Substack cometh, and lo it is good. (Pricing)

Indian culture came to Southeast Asia through Indian people

For nearly 1,000 years the kingdom of Champa occupied the coast of modern Vietnam, from Annam down to the Mekong delta. It was a maritime facing polity whose language was Austronesian. The Chams were closely connected to the people of Maritime Southeast Asia, and part of the broader Indic zone. They were Hindu. Unlike other people in the region, they never fully converted to Theravada Buddhism.

Today most Chams are Muslims due to contact with Islamicized Malays. But a minority in Vietnam remain Hindu. The reason for this is tragic and contingent. The final conquest and absorption of Champa by the expanding Vietnamese in the early modern period occurred during a period of Islamicization. Though the king of the Chams remained Hindu until the end, the court around him was converting to Islam. If the Vietnamese had not conquered the Chams and isolated them from the rest of the world, it is likely all of them would have converted to Islam.

As it is, a Vietnamese program of periodic genocide and forced assimilation produced an involuted Cham culture. That is, it turned inward and became a fossil society. The goal was survival. They stood in place.

Today the Cham are Saivite Hindus in Vietnam, with some elements of the varna system preserved.

This brings us to the question: how did Indian religious and cultural forms become entrenched on the coast of Vietnam 2,000 years ago? Two new preprints shed light on this, Reconstructing the human genetic history of mainland Southeast Asia: insights from genome-wide data from Thailand and Laos, and Indian genetic heritage in Southeast Asian populations. The second preprint is obviously more explicit.

Figure 5 is clear:

From the discussion:

355 Our study revealed substantial South Asian admixture in various populations across Southeast Asia (~2-16% as inferred by qpAdm). We observed South Asian admixture in some populations (Cham, Ede, Giarai, Khmer, Kuy, Nyahkur, and Thai) for whom the admixture was not reported before. Most populations harboring South Asian admixture were heavily influenced by Indian culture in the past or are related to descendants of ancient Indianized states in Southeast Asia. In contrast, we failed to detect South Asian admixture in most “hill tribes” and in present-day hunter-gatherer groups from Thailand. Consequently, the spread of Indian influence in the region can be explained by extensive movement of people from India rather than by cultural diffusion only.

The genetics suggest that the gene flow had to have happened between 0 and 1000 AD. This is basically prehistory in Southeast Asia, as most of our information comes from references in Chinese records. But the presence of R1a in particular across the region indicates to me serious elite migration from India during this period.

In 2000 BC the Sintashta warlord Indra* huddled in his tents some of the Volga, on the cold and open steppe. 2,000 years later his direct paternal descendants were overseeing a nascent thalassocracy on the South China Sea. The history of humanity is more wonderous than anything imagined in your philosophy Horatio…

* I am aware that he is likely a BMAC deity. This is a joke.

0

14 thoughts on “Indian culture came to Southeast Asia through Indian people

  1. Inferences to BMAC are also based on soma being BMAC but the word could also have a PIE root. The BMAC link is probably only cultural accretion. Some kind of borrowing which made this version different from the root. Then again something similar could be said regarding others like Zeus.

  2. Re; “Indian genetic heritage in Southeast Asian populations”, I still wonder about the place of the admixing SA populations on the South Asian clines…

    Best (highest p) fitting models in their supplement table 2 for a subset of populations:

    Cham – Dai:Onge:Hakki_Pikki 75:14:12 (p-value 76.23%)
    Cambodian – Dai:Onge:Pulliyar 78:7:15 (p-value 61.71%).
    Ede – Lahu:Onge:Pulliyar 67:20:13 (p-value 58.63%)
    Giarai – Dai:Onge:Baniya 69:24:7 (p-value 52.85%)
    Khmer – Dai:Onge:Ulladan 73:20:7 (p-value 53.15%).
    Kuy – Dai:Onge:Palliyar 74:18:7 (p-value 32.1%)
    Nyahkur – Dai:Onge:Baniyas 71:23:6 (p-value 17.59%).
    Thai2 – Dai:Onge:Gujjar 89:7:4 (p-value 41.49%).

    I’ve disregarded those target populations where their Table 1 had best fitting East Asian type ancestry in qpGraph as “before Tibetan” or “Tibetan” included in this source (e.g. Burmese, Mon, Thai1), which probably indicates Sino-Tibetan related impulses in ancestry, and also Malay, which seemed like the only passing model was an odd one involving Kalash, possible I would guess because Insular SE Asian has some different ancestry paths.

    Their outgroups involve Siberian groups and Papuans, so this may produce strange fits without populations who are related to these relative to their sources (Atayal, Dai, and Lahu).

