71 Genetic polymorphism is a phenomenon of universal application to bisexually reproducing organisms. Consequently its attributes are predictable in species to which the techniques of experimental breeding have not been applied, including Man in which they are regarded as inappropriate. We can therefore anticipate that the polymorphic phases of human populations will often be associated with super-gene control and be maintained by balanced selective forces involving heterozygous advantage, while the latter concept leads to important conclusions both in anthropology and medicine. It will be useful at the outset to illustrate these situations by means of an example that has been rather fully analysed.
The foetal haemoglobin of Man is wholly replaced by the ordinary juvenile and adult type A within a year of birth. In certain districts, however, another form S partly or completely takes its place in a proportion of the community. These two conditions are controlled by allelic genes HbA and HbS. In the HbSHbS homozygotes A is absent, the haemoglobin being entirely of type S except for a small amount of the foetal form, normally lacking in adults. Such individuals have an extremely poor expectation of life: 4 out of 5 die of a blood disease before the age of reproduction while few of those who survive it do so for long. For, in the absence of the normal haemoglobin, the erythrocytes become 'sickle-shaped'; that is to say, long, curved and irregular. These tend to block the finer capillaries and in consequence are phagocytosed, causing severe anaemia and other dangerous symptoms. Clearly therefore the HbS gene is reduced to mutation level except where subject to powerful heterozygous advantage. 72 This indeed it possesses in certain regions, for the HbAHbS individuals are quite healthy since about half their haemoglobin is of type A and this is sufficient to keep the erythrocytes normal in the blood stream. They, however, assume the sickle-form in a drop of blood if the oxygen tension be reduced. In consequence, the heterozygotes can be distinguished from the normal homozygotes, HbAHbA, and their frequency determined. Yet the fact that half their haemoglobin is of the form S renders them relatively immune to Plasmodium falciparum, which gives rise to malignant tertian malaria (Allison, 1954, 1956).
That fact has been tested experimentally. Allison artificially inoculated 30 adult East African natives, 15 normal, 15 sickle-celled, with P. falciparum. Fourteen of the normal group contracted the disease while only two of those with sickle cells did so. It should be added, moreover, that even when infection of the HbA HbS heterozygotes does occur, the illness takes a mild form.
Thus in malarial regions, but there alone, the heterozygotes are at an advantage over both homozygotes. They comprise 17 per cent of the population in parts of Greece and up to 40 per cent in some East African tribes, in which Allison (1954) finds their fitness is 1.26 times that of normal people. Here, in consequence, HbS persists at a high frequency even though almost completely eliminated in double dose. As expected, the gene is becoming rarer among the Negro population of the U.S.A. where malaria is now largely eliminated.
It is of much interest to notice that in West Africa the situation is complicated by the existence as a polymorphism of yet another haemoglobin type, C, controlled by a gene HbC which behaves as an additional allele of HbA and HbS. Both the HbCHbC and the HbSHbC genotypes suffer from anaemia because they have no haemoglobin A. The HbAHbC heterozygotes are, however, at an advantage in their own African habitat since they also obtain partial protection from a blood parasite, not apparently Plasmodium falciparum.
Here we find ourselves confronted with a situation similar to that already frequently described in other species; the control of polymorphic phases by multiple alleles apparent or real. 73 Having regard to their similarity of action, it seems likely that these represent changes in units located within, or originating from, the same cistron. It is true that genes protecting against malaria in a similar way occur elsewhere on the chromosome. These must not be taken as indicating the type of situation from which a super-gene at the Hb locus has originated. It seems that various conditions, physiologically entirely distinct, prove inimical to the development of this Plasmodium. Thus the gene responsible for a deficiency in glucose-6-phosphate dehydrogenase also protects against malignant tertian malaria; but it is sex-linked, and therefore carried on a different chromosome from Hb, which is autosomal.
There is some ground for thinking that, in general, sex-linked polymorphisms are relatively rare. This would, however, be a difficult proposition to substantiate. It certainly does not apply to Man, with a diploid number of 23 in which the region of X non-homologous with Y is less than the average autosomal length. For in addition to the instance just mentioned, (and of course, sex itself), two other human polymorphisms controlled by X-borne genes have already been discovered: one of the blood groups and red-green colour-blindness. Certain rather dubious conclusions have been reached in regard to the latter condition, and these deserve brief consideration.
