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

The population genomics of South Asia is complicated, and politics doesn’t make it easier


Many people have been sending me links to this article, By rewriting history, Hindu nationalists aim to assert their dominance over India. Here’s a key section:

The RSS asserts that ancestors of all people of Indian origin – including 172 million Muslims – were Hindu and that they must accept their common ancestry as part of Bharat Mata, or Mother India. Modi has been a member of the RSS since childhood. An official biography of Culture Minister Sharma says he too has been a “dedicated follower” of the RSS for many years.

Sharma told Reuters he expects the conclusions of the committee to find their way into school textbooks and academic research. The panel is referred to in government documents as the committee for “holistic study of origin and evolution of Indian culture since 12,000 years before present and its interface with other cultures of the world.”

Sharma said this “Hindu first” version of Indian history will be added to a school curriculum which has long taught that people from central Asia arrived in India much more recently, some 3,000 to 4,000 years ago, and transformed the population

There are several threads here. First, it is a fact that the ancestors of South Asia’s non-Hindus were Hindu. There are minor exceptions, such as the Parsis, who are ~75% Iranian. One can quibble as to whether many tribal and peasant populations were truly Hindu in a formal and explicit sense. But I think this is a semantic dodge. Muslims would recognize these beliefs and practices as Hindu, no matter if one was a Brahmin monk or a member of a tribe which still sacrificed animals.

I’ve looked at the genotypes of a fair amount of South Asians of Muslim background. The overwhelming (usually exclusive) proportion of their ancestry is South Asian. It’s a fact that the ancestors of non-Hindu South Asians were Hindu.

But, the article and a dominant theme in Hindu nationalism today are that distinctive historically important groups like Indo-Aryans are indigenous to South Asia. This is set against a narrative of invasions and migrations from the outside, which is presumed more friendly to a multicultural paradigm (I have a hard time keeping track of the political valence of all these things). To some extent, the reality of invasions and migrations cannot be denied, whether it be Alexander, the Kushans, or the various Muslim groups. But these historical invasions left little genetic imprint.

When 2009’s Reconstructing Indian Population History was published things changed for the impact of the earlier migrations. By the time the ancient Greeks were recording observations of India in Classical Antiquity, it was already noted as the most populous nation in the world. I was initially skeptical about the result in Reconstructing Indian Population History, that there was massive admixture between West Eurasians (ANI) and indigenous South Asians (ASI) because that would imply massive migration. Additionally, phenotypically the pigmentation genes didn’t seem to work out if the source population was European-like.

Nearly 10 years on we have a lot more clarity. Ancient DNA has changed our understanding of the past. Massive migrations were common. And, the pigmentation and genetic profile of modern Europeans is recent, within the last 4,000 years. The source population(s) for “Ancestral North Indians” (ANI) may not have been Europeans in the way we’d understand them. In fact, a follow-up paper, Genetic Evidence for Recent Population Mixture in India, hinted at two admixtures. There’s a fair amount of circumstantial evidence now that one component of “Ancestral North Indian” relates to West Asian populations and another component to the more classical steppe Indo-Aryans. The former is more widespread across the subcontinent than the latter, which is concentrated in the northwest and among upper castes.

I do understand Indians who want to interpret their own history through the lens of their own cultural priors. The problem is that genetic science has proceeded so fast in the last few years that many propositions which were speculative in the 20th century are testable in the 21st century. Some Hindu nationalist friends and acquaintances express embarrassment and worry about the track that Indian nationalists are going on. I don’t know what to say, but Americans have their own delusions and blithe acceptance of propaganda, so I’m not going to be one pointing fingers. Other Indians have told me via Facebook that they “believe in the results from the 2000s” (when they were more congenial to their viewpoints?). I guess that’s one strategy; just keep up with the science until it starts refuting your model.

That being said, with the ubiquity of datasets one can explore questions oneself without too much concern for politics. Additionally, the Indian government may suppress analysis of ancient DNA through soft coercion and negative incentives, but I’m sure at some point Pakistan will let people dig things up (South Asian Muslims seem less angry that they invariably turn out to have very little recent West Asian ancestry; I suspect because they always on some level knew that that was so).

I recently posted my South Asian Genotype Project results. Though the sample size is small, since I have provenance it allowed me to get more clarity on what’s going on in the 1000 Genomes samples for South Asians.

