Endeavour 134-139 (July 1953)
The evolution of adaptations
C. H. WADDINGTON
Current biological belief regards evolution as being primarily the result of the natural selection of random mutations, useful adaptations gradually spreading throughout a race. Professor Waddington regards this as an extreme view, and in this article puts forward a hypothesis to explain how acquired characteristics may become hereditarily fixed by a process of genetic assimilation not invoking the generally discredited theory of direct inheritance.
It is abundantly found in the living world that the structure of an animal or plant is very precisely adapted to the functions which it has to perform. The nature of the processes by which this situation has been brought about during evolution provides one of the major problems for biological theory. The hypothesis of the inheritance of acquired characters suggested that in some way or other the effects of functioning become themselves inherited. It has usually been interpreted to mean that the reaction between the organism and its surroundings has, as one of its results, an effect on the germ-plasm such that new hereditary changes occur, of a kind which determines the development in later generations of individuals suited to these particular conditions of life. Although this idea has recently been revived in a rather nebulous form in the Soviet Union, it has been so completely rejected by the rest of the scientific world that it is hardly considered to be worthy of discussion in most of the important recent works on evolution. The reigning modern view is that, in nature, the direction of mutational change is entirely at random, and that adaptation results solely from the natural selection of mutations which happen to give rise to individuals with suitable characteristics. I want to argue that this theory is an extremist one, and that, in essaying to account for adaptation, it neglects to call to its aid the doctrines emerging in other fields of modern biology which can quite properly be combined with the conclusions of genetics in the strict sense. In the discussion which follows, attention will be confined to animals, but there is no reason to doubt that similar arguments could be advanced in the botanical field.
It will be advisable first to glance briefly at the phenomena which are usually referred to under the heading of adaptation, since they are of several different kinds which must be distinguished from one another.
There is, first, a category in which an animal living under particular circumstances, or behaving in a particular way, itself becomes modified so as to be better fitted for its special circumstances. Examples of such 'exogenous' adaptations are legion. If muscles are continually and intensely used, they become thicker and stronger; if one kidney is removed from a mammal, the other hypertrophies; if the forelegs are absent at birth, or removed shortly afterwards, from rats or dogs, the hind-limbs become modified to suit the hipedal gait which the animals are forced to adopt; if skin is subjected to frequent rubbing and pressing, it thickens and becomes more horny; and one could multiply such instances almost indefinitely.
Secondly, there is a category of what may be called pseudo-exogenous adaptations, in which the animal exhibits characteristics similar to effects which can be called forth as direct exogenous adaptations, but which on investigation are shown to be hereditary, and independent of any particular environmental influence. We shall consider some examples of such adaptation in more detail later, since they pose one of the most striking problems to be solved.
Finally, there is a very large class of adaptations, which, as Medawar  has recently emphasized, need to be distinguished from the previous category, and which are characterized by the fact that the adaptive feature is of a kind which one cannot imagine as having ever been produced in direct response to the environmental conditions or mode of life of the animal. To give two examples only, Medawar mentions the modifications of certain epidermal cells to secrete sweat, and the development, from another part of the skin, of a transparent area which forms the cornea of the eye. It is, as he says, impossible to see how any attempt to peer through an area of opaque skin could tend to cause it to become transparent. The adaptation of the cornea to vision can hardly have arisen in any causal dependence on external factors, and we might therefore give the name of endogenous adaptations to this category.
It is in connection with this third type of adaptation that we can as yet make the least progress beyond the current hypothesis, which is content to rely on the chance occurrence of suitable mutations. But even here there is a little more to be said. In some endogenous adaptations, the usefulness of the character is in connection with factors in the outside world. Another of Medawar's examples, the possession of horns which serve the purposes of aggression or defence, will suffice as an instance. But the transparency of the cornea is adaptive because it is suited to the functioning of another internal part of the organism, namely the retina, sensitive to the light which the cornea allows to enter. Now, I think that we shall often find that the various parts concerned in such internal endogenous adaptations are involved with one another not only during their functioning in the adult animal, but during their development in the embryo. This is certainly true of the cornea, which can be induced from normal epidermis if an eye-cup is transplanted under it at an early enough stage. All the various parts of the eye, which can function efficiently only if they have the correct relations with one another, are interdependent during their development. It was shown many years ago by Ross Harrison  that if the large lens of the axolotl Amblystoma tigrinum is grafted over the eye-cup of the smaller A. punctatum, the lens does not grow to its full size, while the eye-cup provided with the larger lens attains a greater size than usual (figure 1). Similarly, if the eye-cup of A. tigrinum is transplanted under the early embryonic skin of A. punctatum, it induces a lens which is originally of A. punctatum size and therefore relatively too small, but, as growth proceeds, the lens grows faster and the eye-cup more slowly than usual, so that they gradually achieve the normal relative proportions. We might say that the internal adaptation of the lens to its retina, although endogenous in the sense that it arises within the animal, is nevertheless affected by factors from outside the lens itself; namely by the retina. The problem it presents is therefore not wholly different from that offered by adaptations to external factors. Moreover, the adaptation is in part a direct response to the influence of the retina, and is thus similar to the first category distinguished above; but there is clearly also some inherent tendency for the A. tigrinum material to grow faster than the A. punctatum, and in this respect we are reminded of the second category, of pseudo-exogenous adaptations.
