by David Reich
There is a piece of corroborating evidence for the theory that Upper Paleolithic technology developed after the split of the main lineages leading to West Eurasians and East Asians. The Ancient North Eurasians, known earliest from the approximately twenty-four-thousand-year-old remains of the boy from the Mal’ta site in eastern Siberia,25 are on the lineage leading to West Eurasians, which has always been puzzling for geneticists because the Ancient North Eurasians lived geographically closer to East Asia. But it makes sense in light of the geographic distribution of Upper Paleolithic stone tools, which are associated not just with West Eurasians but also North Eurasians and Northeast Asians. Both the distributions of stone tool technology and of genetic ancestry are as expected if Upper Paleolithic technology came into full flower in a population that lived prior to the separation of the lineages leading to Ancient North Eurasians and West Eurasians, but after the separation of the lineage leading to East Asians.
Whatever the reason for the fact that Upper Paleolithic technology never spread to southern East Asia, it is clear from what happened next, and the success these people had in displacing the previously resident populations such as Denisovans, that Upper Paleolithic technology itself was not essential to the successful spread of modern humans into Eurasia after around fifty thousand years ago. It was something more profound than Upper Paleolithic stone tool technology—an inventiveness and adaptability of which the technology was just a manifestation—that allowed these expanding modern humans to prevail everywhere, including in the east.
The Beginnings of Modern East Asia
The first genomic survey of modern East Asian populations was published in 2009, and reported data on nearly two thousand individuals from almost seventy-five populations.26 The authors focused on their finding that human diversity is greater in Southeast Asia than in Northeast Asia. They interpreted this pattern as evidence of a single wave of modern humans reaching Southeast Asia and then spreading from there northward into China and beyond, following a model in which the genetic diversity of present-day populations can be accounted for by a single population moving out of Africa and spreading in all directions, losing genetic diversity as each small pioneer group budded off.27 But we now know that this model is likely to be of limited use. In Europe there have been multiple population replacements and deep mixtures, and we now know from ancient DNA that present-day patterns of diversity in West Eurasia provide a distorted picture of the first modern human migrations into the region.28 The model of a south-to-north migration, losing diversity along the way, is profoundly wrong for East Asia.
In 2015 Chuanchao Wang arrived in my laboratory bearing a treasure: genome-wide data from about four hundred present-day individuals from about forty diverse Chinese populations. China had been sparsely sampled in DNA studies because of regulations limiting the export of biological material. Wang and his colleagues therefore did the genetic work in China, and he brought the data to us electronically. Over the next year and a half, we analyzed these data together with more data from other East Asian countries that had previously been published and with ancient DNA from the Russian Far East generated in our lab. This allowed us to come up with new genetic insights about the deep population history of East Asia and the origins of its current inhabitants.29
By using a principal component analysis, we found that the ancestry of the great majority of East Asians living today can be described by three clusters.
The first cluster is centered on people currently living in the Amur River basin on the boundary between northeastern China and Russia. It includes ancient DNA data that my laboratory and others had obtained from the Amur River basin. So, this region has been inhabited by genetically similar populations for more than eight thousand years.30
The second cluster is located on the Tibetan Plateau, a vast area north of the Himalayas, much of which is at a higher altitude than the tallest of the European Alps.
The third cluster is centered in Southeast Asia, and is most strongly represented by individuals from indigenous populations living on the islands of Hainan and Taiwan off the coast of mainland China.
We used Four Population Test statistics to evaluate models of the possible relationships among present-day populations representing these clusters and Native Americans, Andaman islanders, and New Guineans. The latter three populations have been largely isolated from the ancestors of mainland East Asians at least since the last ice age, and their East Asian–related ancestry effectively serves as ancient DNA from that period.
Our analysis supported a model of population history in which the modern human ancestry of the great majority of mainland East Asians living today derives largely from mixtures—in different proportions—of two lineages that separated very anciently. Members of these two lineages spread in all directions, and their mixture with each other and with some of the populations they encountered transformed the human landscape of East Asia.
