Who We Are and How We Got Here

by David Reich

  Ancient DNA from the Southwest Pacific has continued to produce unexpected findings. When we and Johannes Krause’s laboratory, working independently, analyzed the Papuan ancestry in Vanuatu, we found that it was more closely related to that in groups currently living in the Bismarck Islands near New Guinea than to groups currently living in the Solomon Islands—despite the fact that the Solomon Islands are directly along the ocean sailing path to Vanuatu.52 We also found that the Papuan ancestry present in remote Polynesian islands is not consistent with coming from the same source as that in Vanuatu. Thus there must have been not one, not two, but at least three major migrations into the open Pacific, with the first migration bringing East Asian ancestry and the Lapita pottery culture, and the later migrations bringing at least two different types of Papuan ancestry. So instead of a simple story, the spread of humans into the open Pacific was highly complex.

  Figure 24. Ancient DNA shows that the first people of the southwest Pacific islands had none of the Papuan ancestry ubiquitous in the region today and that first arrived in New Guinea after fifty thousand years ago (top). The pioneer migrants had almost entirely East Asian ancestry (middle), and multiple later streams of migration brought primarily Papuan ancestry (bottom).

  Can we ever hope to reconstruct the details of these migrations? There is every reason to be hopeful. Our picture of how the present-day populations of the Pacific islands formed is becoming increasingly clear thanks to access to ancient DNA from the region, and the fact that some islands have less complex population histories than mainland groups because of their isolation, permitting easier reconstruction of what occurred. Through genome-wide studies of modern and ancient populations, we will soon have a far more accurate picture of how humans moved through this vast region.

  But right now our understanding of what happened in mainland East Asia remains murky and limited. The extraordinary expansion of the Han over the last two thousand years has added one more level of massive mixing to the already complex population structure that must have been established after thousands of years of agriculture in the region, and after the rise and fall of various Stone Age, Copper Age, Bronze Age, and Iron Age groups. This means that any attempt to reconstruct the deep population history of East Asia based on patterns of variation in present-day people must be viewed with great caution.

  But as I write this chapter, the tsunami of the ancient DNA revolution is cresting, and it will shortly crash on East Asian shores. State-of-the-art ancient DNA laboratories have been founded in China, and are turning their powers to investigating collections of skeletal material that have been assembled over decades. Ancient DNA studies of these and other skeletons will reconstruct how the peoples of each ancient mainland East Asian culture relate to each other and to people living today. Our understanding of the deep interrelationships of the ancestral populations of East Asians, and of movements of people since the end of the last ice age, will soon be as clear as our understanding of what happened in Europe.

  But it is hard to predict what ancient DNA studies in East Asia will show. While we are beginning to have a relatively good idea of what happened in Europe, Europe does not provide a good road map for what to expect for East Asia because it was peripheral to some of the great economic and technological advances of the last ten thousand years, whereas China was at the center of changes like the local invention of agriculture. What this means is that while we can be sure that the findings from ancient DNA studies in East Asia will be illuminating, we do not yet know what they will be. All we can be sure of is that ancient DNA studies will change our understanding of the human past in this most populous part of the world.

  9

  Rejoining Africa to the Human Story

  A New Perspective on Our African Homeland

  The recognition that Africa is central to the human story has, paradoxically, distracted attention from the last fifty thousand years of its prehistory. The intensive study of what happened in Africa before fifty thousand years ago is motivated by a universal recognition of the importance of the Middle to Later Stone Age transition in Africa and the Middle to Upper Paleolithic transition at the doorstep of Africa, those great leaps forward in recognizably modern human behavior attested to in the archaeological record. However, scholars have shown limited interest in Africa after this period. When I go to talks, a common slip of the tongue is that “we left Africa,” as if the protagonists of the modern human story must be followed to Eurasia. The mistaken impression is that once Africa gave birth to the ancestral population of non-Africans, the African story ended, and the people who remained on the continent were static relics of the past, jettisoned from the main plot, unchanging over the last fifty thousand years.

  The contrast between the richness of the information we currently have about the human story in Eurasia over the last fifty thousand years and the dearth of information about Africa over the same period is extraordinary. In Europe, where most of the research has been done, archaeologists have documented a detailed series of cultural transformations: from Neanderthals to pre-Aurignacian modern humans, to Aurignacians, to Gravettians, to the people who practiced Mesolithic culture, and then to Stone Age farmers and their successors in the Copper, Bronze, and Iron ages. The ancient DNA revolution—which has disproportionately sampled bones from Eurasia and especially from Europe—has further widened the gap in our understanding of the prehistory of Africa compared to that of Eurasia.

