by Pbo, Svante
Rasmus focused on the regions that were particularly unusual in having deep divergences in Europe relative to Africa. There were seventeen such regions. Ed sent Rasmus the Neanderthal DNA sequences we had in hand from fifteen of these seventeen regions. He wrote back in July with stunning results. At thirteen of the fifteen regions, the Neanderthals did indeed carry the variants now found in Europe and not in Africa. Later, Rasmus refined his analysis and focused on twelve regions that were over 100,000 nucleotides long and found that the Neanderthals carried the variants now found in Europe for ten of these regions. This was indeed an amazing result! I could not imagine any other explanation for this than gene flow from Neanderthals into people outside Africa. Although this was what scientists call a qualitative result, as it didn’t allow us to calculate how much DNA Neanderthals had contributed to people in Europe or Asia, it made vividly clear that such a contribution had occurred. And it was an independent line of support of David and Nick’s other, quantitative, analyses that had led to the same conclusion.
We continued to think about additional ways to test for gene flow. As always, David tended to come up with brilliant ideas. He argued that a trivial reason why a region of the genome today might be close to a Neanderthal is that it accepts very few mutations, either because its mutation rate is low or because it cannot mutate without causing a person to die. If a genomic region in me is similar to a Neanderthal for this reason, then one would expect that region to also be close to other humans today, simply because it rarely changes. But if a region in me is similar to a Neanderthal because my ancestors had inherited it from Neanderthals, then there is no reason why I should be close to other people. In fact, I may even tend to be far from others thanks to the distinct evolutionary history of Neanderthals.
David set out to implement these insights into our analysis. He used the European parts of the human reference genome and divided them into segments. He then plotted the number of differences in those segments from the Neanderthal genome and from another European genome (that of Craig Venter). In general, he saw that the closer the European segments of the reference genome were to the Neanderthals, the closer they were to Craig’s genome, suggesting that the rate with which segments accumulated mutations determined the differences both to the Neanderthal genome and to Craig’s genome. But when he came to segments where the European was very similar to the Neanderthals, the relationship reversed and suddenly became more different from Craig’s genome. I was already convinced from the other analyses that gene flow had occurred. But when David presented these results during a visit to our lab in December 2009, I felt sure that we would be able to convince the world that Neanderthal DNA segments were lingering on in people today. No matter how we looked at the data, we came up with the same result.
We could now turn our full attention to how, when, and where modern humans had interacted intimately with Neanderthals. The first question was the direction in which the gene flow had occurred—that is, whether modern humans had contributed DNA to Neanderthals, Neanderthals had contributed DNA to modern humans, or both. Although one might think that genes would flow equally in both directions when two human groups meet, in real life that is rarely the case. Often one group is socially dominant over the other. A common pattern is then that men from the dominant group sire children with women from the nondominant group and the children remain with their mothers in the nondominant group. Thus, gene flow will tend to be from the socially dominant to the nondominant group. Obvious examples are white slave owners in the American South and British colonialists in Africa and India.
We tend to think that modern humans were dominant over Neanderthals, as Neanderthals eventually disappeared. But our data actually suggested that gene flow had been from Neanderthals into modern humans. David’s last result, for example, showed that the DNA regions where some Europeans were similar to Neanderthals tended to be very different from those found in other Europeans. The implication was that these regions had accumulated differences separately from those of other Europeans for some time before entering the present European gene pool. Presumably this had happened in the Neanderthals. If the contribution had gone in the other direction—from modern humans into Neanderthals—those regions would just have been average parts of the genome with average amounts of differences from other Europeans. For this and other reasons, we concluded that all, or almost all, of the gene flow was from Neanderthals into modern humans.
This did not necessarily mean that children from Neanderthal-modern human unions were never raised by Neanderthals. In 2008, Laurent Excoffier, a Swiss population geneticist who had always taken an interest in data generated by our group, published a paper about gene flow between two populations, after which one expands while the other one doesn’t, or even shrinks in size. In such a case the gene variants exchanged between the populations are more likely to be preserved in the growing population than in the dwindling population. And if the contribution occurs along the “wave front” of an advancing population, where the advancing population is in the process of expanding, then the contributed variants might even reach quite high frequencies. Excoffier had aptly named this phenomenon “allelic surfing,” illustrating the fact that an allele that entered the advancing “wave” of a colonizing population might surge to high frequencies. This meant that interbreeding might have happened in both directions but that we wouldn’t detect it in Neanderthals, because, after the encounter, their population size was likely to have shrunk.
Another more mundane reason why we might not detect gene flow from modern humans into Neanderthals is that the 38,000-year-old Neanderthals from Vindija Cave simply lived before interchange happened. Perhaps we will never really know the details of how Neanderthals and modern humans had interbred, but I am not overly troubled by this. To me, “who had sex with whom” in the Late Pleistocene is a question of secondary importance. What matters is that Neanderthals did in fact contribute genes to people today. That is what matters with respect to the genetic origin of people today.