Since colour-blindness of this type is approximately recessive, and the gene responsible for it is carried in the differential segment of X, affected men and women should be distributed as q : q2. The condition actually occurs in about 8 per cent of men Southern England (but in women seems to be slightly commoner than the expected 0.64 per cent, perhaps because occasional heterozygotes are colour-blind). At this frequency it is of course polymorphic.
There appear in fact to be separate alleles deficiency in red perception and green perception, respectively. This distinction is generally omitted in studies of the subject, in which the two are summed so as to produce an overall estimate of red-green colour-blindness. Since the data do not in general treat the two types separately, while they cannot be well defined even by the most careful testing, it is best to combine them. 74 This is a valid procedure, as they represent subdivisions of a single genetic and sensory condition; consequently the distinction between failures in red and green perception will not be employed here.
Post (1962) reviews the physical anthropology of red-green colour-blindness. He finds that it is rarest in the Aborigines of Australia (1.9 per cent of males), Brazil, Fiji and North America, and commonest in Western Europe and among certain peoples of the Far East.
All the populations in which its frequency is lowest have, or until recently had, primitive cultures. He accordingly suggests that colour-blindness is a greater handicap in communities which live by hunting and food gathering than in those at a higher cultural level who lead a settled life. This no doubt is partially true. Yet, as Post himself mentions, red-green colour-blinds can detect certain differences that are indistinguishable to normal people: a fact used in the Ishihara tests. Accordingly, as I have pointed out in the past (Ford, 1955b, pp. 195-6), it is instructive to accompany a trained but colour-blind naturalist in the field and to observe the preternatural powers of detecting cryptic coloration which he will now and then display.
Post carried his deductions farther. He suggests that the frequency of the gene controlling colour-blindness has increased in civilized communities because the selection against it has been relaxed from about the beginning of the Neolithic Period, 120 generations or so. This he attributes entirely to mutation, which he seems to regard as the sole force balancing the handicap of colour-blindness in primitive peoples. He consequently uses the change in frequency of this trait as a basis for calculating mutation-rate. Here he is in error. Red-green colour-blindness could not have reached the status of a polymorphism unless the gene responsible for the condition had throughout been associated with advantages as well as disadvantages (pp. 12-14). Post does indeed qualify his views when he remarks 'If it (i.e. the gene for red-green colour-blindness) 75 is supposed to have an adaptive advantage in the new habitat, the resulting evolutionary changes would probably be attributed to progressive adaptation not to relaxation (i.e. of selection)'. This misses the point that the gene must have had balanced advantages and disadvantages even in primitive communities. That is to say, the red-green colour-blind situation in Man is comparable with that of the colour or banding phases in the snail Cepaea nemoralis (excluding districts where 'area effects' are predominant) in the sense that the polymorphism is maintained by physiological agencies but its frequencies are adjusted, or partly adjusted, by the severity of the visual handicap also involved.
Clarke (1964, p. 76), commenting on Post's analysis, remarks that in civilized communities red-green colour-blindness now seems to be a neutral trait. This statement needs to be accepted at the precise value of his wording. For the gene responsible for it can be neutral, or is becoming so, in one only of the two possible aspects of the concept. That is to say, it might, on the one hand, be of neutral survival value in the sense that it is not impinged upon by selection when, except in very small populations (up to a few hundred), its rate of frequency-change would be negligible, while its effects could not be adapted. On the other hand, it might be balanced at the point of neutrality by, perhaps powerful, advantages and disadvantages, as in all stable polymorphisms. In this type of neutrality, however, its frequency can be rapidly adjusted to changing conditions, while its effects can be evolved. The latter situation alone accords with the facts. If the disadvantages of red-green colour-blindness are indeed relaxed in Europe, the gene controlling the trait can only become commoner if it still preserves advantageous qualities, though their nature is at present unknown. Before considering likely means for detecting some of the selective agencies involved in this and corresponding situations, it will be useful to give brief consideration to other human polymorphisms in which the action of selection, though less obvious than in the sickle-cell condition, has nevertheless been disclosed. For the methods used are likely to be applicable to polymorphic phases in general.