There are several things I can tell you. For example, I have samples for a Sindhi and Gujarati Lohanna. They cluster near each other, and with Sindhis. Similarly, a Gujarati Muslim sample is also in with the Sindhis.

In contrast, other Gujaratis are placed between Pakistani populations and South Indians. A Vania and Solanki sample helped me label “Gujarati Middle Caste.” The Gujarati Patel samples, are even more shifted toward South Indians, and admixture analysis implies less Indo-Aryan in these samples than other Gujaratis. The Gujarati_ANI_1 is between the middle caste and the Sindhi cluster. I assume Guju Brahmins would be in here, but I don’t have a pure individual from that community.

Gujarat seems to be a state with a lot of connections to Sindh, so you see a wide range of variation in this state on the ANI-ASI cline.

I’ve collected enough Brahmin samples from the four states of Southern India to see that they are very similar genetically (one exception is a Niyogi Brahmin who seemed more ASI-shifted than usual). If they don’t come from the same migration event, they diverged at around the same time. IBD analysis would show that Iyers are closer to each other than they are to Brahmins from Andhra Pradesh surely, but on admixture analysis there’s almost not difference.

Brahmins from North India are different from South Indian Brahmins, in being more ANI-shifted. It seems that to a great extent Brahmins from outside the Gangetic heartland can be modeled as North Indian Brahmins with local admixture. Bengali and Maharashtra Brahmins have shifted away from North Indian Brahmins, but not as much as South Indian Brahmins. Bengali Brahmins are also East Asian shifted, confirming the reality of local admixture.

One result that surprises me is how genetically similar Dalits from North and South India are. The Chamar samples from Uttar Pradesh seem to have some levels of Indo-Aryan admixture, but overall they’re not that different from Dalits from Tamil Nadu.

Additionally, non-Brahmin and non-Dalit individuals from places like Bengal and Tamil Nadu/Sri Lanka, and Andhra Pradesh are very distinct from Dalits. In other words, the caste system is not simply the traditional upper castes vs. everyone else, but it’s deeply structured. The implication here is that caste may predate the Indo-Aryans (this is not a new inference). Or at least not be related them.

Finally, there is a curious pattern where gene flow into southern Pakistan (Sindh) is more shifted toward the Middle East than in northern Pakistan (Punjab, the Pathan areas). I suspect this is due to dynamics which date deep into prehistory, rather than more recent events. Also, Brahmins from North India show a skew where more of their ancestry is steppe and less of it is West Asian. There has been extensive speculation why this might be, so I’ll leave that for the comments.

Addendum: The second ancient DNA paper from Southeast Asia states that the Munda Juang from Odisha are 37% Austro-Asiatic.

22 thoughts on “The population genomics of South Asia is complicated, and politics doesn’t make it easier

  1. The Social’Sciences’ in India have a long hostory of being politicised and these issues are unlikely to be resolved particularly considering that the academia is so closely linked with political groups. While its easy to blame the ‘Hindu’ govt, a large part of the blame must fall on the left leaning academia. While raising the concerns of subaltern social groups is a good thing, weaponising history as a tool to deepen these divides is irresponsible. AIT/AMT would be but an issue of academic interest if it wasn’t used by such ‘liberal’ academics to mean that some groups should be hated by others for being foreign in origin or that Hindu gods/goddess were actual Aryan assassins who murdered local rulers. Perhaps with increasing evidence a significant proportion of pro-hindutva academics might come to accept a PIE homeland outside India? I don’t think govt. positions will affect research much though. The last time the same party was in power hardly stopped the AIT from being taught even though there was far less evidence in its favour then. As for making history Hindu centric, I think its a fair demand since it meaning decreasing the disproportionate importance given to the delhi centric Islamic empires at the cost of regional histories.

  2. i think you make a lot of good points (in particular, i know a little bit of islamic south asian history, and lefty/secular indians seem to want to ‘whitewash’ that aspect in comparison to the british period for political reasons).

  3. Razib,

    Have you considered the possibility that the ancestors of modern Indian Muslims were not only Hindu, but possibly (perhaps even likely) were Jains, Buddhists, or other minor Sramanic faiths just a millenia ago?

