Denisovans and the Middle Paleolithic of India

Sheila Mishra and colleagues have a new paper discussing the antiquity of microblade industries in India, focusing on the site of Mehtakheri in Madhya Pradesh, for which they report new OSL dates on microblade-bearing layers going back some 45,000 years Mishra:2013. Microblades are relatively small, thin flakes of stone, generally intended to be hafted onto a handle of wood, bone or other material. These small tools are not useful by themselves; they break readily and only function when supported by some armature. The production of such small thin flakes requires a fairly specialized series of choices in the reduction sequence, and the stone cores from which these microblades were struck are easily recognized by archaeologists.

Figure 11 from Mishra and colleagues (2013)

Figure 11 from Mishra and colleagues (2013), CC-BY. Original caption: Figure 11. D3, F1, M25, J15 and M23 are microblade cores. J39 is a trimmed nodule. J2 is is a retouched flake on multicoloured chert and M9 is a perforator made on a platform rejuvenation flake. J44, D107, A42 and I41 are hammerstones of various sizes. C26, D26, D78, C3 are broken flakes and M22 and M13 complete flakes. K1 and M2 are from the initial stages of core reduction showing the much larger initial size of the cores. M 2 also retains a part of the crested guiding ridge. doi:10.1371/journal.pone.0069280.g011

I'm going to use this new paper as an occasion to visit the archaeological record of Late Pleistocene India, stretching across a number of posts. In this one, I'll examine Mishra and colleagues' discussion of the Denisova genome.

Readers following this story will remember that Michael Petraglia and colleagues have, through a series of articles, argued for technological and cultural continuity in southern India from as early as 77,000 years ago up to around 38,000 years ago (for example, reviewed in Petraglia:Toba:2012). This time span stretches across the horizon created by ash from the Toba, Sumatra volcanic eruption, one of the most geologically visible events in the Late Pleistocene, and Petraglia and colleagues have shown this continuity at Jwalapuram in levels both under and above the Toba ash. That finding connects to 15-year-long debate in paleoanthropology about the importance of this volcanic event. By the account of Petraglia and colleagues; the Middle Paleolithic industry of south India has technical similarities to Middle Stone Age industries of Africa, and based on its timing and continuity is likely to have been produced by modern humans.

Meanwhile, Paul Mellars and colleagues have argued in a series of articles that the appearance of modern humans in South Asia was accompanied by the first systematic production of microblades as part of a rather more advanced technical repertoire (for example, Mellars:2013). In Mellars and colleagues' view, modern humans in India originated in Africa with a dispersal that occurred sometime after 60,000 years ago. They argue that the microlithic assemblages have clear technical similarities to production of Howieson's Poort toolkits in southern Africa, and suggest that the dispersal carried this cultural information and technical abilities rapidly across southern Asia and ultimately to Australia.

I will investigate both these points of view more fully. As you can see, it is a debate that is fundamentally about the timing of "major dispersal", the importance (or lack of importance) of local technical continuity, and the similarities (or lack of similarities) of South Asian industries with MSA African industries. Both groups have addressed mtDNA evidence for the timing of population growth in South Asia; this evidence is unsatisfactory in one way or another to both.

Why Denisova is important

Neither group has, to date, addressed the relevance of the Denisova genome to the dispersal of humans across South Asia. To my mind, there is a singular point that any hypothesis must accommodate: indigenous Australians and Melanesians have substantial Denisovan ancestry, South and Southeast Asians, and peoples of Java, Sumatra and Borneo have at most a trace of Denisovan ancestry ("How widespread is Denisovan ancestry today?").

We can't easily explain this pattern without multiple waves of population movement into Southeast Asia. The later wave (or waves) of movement must have happened after the initial spread of people who would colonize Australia and Melanesia. This later wave (or waves) must have made up the vast majority of the ancestry of later Southeast Asians, including today's hunter-gatherer peoples of mainland Southeast Asia and the peoples of the Andaman Islands. The source of this later wave must have been from some population that had Neandertal ancestors but did not have Denisovan ancestors.