The Ghost Populations of the Yangtze and Yellow Rivers
One of the handful of places in the world where farming independently began was China. Archaeological evidence shows that starting around nine thousand years ago, farmers started tilling the windblown sediments near the Yellow River in northern China, growing millet and other crops. Around the same time, in the south near the Yangtze River, a different group of farmers began growing other crops, including rice.31 Yangtze River agriculture expanded along two routes—a land route that reached Vietnam and Thailand beginning around five thousand years ago, and a maritime route that reached the island of Taiwan around the same time. In India and in central Asia, Chinese agriculture collided for the first time with the expansion of agriculture from the Near East. Language patterns also hint at the possibility of movements of people. Today the languages of mainland East Asia comprise at least eleven major families: Sino-Tibetan, Tai-Kadai, Austronesian, Austroasiatic, Hmong-Mien, Japonic, Indo-European, Mongolic, Turkic, Tungusic, and Koreanic. Peter Bellwood has argued that the first six correspond to expansions of East Asian agriculturalists disseminating their languages as they moved.32
What can we say based on the genetics? Because of restrictions on exporting skeletal material from China, the information that genetic data currently provide about the deep population history of East Asia is far behind that of West Eurasia or even of America. Nevertheless, Wang learned what he could based on the little ancient DNA data we had and patterns of variation in present-day people.
We found that in Southeast Asia and Taiwan, there are many populations that derive most or all of their ancestry from a homogeneous ancestral population. Since the locations of these populations strongly overlap with the regions where rice farming expanded from the Yangtze River valley, it is tempting to hypothesize that they descend from the people who developed rice agriculture. We do not yet have ancient DNA from the first farmers of the Yangtze River valley, but my guess is that they will match this reconstructed “Yangtze River Ghost Population,” the name that we have given the population that contributed the overwhelming majority of ancestry to present-day Southeast Asians.
But we found that the Han Chinese—the world’s largest group with a census size of more than 1.2 billion—is not consistent with descending directly from the Yangtze River Ghost population. Instead, the Han also have a large proportion of ancestry from another deeply divergent East Asian lineage. The highest proportions of this other ancestry are found in northern Han, consistent with work since 2009 that has shown that the Han harbor subtle differences along a north-to-south gradient.33 This pattern is as expected from a history in which the ancestors of the Han radiated out from the north and mixed with locals as they spread south.34
What could the other ancestry type be? The Han, who unified China in 202 BCE, are believed based on historical sources to have emerged from the earlier Huaxia tribes who themselves emanated from earlier groups in the Yellow River Valley of northern China. This was one of the two Chinese regions where farming originated, and it is also the place from which fa
rming spread to the eastern Tibetan Plateau beginning around thirty-six hundred years ago.35 Since the Han and Tibetans are also linked by their Sino-Tibetan languages, we wondered whether they might share a distinctive type of ancestry as well.
When Wang built his model of deep East Asian population history, he found that the Han and Tibetans both harbored large proportions of their ancestry from a population that no longer exists in unmixed form and that we could exclude as having contributed ancestry to many Southeast Asian populations. Because of the combined evidence of archaeology, language, and genetics, we called this the “Yellow River Ghost Population,” hypothesizing that it developed agriculture in the north while spreading Sino-Tibetan languages. Ancient DNA from the first farmers of the Yellow River Valley will reveal whether this conjecture is correct. Once available, ancient DNA will also make it possible to learn about features of East Asian population history that are impossible to discern based on analysis only of populations living today, whose deep history has been clouded by many additional layers of migration and mixture.
The Great Admixtures at the East Asian Periphery
Once the core agricultural populations of the Chinese plain—the Yangtze and Yellow River ghost populations—formed, they expanded in all directions, mixing with groups that had arrived in earlier millennia.