We may next consider the true exogenous adaptations, in which an animal becomes modified by external factors in such a way as to increase its efficiency in dealing with them. Such adaptation does not, of course, always occur: the environment may overcome the animal and eventually kill it. For instance, specialized breeds of cattle from a temperate zone growing up in the tropics may fail to become acclimatized, and, instead, develop into degenerate forms which have great difficulty in surviving. Even in such cases, however, there are probably always some tendencies towards the development of useful adaptations, though they are not strong enough to be effective. Adaptation to relatively slight changes in environment is, however, usually successful. We still know comparatively little as to how this is brought about, although we can speculate more or less plausibly about the mechanisms involved in particular cases. The hypertrophy of one organ of a pair when the other is removed (as in the case of kidneys) may, for instance, find its explanation in Haddow's suggestion  that each organ requires some special specific substance for its growth, or in Weiss's interesting work  on the effects on the growth of an organ of antisera specific against it (see also Ebert ).
FIGURE I — Transplantation of lens ectoderm from Amblystoma tigrinum to A. punctatum. Curve A: growth of the tigrinum lens when associated with the eye-cup of the same species, expressed as the ratio of the size of tigrinum lens to the size of punctatum lens. Curve B: similar curve for tigrinum lens associated with punctatum eye-cup. Curve C: growth of punctatum eye-cup associated with tigrinum lens, expressed in terms of its size when associated with punctatum lens. Note how in the interspecific combination the lens grows more slowly and the eye-cup faster than they would normally do (after Harrison).
It may be that, in many cases, the chemical processes involved in the functioning of an enzyme or similar substance are related to those by which it is synthesized, so that increased function automatically means increased synthesis. Other examples are certainly more complex. For instance, if newt larvae are kept in water poor in oxygen, not only do the gills grow larger, but the tissue in their walls becomes thinner, allowing a more rapid diffusion of gas . Possibly here we might be able to find some explanation in terms of the effect of oxygen-lack on the rate of blood-flow, and the influence of that in its turn on the vessels through which it is moving. But the phenomenon of adaptive response is so widespread and so general that it seems hardly convincing to explain it only by a series of ad hoc hypotheses, invoking a new one for each different case. At present, we hardly seem able to do better than go to the other extreme, and produce the general argument—too general to be very satisfying—that it is an advantage to animals to be able to adapt successfully to new circumstances, and therefore natural selection will have favoured those animals which by chance had a hereditary endowment which enabled them to do so. One feels that, somewhere between these extremes of particularity and inclusiveness, there should be principles of fairly general application to be discovered—but they still await their Darwin.
It is the remaining category of adaptations, the pseudo-exogenous, which has provoked the most discussion. We are confronted here by phenomena for which an explanation could so easily be found in a direct effect of some environmental factor, were it not that further study demonstrates unequivocally that the structure concerned is determined by the heredity of the organism, and is relatively independent of the environment. The question arises whether we can bring ourselves to believe that the part which the environment can play in mimicking the condition is really irrelevant, and that the evolution of this particular adaptation has resulted from the selection of chance mutations which might have appeared and produced the phenotypes even if the environmental effects had never existed.
Some concrete examples will make the problem clearer. One of the most familiar is that of the thickened skin on the soles of our feet. This thickening is obviously an adaptation to the stresses which this region of the body has to bear; but, as Darwin pointed out, and as Semon  discussed in a full-length paper, the thickening already appears in the embryo, before the foot has ever borne any weight. The structure therefore cannot be a direct response to external pressure, but must be produced by the hereditary constitution independently of the specific external influence to which it is an adaptation. The situation is even more striking when similar thickenings are found on less conventional parts of the body. For instance, the ostrich squats down in such a way that the under surface of the body comes into contact with the ground at its two ends, fore and aft. In just these places a considerable callosity develops in the skin (figure 2), and Duerden  showed that these thickenings make their appearance in the embryo before hatching. The same thing is true of callosities which appear on the wrists of the forelegs of the African wart-hog, which while feeding has a peculiar stance which involves resting on these points (Leche ). Another remarkably clear example, affecting a different organ, is that of the second molar tooth in the dugong. In the adult, the crown of this tooth is more or less flat, with slight transverse ridges. This shape must be regarded as a modification of an originally more conical tooth, and would at first sight appear to be directly related to the use of the tooth for grinding the food. However, Kükenthal  showed that, although the tooth first appears in a conical form in the early embryo, processes of resorption begin to change this into the final flattened form before the tooth is used (figure 3).
