Long-standing dispute about North American prehistory

Pavel Flegontov,CSc. (University of Ostrava, Czech Republic) and his team have recently published an article in Nature that intends to put an end to a long-standing dispute over North American prehistory combining data and knowledge from paleogenetics and linguistics.

Paleogenetics is a rapidly developing scientific discipline at the junction of archeology and genetics. Due to the rapid progress in the methods of sequencing DNA extracted from ancient bones, as well as in the methods of genetic data analysis, archeogenetics is becoming an integral component of research in human prehistory. However, the study of relatively recent history (the last 5 thousand years) by methods of archeogenetics is, oddly enough, methodologically difficult, despite the abundance of bone samples and their usually good preservation. As population density increased and means of transportation developed, mobility of people increased. And while in the long millennia of the Paleolithic a dominant pattern was genetic isolation of small groups of hunters, then from the beginning of the Neolithic migration and population micture became increasingly common. Thus, in order to clarify genetic history of virtually any region, it is necessary to unravel a very complex network of population splits and mixtures, i.e. a graph.

An example of such a complex region is Chukotka and the American Arctic - the vast expanses of tundra and Arctic desert, inhabited by sparse groups of Chukchi, Eskimos, Aleuts, and Inuit. For the first time the tundra zone of Alaska, the Canadian Arctic islands and Greenland was populated by so-called Paleo-Eskimos. This process began about 5,000 years ago with a migration of a small group of caribou, muskox and seal hunters across the Bering Strait. Then a succession of several archaeological cultures culminated in modern Eskimos, Aleuts and Inuit. However, archeology very rarely can find whether the change in material culture was accompanied by mass migration and population replacement, or these were primarily cultural processes. Therefore, for decades, there have been controversies about the history of the Arctic peoples, about the relationship of Paleo-Eskimos and Inuit, as well as about the interaction of Paleo-Eskimos and native Americans who occupied the forests of Alaska and Canada adjacent to the tundra.

An article recently published in the journal Nature (Flegontov et al. 2019),is intended to put an end to a long-standing dispute about North American prehistory: about Paleo-Eskimos and speakers of the Na-Dene language family. This study resulted from a collaboration between three groups of geneticists: under the leadership of Pavel Flegontov (University of Ostrava, Czech Republic), Stefan Schiffels (Max Planck Institute in Jena, Germany) and David Reich (Harvard Medical School, Boston, USA), as well as other geneticists, archaeologists and a linguist Edward Vajda. Pavel Flegontov from the University of Ostrava is a lead author and a co-supervisor of the project. Other co-authors affiliated with the University of Ostrava are Ezgi Altinisik, Piya Changmai Ph.D., and Olga Flegontova Ph.D.

Back in 2012, Reich and co-authors (Reich et al. 2012) showed that the genetic data of the Chippewa people from Canada cannot be explained by descent from a single Asian source common to all Americans, and that there was a second Asian source close to a recently sequenced 4000-years-old individual of the Saqqaq archaeological culture from Greenland (Rasmussen et al. 2010). The same ancestral source, according to Reich’s group (Reich et al. 2012), contributed approximately 50% to the gene pool of the peoples of the Eskimo-Aleut language family (Aleuts, Inuit, Eskimos).

Chippewa speak a language belonging to the Athabaskan sub-group of the Na-Dene language family, and the Saqqaq individual, about 3900 years old, is a typical representative of Paleo-Eskimo hunters who dominated the American tundra from 5000 to about 700 years ago, and then were replaced by the ancestors of modern Inuit and Eskimos. Joseph Greenberg, on the basis of his largely controversial classification of language families (Greenberg et al. 1986), put forward a hypothesis about three waves of American settlement: 1) “first Americans” that gave rise to most American Indians, 2) the peoples who speak Na-Dene languages, 3) Inuit, Eskimos, and Aleuts. According to archeology, the second wave of American settlement is represented by Paleo-Eskimos, which penetrated into Alaska from Chukotka about 5,000 years ago, and the third wave - again by the Eskimos and Inuit (Potter 2010). Greenberg's follower Merritt Ruhlen developed the ideas of earlier authors about the kinship of the Ket language in Siberia and the languages of the Na-Dene family (Ruhlen 1998), and Edward Vaida in 2010 published a large-scale comparative analysis of the Yeniseian family of languages (Ket and a number of extinct languages) and the Na-Dene family, also having come to the conclusion about their relationship (Vajda 2010).

