In the following, we discuss the evidence for, and implications which arise from, the main candidate human polymorphisms to fall within the terms of the malaria hypothesis.
Hemoglobin C
The gene for hemoglobin C is allelic with that for sickle cell hemoglobin but codes for a lysine instead of a glutamate (hemoglobin A) or a valine (hemoglobin S) at position 6 in the beta chain of hemoglobin. In the homozygous carrier, hemoglobin C confers a loss of fitness which is much less than that associated with the sickle cell trait (homozygous hemoglobin S), being apparently comparable to that of a mild form of thalassemia (
119,
120).
Hemoglobin C is found only within certain West African populations, where the frequency of the allele reaches, and sometimes even exceeds, 10 to 20% (
119,
120). There is now little doubt that this polymorphism has been, and indeed is probably still being, selected under the effects of
P. falciparum malaria in West Africa (
133). However, the nature of the protection by hemoglobin C is very different from that associated with hemoglobin S. In the case of hemoglobin S, the heterozygous state confers a high level of protection against death from
P. falciparum malaria. In that of hemoglobin C, similar high levels of protection, greater than 90% against infection and therefore against the risk of death by
P. falciparum malaria, are achieved only in the homozygote (
133). The protective effects of hemoglobin C in a heterozygous combination are, at about 30%, relatively weak.
Now, the rate of selection of an allele whose advantage is expressed mainly in the homogygous state, as is the case for hemoglobin C, is slow compared to that of an allele, such as that for hemoglobin S, which carries a similar advantage but in a heterozygous combination. This principle is discussed more fully in relation to selection for RBC Duffy negativity (see below). Under the influence of the same selective force, therefore, hemoglobin S would be expected at first to achieve higher frequencies than hemoglobin C. However, because the balancing cost of hemoglobin C is low and that of the allelic hemoglobin S is high, hemoglobin C should eventually replace hemoglobin S in populations exposed to selection by
P. falciparum. In West Africa this has clearly not yet happened (
4,
119,
120), and it suggests that
P. falciparum malaria may have arrived only “recently” within the West African population.
When, therefore, might this “recent” arrival of
P. falciparum in West Africa have taken place? The relative fitnesses of the different haemoglobin alleles, A, S, and C, in the West African situation can now be given (
76,
94,
133), and the relevant calculations can be made (
119). We suspect that they would show that the hemoglobin C allele should approach population equilibrium within several thousand years, and almost certainly within less than tens of thousands of years, of selection under
P. falciparum malaria, as, indeed, has already been suggested (
119,
120). Therefore, by this line of argument, because hemoglobin C has not yet reached its expected equilibrium frequency,
P. falciparum has been a selective force in West Africa for less than this time, i.e., probably for less than a few tens of thousands of years.
RBC Duffy Negativity
It is striking and long-recognized fact that most members of indigenous populations of West and Central Africa are completely refractory to infection with
P. vivax malaria (
71). Almost all members of these populations are also homozygous for an FY
null (RBC Duffy-negative) allele of the gene that controls expression of the Duffy antigen on RBCs (
27,
150,
217). The Duffy blood group antigen system is represented mainly by two serologically distinct forms determined by alleles FY*A and FY*B. Each of these alleles also exists in a mutant, unexpressed, or null form as FY*A
null or FY*B
null (
217). The Duffy antigen, which has been identified as a chemokine receptor (
98,
150), is also an essential receptor for
P. vivax merozoites to be able to enter a host RBC (
98,
126,
132). This accounts for the association between complete refractoriness to
P. vivax infection in most West and Central Africans and the almost universal homozygous RBC Duffy negativity in these populations.
The molecular genetic basis for RBC Duffy negativity is a single-nucleotide substitution polymorphism (SNP) in the promoter region of the gene for the Duffy antigen. This promoter controls the expression of the Duffy antigen specifically in RBCs. The same SNP is associated with both the FY*A and the FY*B alleles (
217), leading to the FY*A
null and FY*B
null RBC Duffy-negative alleles. Now, the promoter for a structural gene, such as that for the Duffy antigen, governs its expression on its own chromosome strand only. Therefore, in a heterozygous individual, expression of the Duffy antigen in RBCs should be suppressed only on the strand with an FY
null allele, while from the strand with a Duffy-positive allele, the Duffy antigen should be fully expressed.
