A Troublesome Inheritance: Genes, Race and Human History Read online

  But given that all humans have the same set of genes and that there have been almost no full sweeps that push different alleles to dominance in different races, how have races come to differ from one another? The answer that has dawned on geneticists in the past few years is that you don’t always need a full sweep to change a trait. Many traits, like skin color or height or intelligence, are controlled by a large number of different genes, each of which has alleles that individually make small contributions to the trait. So if just some of these alleles become a little more common in a population, the trait will be significantly affected. This process is called a soft sweep, to contrast it with a full or hard sweep, in which one allele of a gene displaces all the others in a population.

  Pritchard gives the example of height, which is affected by hundreds of genes, because there are so many ways in which height can be increased. Suppose there are 500 such genes and each comes in two forms, with one allele having no effect on height and the other increasing it by 2 millimeters. An individual’s height depends on how many of the height-enhancing alleles he inherits. And that number in turn is determined by the frequency of each type of allele, meaning how common it is in the population. So if each of the height-promoting alleles becomes just 10% more common in the population, almost everyone will inherit more of them, and the average person’s height will increase by 200 millimeters, or 20 centimeters (8 inches).20

  This soft sweep process—a small increase in frequency in many genes—is a much easier way for natural selection to operate than through the hard sweeps—the major jump in frequency of a single allele—that are often assumed to be the main drivers of evolution. The reason is that the hard sweeps depend on a mutation creating a novel allele of great advantage, which happens only very rarely in a population. In a small population, it may take many generations for such a mutation to occur. Soft sweeps, on the other hand, act on alleles that already exist and simply make some of them more common. Soft sweeps can thus begin whenever they are needed.

  So suppose a group of pygmies were to leave their forest habitat and start herding cattle in a hot climate, where it’s advantageous to be tall and thin, like the Nuer and Dinka of the Sudan. The pygmies who were slightly taller would produce more children, and the height-promoting alleles of the genes that affect height would immediately begin to become more common in the population. In each generation, an individual would have a slightly greater chance of inheriting the height-promoting alleles, and the population would quite quickly become considerably taller.

  Consider, on the other hand, a trait in which there is no existing variation, such as the ability to digest milk in adulthood. For most of human existence and still in most people today, the gene for lactase is switched off shortly after weaning. To keep the gene switched on requires a beneficial mutation in the region of promoter DNA that controls it. But the promoter region is some 6,000 DNA units in length and occupies a minuscule fraction of the 3 billion units of the genome. In a small population, it might take many generations for the right mutation to occur in so small a target.

  Thus it seems to have taken around 2,000 years—some 80 generations—after the start of cattle breeding for the right mutation in the lactase promoter region to appear among the people of the Funnel Beaker Culture, cattle herders who occupied northern Europe some 6,000 years ago. Once established, the mutation spread rapidly and is now found at high frequency in northern Europe.

  Three mutations, which differ from one another and from the European mutation but have the same effect, arose independently among pastoralist peoples in eastern Africa and have swept through roughly 50% of the population. In each case, evidently, evolution has had to wait until the right mutation occurred, whereupon the allele grew more common because of the great advantage it conferred.

  In sum, hard sweeps cannot start until the right mutation occurs, and then they may take many generations to sweep through a population. Soft sweeps, based on standing variation in the many genes that control a single trait, can start immediately. For a species that undergoes a sudden expansion in its range and needs to adapt quickly to a succession of different challenges, the soft sweep is likely to be the dominant mechanism of evolutionary change. This explains why so few hard sweeps are visible in the human genome. Soft sweeps are presumably far more common, though at present are very hard to detect. The reason is the difficulty of distinguishing between the minor changes in allele frequency caused by genetic drift and the also minor changes brought about by natural selection through a soft sweep.

  The Genetic Structure of Race

  It is now possible to understand the structure of human variation, at least in broad outline. Different populations don’t have different genes—everyone has the same set. Of the traits specific to one race or another, a few are encoded in hard sweep alleles that have gone almost to fixation, such as the Duffy null allele or some of the alleles involved in shaping skin color, but many more are probably encoded in soft sweeps and hence in mere differences in the frequency of the cluster of alleles that shape each trait.

  The fact that genes work in combination explains how there can be so much variation in the human population and yet so few fixed differences between populations.

  Given the importance of allele frequencies in shaping specific traits, it’s not surprising that they afford a means of identifying an individual’s race. Excluding subjects of a different race is an essential procedure in surveys to detect the alleles that contribute to complex diseases like diabetes and cancer. The idea of these surveys, known as genomewide association studies, is to see if people who are particularly prone to disease are also more likely to carry a particular allele. If so, the allele may be associated with the disease. But the statistics can be confounded if the population being surveyed includes people of more than one race. An apparent association may emerge between the disease state and a particular allele even though the association is really due to some patients belonging to another race, one that naturally has a high frequency of the allele in question.

