by David Reich
What new information does genetics add? Our 2012 study found that the Na-Dene-speaking Chipewyan carry a type of ancestry not shared with many other Native Americans, providing evidence for the later Asian migration theory.44 We estimated that this ancestry constituted only around 10 percent of Chipewyan ancestry, but it was striking all the same. We wondered whether we could use this distinctive strain of ancestry in the Chipewyans as a tracer dye to document an ancestral link between Na-Dene speakers like Chipewyans and individuals from past archaeological cultures who could be studied with ancient DNA.
In 2010, Eske Willerslev and colleagues published genome-wide data from an approximately four-thousand-year-old lock of hair taken from a frozen individual of the Saqqaq culture, the first human culture of Greenland.45 Their analysis showed that this man belonged to a population that had a distinct blend of ancestry compared both to First Americans in the south and the Eskimo-Aleuts who followed them in the Arctic. Willerslev’s group expanded its claim in 2014 when it reported data from several additional “Paleo-Eskimos,” as people who preceded Eskimo-Aleuts are called by archaeologists.46 All these individuals were broadly related, and the authors argued that they represented a distinct migration from Asia that was different from all prior and subsequent ones. They argued that the Paleo-Eskimos largely went extinct without leaving descendants after the arrival of Eskimo-Aleut speakers around fifteen hundred years ago.
In our 2012 study, we tested the idea that the Paleo-Eskimos exemplified by the Saqqaq individual were descended from a distinct migration to the Americas. To our surprise, we found no statistical evidence for a distinct migration. Instead, our tests were consistent with the possibility that the Saqqaq derived their ancestry from the same source that contributed to the Na-Dene-speaking Chipewyans, just in different proportions. Since we know from genetic data that only around 10 percent of the ancestry of many Na-Dene speakers today is from this late Asian migration, it is easy to understand why the clustering analysis used by Willerslev’s team missed the connection to Na-Dene speakers. We proposed that the Na-Dene and Saqqaq might both derive part of their ancestry from the same ancient migration from Asia to the Americas.
In 2017, Pavel Flegontov, Stephan Schiffels, and I confirmed that the Paleo-Eskimo lineage did not die out, and instead lives on in the Na-Dene.47 By examining rare mutations that reflect recent sharing between diverse Native American and Siberian populations, we found evidence for recent common ancestors between the ancient Saqqaq individual and present-day Na-Dene. In fact, the hypothesis that Paleo-Eskimo lineages went extinct after the arrival of Eskimo-Aleut speakers is even more profoundly wrong than I had originally suggested in my 2012 paper.48 The correct way to view the ancestry of present-day speakers of Eskimo-Aleut languages is as a mixture of lineages related to Paleo-Eskimos and First Americans. In other words, far from being extinct, the population that included Paleo-Eskimos lives on in mixed form not just in Na-Dene speakers, but also in Eskimo-Aleut speakers.
Our 2017 work also revealed an entirely new and unifying way to view the deep ancestry of the peoples of the Americas. In this new vision, there were just two ancestral lineages that contributed all Native American ancestry apart from that in Population Y: the First Americans and the population that brought new small stone tools and the first archery equipment to the Americas around five thousand years ago and founded the Paleo-Eskimos.49 We could show this because, mathematically, we can fit a model to the data in which all Native Americans excluding Amazonians with their Population Y ancestry can be described as mixtures of two ancestral populations related differentially to Asians. Mixtures of these two ancestral populations produced the three source populations that migrated from Asia to America and that are associated with Eskimo-Aleut languages, Na-Dene, and all other languages.
A second genetic revelation about Native American population history is clearest in the Chukchi, a population of far northeastern Siberia that speaks a language unrelated to any spoken in the Americas. My analyses revealed that the Chukchi harbor around 40 percent First American ancestry due to backflow from America to Asia.50 For those who are dubious about the idea that descendants of First Americans could have reexpanded out of America and then made a substantial demographic impact on Asia—who are used to thinking about the migratory path between Asia and America as a one-way street—it might be tempting to argue that the genetic affinity of the Chukchi to Native Americans simply reflects that they are the closest cousins of the First Americans in Asia. This bias also impeded my own thinking for more than a year as I tried to make sense of the data we had from diverse Native Americans. But the genetic data clarify that the affinity is due to back-migration, as the Chukchi are more closely related to some populations of entirely First American ancestry than to others, a finding that can only be explained if a sublineage of First Americans that originated well after the initial diversification of First American lineages in North America migrated back to Asia. The explanation for this observation is that the Eskimo-Aleut speakers who established themselves in North America mixed heavily with local Native Americans (who contributed about half their ancestry) and then took their successful way of life back through the Arctic with them to Siberia, contributing not only to the Chukchi but also to local speakers of Eskimo-Aleut languages. The identification of a reflux of First American ancestry into Asia—a type of finding that is difficult to prove with archaeology—is the kind of surprise that genetics is in a unique position to deliver.