  But of course, what all investigations that scratch below the surface show is that the people “left behind” in Africa changed just as much as the descendants of the people who emigrated. The main reason we don’t know as much about the modern human story in Africa is lack of research. Human history over the last tens of thousands of years in Africa is an integral part of the story of our species. Focusing on Africa as the place where our species originated, while it might seem to highlight the importance of Africa, paradoxically does Africa a disservice by drawing attention away from the question of how populations that remained in Africa got to be the way they are today. With ancient and modern DNA, we can rectify this.

  The Deep Mixture That Formed Modern Humans

  In 2012, Sarah Tishkoff and her colleagues studied the biological impact of archaic admixture on the genomes of present-day Africans without access to ancient genomes like those of Neanderthals and Denisovans that had been used to document interbreeding between archaic and modern humans in Eurasia.1

  Tishkoff and her colleagues sequenced genomes from some of the most diverse populations of Africa and analyzed their data to search for a pattern that is predicted when there has been interbreeding with archaic humans: very long stretches of DNA that have a high density of differences compared to the great majority of other genomes, consistent with an origin in a highly divergent population that was isolated until recently from modern humans.2 When they applied this approach to present-day non-Africans, they pulled out stretches of DNA that they found were nearly exact matches to the Neanderthal sequence. Tishkoff and her colleagues also found long stretches of deeply divergent sequences in present-day Africans whose ancestors did not mix with Neanderthals. Since Neanderthals have contributed little if any ancestry to Africans, this was likely to have been the result of mixture with mystery African archaic humans—ghost populations whose genomes have not yet been sequenced.

  Jeffrey Wall and Michael Hammer, using the same types of genetic signatures, attempted to learn something about the relationship of the archaic populations to present-day Africans.3 They estimated that the archaic population separated from the ancestors of present-day humans in Africa about seven hundred thousand years ago and remixed around thirty-five thousand years ago, contributing about 2 percent of the ancestry of some present-day African populations. However, it is important to view these dates and estimated proportions of mixture with caution because of uncertainties about the rate at which mutations occur in humans and because of the limited amount of data Wall and Hammer analyzed.

&nbs
p; The possibility of admixture between modern and archaic humans in sub-Saharan Africa is exciting, and there are even human remains from West Africa dating to as late as eleven thousand years ago with archaic features, providing skeletal evidence in support of the idea that archaic and modern human populations coexisted in Africa until relatively recently.4 Thus, there were ample opportunities for interbreeding with archaic humans as modern humans expanded in Africa, just as in Eurasia.

  If the proportion of admixture with archaic African humans was only around 2 percent as Wall and Hammer estimated, it is likely to have had only a modest biological effect, similar to the effect of the contribution of Neanderthals and Denisovans to the genetic makeup of present-day people outside Africa. However, this does not rule out the possibility of major mixture events in deep African history. The best evidence for deep mixture of modern human populations in sub-Saharan Africa comes from the frequencies of mutations. One generation after a mutation occurs, it is extremely rare as it is present in only a single person. In subsequent generations, the mutation’s frequency fluctuates upward or downward at random, depending on the number of offspring to which it happens to be transmitted. Most mutations never achieve a substantial frequency, as at some point the few individuals carrying them happen not to transmit them to their children, causing them to fluctuate down to 0 percent frequency and disappear forever.

  The effect of this constant pumping into the population of rare new mutations is that there are expected to be fewer common than rare mutations in a population. The frequencies of mutations that are variable in a population are in fact expected to follow an inverse law, with twice as many mutations that occur at 10 percent frequency as those that occur at 20 percent frequency, and twice as many of these in turn as those that occur at 40 percent frequency.

  My colleague Nick Patterson tested this expectation, focusing on mutations present in a large sample of individuals from the Yoruba group of Nigeria that were also present in the Neanderthal genome.5 Patterson’s focus on mutations present in Neanderthals was clever; he knew that mutations discovered in this way were almost certainly frequent in the common ancestral population of humans and Neanderthals, and by implication in their descendants too. Mathematically, the expectation that such mutations would be common is exactly counterbalanced by the inverse law, with the result that mutations meeting these criteria are expected to be equally distributed across all frequencies.

  But the real data showed a different pattern. When Patterson examined sequences from present-day Yoruba, he observed a greatly elevated rate of mutations both at very high and low frequencies, instead of an equal distribution across all frequencies. This “U-shaped” distribution of mutation frequencies is what would be expected in the case of ancient mixture. After two populations separate, random frequency fluctuation occurs in each population, so that the mutations that fluctuate by chance to 0 percent or 100 percent frequency in one population are by and large not expected to be the same as those that do so in the other population. When the populations then remix, the mutations that rose to extreme frequencies in one population but not in the other would be reintroduced as variable genetic types. This would produce peaks of extra mutation density in the mixed population. The first peak corresponding to mutations that rose to extreme frequencies in the first population is expected to start at the proportion of mixture, while the second peak corresponding to mutations that rose to extreme frequencies in the second population is expected to start at 100 percent minus the proportion of mixture. This is exactly the pattern that Patterson found, and he showed that it could be explained if Yoruba descended from a mixture of two highly differentiated human populations in close to equal proportions.