    “There are several threads here. First, it is a fact that the ancestors of South Asia’s non-Hindus were Hindu. There are minor exceptions, such as the Parsis, who are ~75% Iranian. One can quibble as to whether many tribal and peasant populations were truly Hindu in a formal and explicit sense. But I think this is a semantic dodge. Muslims would recognize these beliefs and practices as Hindu, no matter if one was a Brahmin monk or a member of a tribe which still sacrificed animals.
    I’ve looked at the genotypes of a fair amount of South Asians of Muslim background. The overwhelming (usually exclusive) proportion of their ancestry is South Asian. It’s a fact that the ancestors of non-Hindu South Asians were Hindu.”

  4. what is worst is total lack of competence on H right in all these issues, as a result they are being led by crackpots of various types. I assumed Caste system rigidity was post 1-5th AD, as opposed to earlier period having more mixture. Is that still true?.

  5. “If the Koran and Bible are considered as part of history, then what is the problem in accepting our Hindu religious texts as the history of India?” said Sharma.

    Another telling quote from the Reuters article you linked to. The average secular Western historian would dismiss much of the Koran and the Bible too as myth, but this guy is completely unaware of that.

    Just like many alt-righters have this 19th century image of India as akin to what’s portrayed in Indiana Jones and the Temple of Doom, Sharma’ impression of Westerners are colored by what he’s read of the attitudes of British colonial officials in the 19th century Indian Raj. I’ve seen this blind spot in virtually every Hindutvawadi I’ve talked to (including family and friends.)

  6. I think there are issues at play here, one related to the ANI & the other with relation to ASI :-

    1. The ANI has been modelled by the Reich Lab in the paper on Near Eastern Farmers as, part Iran_N & part EHG/Steppe_Eneolithic/Steppe_EMBA.

    Interestingly enough, they remarked in the same paper elsewhere that even the Mesolithic Iran_ Hotu sample could be modelled as part Iran_N & part EHG/ANE or steppe_Eneolithic.

    But what it obviously does not mean is that Iran_Hotu has EHG admixture.

    What appears to be the case is that as we go more to the East & North of Iran_N, the affinity towards EHG or the steppe appears to increase. Could this also not be at play with South Asian populations ?

    There is also a north to South cline of steppe affinity in South Asia similar to the cline from Iran_Hotu to Iran_N.

    Nevertheless, Iran_N also is related to the steppe groups. This is illustrated by the fact that it was modelled as 62 % Basal + 32 % EHG.

    So maybe, it could be that ever since the Mesolithic groups lived across a region stretching from Iran/Caucasus to India who were all related to steppe populations but in different degrees. The groups living in India just had more steppe affinity compared to Iran_N and that is why they can be modelled as part Iran Neolithic & part steppe.

    ——————–

    2. The ASI is also getting quite complicated. The Onge are now clearly more related to SE Asian hunter gatherers. So what exactly is the ASI and how long has it been present in South Asia ? And does Iran_N have any ASI affinity ?

    Moreover, the ASI in South Asians is also likely from multiple sources. Very clearly, the Austroasiatics from Mainland SEA have contributed to various tribal groups across India during the last 4 kya. So how does one resolve the different streams of ASI ancestry in South Asia ?

  7. Have you considered the possibility that the ancestors of modern Indian Muslims were not only Hindu, but possibly (perhaps even likely) were Jains, Buddhists, or other minor Sramanic faiths just a millenia ago?

    these are all part of the dharmic religions. when muslims think of “hindu” they often refer to idol worshipers generally. it’s kind of like how chinese perhaps have a difficulty distinguishing muslims and jews? (historically they did)

    but yes, most of the ancestors of muslims were not orthodox hindus in the way that someone who is steeped in vedanta is. but this sort of ‘sanskritization’ of peasant culture is recent.

    by the same token the folk islam of bengali muslims before the 20th century was very different from normative islam. but ultimately you’d still bracket them as ‘muslims’ even if they are not orthodox.

  8. What appears to be the case is that as we go more to the East & North of Iran_N, the affinity towards EHG or the steppe appears to increase. Could this also not be at play with South Asian populations ?

    yes. i think both proxies may be somewhat off. ancient DNA sampling is thinner in west asia and central eurasia than europe.

    So maybe, it could be that ever since the Mesolithic groups lived across a region stretching from Iran/Caucasus to India who were all related to steppe populations but in different degrees. The groups living in India just had more steppe affinity compared to Iran_N and that is why they can be modelled as part Iran Neolithic & part steppe.

    this is theoretically possible. but your priors should be that this is unlikely. you see HUGE Fst differences btwn iran early neolithic and levant/anatolian neolithic. there was a lot of local structure. i see no reason why the area to the east of iran wouldn’t be like that, as opposed to the more admixed scenario you see today.