In the same year 2010, paleogenetics appeared on the scene. It is noteworthy that the first sequenced genome of the modern human sub-species was precisely that of the Saqqaq culture individual from Greenland. In this pioneering work of Eske Willerslev’s group (Copenhagen, Denmark) DNA was isolated from a lock of hair found in permafrost, and using fairly simple methods of genetic analysis it was shown that Paleo-Eskimos are related to modern Chukchi and Koryaks inhabiting Chukotka and Kamchatka. For some reason, human remains are extremely rare at Paleo-Eskimo campsites, and so far this Saqqaq individual remains the most deeply sequenced representative of Paleo-Eskimos. An important disadvantage of most popular historical genetic methods is that they do not say much about phylogeny, i.e. the order of separation and mixing of populations. In 2012, Reich’s group published a so-called admixture graph method, which allows building a tree of populations and adding “admixture events” onto it, and immediately applied it to analyze the possible contribution of Paleo-Eskimos to the Native American gene pool (see above).

In 2014 and 2015, Eske Willerslev’s group started a controversy with Reich’s group, publishing two papers in the journal Science (Raghavan et al. 2014, 2015), in which it was concluded that Paleo-Eskimos and Na-Dene peoples are not genetically linked, that instead the Inuit admixed with the ancestors of the Na-Dene speakers, and that they might also have mixed with Paleo-Eskimos in Alaska or Chukotka. In 2018, Eske Willerslev’s group continued to explore the history of Paleo-Eskimos, this time actively using graph methods (Moreno-Mayar et al. 2018). The authors constructed a graph explaining the Athabaskan gene pool as a mixture of Native Americans and a group related to the Koryaks (indigenous people of Kamchatka), and the Inuit gene pool as a mixture of Paleo-Eskimos and Native Americans. Thus, according to the latest results by Willerslev’s group, the Koryaks (speakers of Chukotko-Kamchatkan languages) are grouped with the ancestors of Na-Dene, and the Paleo-Eskimos with the ancestors of the Inuit.

Why are genetic analysis results so contradictory? Pavel Flegontov and co-authors believe that several factors came together here. First, all candidates for the second stream of Asian genes in the Na-Dene American Indians (Koryaks, Paleo-Eskimos, Inuit) differ relatively little genetically and apparently became isolated from each other about 5000 thousand years ago. Second, there may have been multiple migrations of these groups back and forth across the Bering Strait and the concomitant population mixing. Under such conditions — small genetic differences between alternative sources and a complex topology of the graph — standard methods of paleogenetic analysis tend to give unreliable results.

To overcome this problem, Flegontov and co-authors collected a wide arsenal of methods, both standard and novel, and also for the first time obtained genomic data for ancient Aleuts (11 individuals with dates from 280 to 2050 years befoe present, YBP), Athabaskans (3 individuals, about 710 YBP) and ancient Eskimos of Chukotka (23 individuals having average dates from 620 to 1770 YBP). Genomic data were also generated for 8 individuals of the Baikal region and for a Paleo-Eskimo dated to about 1760 YBP. The article also presents genetic data for modern populations: Inuit from Alaska, Kets, Nganasans, Enets, and Selkups from West Siberia (93 individuals in total).

Breakthrough results were obtained using two independent graph methods. Initially, all possible branching orders were tested for a sub-tree consisting of Chukotko-Kamchatkan speakers, Saqqaq, Athabaskans, ancient Aleuts, and modern or ancient Eskimos or Inuit. Three topologies "won", i.e. their probabilities did not differ in a statistically significant way. It turned out that a bi-directional “genetic flow” (i.e. admixture) between the Chukotko-Kamchatkan peoples and Eskimos/Inuit significantly improves model’s fit to the data. On the background of the bidirectional flow, all possible orders of branches were tested again, and this time only one model “won”.

Then, on a set of 133 thousand population combinations, two most likely branching orders were tested: “Willerslev’s topology” (Moreno-Mayar et al. 2018) and the topology obtained in the present paper. Finally, several most likely topologies were tested by an independent method based on rare genetic variants, Rarecoal. Unlike the admixture graph method, Rarecoal allows to date events on the graph. For the article under discussion, this method was significantly improved by Stephan Schiffels. Thus, two independent methods based on graphs did not confirm “Willerslev’s topology”, and this result was also supported using other methods.