  Medical geneticists have therefore developed sets of test alleles that can be used to distinguish one race from another. Some alleles, particularly those with large differences in frequency between races, are more useful than others. These race-distinguishing DNA sites are known blandly as AIMs, or ancestry informative markers. Using a set of 326 AIMs, researchers achieved a nearly perfect correspondence between the race that subjects said they belonged to and the race to which they were assigned genetically.21 A set of 128 AIMs suffices to assign people to their continental race of origin, whether European, East Asian, American Indian or African.22 (The fifth continental race, Australian aborigines, could doubtless be identified just as easily, but political restrictions have so far largely blocked the study of aborigine genetics.)

  With greater numbers of markers, more closely related groups can be distinguished, such as the various ethnicities within Europe.

  Some biologists insist that AIMs do not prove the existence of race and that they point instead to geographic origin. But geographic origin correlates very well with race, at least on the continental level.

  Apart from genetic markers like the Duffy null allele, found almost exclusively in people of African ancestry, most AIMs are alleles that are just somewhat more common in one race than in another. A single AIM that occurs in 45% of East Asians and 65% of Europeans says that the carrier is a little more likely to be European, but is hardly definitive. When the results from a string of AIMs are combined, however, an answer with high statistical probability is obtained. This is the same general method used in DNA fingerprinting, except that the 14 sites at which the genome is sampled in forensic DNA analysis are not SNPs but variable runs of DNA repeats.

  The approach of comparing allele frequencies can even be used with people of mixed race to assign component parts of an individual’s genome to their parent’s racial origin. When people of different races marry, their
children are perfect blends of their parents’ genes. But at the genetic level, the chunks of DNA that came from the mother’s and the father’s races remain separate and distinguishable for many generations. Researchers can track along the chromosomes of African Americans, assigning each stretch of DNA to either African or European ancestors. In one recent study, researchers analyzed the genomes of almost 2,000 African Americans and found that 22% of their DNA came from European ancestors and the rest from Africans, a conclusion in line with several previous reports.23

  The same study found evidence that African Americans may already have begun adapting genetically to the American environment in the several generations since their ancestors arrived in the United States. The malaria-protecting genetic variants common in Africans, such as the variation that causes sickle-cell anemia, are no longer a necessity of survival in the United States, so the pressure of natural selection to retain these variants would be relaxed. The authors found some evidence that these variants have indeed declined in frequency in African Americans, while genes that provide protection against influenza have grown more common. The finding, if confirmed, would be a striking instance of evolutionary change within the past few hundred years.

  Over the last 50,000 years, modern humans have been subjected to enormous evolutionary pressures, in part from the consequences of their own social culture. They explored new ranges and climates and developed new social structures. Fast adaptation, particularly to new social structures, was required as each population strove to exploit its own ecological niche and to avoid conquest by its neighbors. The genetic mechanism that made possible this rapid evolutionary change was the soft sweep, the reshaping of existing traits by quick minor adjustments in the sets of alleles that controlled them.

  But what began as a single experiment with the ancestral human population became a set of parallel experiments once the ancestral population had spread throughout the world. These independent evolutionary paths led inevitably to the different human populations or races that inhabit each continent.

  Arguments Against the Existence of Race

  Readers who are by now persuaded that recent human evolution has resulted in the existence of races may wish to proceed to the next chapter. But for those who remain perplexed that so many social scientists and others should argue race does not exist, here is an analysis of some of their contentions.

  Start with Jared Diamond, the geographer and author of Guns, Germs, and Steel, who was quoted in chapter 4 as comparing the idea of race with the belief that the Earth is flat. His principal argument for the nonexistence of race is that there are many different “equally valid procedures” for defining human races, but since all are incompatible, all are equally absurd. One such procedure, Diamond proposes, would be to put Italians, Greeks and Nigerians in one race, and Swedes and Xhosas (a southern African tribe) in another.

  His rationale is that members of the first group carry genes that confer resistance to malaria and those of the second do not. This is just as good a criterion as skin color, the usual way of classifying races, Diamond says, but since the two methods lead to contradictory results, all racial classification of humans is impossible.

  The first flaw in the argument is the implied premise that people are conventionally assigned to races by the single criterion of skin color. In fact, skin color varies widely within continents. In Europe it runs from light-skinned Swedes to the olive complexion of southern Italians. Skin color is thus an ambiguous marker of race. People belong to a race not by virtue of any single trait but by a cluster of criteria that includes the color of skin and hair, and the shape of eyes, nose and skull. It is not necessary for all these criteria to be present: some East Asians, as noted above, lack the EDAR allele for thick hair, but they are still East Asians.

  The single criterion that Diamond proposes as an alternative, genes that confer resistance to malaria, makes no evolutionary sense. Malaria became a significant human disease only very recently, some 6,000 years ago, and each race then independently developed resistance to it. Italians and Greeks resist malaria because of mutations that also cause the blood disease known as thalassemia, whereas Africans resist malaria through a different mutation that causes sickle-cell anemia. The trait of resisting malaria is one that has been acquired secondarily to race, so obviously it is not an appropriate way of classifying the populations. A scholar’s duty is to clarify, but Diamond’s argument seems designed to distract and confuse.