  Patterson tested whether the patterns he observed were consistent with a model in which only Yoruba descend from this mixture but non-Africans do not. But this was contradicted by the data. Instead, all non-Africans—and even divergent African lineages such as San hunter-gatherers—also seem to be descended from a similar mixture. Thus, although Patterson had begun by looking at West Africans, the mixture event he detected was not specific to that population. Rather, it seemed to be a shared event in the ancestry of present-day humans, suggesting that the mixture may have occurred close to the time when anatomically modern human features first appear in the skeletal record after around three hundred thousand years ago.6

  Patterson’s findings resonated with a discovery from the 2011 study by Heng Li and Richard Durbin (discussed in part I) that reconstructed human population size history from a single person’s genome.7 That study compared the genome sequence a person gets from his or her mother to the sequence he or she gets from his or her father. It found fewer locations in the genome where the reconstructed age of the shared ancestor falls between 400,000 and 150,000 years ago than would be expected if the population had been constant in size.8 One possible explanation for this result is that the ancestral population of all modern humans was very large over this period, which would mean that the probability that any two genomes today share a particular ancestor at this time is small (there being many possible ancestors in each generation). But the other possibility was that the ancestral human population consisted of multiple highly divergent groups instead of a single, freely mixing group, and hence the lineages ancestral to present-day people were isolated in separate populations at this time. This pattern could be reflecting the same mixture event that Patterson had highlighted through his study of mutation frequencies. The reconstructed time corresponds to a period when there is skeletal evidence of archaic human forms overlapping with modern human forms in Africa. For example, the Homo naledi skeletons recently discovered in a cave in South Africa had relatively modern human bodies but brains much smaller than those of modern humans, and date to between 340,000 and 230,000 years ago.9

  There was also a third line of evidence for archaic mixture. A commonly held view is that the San hunter-gatherers of southern Africa largely derive from a lineage that branched off the one leading to all other present-day modern human lineages before they separated from one another.10 If so, the San would be expected to share mutations at exactly the same rate with all non–southern Africans. But Pontus Skoglund in my laboratory showed that the San share more mutations with eastern and central African hunter-gatherers than they do with West African populations like the Yoruba of Nigeria.11 This could be explained if the West African populations harbor more ancestry from one of the early-splitting populations than is the case for non-African populations. Perhaps all present-day humans are a mixture of two highly divergent ancestral groups, with the largest proportion in West Africans, but all populations inheriting DNA from both.

  These results suggest the possibility that major mixture in Africa occurred in the time well before around fifty thousand years ago when modern human behavior burst into full flower in the archaeological record. This mixture wasn’t a minor event, such as the approximately 2 percent Neanderthal admixture in non-Africans or the ghost archaic ancestry in Africans found by Wall and Hammer. Because this mixture was closer to 50/50, it is not even clear which one of the source populations should properly be considered archaic and which modern. Perhaps neither was modern, or neither was archaic. Perhaps the mixture itself was essential to forging modern humans, bringing together biological traits from the two mixing populations and combining them in new ways that were advantageous to the newly formed populations.

  Figure 25. The deep relationships among present-day modern human lineages are far from simple. One model that genomic findings suggest is that the oldest modern human split in Africa led to a lineage represented in highest proportion in West Africa, a split that must have occurred before three hundred thousand to two hundred thousand years ago, the date of the separation of ancient East and South African foragers. An expansion of modern humans associated with the Later Stone Age and Upper Paleolithic transitions after around fifty thousand years ago could then have connected all populations in Africa.

  How Ag
riculture Threw a Veil over Africa’s Past

  How can we begin to learn what happened in Africa after the ancestral population of modern humans was forged, and also after the ancestors of present-day non-Africans spread out of Africa and the Near East beginning around fifty thousand years ago? There is a lot of information to work with, as African genome sequences are typically about a third more diverse than non-African ones. Human diversity in Africa is extraordinary not only within but also across populations, as some pairs of African populations have been isolated for up to four times longer than any pairs of populations outside the continent, as reflected by the fact that for some pairs of populations—like San hunter-gatherers from southern Africa and Yoruba from West Africa—the minimum density of mutations separating their genomes is that much greater than that for any pair of genomes outside Africa.12

  But learning about Africans’ deep past from today’s populations is extremely challenging because while much of the ancient variation still exists in people living today, it is all mixed up. The most recent mixing of populations occurred in the last few thousand years due to at least four great expansions, all of which are associated with the spread of language groups, and most of which have been driven by the spread of agriculturalists.13 These expansions have thrown a veil over the African past, moving populations thousands of kilometers from their places of origin, where they displaced or mixed with populations that were widespread before. In this respect, the study of African populations is no different from the study of Eurasian ones, which have also turned over in the last several thousand years.