A third example of what genetics can offer is the story of the arrival of agriculture to the U.S. Southwest from northern Mexico. Today, these regions are linked by a widespread language family called Uto-Aztecan, which linguists have traditionally viewed as having spread from north to south, based on the fact that most of the languages in this group and some plant names that are shared across the languages are typical of the northern end of the present-day Uto-Aztecan distribution. However, others have argued that the languages radiated northward from Mexico, following the spread of maize agriculture. It has been suggested, most forcefully by the archaeologist Peter Bellwood, that languages and peoples tend to move with the spread of agriculture.51 Studying the ancient DNA of people who lived before and after the arrival of maize in the region, along with comparison to the present-day inhabitants, can test this theory at least in part. We are beginning to find some clues in ancient DNA. Studies of ancient maize have now shown that this crop first entered the U.S. Southwest by a highland route (inland, over hills) more than four thousand years ago, and then was replaced by strains of maize of a lowland coastal origin around two thousand years ago.52 This is a remarkable example of how plants, too, have had histories of migration and recurrent mixture, although in the case of domesticated crops the migrations and mixtures are if anything likely to be more dramatic because humans have subjected crops to artificial selection. It will only be a matter of time before we are able to test whether new peoples moved with the new crops.
The dream, of course, is to carry out studies like these more systematically. Modern genetic studies and ancient DNA enable us to discover how Native American cultures are connected by links of migration, and how the spread of languages and technologies corresponded to ancient population movements. Many of these stories have been lost because of the European exploitation that has decimated Native American populations and their culture. Genetics offers the opportunity to rediscover lost stories, and has the potential to promote not just understanding but also healing.
8
The Genomic Origins of East Asians
The Failure of the Southern Route
East Asia—the vast region encompassing China, Japan, and Southeast Asia—is one of the great theaters of human evolution. It harbors more than one third of the world’s population and a similar fraction of its language diversity. Pottery was first invented there at least nineteen thousand years ago.1 It was the jumping-off point for the peopling of the Americas before fifteen thousand years ago. East Asia witnes
sed an independent and early invention of agriculture around nine thousand years ago.
East Asia has been home to the human family for at least around 1.7 million years, the date of the oldest known Homo erectus skeleton found in China.2 The earliest human remains excavated in Indonesia are similarly old.3 Archaic humans—whose skeletal form is not the same as that of humans whose anatomically modern features begin to appear in the African fossil record after around three hundred thousand years ago4—have lived in East Asia continuously since those times. For example, genetic evidence shows that the Denisovans mixed with ancestors of present-day Australians and New Guineans shortly after fifty thousand years ago. And archaeological and skeletal evidence shows that the one-meter-tall “Hobbits” also persisted until around this same time on Flores island in Indonesia.5
There has been intense debate about the extent to which the archaic humans of East Asia contributed genetically to people living today. Chinese and Western geneticists nearly all agree that present-day humans outside of Africa descend from a dispersal after around fifty thousand years ago, which largely displaced previously established human groups.6 Some Chinese anthropologists and archaeologists, on the other hand, have documented similarities in skeletal features and stone tool styles in people who lived in East Asia before and after this time, raising the question of whether there has been some degree of continuity.7 At the time of this writing, our knowledge of East Asian population history is relatively limited compared to that of West Eurasia because less than 5 percent of published ancient DNA data comes from East Asia. The difference reflects the fact that ancient DNA technology was invented in Europe, and it is nearly impossible for researchers to export samples from China and Japan because of government restrictions or a preference that studies be led by local scientists. This has meant that these regions have missed out on the first few years of the ancient DNA revolution.
In the west, the grand narrative is that sometime after around fifty thousand years ago, modern humans began making sophisticated Upper Paleolithic stone tools, which are characterized by narrow stone blades struck in a new way from pre-prepared cores. The Near East is the earliest known site of Upper Paleolithic stone tools, and this technology spread rapidly to Europe and northern Eurasia. It would be natural to expect, given how successful the people who made Upper Paleolithic technology were, that this know-how would have overspread East Asia too. But that is not what happened.
