a more nearly real retracement (forwards or backwards) of steps which, though genically somewhat different, embody essentially the same reaction changes, as judged from the point of view of ordinary embryology, physiology and morphology. For here the steps have not yet become so indispensable; moreover, the thousands of primary gene reactions are necessarily canalized into certain definite channels, that limit the possible effects of their change, as viewed by these methods, which still deal with characters standing relatively far from the gene itself. To some extent, then, reversal, as well as parallelism in evolution, maybe “real”, and to deny the homologies of the resultant forms is to make an arbitrary metaphysical distinction, created to suit the point to be proved. 6. The complex systems of chemical reactions upon which fertility and viability depend become changed by numerous mutations, differing in different populations, that become established in, geologically, a very short time. Some of these mutations, though first indifferent or only an asset, finally become necessary, through the later establishment of other mutations, which without them would be detrimental to fertility or viability. Thereafter, crossing between one of the populations in question and one like the original (or one likewise evolved from the latter) will result in hybrids that are sterile or inviable, owing to the action of these harmful mutant genes, inadequately balanced by the ones that had made them tolerable. Two groups of organisms which are not ordinarily allowed to cross with one another will thus automatically become increasingly immiscible, and their genic, chemical paths of evolution will diverge more and more. This will occur even in cases where their evolution is, from the phenotypic standpoint, strikingly parallel (owing to similar selective conditions and similar developmental and physiological bases for change), or where one of the groups undergoes a striking appearance of reversion towards the other, and even though, in the case of more closely related groups, the parallelism or reversion may involve physiological and ontogenetic processes lying on a relatively deep plane of analysis. There must also be a hidden shift in the chemical, genic basis of a population which, phenotypically, remains relatively constant. But although these deeper‐lying genic changes may for a long time remain cryptic, they will eventually find more and more expression in “unnecessary” features of the life processes, discoverable by the chemist, the physiologist, the embryologist, or the morphologist, and an ever more different basis will be laid conditioning the future evolutionary possibilities.