Setting aside the 'private' antigens, thirteen blood group systems, including Auberger and Ii, are recognized at the present time. Six of them are already known to be controlled by multiple alleles, apparent or real; a situation likely to be detected in the others, several of which are but recently discovered and are imperfectly known.
It seems hardly profitable to speculate upon the origin of these clusters of alleles since at the present time the evidence for forming an opinion on that subject is, in general, lacking: for, as the characters involved are not recessives, the concepts of complementarity and of their respective reactions when controlled by genes in the cis and trans positions, are difficult to apply. However, Race and Sanger (1962) discuss the evidence which indicates that the anti-C (B) and anti-E (B) of the Rhesus series react with red cells of the CACB, EAEB double heterozygotes when in the CAEA/CBEB, but not in the CAEB/CBEA relationship. It looks therefore as if CBEB can produce a compound antigen when these genes are on the same chromosome but not when they are on homologous ones. This certainly suggests that they lie within the same cistron. Whether or not D is included with them remains uncertain. At any rate, all the evidence shows that it does not lie between C and E.
In general, the similarity of action which distinguishes the genes controlling phases of the same system, would suggest true allelism (intra-cistronic changes). 77 It may be that super-genes have evolved in some instances, but it seems likely that in view of the fact that they control different forms of the same antigen, such major genes, if they exist, have originated from one cistron. Yet some caution must be exercised here. The occurrence of unlinked genes affecting the same blood group system has without doubt been established. Thus it is now clear, as Race has kindly informed me, that the gene controlling the antigen recognized by anti-Rhesus sera as stimulated in the Rhesus monkey is at a separate locus unlinked with the DCE Rhesus genes with which, however, it interacts. Similarly the P series is controlled from three loci two of which are certainly not linked. Such situations could at least in theory indicate the type of separate loci from which super-genes have been compounded.
A new light was thrown on the significance of the blood groups when it was recognized that they constitute genetic polymorphisms, with all that this implies (Ford, 1942). Up to that time a rather curious assumption was current. It was held that the different serological types were of neutral survival value (e.g. Wright, 1940; Boyd, 1940), although it had long been known that they could be responsible for fatal reactions to transfusion. Already indeed by March 1940 the conclusion had been published that they must be maintained by powerful selective forces (1940a), though this was before the discovery of Rhesus incompatibility by Levene and Katzin in the autumn of that year.
We may set aside certain rare types such as A3 and Ax, the latter apparently heterogeneous, which do not seem to be polymorphisms at all; on the contrary, they are probably eliminated by selection and maintained merely by mutation, like any of the disadvantageous heterozygotes which may give rise to abnormalities and disease. Apart from such as these, the alternative phases of the various blood groups being unifactorial and maintained at high frequencies, must be polymorphic. They could not have reached such proportions unless favoured by some advantage; they could not have stabilize at them unless balanced by opposed disadvantages.
The genes and super-genes controlling the human blood groups have no effect upon anatomy or upon choice in marriage. 78 Their selective importance must therefore influence fertility, general viability, pre-natal survival or susceptibility to disease. Considerations along these lines lead to the prediction that the blood groups and other human polymorphisms must be associated with liability to, and protection against, specific diseases (Ford, 1945). The first instance of the kind was reported six years later, when Struthers (1951) found that a larger proportion of group A than group O babies died of broncho-pneumonia during the first two years of life. This result, however, has not been confirmed; probably due to the greater use of antibiotics in recent years.