Having confirmed David and Nick’s findings, we explored the question of how much of the genomes of people outside Africa had come from Neanderthals. This could not be directly estimated from the SNP matching frequencies because the number of extra matches between Neanderthals and people outside Africa depended on a number of other variables. One variable was when the common ancestor of Neanderthals and modern humans lived, another was when they had mixed with each other, and a third was how large the populations of Neanderthals had been. Monty Slatkin estimated the fraction of Neanderthal DNA in present-day people by modeling the population history of Neanderthals and modern humans. His results suggested that people who are of European or Asian ancestry have inherited between 1 and 4 percent of their DNA from Neanderthals. David and Nick did a different analysis where they essentially asked how far Europeans and Asians are toward being 100 percent Neanderthals. The answer varied between 1.3 and 2.7 percent. Thus we concluded that less than 5 percent of the DNA of people outside Africa came from Neanderthals—a small but clearly discernible proportion.
The final question for that round of work was how Neanderthal DNA wound up not only in Europeans but also in the Chinese and the Papuans. As far as we knew, Neanderthals had never been in China and surely they had never made it to Papua New Guinea, which led us to infer that Neanderthals and the ancestors of the Chinese and the Papuans must have met somewhere farther west.
I kept my Middle Eastern idea to myself as we sat huddled around the speaker phone in my Leipzig office during our weekly phone meeting so that the sharp minds of the people in the consortium would explore all possibilities. Monty came up with a complicated scenario to explain the patterns in variation we saw. First, he assumed that Neanderthal ancestors originated in some corner of Africa and then left Africa and evolved into Neanderthals in western Eurasia sometime around 300,000 to 400,000 years ago. Second, if the place where Neanderthal ancestors originated in Africa was the same as the place where 200,000 o
r more years later the ancestors of modern humans originated; and if populations in Africa over that time had remained subdivided so that there were differences in allele frequencies that persisted from the time the Neanderthal ancestors left until the time the modern human ancestors started spreading; and if, when modern humans originated, they swept not only out of Africa but also across Africa and incorporated variants that existed there through interbreeding with archaic African humans, then the result would be that Neanderthals were more similar to people outside Africa than inside Africa, just as we were seeing.
Although this scenario was theoretically possible, it required the persistence of a stable subdivision of the Africa population for hundreds of thousands of years. As Monty himself pointed out, this seemed unlikely since humans are quite prone to moving around. The bigger problem was its complexity. For reconstructing the past, it is considered best to favor the simplest scenario that can account for the patterns seen, even though many other, more complex scenarios are also possible. The principle of preferring the simplest explanation is called parsimony. For example, others had assumed that the ancestors of modern humans and Neanderthals had originated in Asia and that the ancestors of modern humans went to Africa, leaving no descendants in Eurasia, and then expanded out again to replace the Neanderthals. This hypothesis is indeed compatible with all observations, but it requires more population movements and population extinctions than the simpler assumption that Neanderthals had originated in Africa. So the Asian-origin scenario is less parsimonious, and therefore inferior as an explanation, than the African-origin scenario. Thus we noted the African substructure scenario as a possible explanation for our data but regarded it as unlikely, since there was a simpler, more obvious explanation that was in fact so obvious that several of us had thought of it independently: the Middle Eastern scenario.
Chapter 19
The Replacement Crowd
__________________________________
The earliest remains of modern humans yet found outside Africa have been discovered in the Carmel mountain range in Israel. Bones found in two caves named Skhul and Qafzeh are more than 100,000 years old. And at two other sites just a few kilometers away, Tabun Cave and Kebara Cave, skeletons of Neanderthals that are about 45,000 years old have been found. These finds don’t necessarily mean that Neanderthals and modern humans lived side by side in the Carmel Mountains for over 50,000 years; in fact, many paleontologists suggest that modern humans from the south lived in the area when the climate was warmer and that Neanderthals from the north moved in, and the modern humans moved out, during colder periods. It has also been suggested that the modern humans from Skhul and Qafzeh died out without leaving any descendants. But even if they did not leave descendants, they probably had relatives. And even if they weren’t constant neighbors, the two groups must have made contact, over periods of thousands of years, even as changes in climate might have shifted the zone of contact, sometimes north, sometimes south. That, in brief, is the Middle East scenario.
The Middle East, as I learned from talking to paleontologists, particularly Jean-Jacques Hublin, the French scientist who joined our institute as the director of the department of human evolution in 2004, was an attractive place for modern humans and Neanderthals to have mixed 50,000 to 100,000 years ago. One reason is that it is the only area in the world where we know that Neanderthals and modern humans were, at least potentially, in contact for such a long time. Another reason is that neither of the two groups appears to have been clearly dominant during that period. For example, the stone tools used by both groups were the same. In fact, since their tool kits are identical, the only way one can know for certain whether a Middle Eastern archaeological site from that time period was occupied by Neanderthals or modern humans is if skeletal remains are present.