So what is the historical model proposed by the authors and how does it fit with linguistics and archeology? In the figure the best topology of the Paleo-Eskimo sub-tree is superimposed on a geographical map. More than 5,000 years ago, Paleo-Eskimo and Chukotko-Kamchatkan populations were separated due to the migration of the former across the Bering Strait. About 4800 years ago, i.e. soon after the migration, Paleo-Eskimos mixed with two groups of “First Americans”. One of these admixture events (30–40% of Paleo-Eskimo ancestry) gave rise to the peoples of the Na-Dene language family. Another (about 50% of Paleo-Eskimo admixture) gave rise to the peoples of the Eskimo-Aleut language family. It is noteworthy that Paleo-Eskimo ancestry was found in all branches of the Na-Dene family, and its level is noticeably lower among neighboring peoples of other language families. Although the genetic results do not allow an unambiguous conclusion about the original homeland of the Na-Dene family (in America or Siberia), we can assume that Paleo-Eskimos mixed with an American population that was a common ancestor of all the Na-Dene peoples. Given a relatively close relationship of Paleo-Eskimos with modern Siberian Kets (Flegontov et al. 2016 and unpublished results by Flegontov’s group), Vajda’s hypothesis about the relationship of the Yeniseian and Na-Dene language families no longer looks incredible.

As for the history of Eskimos and Aleuts, the mixture of Amerindians and Paleo-Eskimos dates back to 4900-4400 YBP. Although we do not have samples of this age from Alaska, combining genetic and archaeological results, we can assume that this event played a key role in the ethnogenesis of the Eskimo-Aleut speakers and occurred in southern Alaska: in the Kodiak archipelago and the Alaskan peninsula. It was there that the earliest communities relying almost entirely on marine resources were found, as well as some features of material culture typical of the later Eskimos and Aleuts (indicated by a brown oval in the figure). Shortly after mixing with Amerindians, the Aleut ancestors probably migrated to the Aleutian Islands (the thin blue arrow pointing down in the figure) and then remained in relative isolation, which well explains the absence of Chukotko-Kamchatkan admixture among Aleuts. The other group probably migrated north towards the Bering Strait and then for some reason returned to Chukotka (the thin blue arrows in the figure), where about 2,200 years ago communities having typical Eskimo-Inuit traits emerged. About 2,000 years ago, a two-way mixture of Eskimos and ancestors of the Chukchi and Koryaks occurred (the purple arrows in the figure). Mastering whale hunting led to a dramatic population growth and expansion of Inuit to Alaska approximately 1150 BC. Then the Inuit spread throughout the American Arctic and ousted Paleo-Eskimos (the thick blue arrow in the figure).

YBP – years before present, i.e. before 1950.

Literature cited

Flegontov, P. et al. Genomic study of the Ket: A Paleo-Eskimo-related ethnic group with significant ancient North Eurasian ancestry. Scientific Reports 6, 20768 (2016).

Flegontov, P. et al. Paleo-Eskimo genetic ancestry and the peopling of Chukotka and North America. Nature (2019).

Moreno-Mayar, J. V. et al. Terminal Pleistocene Alaskan genome reveals first founding population of Native Americans. Nature 553, 203–207 (2018).

Potter, B. A. Archaeological patterning in Northeast Asia and Northwest North America: an examination of the Dene-Yeniseian hypothesis. The Dene-Yeniseian Connection, ed. Kari, J., Potter, B. A. Anthropological Papers of the University of Alaska: New Series 5, 138–167 (2010).

Raghavan, M. et al. The genetic prehistory of the New World Arctic. Science 345, 1255832 (2014).

Raghavan, M. al. Genomic evidence for the Pleistocene and recent population history of Native Americans. Science 349, 1–20 (2015).

Rasmussen, M. et al. Ancient human genome sequence of an extinct Palaeo-Eskimo. Nature 463, 757–762 (2010).

Reich, D. et al. Reconstructing Native American population history. Nature 488, 370–374 (2012).

Ruhlen, M. The origin of the Na-Dene. Proc. Natl. Acad. Sci. USA 95, 13994–13996 (1998).

Vajda, E. J. Siberian link with Na-Dene languages. The Dene-Yeniseian Connection, ed. Kari, J., Potter, B. A. Anthropological Papers of the University of Alaska: New Series 5, 33–99 (2010).

Archaeological and geographical interpretation of the model proposed by Flegontov and co-authors.

a) The “Paleo-Eskimo tree” superimposed on the map. The authors suggested that the contact between Paleo-Eskimos and the ancestors of Na-Dene occurred at the border between the Arctic Small Tools and Northern Archaic archaeological areas in Alaska. Abbreviations: E-A, ancestors of the Eskimo-Aleut speakers; P-E, Paleo-Eskimos; N. First Peoples, American Indians of the northern group. b) A model of the history of the Eskimo-Aleut peoples built on the basis of genetic and archaeological results. The movements of ancestral groups through the Bering Strait to Chukotka and back, as well as the bi-directional admixture in Chukotka between the ancestors of Yupik and Inuit (Old Bering Sea archaeological culture, OBS) and Chukotko-Kamchatkan speakers (C-K) are shown. In both panels, the earliest dates in years before present (YBP) are shown for archaeological cultures and migrations. Some migration paths are shown only approximately.

Updated: 07. 06. 2019