In accordance with this expectation, Duffy antigen expression on RBCs from individuals from Papua New Guinea who were heterozygous for the FY*A
null allele was approximately half of that on RBCs from homozygous Duffy-positive individuals (
217). In the same human population in Papua New Guinea, a 50%, but statistically insignificant, reduction was noted in the prevalence of
P. vivax infections among heterozygous FY*A/FY*A
null individuals compared to those who were homozygous Duffy positive (
217). The characteristics of human RBCs with genetically reduced RBC Duffy antigen expression (Fy
xFy
x, according to serotype nomenclature [
126]) and weak reaction with Duffy antigen antisera was investigated using the simian malaria parasite
P. knowlesi, which, like
P. vivax, is absolutely dependent on the Duffy antigen for the ability to invade human RBCs. Merozoites of
P. knowlesi invaded RBCs from such individuals with about half of the efficiency that they invaded RBCs from individuals with full Duffy antigen expression (
126). However, RBCs from heterozygotes for RBC Duffy negativity were invaded as efficiently as were RBCs from normal RBC Duffy-positive individuals (J. Barnwell, personal communication). Susceptibility to infection with
P. vivax malaria of FY
null heterozygotes may be partially reduced, but the effect is probably not very strong, certainly by comparison with the total refractoriness to
P. vivax malaria of individuals who are homozygous for an FY
null allele.
Throughout West and Central Africa, the frequency of the FY*B
null allele is, as already noted, close to fixation. In most of these populations, its frequency exceeds 97% (
27,
135), leading to almost universal RBC Duffy negativity in these populations. Not surprisingly,
P. vivax malaria is very rare throughout the region (
71,
131) (Table
1). It might seem natural to conclude that
P. vivax malaria must have been the selective agent for the near fixation of the FY*B
null allele in these African populations, to the point that
P. vivax was itself virtually eliminated. While this is our view, it is not universally held (
217). Nor is the case, as we will now present it, straightforward.
In contrast to the frequently and directly lethal
P. falciparum, it is often assumed that
P. vivax exerts little or no selective pressure on a human population. On these grounds, therefore, and notwithstanding that the RBC Duffy-negative condition confers no evident disadvantage on a carrier (
87,
121),
P. vivax might be discounted as a selective force for the near fixation of RBC Duffy negativity in West and Central Africa. If, on the other hand,
P. vivax can, and does, exert a significant selective force, we are confronted with a paradox. Why, in this case, has selection for RBC Duffy negativity not taken place in other parts of the world, in southern Asia and the Western Pacific rim (
27,
135), where
P. vivax has certainly been present for several thousand years? It has recently been suggested, in a report of the presence of an FY*A
null allele in a single population Papua New Guinea, that such selection may indeed be happening (
217). However, no previous studies appear to have detected this allele in Papua New Guinea, where FY*A is otherwise at virtual fixation (
27). Moreover, at around 2%, the FY*A
null allele in this single population in Papua New Guinea is still far below the frequencies of the FY*B
null allele, which, throughout tropical Africa, are rarely less than 50% and more often in excess of 95% (
27).
Concerning the first of the points above, there is strong evidence that
P. vivax has, in fact, often placed a heavy burden of mortality and loss of fecundity on the populations that it afflicted (
47,
176). Its effects are greatest under conditions of relatively low and unstable malaria inoculation rates. These were the conditions that probably prevailed in Africa before 5,000 to 10,000 years ago (
118) and possibly throughout much of the preceding 100,000 years. Had
P. vivax been prevalent within human populations in West and Central Africa during this period, and we propose that it was, we would expect that the homozygous RBC Duffy-negative condition would have carried a considerable selective advantage and the heterozygous condition a slight one.
We would, however, expect that selection for raised frequencies of an FYnull allele in a population under P. vivax pressure should also take a “long” time. This expectation follows if mainly the FYnull homozygotes are refractory to P. vivax infection but rather little selective advantage is associated with the heterozygous condition. This, as we have just discussed, is quite likely to be the case.
