The Stereotype of a Beringian Refugium

The new paper “Mitochondrial DNA Signals of Late Glacial Recolonization of Europe from Near Eastern Refugia” by Maria Pala et al. (2012) seemingly has nothing to do with the “peopling of the Americas.” It argues, on the basis of a modern European mtDNA sample that mtDNA J and T subclades entered Europe prior to the Neolithic, with the post-LGM population re-expansion from a Near Eastern refugium. But in the opening of the paper they write the following:

“The last Ice Age, which ended 11.5 thousand years (ka) ago, was an era of great climatic uncertainty, with dispersed populations in some regions driven into safe havens at times of greatest stress such as the Last Glacial Maximum (LGM), ~26–19 ka ago. Investigating these glacial refugia has long been a favorite pursuit of phylogeographers. The genetic diversities of many species across a huge taxonomic range have been mapped to various putative refugia, often but not always at low latitudes. Perhaps the clearest example is the Beringian refugium, from which modern humans spread into the Americas.”

For the idea of a Beringian refugium, they cite Perego et al. (2009), Perego et al. (2010) and Tamm et al. (2007). The Beringian refugium is a legitimate hypothesis trying to explain the origin of American Indian linguistic and genetic variation. But Pala et al (2012) make it look like it’s an established fact that can be exported “as is,” without qualification, into a study of West Eurasian glacial refugia. It’s of course possible that some human populations were stuck in Beringia and then expanded out of it as the ice melted and Beringia submerged. But Pala et al. use Beringian refugium to refer to a geographic area that supposedly explains the origin of all of the American Indian populations. If only a few populations inhabited a Beringian refugium during the most recent Ice Age it would be much harder to detect their post-glacial spread than if the progenitors of all of New World populations were locked there.

What one would expect to find in the modern human linguistic and genetic variation as well as in archaeological data that might suggest that a glacial refugium existed and it left an indelible mark on modern populations?

Linguistically, one might expect to see an excess in diversity, as measured by language stocks or typological features, in the areas adjacent to the refugium. In the case of Beringia this is not the case: northern North America and Northeast Asia are not richer in language families or typological features than southern North America, Mesoamerica and South America. The only linguistic classification that suggests higher stock diversity in northern North America and Northeast Asia is Greenberg’s tripartite classification of American Indian languages into Eskimo-Aleut, Na-Dene and Amerind. Members of all three families are found in the geographic areas adjacent to Alaska, while Mesoamerica and South America have languages belonging to the Amerind phylum only. However, Greenberg’s classification is rejected by all specialists in American Indian languages and is currently off the table.

Genetically, again, one might expect to find an excess in genetic diversity in the areas adjacent to a refugium. In the case of Beringia, this is again not the case. For example, mtDNA hgs X and B are rare to non-existent in northern North America and Northeast Asia, while they are wide-spread further down south and are frequently found in combination with hgs A, C and D. Northern Amerindian populations tend to have a limited number of mtDNA lineages at high frequencies. Northern Athabascans are mostly hg A, Eskimo-Aleuts are hgs A and D. This suggests that the modern New World populations living in the areas adjacent to Beringia colonized these territories from elsewhere in the Americas.

Archaeologically, one should expect to find in the geographic area that served as a refugium during the last Ice Age technological prototypes of toolkits observed in the younger strata of the areas colonized from this refugium. As of now, there’s simply no archaeological evidence for a Beringian standstill happening before the founding migration down the coast or an ice-free corridor into the southern areas of the New World. This evidence may be submerged under water but it’s definitely not there now. Instead, what archaeology has been telling us in the past 15 years is that human populations were present in the New World south of the ice shield (e.g., Monte Verde in Chile), that Clovis originated in southern North America (Buttermilk complex in Texas) and that Clovis-type points moved north into Alaska and Northeast Asia (Mesa, Serpentine Hot Springs, Uptar). Instead of thinking of Beringia as a refugium for the New World, we need to start thinking about the New World as a Pleistocene refugium for the post-glacial Circumpolar spread zone. In a number of publications Richard A. Rogers et al. provided precisely such a model: they showed a surprisingly good fit between the distribution of modern language families and the Pleistocene ecological zones (see below, Fig. 7 from Rogers et al. “Ice-Age Geography and the Distribution of Native North American Languages,” Journal of Biogeography (1990) 17, 131-143) and concluded on the basis of this isomorphism that humans must have been present south of the ice shield in America.