  3. Completely unfounded idea, but one that is fun to consider: The Ramayana is extremely popular in SE Asia, and in terms of material culture, there are more depictions of it across SE Asia than in India itself. You also find more episodes that seem to be lost in the Indian versions.

    What if the original myth involved a frontier group moving into SE Asia? Lanka simply meant an island, and the Lanka of the epic may have been an SE Asian Island before it was reapplied to Sinhala/Elu/Illankai. The epic was reintroduced to India, but was then reinterpreted to an Indian geography.

  4. Question, off topic about Finnish DNA. My half brother, finally got DNA tested, as a request from me. My dad was illegitimate, and it took me 6 years to find out who his father was. I did, thanks to an 88 year old lady who showed up as a possible, first cousin, which would be impossible. She had a question mark on her birth certificate, and draft registration documents, which led to proof that her father, was my great grandfather. My question is about my great grandfather, grandfather, father’s, brother’s Y DNA, will it be forever, Finnish, though not actually carrying any Finnish DNA?

  5. @ Matt

    I have a hard time imagining the source population for the South Asian in modern Southeast Asia was some South Indian tribal and/or low caste group. Part of this is because there’s very low levels of R1a among those groups (I believe less than 10%). But the major reason is just simple historiography – that any group which was able to expand overseas into the Southeast Asian littoral must have been high enough status to be able to mount major trading and/or military expeditions. Tribals and untouchables don’t have that.

    I wonder if part of the issue is Onge isn’t the best fit for the original negrito-type ancestry of these groups. The Andaman Islands were probably settled from Burma over a now-submerged land bridge, meaning they should be reasonably close fit for the ancestry in Burmese people, but they may have been more distinct from the indigenous population in other areas. As a result, some of the negrito-type ancestry may be getting lumped into the South Asian cluster, which is making that component seem more AASI-shifted than it was in reality.

    It was probably a South Indian group however. Most of the states with the most developed maritime traditions were around modern Kerala and Tamil Nadu. I believe in most cases their full cultural flowering was a bit later than the first century CE, but this may be more an issue of spottiness of records than anything.

  6. @Karl, my impression from PCA is that SEA samples with S Asian ancestry “point” away from SE samples w/out S Asian ancestry towards middle/”most ASI” end of S Asian Cline. I don’t think this is a composite of the early Hoainbhian Man Bac like ancestry. I don’t know what the group(s’) status might have been.

    There’s obviously still very uneven proportions of this ancestry, in those samples that have it, today. Not evenly distributed at all.

  7. I’m surprised they didn’t try models with their Mania population rather than Onge though. Might have worked and resolved any concerns in that direction issues.

  8. Can also see that in ADMIXTURE plots, there’s no preference for any excess Maniq component on top of relatively low ASI component and high West Eurasian component, and instead just high ASI low West Eurasian. Of course, ADMIXTURE is not perfect measure of these things as likely Maniq have strong bottleneck effect. So worth some formal qpadm testing with them, using their outgroups which are insensitive to Maniq specific drift. Might also allow for Onge in the outgroups. I will suggest to authors.

    Overall, I mean, I think there could some elite process, but think it also would’ve been accompanied by retinue (military / commercial) of ordinary South Asians, and their families. Even Conquistadors were not just elite Iberians but accompanied by the ordinary “sweepings of the streets”, Conversos etc? Who all contributed to present day Latin America. And in SE Asia it probably would’ve been a composite process over time, probably throughout entire last 2000 years. Not one pulse. It does seem likely that elite associated male lineages introduced would show more survival over time though, but I think getting from “What y-dna survives” to the exact autosomal construction of the migrating populations, I’m not sure about that. Y-dna can replicate fast but can “get away” from main autosomal contributions. Say you could’ve had elite Indians who really spread y-dna fast by marrying local gels, but really autosomally diluted fast, while at same time a greater mass of ordinary Indians migrates and make less y-dna impact but lasts more in autosome…

    Re y-dna when we get really finely grained large scale SE Asian and S Asia y dna, probably whole-genome, we can see if the R1a lineages look like they diverge 2000 years ago from any sequences identified in India, often more like 500 years etc. It’s also likely that these will show some good tight group structure in S Asia (some subclades *only* in found in particular jati populations), which is good for working out the population history. That’s at least very testable, just needs lots of data. So perhaps another 10 years (or more?) if I’m pessimistic.