  A more serious and influential argument, also designed to banish race from the political and scientific vocabulary, is one first advanced by the population geneticist Richard Lewontin in 1972. Lewontin measured a property of 17 proteins from people of various different races and calculated a measure of variation known as Wright’s fixation index. The index is designed to measure how much of the variation in a population resides in the population as a whole and how much is due to differences between specific subpopulations.

  Lewontin’s answer came out to 6.3%, meaning that of all the variations in the 17 kinds of protein he had looked at, only 6.3% lay between races, while a further 8.3% lay between ethnic groups within races. These two sources of variation add up to around 15%, leaving the rest as common to the population as a whole. “Of all human variation, 85% is between individual people within a nation or tribe,” Lewontin stated. He concluded on this basis that “human races and individuals are remarkably similar to each other, with the largest part by far of human variation being accounted for by the differences between individuals.”

  He went on to say that “Human racial classification is of no social value and is positively destructive of social and human relations. Since such racial classification is now seen to be of virtually no genetic or taxonomic significance either, no justification can be offered for its continuance.” 24

  Lewontin’s thesis immediately became the central genetic plank of those who believe that denying the existence of race is an effective way to combat racism. It is prominently cited in Man’s Most Dangerous Myth: The Fallacy of Race, an influential book written by the anthropologist Ashley Montagu with the aim of eliminating race from the political and scientific vocabulary. Lewontin’s statement is quoted at the beginning of the American Anthropological Association’s statement on race and is a founding principle of the assertion by sociologists that race is a social construct, not a biological one.

  But despite all the weight that continues to be placed on it, Lewontin’s statement is incorrect. It’s not the basic finding that is wrong. Many other studies have confirmed that roughly 85% of human variation is among individuals and 15% between populations. This is just what would be expected, given that each race has inherited its genetic patrimony from the same ancestral population that existed in the comparatively recent past.

  What is in error is Lewontin’s assertion that the amount of variation between populations is so small as to be negligible. In fact it’s quite significant. Sewall Wright, an eminent population geneticist, said that a fixation index of 5% to 15% indicates “moderate genetic differentiation” and that even with an index of 5% or less, “differentiation is by no means negligible.” 25 If differences of 10 to 15% were seen in any other than the human species they would be called subspecies, in Wright’s view.26

  Why should Wright’s judgment that a fixation index of 15% between races is significant be preferred over Lewontin’s assertion that it is negligible? Three reasons: (1) Wright was one of the three founders of population genetics, the relevant discipline; (2) Wright invented the fixation index, which is named after him; (3) Wright, unlike Lewontin, had no political stake in the issue.

  Lewontin’s argument has other problems, including a subtle error of statistical reasoning named Lewontin’s fallacy.27 The fallacy is to assume that the genetic differences between populations are uncorrelated with one another; if they are correlated, they become much more significant. As the geneticist A.W.F. Edwards wrote, “Most of the infor
mation that distinguishes populations is hidden in the correlation structure of the data.” The 15% genetic difference between races, in other words, is not random noise but contains information about how individuals are more closely related to members of the same race than those of other races. This information is brought to light by the cluster analyses, described earlier in this chapter, which group people into populations that correspond at the highest level to the major races.

  Despite the misleading political twist on Lewontin’s argument, it became the centerpiece of the view that racial differences were too slight to be worth scientific attention. The assertion left the ugly implication that anyone who thought otherwise must be some kind of a racist. The subject of human race soon became too daunting for all but the most courageous and academically secure of researchers to touch.

  A frequent assertion of those who seek to airbrush race out of human variation is that no distinct boundaries can be drawn between one race and another, leaving the implication that races cannot exist. “Humanity cannot be classified into discrete geographic categories with absolute boundaries,” proclaims the American Association of Physical Anthropologists in its statement on race.28 True, races are not discrete entities and have no absolute boundaries, as already discussed, but that doesn’t mean they don’t exist. The classification of humans into five continental based races is perfectly reasonable and is supported by genome clustering studies. In addition, classification into the three major races of African, East Asian and European is supported by the physical anthropology of human skull types and dentition.

  A variation on the no distinct boundary argument is the objection that the features deemed distinctive of a particular race, like dark skin or hair type, are often inherited independently and appear in various combinations. “These facts render any attempt to establish lines of division among biological populations both arbitrary and subjective,” states the American Anthropological Association’s statement on race.29 But as already noted, races are identified by clusters of traits, and to belong to a certain race, it’s not necessary to possess all of the identifying traits. To take a practical example of what the anthropologists are talking about, most East Asians have the sinodont form of dentition, but not all do. Most have the EDAR-V370A allele of the EDAR gene, but not all do. Most have the dry earwax allele of the ABCC11 gene, but all not all do. Nonetheless, East Asian is a perfectly valid racial category, and most people in East Asia can be assigned to it.