    (otoh east africa does show a cline, so it’s possible)

    2. The ASI is also getting quite complicated. The Onge are now clearly more related to SE Asian hunter gatherers. So what exactly is the ASI and how long has it been present in South Asia ? And does Iran_N have any ASI affinity ?

    even the original reich paper assumed divergence 20-30,000 years ago.

    Iran_N has no east eurasian affinity. that’s pretty obvious. and basal eurasian is a clade that EXCLUDES west eurasians, ANI, and east eurasians. no ASI outside of south asia aside from recent stuff it looks like.

    Moreover, the ASI in South Asians is also likely from multiple sources. Very clearly, the Austroasiatics from Mainland SEA have contributed to various tribal groups across India during the last 4 kya. So how does one resolve the different streams of ASI ancestry in South Asia ?

    more ancient DNA.

    the issue that suggests to me old ASI in south asia is how diverse haplogroup M (mtdna) is. OTOH, that diversity can happen through multiple migrations too.

  9. I am pretty sure you have seen this, but just to seek your opinion – is this true that Rakhigarhi DNA is pointing to L-M20.

    i have no inside information but it seems plausible.

    I assumed Caste system rigidity was post 1-5th AD, as opposed to earlier period having more mixture. Is that still true?.

    right to a first approximation. priya moorjani has told me that she knows factually there were ‘pure ASI’ and ‘pure ANI’ 4,000 years ago based on LD admixture rates.

  10. this is theoretically possible. but your priors should be that this is unlikely. you see HUGE Fst differences btwn iran early neolithic and levant/anatolian neolithic. there was a lot of local structure. i see no reason why the area to the east of iran wouldn’t be like that, as opposed to the more admixed scenario you see today.

    Here there are few things to consider :-

    If there was significant population differentiation between Iran Neolithic and groups towards to its east (just as is the case between Levant, Anatolian & Iran Neolithic) what would that population be like ? ASI ? Or some sort of East EurAsian in Central Asia ? ASI is more concentrated in South India and has SE Asian affinities so I highly doubt that it was originally spread across NW India.

    Moreover the Mehrgarh Neolithic is clearly related to Iran Neolithic. This has been not by its excavator J F Jarrige. This is also the basis of the argument that South Asian Neolithic is derived from Iran Neolithic. In contrast the Levant Neolithic and Anatolian Neolithic are poles apart from Iran Neolithic.

    One can argue that the Mehrgarh people came from Iran while the indigenous hunter gatherers of South Asia would have been like ASI. But there is no clear evidence for this. Mehrgarh Neolithic had indigenous Zebu cattle which is lacking in Iran_N. Also South Asian Mesolithic groups in the Gangetic plains are known to have been quite robust and tall, much taller than the European Mesolithic groups.

    More importantly, Iranians and South Asians share several haplogroups which show divergence dating to the end of LGM such as ydna Q3, J2 & mtDNA U7. Metspalu et al already noted that the ANI like ancestry in South Asians is already separated from its peers in the Caucasus and Iran by atleast 12500 years.

    Lastly we may also consider how CHG, Iran_N and Iran_Hotu are so geographically spread out and yet are so close to each other in relation to other Mesolithic groups from the Levant and Europe. We also have this evidence that there was some eastern admixture in the Villabruna with affinities to CHG and ANE. So it could be that a meta group broke up during the Mesolithic and its various descendants spread out leading to CHG, Iran_Hotu, Iran_N etc. Some of these descendants may have been in South Asia with greater EHG affinities.

  11. If there was significant population differentiation between Iran Neolithic and groups towards to its east (just as is the case between Levant, Anatolian & Iran Neolithic) what would that population be like ? ASI ? Or some sort of East EurAsian in Central Asia ? ASI is more concentrated in South India and has SE Asian affinities so I highly doubt that it was originally spread across NW India.

    ANE is found on the fringes of central asia, and it’s nested within the “west eurasian” clade which excludes east eurasians, and, basal eurasians.

    to me it looks that central asia was very low pop density, creating the gene flow barrier btwn east and west eurasians that is so striking.