On the other hand, it was not long before observations of this kind were fully established. The first was provided by Aird, Bentall and Fraser-Roberts (1953), who demonstrated a significant excess of blood group A among those who develop cancer of the stomach. Other examples have followed and accounts of these associations, which are opening up a new aspect of medicine, are available in the literature (e.g. Clarke, 1964; Ford, 1964). One general point in regard to them may, however, be taken up here. Some, though by no means all, of the diseases concerned occur principally from middle age onwards probably do so because liability to them has been pushed back to an age beyond which selection cannot influence it: beyond the effective period of reproduction, that is to say; and to that group the majority of cancers almost certainly belong (Ford, 1949). However, the association of polymorphisms with disease of any kind provides an example of the fact that the controlling genes have important effects in addition to those by which they are normally recognized: thus the GA gene not only, as already mentioned, controls a blood group and raises the susceptibility to one type of cancer, but it also increases liability to pernicious anaemia, which is not restricted to later life.
79 An association between blood grouping and general viability is an obvious deduction from the theory of polymorphism. I am not aware that any data have yet been amassed upon the serological types of the aged; those of 90 years and upwards, for example. It is more than likely that such people, the men in particular, will prove not to represent a random frequency distribution of the blood groups of the community to which they belong. So too at the other end of life, during the pre-natal period; and here information demonstrating selective elimination of the blood group phases has already been obtained (Chung and Morton, 1961).
As already mentioned, a knowledge of the blood groups and their effects is here assumed. However, the results of Rhesus incompatibility raise quite exceptional difficulties in regard to the control of polymorphism. As so fully discussed in this Monograph, polymorphic phases are most usually maintained by heterozygous advantage, and yet in the Rhesus groups we meet the extraordinary situation that the heterozygote is differentially and heavily eliminated.
Any Rhesus-positive foetus born of a Rhesus-negative mother must be heterozygous, and in danger of suffering from erythroblastosis which, apart from the modern technique of transfusion, has generally been fatal. The condition does not arise invariably but it constitutes a heavy heterozygous disadvantage, together with a differential destruction of whichever of the controlling super-genes is the rarer; in this instance that for the Rhesus-negative condition. The question arises then, why has not this super-gene been eliminated?
Various suggestions have been put forward to account for this. It has been maintained that those who lose a child will over-compensate in child-bearing and have more children than those who have lost none. There is, moreover, an interesting interaction which, though to a slight extent, tends to preserve Rhesus.
Polymorphism may of course be maintained by any system which opposes uniformity, such as the disassortative mating of Panaxia dominula (pp. 42-3). 80 A condition which does so in Man, in addition to sex which is also of this kind, is provided by a special type of reaction between the blood groups. For the incompatible children of those who are incompatibly mated with respect to the OAB system are better protected than compatibles against haemolytic disease of the new born when due to Rhesus, to Kell and probably to other groups also.
It seems clear, however, that neither this situation nor over-compensation, if it occurs, can possibly explain the high frequency of the Rhesus super-gene in Man. By far the most likely interpretation of this is to be found in some, at present unknown, advantage of the heterozygote. Indeed there are now rumours, as Clarke (1964, p. 82) remarks, that as with sickle-cell this may consist in relative protection against malaria.
'By a curious inversion of logical thought, it was held that their occurrence in distinct and characteristic proportions in the different races of mankind was especially important because the variation involved was selectively neutral. Precisely the contrary is true. The fact that the genes concerned are balanced by selection at optimum frequencies, which differ from race to race, is the one which gives them significance as a criterion of relationship. It does so because in these circumstances their proportions are influenced by the average genotype of the population in which they occur.'
I wish here to draw attention to the clear-sighted attitude of one of the most distinguished physical anthropologists who, when the impact of polymorphism theory upon serology became apparent, wrote 81 'Until recently it was the fashion ... to state that characteristics suitable for the classification of Man into races should be non-adaptive, meaning not influenced by selection, and the present author ... was maintaining in 1940 (Boyd, 1940) that the blood groups were suitable for racial classification partly because they were non-adaptive. That point of view has been completely abandoned' (Boyd, 1955).
