When agriculture was new, natural selection must have operated with the genetic variation that already existed, just as it does in small-scale artificial-selection experiments. Such experiments cause changes in the frequency of existing alleles.
Most preexisting genetic variation must have taken the form of a few neutral variants of each gene—variants that are not significantly different from each other. They may well do something, but the neutral alleles all do the same thing. We doubt if many of those neutral genes turned out to be the solution for the problems faced by the future farmers of Eurasia. More likely, preexisting functional variation mattered more. For example, there is a gene whose ancestral form helps people to conserve salt. Since humans spent most of their history in hot climates, this variant was generally useful. A high frequency of this ancestral allele among African Americans probably plays a role in their increased risk of high blood pressure today. In tropical Africa, in fact, almost everyone has the ancestral version of the gene. In Eurasia, a null variant (one that does nothing at all) becomes more and more common as one moves north.4 Perhaps the gene's action of promoting salt conservation becomes harmful—by causing higher blood pressure—in cooler areas, where people sweat less and lose less salt.
Significantly, the null allele is the same in both Europe and eastern Asia—which suggests that it originated in Africa and is ancient. If it had separate European and Asian origins, then we would expect to see different versions in the two regions, just as different broken pigment genes lead to light skin in the two regions.
The most reasonable explanation for this dud salt-conservation gene is that parts of Africa (before the expansion out of Africa) were cool enough that salt retention was not a major concern, so that in these regions an inactive form of the gene was in fact advantageous. This might have happened in Ethiopia during glacial periods, considering that the climate on the Ethiopian plateau is moderate even today. If so, the null allele would represent preexisting adaptive variation caused by environmental variations inside Africa rather than neutral variation. Such internal variation inside Africa must have helped prepare humans for environments outside Africa.
Another kind of preexisting genetic variation would have consisted of balanced polymorphisms. Balanced polymorphisms occur within a population when the population maintains two different alleles of a gene, and the reason the polymorphism can be stable is that heterozygous individuals will have greater fitness than homozygous individuals. A heterozygote advantage exists, for example, in sickle cell and other malaria defenses. There are also alleles that have positive effects when rare, but whose advantages decrease as they become common, eventually becoming negative (this is called frequency-dependent selection). Some of the most interesting examples involve behavior and lend themselves to a game-theory analysis.
The best-known model is the hawk-dove game, where some individuals are genetically aggressive while others are genetically peaceful. When hawks are rare, they easily defeat doves and have higher fitness. As they become more common, however, they run into other hawks more often and have costly fights that decrease their fitness. At some frequency, the fitness of hawks and doves is the same, leading to a balanced polymorphism.5
Balanced behavioral polymorphisms could respond quickly to new selective pressures. If the original mix was 50 percent doves and 50 percent hawks, an environmental change that raised the costs of aggressive behavior would lead to a shift in frequency—say to 70 percent doves and 30 percent hawks. This kind of evolutionary change is very rapid, especially when compared to new sweeping genes, which are rare in the beginning and take thousands of years to reach frequencies of 20 percent or more. If the doves acquired a selective advantage of 5 percent, that change (from 50 percent to 70 percent) could occur in less than ten generations.
Human genetic variation was limited in the days before agriculture, in part because populations were small, and it was often not useful, since many of the changes that were favored among agriculturalists would actually have been deleterious among their hunter-gatherer ancestors. This means that some of the alleles with the right effects in farmers would have been extremely rare or nonexistent in their hunter-gatherer ancestors. For example, variants of G6PD (for glucose-6-phosphate dehydrogenase) with reduced function protect against falci-parum malaria but also have negative effects, especially in men. Today, those G6PD variants have a net positive effect in malarious regions and have become common in many populations. Before the spread of falciparum malaria, those variants likely had a net negative effect in all populations, and so were extremely rare.
Therefore, new mutations must have played a major role in the evolutionary response to agriculture—and as luck would have it, there was a vast increase in the supply of those mutations just around this time because of the population increase associated with agriculture. We're not saying that the advent of agriculture somehow called forth mutations from the vasty deep that fitted people to the new order of things. Mutations are random, and as always, the overwhelming majority of them had neutral or negative effects. But more mutations occurred in large populations, some of them beneficial. Increased population size increased the supply of beneficial mutations just as buying many lottery tickets increases your chance of winning the prize.
By the beginnings of recorded history some 5,000 years ago, new adaptive mutations were coming into existence at a tremendous rate, roughly 100 times more rapidly than in the Pleistocene. This means that recent human evolution differs qualitatively from typical artificial selection acting on domesticated animals. It is simply a matter of scale. In the artificial-selection experiments, which typically involve no more than tens or hundreds of animals, very few new favorable mutations occur, and selection must act primarily on preexisting genetic variation. In recent human evolution, we're talking anywhere from millions to hundreds of millions of individuals, all of them potential mutants, so most of the advantageous variants would have been new.
You might think that alleles that were already common would be more likely than new variants to grow to high frequency under agriculture. It stands to reason that the new mutations, which would start out with a single copy, would face disadvantages. But that reasoning underestimates the effect of the advantage that the mutation conferred on the individual who carried it and his or her descendants. Even a single copy of an advantageous gene has a fair chance of succeeding (10 percent for a gene with a 5 percent advantage), and exponential growth allows it to spread rapidly. Many new mutations must have occurred in those large farming populations, and the great majority of the sweeping genes must have been new.
Not only did post-agricultural evolution involve much higher numbers than would be possible in any artificial-selection experiment, it also involved a much longer time frame. Post-agricultural evolution occurred over some 400 generations, which would be impractical for selection experiments using mammals. That long time scale also makes for a qualitative difference, since it is long enough to allow new mutations to rise to high frequency and make up a major part of adaptive variation.
Recent studies have found hundreds of ongoing sweeps— sweeps begun thousands of years ago that are still in progress today. Some alleles have gone to fixation, more have intermediate frequencies, and most are regional. Many are very recent: The rate of origination peaks at about 5,000 years ago in the European and Chinese samples, and at about 8,500 years ago in the African sample. There are so many sweeps under way, in fact, that we can do some useful statistical analysis. Often we have some idea of a gene's function—for example, by seeing what tissues it is highly expressed in, or by knowing what goes wrong when it's inactivated. Using that information, we can look at the hundreds of genes undergoing sweeps and see what kinds of jobs they do. And when we do that kind of analysis, we see that most of the sweeping alleles fall into a few functional categories: Many involve changes in metabolism and digestion, in defenses against infectious disease, in reproduction, in DNA repair, or in the central nervous system.
Was this article helpful?