Let’s start at the beginning. If you read a book about Indian history in the 1980s it might begin with this sort of stylized narrative: in the beginning were the Mundas. Then there were the Dravidians, then finally the Aryans (and as an afterthought various East Asian groups on the fringes of northern and eastern Aryavarta). The thesis, broadly, was that the Munda people, who speak an Austro-Asiatic language, were the closest that the Indian subcontinent had to genuine aboriginals. The oldest of old. Supporting this contention is the fact that the languages of the Munda people, with distant affinities to Cambodian and Vietnamese, are very alien in comparison to Dravidian and Indo-Aryan (if Dravidian has any connections outside of the subcontinent, they are always posited to the west, in ancient Iran. The Munda languages clearly have eastern connections). The supposition then was that from the Munda arose various peoples of eastern Eurasia. To cut to the chase this model is probably wrong. The genetic structure of South Asia seems to have arrived at its current outlines relatively recently. In regards to the Munda people their origin is in Southeast Asia. They are not the progenitors of Southeast Asians, they are in part derived from Southeast Asians. Part of the broader expansion of “first farmers” in Southeast Asia from southern China. Their Y chromosomal lineages and autosomal heritage both imply this. Additionally, they carry the Northeast Asian derived variant of EDAR. Though much of their culture is almost certainly exogenous, and of relatively recent vintage, they are clearly highly admixed with the South Asian substrate. In particular, the fusing of an ancient West Eurasian population (“Ancestral North Indians”, or ANI) and a deeply rooted indigenous group with distant affinities further east (“Ancestral South Indians”, or ASI).
One of the reasons that the ancient character of Munda residence in South Asia was persuasive is that they are resident in upland zones, which perhaps refuges after being marginalized by later arrivals. Their fragmented distribution is a tell that they occupied wider territories than is the case today. One thesis is that the Gangetic plain was inhabited by Munda people before the Indo-Aryans arrived. Rather than Dravidians, the indigenes in the Vedas may have been Mundas. But I’m interested in a more parochial question: can Munda ancestry explain the high fraction of East Asian ancestry in Bengalis, particular eastern Bengalis?
We can address this question a bit with genetics thanks to the resources we have in terms of population coverage. As readers know I’ve started to work with the 1000 Genomes data set. Luckily it has a large number of Bengalis within it. Meanwhile, the Estonian Biocentre has put its genotype data online, and there are Munda samples in there. I merged the data together, and removed pretty much all missing alleles. At the end of it I had 185,000 SNPs. To explore the questions I had in mind I decided to look at several populations. Bengalis and Telegu speakers (their genetic position would put them as “middle castes”; not Brahmins, but not Dalits or tribals). Georgians (from the Caucasus) as an outgroup. For Southeast Asian groups, Burmese, Cambodians, Filipinos and Dai. Finally, a small number of Munda. I plotted them on a PCA and removed those individuals who were not easily assigned to a cluster. The first PCA:
This isn’t really telling you much you don’t know. Let’s look at PC 3 now:
As you can see the Munda show a cline toward the Cambodians. This makes sense if the Munda descend from Austro-Asiatic agriculturalists. The Austro-Asiatic expansion in Southeast Asia probably dates to 4,000 years ago or so. Peter Bellwood has stated that archaeologists have excavated villages in northern Vietnam which catch the process of ethnic transition in action at this date (e.g., 75% of the burials are of gracile individuals, whille 25% very robust individuals). Such dates might put a ceiling on how early the Munda arrived inthe Indian subcontinen. In these results the Filipinos are representative of Austronesians, who have their roots in Taiwan and the Fujian coast, while the Dai are the forerunners of the Thai who arrived in Southeast Asia over the last few thousand years, taking over the uplands of Burma (Shan) and Laos (Lao), and swallowing the Khmer civilization which once flourished in the Chao Phraya basin (becoming Thailand). But it’s hard to make out what’s going on with the Bengalis…to me it isn’t clear that they’re shifted as much toward the Cambodians as they should be if the Asian ancestry was due to Munda being absorbed by Indo-Aryan speaking farmers.