They argued for this at a time when archaeology did not have fully legitimate pre-Clovis sites and they were right. Note that they assigned only Eskimo-Aleuts to a Beringian refugium (or the Saiga Faunal Province), with Na-Dene speakers, Haida and Tsimshian occupying a coastal one and the rest of American Indian language families maintaining their distribution south of the glaciated areas of North America, with Algonquians inhabiting the Symbos-Cervalces Faunal Province, the Uto-Aztecans and Tanoans the Camelopes Faunal Province, the Gulf languages the Chlamyrhere-Clyprodont Faunal Province, the Siouan, Caddoan and Iroquoian languages the Odocoileus-Pitymys Faunal Province, etc. They arrived at the following model of population expansion after the break-up of the ice shield:

It makes good sense and offers an unacknowledged compromise between the various models of the peopling of the Americas and linguistic classifications. Eskimo-Aleuts and Na-Dene re-expanded into their present territories from glacial refugias, one of which was in Beringia. Salishan, Caddoan, Siouan, Algonquian and Iroquoian languages pushed up north diffusing Clovis-type lithic technologies into Na-Dene, Eskimo-Aleut ranges and further out into Northeast Asia (Uptar site at 8,600 YBP). In an earlier paper (“Language, Human Subspeciation, and Ice Age Barriers in Northern Siberia,” Canadian Journal of Anthropology 5 (1), 1986, pp. 11-22), Richard Rogers applied the same logic to Northeast Asian language families (Chukchee-Kamchatkan, Uralic, Altaic and Yeniseian) and arrived at the following map.

Historically and arguably presently, American Indians have been subject to a host of popular stereotypes. (See, e.g., “The White Man’s Indian,” by Robert Berkhofer.) Scientists are not immune to these stereotypes. Pala et al. (2012) uncritically interpreted some publications advocating for a Beringian refugium as a jumping board for the peopling of the Americas in order to provide a firm foothold for their interpretation of the peopling of Europe. The assumption is that Beringian refugium is a fact. But the truth is that the Beringian refugium is a conceptual device used to make the old Clovis I model work in the face of a growing body of cross-disciplinary evidence against it. It also serves the purpose of finding a strategic balance between a single origin model and a multiple-migration model for the peopling of the Americas: one migration brought the founding population to the Beringian refugium from which subpopulations began to be dispatched into the new continent as deglaciation allowed. But these are conceptual compromises that will change once the contradicting data reaches a tipping point to overturn the paradigm. They are unrealiable as models to be exported as proven case studies to the Old World regions. One may also recall the early studies of worldwide genetic variation by Luca L. Cavalli-Sforza who assumed that the recent peopling of the Americas was a fact set in stone by archaeology. He then noticed that genetically American Indians are the exact opposites of Africans and concluded that Africans must be the oldest, since American Indians are the youngest. The truth is archaeology doesn’t have the facts to prove the 12-15,000 year old colonization of the New World. It’s an intellectual model at best and a stereotype at worst that cannot be reliably used to kickstart the far-reaching interpretations of human genetic history from the genetic perspective.

At the same time, innovative interdisciplinary approaches such as Richard Rogers’s need to be experimented with beyond the North American and Northeast Asian language families, and Western Eurasia analyzed genetically by Pala et al. (2012) is a very suitable region for the study of isomorphisms between the distribution of language families and ancient biogeographies.