    Metspalu et al already noted that the ANI like ancestry in South Asians is already separated from its peers in the Caucasus and Iran by atleast 12500 years.

    what is the paper you are citing? (he has multiple) i am generally skeptical of taking coalescence dates too seriously tbh.

  12. What’s the reason of Bengalis having East Eurasian dna?

    it’s a pulse admixture ~1,500 years ago (to a first approx.). no one knows the details, but i think it might have something to do with the tai migrations (the admixture isn’t just austro-asiatic).

  13. what is the paper you are citing?

    https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3234374/

    i am generally skeptical of taking coalescence dates too seriously tbh.

    But it does give us something to work with. And the coalescence date I m arguing is not just based on the above paper but also other papers such the ones on Indian ydna J2, ydna Q3 by the team of Balanovsky, mtdna U7 as well as the Silva et al paper which argued that the earliest expansion date of a West EurAsian mtDNA lineage in South Asia is about 21 kya.

    So when CHG was discovered I recalled this date by the Metspalu team and surmised that the ANI was probably already separated from CHG by that time.

  14. “First, it is a fact that the ancestors of South Asia’s non-Hindus were Hindu.”

    Not necessarily. Sindh (including most of modern day Balochistan and southern Punjab) for example was mostly Hinayanic Buddhist prior to Islam.

  15. Buddhism is as different from Hinduism as Christianity is from Judaism. If a non-Indian converts to Buddhism, they become a true and full Buddhist. But no non-Indian may truly convert to Hinduism (I tried!). For example Japanese Zen is very true to Buddha’s teachings, and to later Indian Buddhist acharyas such as Shantideva. But no way you would describe Japanese Zen as Hindu.

    One of the key Tibetan teachings is Lam-rim:
    https://en.wikipedia.org/wiki/Lamrim

    – Taught by Atisha, who was a Bengali. In fact his concepts of ‘madhyam adhikari’ are central to Gaudiya Vaishnavism – the main Bangladesh form of ‘Hinduism’. The Bengali Narrottamadas converted the Manipuris to the same religion, whose founder Chaitanya was partly modelled on Buddha, filtered through Shankara’s Vedanta.

    Before Shankara it seems much of northeast India at least was Buddhist. The first texts in Bengali, Charyapada, were Buddhist. Bengali Hindu tradition is no more than a few centuries old. I have lived with traditional Gaudiya Vaishnava ‘Babaji’ monks whose practices and beliefs are a mix of Buddhism and Sufism, while revering some Sanskrit texts.

    Scholars have recently noted that Buddha was interpreting and quoting the Vedas and IE traditions. At a time when modern Hinduism did not exist. Using the word Hindu to describe all this can be just wrong. India was as much or more Buddhist than ‘Hindu’ even in the cultural sense, only Brahmins were truly Hindu. Yet to describe India as ‘formerly Buddhist’ doesn’t help either, I concede.

    There was a once popular book, ‘Imagining India’ in which the author concludes that Indians are united by a unique bathing/washing routine. It’s true, even south Indian Cistercian monks have told me they can not tolerate the non-bathing of their European brothers. The whole mouth-washing ritual at dawn can be witnessed in any village. Cup and bucket bathing…

  16. “Buddhism is as different from Hinduism as Christianity is from Judaism.”

    This is rather debatable. Buddhism lacks some of the core pillars of Hinduism, most notably the presence of ‘varna’ or Brahmanic caste divisions as a pre-requisite.

    I don’t think it’s fair to say that all South Asian Muslim’s ancestors were Hindus just because ‘Muslims would have recognized their practices and beliefs as Hindu’ (even though many a times this isn’t the case).
    Would the religion of the Kalash be considered ‘Hindu’ for example?

    Many regions of the south asian North-west around the Indus Valley were considered mlechha (foreign) by Brahmins where ‘dharma’ was not practiced and Brahmins were advised not to go. This is mentioned in many texts such as the Mahabharata, Ramayana, Puranas etc.

  17. “Gujarat seems to be a state with a lot of connections to Sindh, so you see a wide range of variation in this state on the ANI-ASI cline.”

    Lohanas are migrants to Gujarat from Sindh (this is well attested to by their history and now genetics as well). Average Gujarati Patels, Vanias, Solankis etc. cluster with mainland Indians or south Indians. Historically, Sindh was quite seperate and isolated from Gujarat with the marshlands of Kutch being a firm divider between the two.

Comments are closed.