So next I ran Admixture. I ran supervised and unsupervised and they showed the same qualitative result. Below is a bar plot of the unsupervised result, K = 5. ￼
The Munda ancestry which is Southeast Asian here is overwhelmingly Austro-Asiatic. That is not the case with the Bengalis, who exhibit a range of fractions. There is very little Austronesian ancestry, which is something one might expect. But, there’s a balance of Austro-Asiatic and Daic ancestry in many individuals, though there is inter-individual variation (my mother has one of the strongest Austro-Asiatic skews among the Bengalis, while my father is among the most Daic; previous runs of admixture consistently show that her eastern Eurasian is more Southeast Asian than his, which has suggestions of Northeast Asian). This is not consistent with Munda being the sole source of East Asian ancestry in modern Bengalis.1 Using rolloff based methods researchers have estimated that admixture into Bengalis occurred on the order of 1,000 years ago. There’s nothing here that would contradict that, and the admixture can easily be explained by the Burmese in the data above, or Khasi and Garo people, who live to north and and east of Bengalis.
Finally, I ran TreeMix on the data. I removed the Georgians and Filipinos because they didn’t add much. Additionally, for kicks I broke apart the Bangladeshis into two groups defined by the 25 most Daic, and 25 least Daic. Below are the ten plots from the ten runs.
I don’t think breaking apart the Bengalis did any good. There were runs with the full fused sample, and the results were similar. It is clear that TreeMix also suggests that Munda are not a singular donor to Bengalis of their East Asian heritage. The source of the donor migration arrow is always shifted more toward Southeast Asian groups proper. Breaking apart the Bengalis into Austro-Asiatic and Daic skewed groups did result in the source of the gene flow being somewhat different. But not appreciably. I also ran the f3 and f4 statistics. There’s nothing surprising about who mixes with who…though it is notable in these and the above results that Burmese show nearly as high a gene flow from South Asians as Bengalis show from Southeast Asians. There have long been suggestions of gene flow from India to Cambodia, perhaps associated with the ancient mediation of South Asian cultural forms across Southeast Asia. But the Burmese evidence of gene flow is tragically ironic in light of the fact that modern Burmese are virulently racist toward dark skinned Muslims who clearly have South Asian origins.
So what happened in Bengal? At the top of the post I have an illustration of The Rise of Islam and the Bengal Frontier, 1204-1760. The thesis of this book is that the Islamic nature of eastern Bengal is in large part due to its relatively recent settlement by Indo-Aryan farmers. Though Bengal has always been a marchland, on the fringes of Aryavarta, before the Islamic conquest of the 13th century its center of gravity, culturally and demographically, was in what is today in the Indian state of Bengal, to the west of Bangladesh. The author of The Rise of Islam and the Bengal Frontier, 1204-1760 suggests that Islamic elites were instrumental in opening up the lands to the east of the old core, and the peasant cultivators who came to cultivate the new territory under their leadership identified vaguely with the religious identity of this new elite (though in general practicing on a day to day level their own folk beliefs), as the old organically developed institutions of Hinduism and Buddhism were poorly moored in the virgin lands. To me this is reminiscent of Michelle Salzman’s data in The Making of a Christian Aristocracy, which suggests that Christian elites arose on the frontiers, rather than the old cores, under patronage of the new religious dispensation (these data are predicted by Peter Turchin in War and Peace and War). In contrast the old Roman elite was relatively late to Christianity, as they were attached their own customs and traditions, which had thick and deep roots in the heartlands of the Roman world. Similarly, Hinduism (or what became what we term Hinduism) between the Doab and western Bengal seems to have resisted Islam’s attempt to destabilize local institutions and interpose itself as the dominant religious ethos of the sub-elites. Only on the destabilized fringes of the west and east, where old orders did not exist or were totally torn down, did Islam find purchase as a majority dispensation.
Finally, the high component of East Asian ancestry among the peasants of eastern Bengal is probably a function of the fact that there were groups from the east also pushing into the fertile territory. If the initial population density was low then a modest inflow at the early stages, ~1,000 years ago, could have a major long term impact. The crushing population densities of “Golden Bengal” was centuries into the future. A lack of cultural memory of this admixture is curious, but to a great extent shifting to the new religion meant that the proto-east Bengalis were creating a new identity. Things get lost….