Who We Are and How We Got Here

by David Reich

  The finding that Oase was from a dead-end population accords with the archaeological record of the first modern humans of Europe. The stone tools these humans made came in a variety of styles, but like the population of Oase himself, most were dead ends in the sense that they disappeared from the archaeological record after a few thousand years. However, one style known as the Protoaurignacian—thought to derive from the earlier Ahmarian of the Near East—persisted after thirty-nine thousand years ago and likely developed into the Aurignacian, the first widespread modern human culture in Europe.30 These patterns could be explained if the makers of Aurignacian tools derived from a different migration into Europe compared to other early modern humans like Oase. This scenario could explain how it could be that Oase’s population interbred heavily with local European Neanderthals, and yet the Neanderthal ancestry in Europeans today is not from Europe.

  Two Groups at the Edge of Compatibility

  The low fertility of hybrids may also have reduced Neanderthal ancestry in the DNA of people living today. This possibility was first advanced by Laurent Excoffier, who knew from studies of animals and plants that when one population moves into a region occupied by another population with which it can interbreed, even a small rate of interbreeding is enough to produce high proportions of mixture in the descendants—far more than the approximately 2 percent Neanderthal ancestry seen in non-Africans today. Excoffier argued that the only way that the modern human genome could have ended up with so little Neanderthal ancestry was if expanding modern humans had offspring with other modern humans at least fifty times more often than they did with the Neanderthals living in their midst.31 He thought that the most likely explanation for this was that Neanderthals and modern human offspring were much less fertile than the offspring of matings between pairs of modern humans.

  I wasn’t convinced by this argument. Rather than low hybrid fertility, I favored the explanation that there simply wasn’t much interbreeding for social reasons. Even today, many groups of modern humans keep largely to themselves because of cultural, religious, or caste barriers. Why should it have been any different for modern humans and Neanderthals when they encountered one another?

  But Excoffier got something important right. This became evident when we and others analyzed the bits of Neanderthal DNA that entered into the modern human population and mapped their positions in the genome. To do this, Sriram Sankararaman in my laboratory searched for mutations that were present in the sequenced Neanderthals but were rare or absent in sub-Saharan Africans. By studying stretches of such mutations, we were able to find a substantial fraction of all the Neanderthal ancestry fragments in each non-African. Looking at where in the genome these Neanderthal ancestry fragments occurred, it became clear that the impact of Neanderthal interbreeding varied dramatically across the genome of non-African people today. The average proportion of Neanderthal ancestry in non-African populations is around 2 percent, but it is not spread evenly. In more than half the genome, no Neanderthal ancestry has been detected in anyone. But in some unusual places in the genome, more than 50 percent of DNA sequences are from Neanderthals.32

  A critical clue that helped us to understand how this pattern had formed came from studying the places in non-African genomes where Neanderthal ancestry is rare. In any one stretch of DNA, an absence of Neanderthal ancestry in the population can happen by chance, as we think is the case for mitochondrial DNA. However, it is improbable that a substantial subset of the genome with particular biological functions will be systematically depleted of Neanderthal ancestry unless natural selection systematically worked to remove it.

  But evidence of systematic removal of Neanderthal ancestry is exactly what we found—and, remarkably, we found a particularly intense depletion of Neanderthal ancestry by natural selection in two parts of the genome known to be relevant to the fertility of hybrids.

  The first place of reduced Neanderthal ancestry was on chromosome X, one of the two sex chromosomes. This reminded me of a pattern that Nick Patterson and I had run into in our work on the separation of human and chimpanzee ancestors in a study we had carried out together and published years before.33 There are only three copies of chromosome X in any population for every four other chromosomes (because females carry two copies and males only one, in contrast to two copies in each sex for most of the rest of the chromosomes). This means that in any one generation, the probability that any two X chromosomes share a common parent is four-thirds the probability that any two of one of the other chromosomes share a common parent. It follows that the expected time since any pair of X chromosome sequences descend from a common ancestral sequence is about four-thirds of that in the rest of the genome. In fact, though, the real data suggest a number that is around half or even less.34 In our study of the common ancestral population of humans and chimpanzees, we had not been able to identify any history that could explain this pattern, such as a lower rate of females moving among groups than males, or a more variable number of children in females than in males, or population expansion or contraction. However, the patterns could be explained by a history in which the ancestors of humans and chimpanzees initially separated, then came together to form either human or chimpanzee ancestors before the final separation of the two lineages.

  How is it that hybridization can lead to so much less genetic variation on chromosome X than on the rest of the genome? From studies of a variety of species across the animal kingdom, it is known that when two populations are separated for long enough, hybrid offspring have reduced fertility. In mammals like us, reduced fertility is much more common in males, and the genetic factors contributing to this reduced fertility are concentrated on chromosome X.35 So when two populations are so separated that their offspring have reduced fertility, but nevertheless mix together to produce hybrids, it is expected that there will be intense natural selection to remove the factors contributing to reduced fertility. This process will be especially evident on chromosome X because of the concentration of genes contributing to infertility on it. As a result, there tends to be natural selection on chromosome X for stretches of DNA from the population that contributed most of the hybrid population’s ancestry. This causes the hybrid population to derive its chromosome X almost entirely from the majority population, leading to an anomalously low genetic divergence on chromosome X between the hybrid population and one of the hybridizing populations, consistent with the pattern seen in humans and chimpanzees.