The archaeological pattern in the east does not conform to that in the west. Around forty thousand years ago and across a vast tract of land in China and east of India there is indeed archaeological evidence of great behavioral change associated with the arrival of modern humans, including the use of sophisticated bone tools, shell beads or perforated teeth for body decoration, and the world’s earliest known cave art.8 In Australia, archaeological evidence of human campsites makes it clear that modern humans arrived there at least by about forty-seven thousand years ago,9 which is about as old as the earliest evidence for modern humans in Europe.10 So it is absolutely clear that modern humans arrived in East Asia and Australia around the same time as they came to Europe. But, puzzlingly, the first modern humans in central and southern East Asia, and those in Australia, did not use Upper Paleolithic stone tools. Instead, they used other technologies, some of which were more similar to those used by modern humans in Africa tens of thousands of years earlier.11
Prompted by these observations, the archaeologists Marta Mirazon Lahr and Robert Foley argued that the first humans in Australia might derive from a migration of modern humans out of Africa and the Near East prior to the development of Upper Paleolithic technology in the west. According to this “Southern Route” hypothesis, the migrants left Africa well before fifty thousand years ago and skirted along the coast of the Indian Ocean, leaving descendants today among the indigenous people of Australia, New Guinea, the Philippines, Malaysia, and the Andaman Islands.12 The anthropologist Katerina Harvati and colleagues also documented skeletal similarities between Australian Aborigines and Africans that, they argued, provide evidence for this model.13
The Southern Route hypothesis was far more than a claim that there were modern humans outside of Africa well before fifty thousand years ago—a fact that every serious scholar now accepts.14 Evidence of early modern humans outside of Africa well before fifty thousand years ago includes the morphologically modern skeletons in Skhul and Qafzeh in present-day Israel that date to between around 130,000 to 100,000 years ago.15 Stone tools found at the site of Jebel Faya from around 130,000 years ago are similar to ones found in northeast Africa from around the same time, suggesting that modern humans made an early crossing of the Red Sea into Arabia.16 There is also tentative genetic evidence of an early impact of modern humans outside Africa, with Neanderthal genomes harboring a couple of percent of ancestry that may derive from interbreeding with a modern human lineage that separated a couple of hundred thousand years ago from present-day human lineages, as expected if a modern human population possibly related to that in Skhul and Qafzeh interbred with Neanderthal ancestors.17 Although many geneticists, including me, are still on the fence about whether this finding of earlier interbreeding between modern humans and Neanderthals is compelling, the key point is that almost all scholars now agree that there were early dispersals of modern humans into Asia that preceded the widely accepted dispersals after fifty thousand years ago that contributed in a major way to all present-day non-Africans. The outstanding question raised by the Southern Route hypothesis is not whether such expansions occurred, but whether they had an important long-term impact on humans living today.
In 2011, Eske Willerslev led a study that seemed to show that the early expansions indeed left an impact.18 He and his colleagues reported a Four Population Test showing that Europeans share more mutations with East Asians than with Aboriginal Australians, as would be expected from a Southern Route contribution to the lineage of Australians. Applying a Southern Route migration model to the genomic data, they estimated that Australian Aborigines harbor ancestry from a modern human population that split from present-day Europeans at twice the time depth that East Asian ancestors split from Europeans (seventy-five thousand to sixty-two thousand years ago versus thirty-eight thousand to twenty-five thousand years ago).
There was a problem, though, which is that the analysis did not account for the 3 to 6 percent of ancestry that Australians inherited from archaic Denisovans.19 Because Denisovans were so divergent from modern humans, mixture from them could cause Europeans to share more mutations with Chinese than with Australian Aborigines. Indeed, this explained the findings. My laboratory showed that after accounting for Denisovan mixture, Europeans do not share more mutations with Chinese than with Australians, and so Chinese and Australians derive almost all their ancestry from a homogeneous population whose ancestors separated earlier from the ancestors of Europeans.20 This revealed that a series of major population splits in the history of non-Africans occurred in an exceptionally short time span—beginning with the separation of the lineages leading to West Eurasians and East Eurasians, and ending with the split of the ancestors of Australian Aborigines from the ancestors of many mainland East Eurasians. These population splits all occurred after the time when Neanderthals interbred with the ancestors of non-Africans fifty-four to forty-nine thousand years ago, and before the time when Denisovans and the ancestors of Australians mixed, genetically estimated to be 12 percent more recent than the Neanderthal/modern human admixture, that is, forty-nine to forty-four thousand years ago.21
Figure 22. Two major splits were sandwiched in an approximately five-thousand-year period between the Neanderthal and Denisovan interbreeding events.
The rapid succession of lineage separations during the relatively short interval between Neanderthal and Denisovan interbreeding with modern humans suggests that throughout Eurasia, modern humans were moving into new environments where their technology or lifestyle allowed them to expand, displacing the previously residen
t groups. The spread was so fast that it is hard to imagine that archaic humans who had already been resident there for close to two million years, and who we know were also there when modern humans expanded based on the evidence of interbreeding with Denisovans, put up much resistance. Even if early modern humans expanded into East Asia via a Southern Route, they were likely also replaced by later waves of human migrants and can be ruled out as having contributed more than a very small percentage of the ancestry of present-day people.22 In East Asia as in West Eurasia, the expansion of modern humans out of Africa and the Near East had an effect akin to the erasing of a blackboard, creating a blank slate for the new people. The old populations of Eurasia collapsed, and in their place came new groups that swiftly inhabited the landscape. There is no genetic evidence of any substantial ancestry from these earlier populations in present East Asians.23
So if essentially all modern human ancestry in East Asia and Australia today derives from the same group that contributed to West Eurasians, what explains how Southeast Asians and Australians missed out on the Upper Paleolithic technology that is so tightly linked with the spread of modern human populations into the Near East and Europe?
The first long-bladed stone tools characteristic of Upper Paleolithic technology in the archaeological record date to between fifty thousand and forty-six thousand years ago.24 But genetically, the split of the lineages leading to West Eurasians and East Asians may have been more ancient since, as I have discussed, it almost certainly occurred within a few thousand years after the admixture of modern humans with Neanderthals fifty-four thousand to forty-nine thousand years ago. So the main split of West Eurasian and East Asian ancestors could have occurred before the development of Upper Paleolithic technology, and the geographic distribution of this technology could just reflect the spread of the population that invented it.