This all changed shortly after 50,000 years ago. At that time, modern humans firmly established themselves outside of Africa and began to rapidly fan out over the Old World, reaching Australia in just a few thousand years. By then, the way in which they interacted with Neanderthals seems to have changed. In Europe, where the fossil record is particularly well studied, it seems clear that when modern humans appeared in an area, Neanderthals disappeared either immediately or shortly thereafter. The same eventually happened all around the world: wherever modern humans appeared, earlier forms of humans, sooner or later, disappeared.
To distinguish these expansive and ambitious modern humans from the modern humans who were hanging around in Africa and the Middle East between 100,000 and 50,000 years ago, I like to call them the “replacement crowd.” They had developed a more sophisticated tool culture, called the Aurignacian by archaeologists, characterized by flint tools of different sorts, including a variety of blades. Points for spears and arrows made of bone are often found at Aurignacian sites, representing what some archaeologists believe to be the first instances of projectile weapons. If true, this invention, which for the first time allowed humans to kill animals and enemies at a distance, may in itself have tipped the balance in their favor when they met Neanderthals and other earlier forms of humans. The Aurignacian culture also produced the first cave art and the first figurines of animals, including mythical figures that are half-human and half-animal, suggesting that they possessed a rich inner life that they wanted to communicate with others in their group. The “replacement crowd” thus exhibited behaviors that were only occasionally or not at all seen among Neanderthals and among the earlier modern humans from Skhul and Qafzeh.
We don’t know where the “replacement crowd” came from. In fact, they could even have been the descendants of the same humans who had already been living in the Middle East, simply accumulating the cultural inventions and proclivities that enabled “replacement,” but it is more likely that they came from somewhere in Africa. In any event, the “replacement crowd” must have spent time in the Middle East.
As the replacement crowd moved into the Middle East they may well have incorporated the modern humans already there into their groups. Those humans, in turn, may have already mated with Neanderthals, so that the Neanderthal DNA passed through them into the replacement crowd, and then on to us today. Such a model may seem more complex, and thus less parsimonious, than is ideal. A major problem facing a direct model, where the replacement crowd mated with Neanderthals, is the question of why, if they were willing to rear children with Neanderthals in the Middle East, the replacement crowd didn’t also do so later when they met and replaced them in central and western Europe. If they had, Europeans should have more Neanderthal DNA than Asians. What this indirect scenario suggests is that perhaps the replacement crowd never mixed with Neanderthals but instead received the Neanderthal contribution through these other modern humans, those whose remains have been found in Skhul and Qafzeh. These very early modern humans, having a culture very similar to Neanderthals and having lived next to them for tens of thousands of years, may have been more inclined to mate with Neanderthals than to “replace” them.
This indirect model is admittedly pure speculation. It may be that we don’t see an additional contribution in Europe because it is too small for us to detect. It may also be that the interbreeding in the Middle East was followed by a particularly large growth of the population that had mixed with the Neanderthals. If so, this would mean that we are particularly able to detect that event because of the “surfing” that Excoffier had described, and less able to detect later events that were not followed by any equally large population expansion. Or perhaps there was later migration from Africa into Europe that “diluted” the extra Neanderthal contribution in Europe. I hope that direct evidence in the future can address this. Should it prove possible to study the DNA from the Skhul and Qafzeh people, then it will be possible to see if they had mixed with the Neanderthals, perhaps to a large extent. It should then also become possible to see if they carried the very same Neanderthal DNA fragments that today exist in people in Europe and Asia.
For the moment, the simples
t—the most parsimonious—scenario is that the replacement crowd met and mated with Neanderthals somewhere in the Middle East and raised the children who resulted from those unions. Those children, part modern human, part Neanderthal, became members of the replacement crowd, carrying the DNA of the Neanderthals and their descendants as an internal fossil of sorts. Today, those internal Neanderthal remains have reached even the southern tip of South America, Tierra del Fuego, and Easter Island in the middle of the Pacific. The Neanderthals live on in many of us today.
When we had come this far in our analyses, I began to worry about what the social implications of our findings might be. Of course, scientists need to communicate the truth to the public, but I feel that they should do so in ways that minimize the chance for it to be misused. This is especially the case when it comes to human history and human genetic variation, when we need to ask ourselves: Do our findings feed into prejudices that exist in society? Can our findings be misrepresented to serve racists’ purposes? Can they be deliberately or unintentionally misused in some other way?
I could imagine a few scenarios. It is generally not a compliment to be called a “Neanderthal,” and I wondered if some individuals might link Neanderthal DNA with aggressiveness or other behaviors associated with the colonialist expansion out of Europe. I didn’t see this as too much of a threat, however, as such “reverse racism” against Europeans would most likely not be very virulent. A more serious issue was what it meant that Africans lacked this component. Were they not part of the “replacement crowd”? Was their history somehow fundamentally different?