This is a minimum. If we imagine Denisovans may have lived in China, or in South Asia, then the extent of later waves of movement must have been even greater, coming from even farther away. If we imagine that the Denisovans had lived in South Asia, then the later wave of movement must have followed very soon after 60,000 years ago, given the evidence for early population growth of the ancestors of current South Asian populations Atkinson:2008.

Can we escape from this problem by hypothesizing that the Denisovans only lived in Java, and mixed with the ancestors of aboriginal Australians there? Setting aside the fact that the Denisova genome itself was found more than 5000 kilometers away, we must still deal with the apparent lack of Denisovan ancestry in any of the populations of Indonesia west of the Wallace Line. We must have a second migration, and given the diversity of today's populations, that second migration must long predate the Neolithic population expansions that also characterize the region.

In reality, talking about "waves of migration" is unnecessarily specific. I think every population movement included newcomers interacting with the previous inhabitants. We may be talking about a continual flux of people across southern Asia. In the end, we are restricted by only a single fact: today's degree of Denisovan ancestry appears to be nearly uniformly low or zero everywhere that contacted the Asian mainland during the Late Pleistocene. If that apparent fact turns out to be wrong, we'll have to revisit the rest of these assertions.

Microblades and moderns

Mishra and colleagues Mishra:2013, demonstrate that the microlithic assemblages in India extend as far back as 45,000 years ago. At first glance, this might seem to agree with the Mellars interpretation, but in an e-mail, Mishra indicated to me that she essentially agrees that the earlier non-microblade tradition is also modern humans, but that this industry does not share close technical links with Arabia or West Asia. They propose that the microblade industries mark a connection between South Africa and South Asia as a result of migration from Africa within the last 60,000 years. (this paragraph reflects updates relative to the original post, 22-07-2013).

By doing so, they create a mystery: Why did technology mark this later period of movement, but not earlier dispersals? If there really are technical and biological ties between Arabia and Nubia ("Jebel Faya and early-stage reduction"), why did these people not spread further, along the coast of southern Asia? And how did modern humans arrive at such an early date in Australia, if they were unable to penetrate India?

To answer these questions, Mishra and colleague propose an archaic population of South Asia that was highly competent and competitive with modern humans. They draw upon the hypothesis that Neandertals replaced modern humans in the Levant after the initial habitations evidenced at Skhul and Qafzeh. This example may show that modern humans were not well suited to certain climatic and ecological challenges before the appearance of the microblade toolkit.

Sharp differences in the stone tool technology of modern humans in the Indian Subcontinent and Southeast Asia exist throughout the Late Pleistocene [19]. This would not be the case if modern humans had reached there from the Indian Subcontinent. Although the evidence is still not conclusive, it appears that modern humans reached Southeast Asia during MIS 5 from a different route and earlier than the Indian Subcontinent. Given the rapidly accumulating evidence for the presence of modern humans in Arabia during MIS 5 [26,27,39,40,41,42,43,44,45,46,47,48], an explanation for their failure to disperse into the Indian Subcontinent at that time, is required. We suggest that the Indian Subcontinent during MIS 5 times was occupied by a population derived from Homo erectus adapted to the Indian environment from Lower Pleistocene times onwards. This population would be archaic, and the Narmada hominin would be ancestral or a representative of it. Competition between Indian archaics and modern humans would have been intense since they were adapted to similar environments. Failure of modern humans to disperse into the Indian Subcontinent during MIS 5 was probably due to their failure to successfully compete with the Indian archaics during a period when the climatic conditions were favourable to both. However during the MIS 4 times, when the desert zones of Africa and Arabia were abandoned and more favourable zones in the Middle East such the Levant and Iran were occupied by Neanderthals, modern humans had more success in entering India and a major change in the Indian Palaeolithic record then occurred. The expansion of modern humans into India therefore coincides with the expansion of Neanderthals into the Middle East at the expense of modern humans and into Central Asia possibly at the expense of Denisovans.

In the place of a "southern route" hypothesis for the colonization of Australia, Mishra and colleagues present an alternative: Migration around the northern tier of South Asia, possibly through southern Siberia and China.

