Pittfalls of Multidisciplinarity: Vasil’ev et al. (2014)
Cultural Developments in the Eurasian Paleolithic and the Origin of Anatomically Modern Humans: Proceedings of the International Symposium “Cultural Developments in the Eurasian Paleolithic and the Origin of Anatomically Modern Humans” (July 1–7, 2014, Denisova Cave, Altai), edited by А.P. Derevianko, М.V. Shunkov. Pp. 165-171. Novosibirsk: Publishing Department of the Institute of Archaeology and Ethnography SB RAS, 2014.
First People and Prehistoric Migrations in the New World: Some Preliminary Results of a Multidisciplinary Study
Vasil’ev, S.A., Y.E. Berezkin, A.G. Kozintsev, I.I. Peiros, S.B. Slobodin, and A.V. Tabarev.
There’s a general agreement among scholars of human origins and dispersals that interdisciplinarity (cross-disciplinarity) is a good thing. Ultimately the goal is to have archaeology, genetics, linguistics, paleobiology, craniology, etc. tell the same story of human origins and dispersals because it’s exactly the same historical people whose bodies, tongues and brains form populations, speak languages, create and then leave behind artifacts of their daily lives. Patterns of kinship between populations gleaned from genetic data may hold cues to previously-unknown linguistic groupings, while the nature of linguistic diversity on a continent may shed light on the origins of genetic diversity exhibited by the populations that occupy it. However, there are no best practices in academia as to how to conduct interdisciplinary studies. Just like complex investment products such as derivatives depend on the performance of underlying equities and the recent financial crisis stemmed from the toxic mortgage assets rolled into more complex contracts, the success of the integration of scientific disciplines depends on the quality of the underlying material. As noble as the interdisciplinary goal is, there are cases when merging different disciplines – each having its own intradisciplinary epistemological standards, assumptions, politics and materials – leads to flawed models and bad science. Greenberg, Turner and Zegura (1986) brought together linguistics, odontology, classical genetic markers and the Clovis-I archaeological model to argue for three waves of migration into the New World in the past 12,000 years. Since then mtDNA and Y-DNA studies found little support for multiple migrations into the New World, odontologists found greater diversity of teeth patterns in ancient and modern Amerindians than Turner’s “Sinodonty type,” while linguists utterly dismissed Greenberg’s tripartite classification of New World languages into Amerind, Na-Dene and Eskimo-Aleut stocks. Long passionate about the co-evolution of genes, people and languages, Cavalli-Sforza (1997) attempted to align allele frequencies and language families globally but he utilized unrecognized linguistic taxa such as “Amerind,” “Khoisan” and “Indo-Pacific” and attributed to world’s linguistic families a phylogenetic structure derived from classical genetic markers (Africans branching off first), a structure that linguistic evidence does not independently verify.
The conference paper from a team of Russian scholars made up of archaeologists (Vasil’ev, Slobodin, Tabarev), a craniologist (Alexander Kozintsev, my former professor of physical anthropology at St. Petersburg University), a linguistic long-ranger once associated with the Santa Fe Instutute (Ilya Peiros) and a folklorist (Yuri Berezkin, my collaborator on the articles devoted to American Indian cultures for the Great Russian Encyclopedia) is a summary of a larger study to appear as a monograph. Although it brings together several distinct disciplines, the authors apparently do not seek to establish any substantial logical continuities between them. Archaeology tells its own story, linguistics its own, physical anthropology its own, and folklore still another one.
On the archaeological side, just 10 years ago this very team insisted that Clovis-I was the only valid scientific paradigm for the study of the peopling of the Americas. Now that evidence has accrued that has firmly disproved this naive point of view, Vasil’ev et al. (2014) have dug in at the next latest frontier:
“The blade and biface industry from Cactus Hill discovered below the Clovis component is the unambiguous evidence for Pre-Clovis habitation with surprisingly early dates, up to ca. 20,000 yr BP.”
The new date of 20,000 YBP leaves enough room for a back migration out of America at the end of the last Ice Age. Vasil’ev et al. (2014) indeed report that the dates of the Western Beringian tradition are older in Alaska than in the adjacent areas of northeast Asia (Chukotka and Kamchatka) but they quickly manipulate this data point to declare that the Western Beringian tradition must have originated in northeast Asia even earlier.
“In Kamchatka and Chukotka, the sites belonging to this tradition are so far dated by 14C to only 13,000–10,000 cal. yr BP (Dikov, 2004; Kiryak, Bland, Kuzmin, 2010), but considering that the Alaskan record extends back to 14,000 cal. yr BP, it likely began in Western Beringia around
16,000–15,000 cal. yr BP.”