Now, the homozygous combination of a rare allele, as the FY*B
null gene in Africa would at first have been, is itself almost vanishingly rare. Moreover, even when the homozygous combination of such an allele had arisen, it would, in the next generation, almost invariably have been diluted again to the weakly advantageous heterozygous state in combination with one of the high-frequency FY*B or FY*B RBC Duffy-positive alleles. Thus, a selection which involved mainly the persistently rare FY*B
null/FY*B
null homozygotes would inevitably proceed “slowly.” It would proceed slowly, that is, compared, with the speed of selection for a gene such as the allele for hemoglobin S (Hb*S), in which the heterozygous condition, Hb*S/Hb*A, carries the main selective advantage and which, in the early stages of selection, arises as frequently as the Hb*S allele itself. Moreover, the speed of selection for Hb*S would be further increased above that for the FY*B
null allele, because
P. falciparum is probably much more dangerous than
P. vivax under any conditions of endemicity. Hb*S has been estimated to approach a balanced equilibrium after around 2,000 years of selection under
P. falciparum malaria (
119). The “longer” time expected for
P. vivax malaria to select for the near fixation of the FY*B
null allele could, therefore, be in the order of 10,000 to several tens of thousands of years.
While we may infer that the process of selection for near fixation of the FY*Bnull allele in Africa under P. vivax would have taken several tens of thousands of years, we cannot, from the above line of argument, even hazard a guess as to the times when this process may have begun or ended. These may, however, be partly estimated from other evidence.
Modern human migrations out of Africa are believed to have taken place largely, if not entirely, within the past 100,000 years. Now, RBC Duffy negativity is, as already noted, an apparently harmless genetic condition. Therefore, once an FY
null allele has been selected in a human population, there should be little or no loss of frequency of the gene, especially from a population in which it had become fixed. Outside Africa, RBC Duffy negativity is found in declining frequencies through the Arabian peninsular, across the Middle East, and to the edges of Central Asia, but beyond these areas it is, with rare exceptions, virtually absent (
27). Had the high frequencies of the FY*B
null allele been selected before the main dispersals of modern humans began, this allele should be common in all human populations. Since it is not, its selection in Africa must have been completed only after these dispersals had taken place, i.e., within less than the past 100,000 years. Indeed, since human migrations out of Africa across the Old World and into the Western Pacific have probably continued into much more recent times and have carried almost no trace of the FY*B
null allele with them, we can probably reduce the time by which selection for African FY*B
null would have been completed to less than 50,000 years, which is the approximate period of the migrations to Melanesia and Australia. And if, as we do, we take it that the force for the selection of the FY*B
null allele in Africa was
P. vivax malaria, then the fact that at least 5,000 years of exposure to
P. vivax in southern Asia and China and the Western Pacific has not led to selection for high levels of RBC Duffy negativity anywhere in this region suggests that the process also takes longer than 5,000 years.
We now have a case that the selection for near fixation of RBC Duffy negativity takes at least 5,000 years and that it was completed in Africa less than 50,000 years ago. Unfortunately, we still cannot place an upper limit upon how long the process takes. The
P. vivax pressure could, for example, have begun several hundred thousand years ago and reached completion in a final rapid spurt at any time within the past 50,000 years, or it could have begun in Africa only 6,000 years ago and been completed in the last few hundred years, just in time for us to record the outcome. However, we now have corroborating evidence for the rough period within which the beginning of the selection would have taken place. In a study of haplotypes associated with the FY*B
null allele in African and European populations, Hamblin and Di Rienzo (
87) have proposed that a selective sweep towards the near fixation of RBC Duffy negativity in the African populations began between 97,200 and 6,500 years ago within 95% confidence limits.
We suggest that the most likely period for the selection of RBC Duffy negativity in Africa to have taken place, and therefore for strong selective pressure from
P. vivax to have been still active on these populations, lies somewhere between perhaps 10,000 and 5,000 years ago. This falls at the height of the last major glacial period, when equatorial Africa would have been much cooler than today and when it would have been infested with
Anopheles mosquitoes, which were at the time relatively inefficient vectors of malaria (
39,
118). These are conditions which would have tended to support low levels of unstable, and, therefore, severely life-degrading,
P. vivax transmission. They are the conditions which would have strongly favored selection for RBC Duffy negativity in the affected human populations in Africa.