  This theoretical prediction might sound fanciful, but in fact it is borne out in hybrids of the western European and eastern European house mouse species in a band of territory that runs in a north-to-south direction through central Europe, roughly along the line of the former Cold War Iron Curtain. While the density of mutations separating the hybrid mice from western European mice is high in most of the genome because the hybrid mice carry DNA not just from western European mice but also from highly divergent eastern European mice, the density on the X chromosome is far less because the hybrid mice harbor very little DNA from the eastern European population whose X chromosomes are known to cause infertility in male hybrids.36

  Since the publication of our paper in 2006 suggesting that either humans or chimpanzees may derive from an ancient major hybridization, the evidence for ancient major hybridization in the ancestry of humans and chimpanzees has, if anything, become even stronger. In 2012 Mikkel Schierup, Thomas Mailund, and colleagues developed a new method to estimate the suddenness of separation of the ancestors of two present-day species from genetic data, based on principles similar to the Li and Durbin approach described in chapter one.37 When they applied the method to study the separation time of common chimpanzees and their distant cousins, bonobos, they found evidence that the separation was very sudden, consistent with the hypothesis that the species were separated by a huge river (the Congo) that formed rather suddenly one to two million years ago. In contrast, when they applied the method to study humans and chimpanzees, they found evidence for an extended period
of genetic interchange after population differentiation began, as expected for hybridization.38

  An even more important piece of evidence came from a paper Schierup and Mailund published in 2015, when together with other colleagues, they showed that the regions that are denuded of Neanderthal mixture on chromosome X in non-Africans are to a large extent the same regions that are driving the low genetic divergence between humans and chimpanzees.39 This is what would be expected if mutations that contribute to reduced fertility when they occur in a hybrid individual tend to be concentrated not just on chromosome X, but in particular regions along chromosome X, causing the minority ancestry to be removed from the population by natural selection against the male hybrids who carry it. The evidence of selection to remove Neanderthal DNA from chromosome X was a tell-tale sign that male hybrids had reduced fertility.

  We also found a second line of evidence for infertility in hybrids of Neanderthals and modern humans—a line of evidence that had nothing to do with the X chromosome. When reduced fertility is observed in hybrid males, the genes responsible tend to be highly active in the male reproductive tissue, causing malfunctions of sperm. So a prediction of the hypothesis of male hybrid infertility suggested to me by evolutionary biologist Daven Presgraves after I showed him the X chromosome evidence is that genes unusually active in the germ cells of a man’s testicles will have less Neanderthal ancestry on average than genes that are most active in other body tissues. When we looked in real data, Presgraves’s prediction was exactly borne out.40

  The problems faced by modern humans with Neanderthal ancestry went beyond reduced fertility, as it turns out that Neanderthal ancestry is not just reduced on the X chromosome and around genes important in male reproduction, but is also reduced around the great majority of genes (there is far more Neanderthal ancestry in “junk” parts of the genome with few biological functions). The clearest evidence for this came from a study in 2016, in which we published a genome-wide ancient DNA dataset from more than fifty Eurasians spread over the last forty-five thousand years.41 We showed that Neanderthal ancestry decreased continually from 3 to 6 percent in most of the samples we analyzed from earlier times to its present-day value of around 2 percent at later times and that this was driven by widespread natural selection against Neanderthal DNA.

  A large part of the Neanderthal range was in a region where ice ages caused periodic collapses of the animal and plant populations that Neanderthals depended on, a problem that may not have afflicted modern human ancestors in tropical Africa to the same extent. There is genetic confirmation for smaller Neanderthal than modern human population sizes from the fact that the diversity of their genomes was about four times smaller. A history of small size is problematic for the genetic health of a population, because the fluctuations in mutation frequency that occur every generation are substantial enough to allow some mutations to spread through the population even in the face of the prevailing wind of natural selection that tends to reduce their frequencies.42 So in the half million years since Neanderthals and modern humans separated, Neanderthal genomes accumulated mutations that would prove detrimental when later, Neanderthal/modern human interbreeding occurred.

  The problematic mutations in the Neanderthal genome form a sharp contrast with more recent mixtures of divergent modern human populations where there is no evidence for such effects. For example, among African Americans, in studies of about thirty thousand people, we have found no evidence for natural selection against African or European ancestry.43 One explanation for this is that when Neanderthals and modern humans mixed they had been separated for about ten times longer than had West Africans and Europeans, giving that much more time for biological incompatibilities to develop. A second explanation relates to the observation, from studies of many species, that when infertility arises between populations, it is often due to interactions between two genes in different parts of the genome. Since two changes are required to produce such an incompatibility, the rate of infertility increases with the square of population separation time, so a ten-times-larger population separation translates to one hundred times more genetic incompatibility. In light of this the lack of infertility in hybrids of present-day humans may no longer seem so surprising.