  9. @ Matt,

    It seems to me there might be an easy way to figure this out is if there is any fractional trace of steppe ancestry in Southeast Asia. Razib said over on his substack there is 0% steppe in South Indian Tribals and Dalits. But up to 5% in middle castes, and more in Brahmins. So if you can do an admixture run and find any Sintashta-like component in Southeast Asian populations (even if only around 1%), that would be solid evidence the migration was not of low-caste South Indians.

  10. @Karl, I don’t know about that really, as from what I can remember from Narasimhan’s paper there are some groups with relatively higher ratios of steppe ancestry who also have high levels of ASI.

    But the ancestry in SE Asians is like 2-16% from S Asia, so you’re talking about trying to find the difference between 5% (Middle Caste)* 9% (median SE Asian S Asia ancestry), or around 0.5% ancestry. I’m not sure how easily I could get confidence that a model could tell 0.5% ancestry from 0% ancestry!

    It seems easier just to focus on reconstructing working out what real West Eurasian:AASI ancestry level is!

  11. @Karl, had some time so did try to see if any of the Global25 data with Vahaduo detected any of either Sintashta ancestry or a clear ratio of Kalash:Irula+Onge ancestry (these are the ANI and ASI present day maxima as described by Narasimhan 2019, and Onge for an “excess AASI” point; G25 is as far as I can tell not designed to capture drift specific to any of them apart from clines, although may unavoidably pick up a little bit of ancestry specific to them).

    Some results: https://imgur.com/a/FgPBhW4

    I didn’t get a real clear signal out of it really; with the methods I used it looks like the admix in Burmese samples is South Indian like, while the other groups are more ambiguous, and *could be* ratios more typical of North India / Upper Caste groups (Thai, Cambodian). There could be some different sources – I know Burma has had ongoing migration of people in colonial era, and it doesn’t seem implausible to me for there to be ongoing migration to there from Bangladesh / NE India.

    Btw, where I said samples looked like populations were “pointing” at more South Indian/Middle-Low populations in their internal diversity/clines, here’s a graphic illustration of what I mean: https://imgur.com/a/qYUQfpg

    Using the Eurogenes Global 25 data again I’ve taken some frame population averages to generate more PCA, which have a PC1 distinguishing South Asian from East Asian, PC2 splitting North China like/Sino-Tibetan populations from ISEA populations (most North China like/Sino-Tibetan is Naga followed by Korean/Nyishi, then Han, most ISEA is Lebbo, followed by Indonesians from Bali), then PC3 splitting ANI from ASI (Kalash most ANI, Paniya most ASI).

    Then I take the main population cline for the Burmese, Thai, Malay using regression and project it on. The “point” on the intra-East Asian clines tends to look about right (e.g. Thai point is close to Dai, Malay is close to Indonesian Bali, Burmese close to Central China). It doesn’t look to be pointing to a two way model between a populations anywhere on East Asian clines, and a population that has much of a North Indian like structure.

    That may be due to the compound effects Razib talks about though, where you really have a complex mixes between Mlabri like populations, some varied set of populations on the South Asian cline, and then some populations on the intra-East Asian cline.

    For the Thai population the clines suggest they are can be approximated as a two-way ongoing admixture compound between a mainly Dai+Taiwan Aboriginal like population with a population that is 57:25:18 Mlabri:Irula:Kalash. (Irula:Kalash ratio 58:42, which is about equally between Brahmin populations and Piramalai Kallars). For the Burmese a mainly Naga like population mixes with a point that is 30:48:22 Mlabri:Irula:Kalash (68:32 Irula:Kalash is about the same as Bengalis/Piramalai Kallars).

    Obviously this sort of cline reconstruction runs risks; say you have an idealized situation where you have a cosmpolitan population that accepts migrations from varying sources, and a isolated population that doesn’t, then these two admix later. (That’s plausible in SE Asia, with some mix of isolated “hill tribes” and “civilization”). Well, you will find a two way cline, but if you extrapolate one end of the cline at the cosmopolitan end you may end up with a population that never existed as a single population! I do think its worth trying to use the information from differences within population though, rather than throw them all into one big bag and work out its median proportions.

    Anyway, although happy to try the question with the tools I have to work with and I think there are some interesting directions, I don’t have too much confidence in these methods really, based on PCA. Maybe someone can check it out with admixtools (my PC is a bit weak to even run the new Windows version and I need to upgrade; maybe I will once I do).

    Also, btw, authors seemed receptive to my Biorxiv comment on testing qpAdm models with the Maniq (75% HG) population in place of Onge. Maybe if they find something interesting they will add it to the supplement for the print, or maybe not, but cool to get a response anyway.

Comments are closed.