They go ahead to observe that such cultures as Ushki in northeast Asia, Nenana, Mesa and others in Alaska show technological similarities with, respectively, the Western Stemmed Tradition, Clovis and Agate Basin and may reflect both north-to-south and south-to-north movements along the Mackenzie ice corridor. This increased level of comfort with the idea of the bi-directionality of movement in and out of America experienced by a conservative academic team is a significant change of mind but they fall short of connecting the dots between the formal technological links between Clovis and Nenana, on the one hand, and the proposed origin of Clovis in the chronologically earlier Buttermilk Complex in Texas. It’s understandably hard for scholars who have long believed in the origin of Clovis (which at that time meant the very first Amerindians) from Nenana to acknowledge that, if one follows data and not beliefs, the opposite development must be postulated.
The review of craniological evidence by Vasil’ev et al. (2014), too, is fraught with inconsistencies and logical flaws. On the one hand, it’s proposed that
“The totality of data, genetic and morphological alike, suggest that American natives are members of the Mongoloid race, relics of the early stages of its formation.”
This statement is demonstrably untrue. But let facts not distracted us from the authors’ story. They further observe that
“Mongoloid traits in Asian populations become more and more pronounced from Southeast to Northeast Asia, whereas the frequency of Mongoloid markers in America decreases from Alaska to Central and South America.”
The conclusion from these poorly supported arguments, namely that evidence “unambiguously demonstrates that the source of the later Mongoloid admixture in the New World was Siberia,” is a logical oddity because more ancestral forms (“relics of the early stages of its formation”) can’t be derived from a region in which the most derived forms are recorded. It’s also not clear how this conclusion can be supported by Y-DNA data, which shows no traces of most common East Eurasian haplogroups, namely N and O, in the Americas, or with blood groups, which, again, shows no evidence for blood group B in the Americas. Finally, a “later Mongoloid admixture in the New World” runs counter to the claim from recent whole-genome studies (e.g., Raghavan et al. 2013) that it’s West Eurasian gene flow that entered the New World following the basic East Eurasian (“Mongoloid”) population flow. This latter claim may not be true in its own right but to argue for a “later” Mongoloid gene flow into the New World from Siberia requires an explanation of the evidence for a unique West Eurasian-Amerindian genetic link that must have postdated any founding “Mongoloid” migration, if New World genetic diversity is to be derived from a mix of Old World populations in the first place. There’s no win here for the proponents of either migration, in my opinion, but at least scholars who see Amerindians as phenotypically “Mongoloid” should account for their deep West Eurasian genetic links that are already apparent in the Mal’ta boy genome (24,000 YBP), which shows such a key pre- and non-Mongoloid marker as the underived state of EDAR gene, and in the Anzick genome (12,600 YBP) found in association with Clovis tools. If the West Eurasian-Amerindian link predates the Amerindian-East Asian link (as in the New World was colonized from West Eurasia, or by a genetically West Eurasian population with zero East Asian ancestry, some 25-20,000 years and then a secondary migration happened that brought in a much more recent “Mongoloid” phenotype that partially displaced original Amerindians), then Amerindians, again, can’t be considered “members of the Mongoloid race.” It also worth reminding ourselves that the earliest skulls with generalized Mongoloid morphology appear in the New World paleobiological record, and not in Asia. Consistent with this observation is the finding of Sinodonty teeth in the Upper Sun burial in Alaska (11,500 YBP), the earliest attested occurrence of a highly specific dental type that’s shared between modern Amerindians and East Asians. One of the defining traits of Sinodonty is incisor shoveling, which is coded for by the derived state of EDAR gene. All these facts together suggest that Mongoloid phenotype recently expanded from North America into East Asia as well as further down south in the New World in post-Mal’ta times and admixed with pre-existing populations in both the Old World and the New World. The phenomenon of the weakening of the “Mongoloid” signal in both Asia and America from north-to-south, which Vasil’ev et al. (2014) mention, is fully consistent with this admixture hypothesis.
Undeterred by the missing facts and lapsing logic, Vasil’ev et al. (2014) proceed to linguistics and (thanks to Ilya Peiros) announce the presence of evidence for at least 6 waves of migration to the New World. As compared to Greenberg’s tripartite classification, Pieros has retained Eskimo-Aleut and Na-Dene intact but split Amerind into Mosan, Algic, East-Amerindian and the rest. He also acknowledged that Siouan, Iroquoian, Muskogean and a number of others may represent additional migrations. Only Eskimo-Aleut and Na-Dene have plausible linguistic relatives in the Old World, with Sino-Caucasian languages stretching all the way across Eurasia. Neither archaeology, nor craniology show anything similar to the geographic signature of Na-Dene and their putative Old World linguistic relatives, so it’s not clear why Vasil’ev et al. need linguistic material in the first place.