  Thesis, Antithesis, Synthesis

  An important strand in continental European philosophy beginning in the eighteenth century was that the march of ideas proceeds in a “dialectic”: a clash of opposed perspectives that leads to a synthesis.44 The dialectic begins with a “thesis,” followed by an “antithesis.” Progress is achieved through a resolution, or “synthesis,” which transcends the two-sided debate that engendered it.

  So it has been with our understanding of modern human origins. For a long time, many anthropologists favored multiregionalism, the theory that modern humans in any given place in the world descend substantially from archaic humans who lived in the same geographical region. Thus Europeans were thought to derive large proportions of their ancestry from Neanderthals, East Asians from humans who dispersed to eastern Eurasia more than a million years ago, and Africans from African archaic forms. The biological differences among modern human populations would then have extremely deep roots.

  Multiregionalism soon encountered its antithesis, the out-of-Africa theory. In this theory, modern humans did not evolve in each location in the world separately from local archaic forms. Instead, modern humans everywhere derive from a relatively recent migration from Africa and the Near East beginning around fifty thousand years ago. The recent date of “Mitochondrial Eve” compared with the deep divergence of Neanderthal mitochondrial DNA provided some of the best evidence for this theory. In opposition to the multiregional hypothesis, the out-of-Africa theory emphasizes the recent origin of the differences among present-day human populations, relative to the multimillion-year time depth of the human skeletal record.

  Yet the out-of-Africa argument is not entirely right either. We now have a synthesis, driven by the finding of gene flow between Neanderthals and modern humans based on ancient DNA. This affirms a “mostly out-of-Africa” theory, and also reveals something profound about the culture of those modern humans who must have known Neanderthals intimately. While it is clear from the genetic data that modern humans outside of Africa descend from the expansion of an African-origin group that swept around the world, we now know that some interbreeding occurred. This must make us think differently about our ancestors and the archaic humans they encountered. The Neanderthals were more like us than we had imagined, perhaps capable of many behaviors that we typically associate with modern humans. There must have been cultural exchange that accompanied the mixture—the novels by William Golding and Jean Auel were right to dramatize these encounters. We also know that there has been a biological legacy bequeathed by Neanderthals to non-Africans, including genes for adapting to different Eurasian environments, a topic to which I will return in the next chapter.

  At the conclusion of the Neanderthal genome project, I am still amazed by the surprises we encountered. Having found the first evidence of interbreeding between Neanderthals and modern humans, I continue to have nightmares that the finding is some kind of mistake. But the data are sternly consistent: the evidence for Neanderthal interbreeding turns out to be everywhere. As we continue to do genetic work, we keep encountering more and more patterns that reflect the extraordinary impact this interbreeding has had on the genomes of people living today.

  So the genetic record has forced our hand. Instead of confirming scientists’ expectations, it has produced surprises. We now know that Neanderthal/modern human hybrid populations were living in Europe and across Eurasia, and that while many hybrid populations eventually died out, some survived and gave rise to large numbers of people today. We now know approximately when the modern human and Neanderthal lineages separated. We now also know that when these lineages reencountered each other, they had evolved to such an extent that they were at the very limit of biological compatibility. This raises a question: Were
the Neanderthals the only archaic humans who interbred with our ancestors? Or were there other major hybridizations in our past?

  3

  Ancient DNA Opens the Floodgates

  A Surprise from the East

  In 2008, Russian archaeologists dug up a pinky bone at Denisova Cave in the Altai Mountains of southern Siberia, named after an eighteenth-century Russian hermit named Denis who had made his home there. The bone’s growth plates were not fused, showing that the bone came from a child. Its date was uncertain, as it was too small to be dated by radiocarbon analysis, and it was found in a mixed-up soil layer of the cave that contained artifacts dating to both less than thirty thousand and more than fifty thousand years ago. The leader of the excavation, Anatoly Derevianko, reasoned that the bone’s owner could have been a modern human, and the sample was so labeled. Alternatively, could the bone’s owner have been a Neanderthal, as Neanderthal remains were also found near the cave?1 Derevianko sent part of the bone to Svante Pääbo in Germany.

  Pääbo’s team, led by Johannes Krause, was successful in extracting mitochondrial DNA from the Denisova Cave bone.2 Its sequence was of a type that had never before been observed in more than ten thousand modern human and seven Neanderthal sequences. There are around two hundred mutational differences separating the mitochondrial DNA of people living today from that of Neanderthals. The new mitochondrial DNA from the Denisova finger bone featured nearly four hundred differences from the mitochondrial DNA of both present-day humans and Neanderthals. Based on the rate at which mutations accumulate, mitochondrial DNA sequences from present-day humans and Neanderthals are estimated to have separated from each other 470,000 to 360,000 years ago.3 The number of mutational differences found in the mitochondrial DNA from the Denisova finger bone suggested a separation time of roughly eight hundred thousand to one million years ago. This suggested that the finger bone might belong to a member of a never-before-sampled group of archaic humans.4