We may generalize by saying that polymorphic genes stabilized at given frequencies in one population may be maintained with equal precision at quite different proportions in another. That situation is determined partly by an adjustment to the balanced gene-complex of each race and partly by the necessity to interact favourably with the environment. I have in the past illustrated these points with reference to group B of the OAB series. This becomes progressively commoner as we pass south-eastwards from Western Europe (about 6 to 8 per cent according to race) to Northern India (about 37 per cent), where it reaches its highest value. Such a cline must in part be an adjustment to certain aspects of the environment since the high values of B in Southern Asia are found in both Caucasoid and Mongoloid stocks. On the other hand, it is clear that these graded frequencies are partly adapted to the genetic structure of the peoples concerned; for a Hindu race, the Gypsies, has preserved its North Indian blood groups practically unaltered after living for hundreds of years in Central and Western Europe (Mourant et al., 1958).
Indeed all migrant populations tend to retain their original blood groups, which may differ markedly from those of the people among whom they live. For example, Hart (1944) showed that in rural Ulster, blood donors with English names have the English blood group frequencies, though it is more than 400 years since their ancestors settled there. The rest of the Ulster population has the distinctive Scottish and Icelandic serology, suggesting Viking influence (Fisher and Taylor, 1940).
What has been said of the selective control of the blood groups applies to all other aspects of human polymorphism. In some of them, evidence which indicates the impact of selection has already been obtained. Of this, two illustrative instances may be given.
First, the ability to taste phenyl-thio-urea, 82 extremely bitter to those who can detect it, is possessed by about 70 per cent of the population in Europe and the Middle East. The condition is a dominant and its frequency is very different in some other races affecting, for example, over 95 per cent of African negroes. The character would in itself seem to be trivial in the highest degree: indeed no one had the opportunity of tasting the substance until the present century. That the gene itself is of importance is, however, indicated by the stable proportions in which it is maintained in distinct populations. It appeared likely that evidence could be obtained to show that this polymorphism antedates the separation of the human stock from the anthropoid apes. That possibility was successfully tested by Fisher, Ford and Huxley (1939) who found that tasters and non-tasters of phenyl-thio-urea exist in the Orang-utan and Chimpanzee. The animals could be classed for this character without risk of confusion. We examined 27 Chimpanzees and 7 of them (26 per cent) proved to be non-tasters, a result which strongly suggests a balanced polymorphism in this species also. As I have pointed out (Ford, 1961), it seems hardly conceivable that such a sensory distinction should have arisen independently in the anthropoid and hominoid stocks. On the contrary, the facts strongly suggest that heterozygous advantage of the gene concerned had already been evolved in the common ancestors of Man and the Chimpanzee, and preserved in the evolving lineage of both up to the present day. The nature of that advantage is unknown, but its existence is further indicated by the fact that it seems to affect the type of thyroid disease to which an individual is liable. That is to say, toxic diffuse goitre (Graves's Disease) appears to be much commoner in 'tasters' who, however, are less subject than 'non-tasters' to goitre of the simple adenomatous form; especially in men, in whom the association is closer than in women. These findings have not been confirmed by the work of G. R. Fraser who, however, reports a heavy excess of non-tasters in children suffering from athyreotic cretinism. (For a review of the association between the taste-test dimorphism and disorders of the thyroid, see Clarke, 1964, pp. 217-19).
Secondly, the ability to secrete the antigenes of the OAB blood groups into the saliva is itself a polymorphism. 83 The non-secreting type is recessive and affects about 22 per cent of Western Europeans. Its frequency is different in other races: for example, non-secretors are nearly absent among North American Indians. That the genes are concerned are of selective importance is indicated by the fact that non-secretors are about 40 per cent more liable to develop duodenal ulcers than are secretors (Clarke et al., 1959).
The secretor gene and that for the Lutheran blood groups are fairly close together on the same chromosome, the recombination fraction is about 9 per cent in England.* This association may be fortuitous. On the other hand, there may be an advantage in keeping these two genes together in certain circumstances. Further linkage studies should now be carried out in a few very different races, Negroes, Hindus or Japanese for example, so as to provide a marked contrast with the information already available on this subject in Western Europe. A significant racial difference in crossing-over between the same two genes might thus be established. This would be a matter of great interest demonstrating, as it would do, the evolution of a super-gene in Man.
* Race and Sanger (1958, 3rd edition) Blood Groups in Man, Addenda; Blackwells, Oxford.