To this amalgam of interdisciplinary material, Berezkin adds the distribution of mythological patterns reflecting, in his opinion, several “episodes” or “moments” in the peopling of the Americas. He finds the earliest Amerindian motifs in South America east of the Andes. They show formal similarity with South Asia and the Sahul. Berezkin connects Earth-Diver with Clovis, while some remarkable similarities between South Siberian and North American mythologies with mtDNA hgs X2 and C4c. Berezkin is driven by a desire to relate his folkloristic findings to presumably more robust archaeological and genetic props but there is no rhyme or reason to his effort. mtDNA hg X2 in South Siberia is the product of a very recent backflow from the Caucasus and can’t be seen as a shared South Siberian-Northern Amerindian legacy. The geographic signatures of Clovis and Earth-Diver are too dissimilar and, even if we assume they are “similar,” Clovis’s connection to an earlier Buttermilk antecedent in the south and to the contemporaneous Nenana complex in the north would suggest that Earth-Diver expanded out of the Americas (as I previously observed) and not into the Americas, as Berezkin wants to believe. Finally, Berezkin sketches out another migration from Alaska down the western coast of Canada related to the Western Stemmed tradition but with “vague Asiatic roots.”
Human origins research has entered its Big Data stage. And, as it’s the case with Big Data elsewhere, its exponential growth is not accompanied by the increased capacity of scholars to organize, process and make inferences from the data. Vasil’ev et al. (2014) contradict themselves, manipulate inconvenient data to fit agendas, freely walk away from previously firmly held views, generate “moments,” “episodes” and “scenarios” of migrations at will, claim random connections between isolated data patterns from one discipline and isolated data patterns from another discipline and do everything else to foster a strong impression that the peopling of the Americas is entirely beyond their comprehension, regardless of what kind of evidence they are looking at.
Meanwhile, genetics, craniology, archaeology, linguistics and mythology do tell the same story. Phenotypically and genetically New World populations show worldwide greatest intergroup (between-group) diversity values and world smallest intragroup (within-group) diversity values. (Vasil’ev et al. (2014) confirm this when they refer to “the between-group variation in Native Americans in some traits including anthropometry and dermatoglyphics approaches that between geographic
races.”) This pattern suggests long-term isolation, genetic drift, high levels of inbreeding and small population size. Correspondingly, the archaeological signatures of such dispersed and small populations are elusive and take longer to recover. There is no reason whatsoever to assume that every early find in the New World is the absolutely earliest one. New World demography and inferences of long-term population size from genetic data makes this assumption unwarranted. The absence of clear East Asian antecedents of the earliest New World archaeological traditions suggest that it’s the rate of archaeological recovery in the New World that’s at stake here and not the objective absence of humans in the New World up until 12,000 or, now thankfully, 20,000 years ago. The currently available demographic and genetic picture fits well with extensive linguistic (as measured in the number of independent stocks and pan-Eurasian affinities for such stocks as Na-Dene) and folkloric diversity (as measured in the number of motifs and their pan-Old World similarities) found in the Americas. At the same time, all New World populations cluster together on PCA plots, share most genetic drift with each other and share unique markers at some genetic loci (such as the 9RA marker mentioned by Vasil’ev et al. 2014) – facts that are incompatible with a multiple migration scenario, which Vasil’ev et al. 2014 favor when it comes to interpreting the evidence from folklore and linguistics. An out-of-America scenario remains the strongest alternative to the all-around confusion presented by Vasil’ev et al. 2014. This confusion stems from forcing underlying intradisciplinary evidence into an interpretative straitjacket and hoping that adherence to cultural beliefs can substitute for the real engagement with available interdisciplinary evidence.
“Only Eskimo-Aleut and Na-Dene have plausible linguistic relatives in the Old World”
How do you personally regard proposals like Eskimo-Uralic or Eskimo-“Altaic”?
Nothing is firmly established in this area. The Dene-Ket connection proposed by Vajda is the only proposal of linguistic kinship between the New and the Old Worlds that has met the standards of provability accepted in the linguistic community. I consider the evidence marshaled by Uwe Seefloth for Uralo-Eskimo-Chukchi-Kamchatkan (Seefloth, Uwe. 2000. “Die Entstehung polypersonaler Paradigmen im Uralo-Sibirischen.” Zentralasiatische Studien 30, 163-191) as very strong. But this opens a can of worms as other people would consider evidence for Nostratic (Uralic, IE, Dravidian, Kartvelian) as no weaker (and I’m cautious about it), so we end up with Greenberg’s Eurasiatic family of which Eskimo-Aleut is a part and I seriously doubt its reality. The situation is similar to Dene-Caucasian: the Ket-Na-Dene link is solid but everything else further out West all the way to Basque is potentially interesting but far from solid.
To sum up, let me put it this way: I’d be LEAST SKEPTICAL about proposals that unite 1) Eskimo-Aleut with Chukchi-Kamchatkan, Uralic, Altaic and Indo-European to the exclusion of Dravidian and Kartvelian; 2) Na-Dene, Ket and Sino-Tibetan to the exclusion of Burushaski, North Caucasian and Basque.
But the methods of historical linguistics are very slow in building evidence for higher-order taxa, and currently it’s all mostly speculation. There’s nothing wrong about it, we just have to be mindful about the propensity of all those theories to be dismissed over time. We can be much more certain about the relative degree of linguistic differentiation between the Old World and the New World and about the worldwide clusters of grammatical traits.