It certainly seems very far-fetched to attempt to explain such phenomena without bringing in the fact that the environment might be expected to produce similar effects. Let us consider, therefore, what might happen to an ostrich in which the appropriate callosities were not hereditarily determined. Presumably its skin, like that of most other animals, would react directly to external pressure and rubbing by becoming thicker. Now the point which seems to have been overlooked in previous discussions of the matter is that this capacity to react must itself be dependent on genes. Since populations of animals are never quite uniform in any character, we must expect that the ostrich ancestors varied in their capacity to produce the most suitable callosities; and there could he effective natural selection for those which performed the most satisfactory exogenous adaptation. A race would evolve in which the stresses set up by squatting in a particular way would call forth the development of appropriate adaptive thickenings of the skin.
FIGURE 2 — The underside of an ostrich's body, showing the two callosities (redrawn from a photograph published by Duerden).
FIGURE 3 — Two stages in the development of the second upper molar of the dugong. In the young embryonic stage on the left, the tooth has three conical protuberances, in which resorption is just beginning. In the older but still embryonic stage on the right, more or less flat faces have appeared (after Kükenthal).
At this stage, the thickenings would still not be hereditary and independent of the pressure and rubbing; they would still be acquired characters in the conventional sense. We can find a hypothesis of how they might come to be hereditarily fixed if we turn to consider another aspect of the matter. The callosities are the results of developmental processes. Now, one of the main characteristics of animal development is that it tends to be canalized or buffered, so that the optimum endresult is produced even if there are minor variations from the normal conditions while the process is going on . Natural selection, in fact, does not merely ensure that only those animals survive which have something near the optimum characteristics, but favours those genotypes which tend to produce such animals under any conditions. It gradually builds up efficient cybernetic mechanisms, to use a fashionable phrase. Thus we may expect to reach a stage in which our ostriches nearly always develop callosities of just the right size and position, even in those individuals which, to put it crudely, sit down very seldom or those which loll about the whole time.
Once such a cybernetic developmental mechanism has been built up, it will be rather like a gun set to go off when the trigger is pulled. The development of the callosities will proceed quite autonomously, once the process can be started. The initial stimulus, which may be a greater or lesser amount of external pressure, has become a relatively minor factor in the whole situation. It may then not be too difficult for a gene mutation to occur which will modify some other nearby region of the embryo in such a way that it takes over the function of the external pressure, interacting with the skin so as to 'pull the trigger' and set off the development of the callosities .
This may have seemed too long a train of argument to be very convincing, but a good deal of experimental evidence can now be produced to support it. In the first place, there seems to be no doubt that the formation of callosities on the sole of the foot is actually brought about by developmental interactions which go on in the embryonic limb-bud, as was suggested in the last paragraph. In animals which are hereditarily polydactylous, the genetic constitution causes an increase in the number not only of skeletal elements, but of the associated structures such as muscles, and, in particular, callosities (figure 4). These thickenings are therefore an integral part of the presumably complex but as yet unanalysed actions and reactions which mould the early limb-bud into the adult foot.
FIGURE 4 — Right hind foot of a normal guinea-pig, above, and of a polydactylous guinea-pig, below (after S. Wright).
FIGURE 5 — Defects of the posterior cross-vein in the wings of Drosophila melanogaster which had been given a temperature shock in early pupal life.
A more comprehensive check on the whole train of thought may be obtained if we imitate the postulated action of natural selection by artificially selecting for the ability to respond in some definite way to an environmental stimulus. No such work has been done on the formation of callosities, but an experiment has been made on the rapidly breeding fruit-fly Drosophila melanogaster [14, 15]. In order to make the situation as clear-cut as possible, an environmental stimulus was used which produced a phenotype which did not normally occur at all in the stock used. If pupae aged about 21-23 hours are subjected to a temperature of 40° C for four hours, a proportion of them develop wings in which the posterior cross-vein (and occasionally the anterior one) is broken or missing (figure 5). Two selection-lines were set up, in one of which only flies with broken cross-veins were used as parents; in the other, the flies which failed to react were selected. In both lines, the percentages of reactive individuals changed as time went on, increasing in the upward selected line and decreasing in the other. It is therefore actually possible to select for the capacity to respond to the environment (figure 6).