Figure 23. Between fifty thousand and ten thousand years ago, hunter-gatherer groups diversified and spread northeast toward the Americas and southeast toward Australia. By nine thousand years ago, two very divergent populations from this initial radiation—one centered on the northern Yellow River and one on the Yangtze River—independently developed agriculture, and then by five thousand years ago spread in all directions. In China, their collision created the gradient of northern and southern ancestry seen in the Han today.
The peoples of the Tibetan Plateau—who harbor a mixture of about two-thirds of their ancestry from the same Yellow River ghost population that contributed to the Han—are one example of this expansion. They likely brought farming for the first time to the region, as well as about one-third of their ancestry from an early branch of East Asians that plausibly corresponds to Tibet’s indigenous hunter-gatherers.36
Another example is the Japanese. For many tens of thousands of years, the Japanese archipelago was dominated by hunter-gatherers, but after around twenty-three hundred years ago, mainland-derived agriculture began to be practiced and was associated with an archaeological culture with clear similarities to contemporary cultures on the Korean peninsula. The genetic data confirm that the spread of farming to the islands was mediated by migration. Modeling present-day Japanese as a mixture of two anciently divergent populations of entirely East Asian origin—one related to present-day Koreans and one related to the Ainu who today are restricted to the northernmost Japanese island and whose DNA is similar to that of pre-farming hunter-gatherers37—Naruya Saitou and colleagues estimated that present-day Japanese have about 80 percent farmer and 20 percent hunter-gatherer ancestry. Relying on the sizes of segments of farmer-related ancestry in present-day Japanese, we and Saitou estimated the average date of mixture to be around sixteen hundred years ago.38 This date is far later than the first arrival of farmers to the region and suggests that after their arrival, it may have taken hundreds of years for social segregation between hunter-gatherers and farmers to break down. The date corresponds to the Kofun period, the first time when many Japanese islands were united under a single rule, perhaps marking the beginnings of the homogeneity that characterizes much of Japan today.
Ancient DNA is also revealing the deep history of humans in mainland Southeast Asia. In 2017, my laboratory extracted DNA from ancient humans at the almost four-thousand-year-old site of Man Bac in Vietnam, where people with skeletons similar in shape to those of Yangtze River agriculturalists and East Asians today were buried side by side with individuals with skeletons more similar to those of the previously resident hunter-gatherers.39 Mark Lipson in my laboratory showed that in ancient Vietnam, all the samples we analyzed were a mixture of an early splitting lineage of East Eurasians and the Yangtze River Ghost Population, with the proportion of the Yangtze River Ghost Population higher in some of the Man Bac farmers we analyzed than in others. The main group of Man Bac farmers also had proportions of ancestry from these two lineages that were similar to those seen in present-day speakers of isolated Austroasiatic languages. These findings are consistent with the theory that Austroasiatic languages were spread by a movement of rice farmers from southern China who interbred with local hunter-gatherers.40 Even today, large Austroasiatic-speaking populations in Cambodia and Vietnam harbor substantial albeit smaller proportions of this hunter-gatherer ancestry.
The genetic impact of the population spread that also dispersed Austroasiatic languages went beyond places where these languages are spoken today. In another study, Lipson showed that in western Indonesia where Austronesian languages are predominant, a substantial share of the ancestry comes from a population that derives from the same lineages as some Austroasiatic speakers on the mainland.41 Lipson’s discovery suggested that Austroasiatic speakers may have come first to western Indonesia, followed by Austronesian speakers with very different ancestry. This might explain why linguists Alexander Adelaar and Roger Blench noticed the presence of Austroasiatic loan words (words with an origin in another language group) in the Austronesian languages spoken on the island of Borneo.42 Alternatively, Lipson’s findings could be explained if Austronesian-speaking farmers took a detour through the mainland, mixing with local Austroasiatic-speaking populations there before spreading farther to western Indonesia.