The contrast between the technology associated with modern humans in the Indian Subcontinent and Southeast Asia is present right from the earliest presence of modern humans in the two regions thus making it more likely that modern humans reached Southeast Asia from Southern China rather than that the differences emerged after modern humans reached Southeast Asia from India. The variation in the degree of Neanderthal and Denisovan ancestry in present day populations can also be explained by an earlier dispersal of modern humans to SE Asia via China rather than the Indian subcontinent. Denisovan ancestry is significant in Island but not mainland SE Asia [72]. This is explained if populations in Island SE Asia are descended from populations which spread through China when Denisovan populations were still present. Present day Chinese have different and greater amounts of Neanderthal ancestry than populations elsewhere [73]. but lack significant Denisovan ancestry [72]. This could be due to admixture with Neanderthals in Central Asia when modern humans expanded from there into China during MIS 3 times after Denisovians had become extinct.

That seems at first glance like a highly counter-intuitive take on the evidence. The hypothesis: Modern humans first reached Southeast Asia via China, with little or no input from India. Three elements of this hypothesis accord well with the data as they stand:

  1. The earliest modern humans now known from China appear to substantially predate the appearance of microblade toolkits in the Indian subcontinent. These include Zhirendong ("Zhirendong puts the chin in China") and plausibly Liujiang. Liujiang has been associated with a minimal date of 68,000 years ago, and arguably is twice that age Shen:Liujiang:2002, but remains debatable because of the circumstances of its discovery Shen:Laibin:2007. In addition to these, there are several sites with teeth or more fragmentary remains that also might represent early modern humans in China, and there are some very early occupation sites in Southeast Asia. I'll review these in a later post.

  2. As Mishra and colleagues suggest, the Southeast Asian archaeological record seems to follow a different trajectory than the South Asian record across much of this time period. That is also true of the Indonesian archaeological record of the Late Pleistocene.

  3. The lack of Denisovans in South Asia helps to contain the problem of no Denisovan ancestry to Southeast Asia. This is possibly the most minimal scenario. Further, the later immigration of microblade-using peoples into South Asia would provide a source population for further migrations into Southeast Asia and Indonesia that would lack Denisovan ancestry.

Can we avoid South Asia?

I have my doubts. I think Mishra and colleagues are on to something very interesting, but I am not convinced it has to do with the earliest movement of modern humans.

Most important, I am not convinced that the technical similarities among industries were produced by the biological movements of peoples. For example, Howieson's Poort was a localized and short-term industry without compelling technical connections to later peoples of southern Africa. Sure, it is self-evident that the existence of Howieson's Poort shows that its makers were capable of making it. But why would a microblade tradition take hold in India but not other destinations of these people, including within Africa? And why did the technology persist for so long in India when it was far more ephemeral elsewhere? These questions necessitate local ecological and cultural answers. The strength of ecology and culture history in maintaining these industries in India must have outweighed the importance of drift and ecological factors in other regions. But if so, then surely the microblade industries are much more likely instances of convergent technical solutions than lineal cultural relations.

I recognize that this is a general argument against the use of technical similarities as markers of population relationships. It is just as applicable to the similarities of Jebel Faya and the Nubian complex, for example, or to the linking of Indian Middle Paleolithic and MSA African industries proposed by Petraglia and colleagues.

But the evident diversity of the MSA record of Africa, with a rapid invention, local proliferation, and frequent disappearance of such interesting elements, makes it very likely that technical links will appear by random chance with toolkits elsewhere. Let's consider the technical similarities of Mousterian Neandertals -- including ornament production, pigment use, and blade production -- with MSA peoples of southern Africa. Should we believe that these ancient people were linked by an unbroken chain of ideas, so much so that we can infer they were members of a single migrating population?

Probably not.

But we focus too much, perhaps, on similarities. The differences between southern African and Neandertal industries are meaningful, both in terms of ecology and cultural connections. And I think the most interesting part of the current paper is the way Mishra and colleagues reflect upon the technical differences between South and Southeast Asia and China. This means something. The peoples of these regions retained a strong pattern of technical difference in the wake of a time when modern humans had, by some accounts, cruised rapidly across the southern tier of Asia on their way to Australia. These differences may disprove the model and require a more complex scenario of population movement and mixture.

That more complex model may help us make sense of the Denisova genome and its legacy among living peoples. The archaeological evidence is going to help us sort this story, and I am pleased to see it developing.