As the experiment proceeded, it confirmed further points in the theory given above. Stabilization of the type with broken or missing cross-veins did in fact take place, and proceeded so far that, as was predicted, strains could be produced which lacked the cross-vein even when they had been reared for their whole life under the standard normal conditions, and had never been subjected to the high-temperature treatment of their pupae. The cross-vein effect, which in the original stock had appeared only as a response to an external stimulus, is in these selected races quite independent of any special feature of the environment. We may say that the acquired character has been 'assimilated' by the genotype. Presumably, the lack of the vein is due now to some modification of the reactions which go on autonomously within the developing wing, but we do not know enough about the physiology of wing development in Drosophila to guess exactly what the reaction might be.
The gradual increase in the frequency of cross-veinless flies shows quite definitely that selection has been at work. It may, however, still be asked whether all that happened was the concentration of genic variants which were present in the population to begin with, or whether new mutations tending to break the cross-vein have arisen during the course of the selection, and, if so, whether we can suppose that the treatment has itself caused them to appear. The experiments which have been made so far do not allow of a final answer. It may be pointed out that the selected cross-veinless lines differ from the foundation stock in quite a large number of genes; we are not dealing with only a single gene, and if what we have collected are new mutations, there must have been a large number of them. Moreover, a considerable number of flies, of the order of a thousand, were involved in each generation, and it seems unlikely that new mutations could have been sorted out from such numbers in the comparatively few generations of the experiment. One cannot, however, absolutely rule out the possibility that the treatment provoked appropriate mutations. The main point is that there is no need to make this hypothesis, which is contrary to everything that we know about the mutation process. Selection of already existing genes which affect reponsiveness to the environment, and stabilization of their effect, together provide a plausible account of the result in terms of orthodox genetic and embryological mechanisms.
FIGURE 6 — The results of selection for and against the ability to respond to a temperature shock by the formation of a cross-veinless phenotype.
It is probable that this process of genetic assimilation of an environmental effect has played a very important part in evolution. Pseudo-exogenous adaptations are very common in the animal, and also in the plant, world. Instances such as the ostrich's callosities and the dugong's tooth, which were discussed above, were chosen as particularly striking and impressive examples, not because they are typical; usually such adaptations have a much more ordinary character. One such case, which will also serve to illustrate some of the problems which remain to be cleared up, is that of the mountain forms of the common water snail Lymnaea peregra . In certain tarns and loughs in Scotland and Ireland having very soft water, this snail occurs with an unusually thin shell, a form to which the name L. praetenuis has been applied. The thinness of the shell is clearly an effect of the lack of calcium in the water. It is found that when these forms are bred in water of normal hardness, they revert more or less completely to the normal; the effect in this case therefore is entirely, or at least mainly, an exogenous one which has not been genetically assimilated. But there is another method of economizing in shell-forming materials, which involves coiling the shell more nearly in one plane, so that the spire is reduced. Some lakes contain races which have adopted this shape, usually combined with a certain thinning of the shell; they have been given the varietal names burnetti and involuta. When they are bred in hard water, many of the races revert to normal, but in some of them the involute character is persistent, and must have been genetically assimilated. A whole series of questions suggest themselves. Why is the praetenuis form apparently never genetically fixed, while the involuta form sometimes is? Is it because the direct action of the environment is more effective in producing a thin shell of normal shape than a shell of altered configuration? Or is the thinness of the shell in praetenuis not truly an adaptation which is advantageous to the animal, but merely a necessary consequence of the lack of calcium, while the altered shell form of involuta has a selective advantage? And why has involution been genetically assimilated in some races but not in others? Possibly the length of time for which the lake has been colonized, or the degree of involution attained, may be important in this connection, but we still know too little to say.
It is clear that the theory of genetic assimilation may have wide applications, but before it can be used with confidence it requires much more experimental support than it yet has. The first thorough experiment made to test it gave the expected result, but that is hardly sufficient. Attempts are now being made with Drosophila to select for the capacity to give other responses to various environmental stimuli. In some cases, the experience with the cross-veinless character seems to be being repeated, but in others there has been little response to selection over the first few generations, and it may turn out that the original populations have little available variation in capacity to respond in the necessary way, in which case genetic assimilation would have no chance to operate. Until such limitations on the process have been more fully worked out, some caution is called for in applying the theory to all the phenomena for which it seems able to provide an explanation. However, it is at least not unimportant that we now have a hypothesis which gives a more or less plausible explanation of this large class of adaptations without invoking the discredited theory of the direct inheritance of acquired characters, and in a more convincing way than by a mere reliance on the occurrence of a suitable chance mutation.