The most impressive example of the movements of farmers from the East Asian heartland to the periphery is the Austronesian expansion. Today, Austronesian languages are spread across a vast region including hundreds of remote Pacific islands. Archaeological, linguistic, and genetic data taken together have suggested that around five thousand years ago, mainland East Asian farming spread to Taiwan, where the deepest branches of the Austronesian language family are found. These farmers spread southward to the Philippines about four thousand years ago, and farther south around the large island of New Guinea and into the smaller islands to its east.43 At about the time they spread from Taiwan they probably invented outrigger canoes, boats with logs propped on the side that increase their stability in rough waters, making it possible to navigate the open seas. After thirty-three hundred years ago, ancient peoples making pottery in a style called Lapita appeared just to the east of New Guinea and soon afterward started expanding farther into the Pacific, quickly reaching Vanuatu three thousand kilometers from New Guinea. It took only a few hundred more years for them to spread through the western Polynesian islands including Tonga and Samoa, and then, after a long pause lasting until around twelve hundred years ago, they spread to the last habitable Pacific islands of New Zealand, Hawaii, and Easter Island by eight hundred years ago. The Austronesian expansion to the west was equally impressive, reaching Madagascar off the coast of Africa nine thousand kilometers to the west of the Philippines at least thirteen hundred years ago, and explaining why almost all Indonesians today as well as people from Madagascar speak Austronesian languages.44
Mark Lipson in my laboratory identified a genetic tracer dye for the Austronesian expansion—a type of ancestry that is nearly always present in peoples who today speak Austronesian languages. Lipson found that nearly all people who speak these languages harbor at least part of their ancestry from a population that is more closely related to aboriginal Taiwanese than it is to any mainland East Asian population. This supports the theory of an expansion from the region of Taiwan.45
Although there are genetic, linguistic, and archaeological common threads that make a compelling case for the Austronesian expansion, some geneticists balked at the suggestion that the first humans who peopled the remote islands of the Southwest Pacific during the Lapita dispersal were unmixed descendants of farmers from Taiwan.46 How could these migrants have
passed over the region of Papua New Guinea, occupied for more than forty thousand years, while mixing little with its inhabitants? Such a scenario seemed improbable in light of the fact that today, all Pacific islanders east of Papua New Guinea have at least 25 percent Papuan ancestry and up to around 90 percent.47 How could this fit with the prevailing hypothesis that the Lapita archaeological culture was forged during a period of intense exchange between people ultimately originating in the farming center of China (via Taiwan) and New Guineans?
In 2016, ancient DNA struck again, disproving the view that had prevailed until then in the genetic literature. Well-preserved ancient DNA is hard to find in tropical climates like those of the southern Pacific. But the ability to get working DNA from the Pacific changed when, as described earlier, Ron Pinhasi and colleagues showed that DNA from the dense petrous bone of the skull containing the structures of the inner ear sometimes preserves up to one hundred times more DNA than can typically be obtained from other bones.48 We initially struggled to study samples from the Pacific, but when we tried petrous bones, our luck changed.49
We succeeded at getting DNA from ancient people associated with the Lapita pottery culture in the Pacific islands of Vanuatu and Tonga who lived from around three thousand to twenty-five hundred years ago. Far from having substantial proportions of Papuan ancestry, we found that in fact they had little or none.50 This showed that there must have been a later major migration from the New Guinea region into the remote Pacific. The late migration must have begun by at least twenty-four hundred years ago, as all the Vanuatu samples we have analyzed from that time and afterward had at least 90 percent Papuan ancestry.51 How this later wave could have so comprehensively replaced the descendants of the original people who made Lapita pottery and yet retained the languages these people probably spoke remains a mystery. But the genetic data show that this is what happened. This is the kind of result that only genetics can deliver—the definitive documentation that major movements of people occurred. This proof of interaction between highly divergent peoples puts the ball back into the court of archaeologists to explain the nature and effects of those migrations.
