Genetic variation across many South Asian communities

Someone in the comments posted the results from The Genomic Formation of South and Central Asia. I put the percentages with a few ratios in a Google doc. I don’t know what a lot of these groups are. Can readers illuminate? We need to be careful about the sample size, but I think there are a lot of interesting patterns in there.

Remember that “Steppe”, “Indus Periphery” and “Onge” are populations artifacts within a model. The way I explain it to people is that rather than focusing on the percentage, look at how the populations vary across the parameters. That is a pretty robust result. No matter what outgroups you’re going to use, Brahmins in most of South Asia seem to have more “West Eurasian” type ancestry than other populations (except in the NW). Because “Indus Periphery” has a minority of “Ancient Ancestral South Indians” (AASI) as part of its ancestry, the “Onge” fraction should be seen as a floor on AASI ancestry (the Onge ancestors diverged from the AASI ~40,000 years ago, so it’s a very large difference).

To me this part of the preprint really jumped out:

Later samples from the Swat time transect from the 1 456 st millennium BCE had higher proportions of Steppe and AASI derived ancestry more similar to that found on the Indian Cline, showing that there was an increasing percolation of Steppe derived ancestry into the region and additional admixture with the ASI through time

Here is a subset of the populations in the attached sheet. I’ve clearly differentiated four different groups:

Group Steppe Indus Periphery Onge
Sikh_Jatt 0.242 0.55 0.208
Bhumihar_Bihar 0.254 0.545 0.201
Brahmin_UP 0.244 0.522 0.234
Brahmin_Tiwari 0.25 0.53 0.22
Lohana 0.251 0.64 0.109
GujaratiA 0.245 0.636 0.119
Irula 0.023 0.409 0.568
Ulladan 0.014 0.378 0.608
Palliyar 0.012 0.35 0.638
Naidu 0.065 0.571 0.364
Nadar 0.052 0.573 0.375
Reddy_Telangana 0.083 0.582 0.335

What you see is enriched Onge in high caste groups with a lot of Steppe in North India compared to Sindh and Gujarat. What I think this suggests is that the matrix of the Indus Valley Civilization broke, and Aryan agro-pastoralists expanded across a landscape where there wasn’t necessarily preexistent complex society. On the eastern frontier in the Gangetic valley, relatively unmixed Aryans may have interacted with local AASI tribes, and assimilated them.

In contrast, in Sindh and Gujarat, the IVC matrix was denser, and the AASI had long been marginalized. Though Aryans moved into these regions, a huge proportion of the population remained classic “Indus Periphery”, and the AASI tribal populations were much more marginal than they were in the Gangetic valley.

Meanwhile, in South India, you have the phenomenon of large differences between lower caste and higher caste populations.

With all these data, I think the most plausible scenario is that we can imagine that the period between 1900 BC, when the IVC went into sharp decline, and the reemergence of South Asia into “history” around 500 BC, is one of massive churn, interaction, migration, and eventually social-cultural integration. Also, I assume that the Aryans must have come via a “northern route.” Steppe ancestry always seems higher in northern groups in Pakistan than southern ones (though not necessarily AASI ancestry), so a simple explanation is that this is where Aryans first entered the subcontinent. But how does this work with pastoralists?

Anyway, please give me your opinion on the Google doc.

(also, I’d kindly ask you to minimize long rants…otherwise I will just delete the comment anyhow)


62 Replies to “Genetic variation across many South Asian communities”

  1. On a personal level, those below are the most interesting. I am a Jain Vania, and I cluster between Gujarati B and Gujarati C, slightly closer to Gujarati C. So the Baniya (baniyas are called vanias in Gujarat) lines up pretty perfectly.

    The lower Jain steppe number I think is a testament to the fact that Jainism had a lot of converts from many different backgrounds. I understand that I am using the phrase “a lot” in an ironic way, insofar as Jains are less than 1% of the current total Indian population.

    GujaratiB 0.210 0.577 0.213

    Baniya 0.182 0.617 0.201

    GujaratiC 0.162 0.587 0.251

    Jain 0.143 0.555 0.302

    What I find intriguing is that the peasant class of most of the NW, especially Punjab, are quite Steppe shifted relative to the dalits of the region. On the other hand, the peasant castes of Gujarat, the Patels, are relatively not as far shifted in that direction. IMHO, this adds further credence to Razib’s rationale of a denser IVC matrix in Gujarat and Sindh.

    I also cluster with S Indian Brahmins on Harrapa as my first match on the oracle on GED match

    Brahmin_Karnataka 0.186 0.575 0.239

    1. Gujarati Patels also have less Steppe and less Onge components than MP and UP Kurmis. (Patels are basically landed gujarati kurmis)

      Further supporting dense IVC matrix.

  2. One additional point of note

    The proxy used for Indus Periphery is based on the ancient DNA of only three individuals. Those individuals also averaged 25% Onge, though one of three had a good bit more Onge than the other two did.

    Therefore, the Indus Periphery includes a 25% component of Onge. I know a straight ratio like that is an oversimplification. But I am not apt enough at the technical details to adequately understand the actual significance of this proportions and therefore certainly not sophisticated to explain them properly. Hopefully, someone with more expertise can below.

  3. Some of this data seems quite strange.

    UP Chamars having more Steppe than Punjabis? Lohana having more Steppe than Sikh Jatts?

    I’ve said it before, but there doesn’t seem to be a consistent way people are using ancient DNA for these admixtures, with the results being somewhat inconsistent.

    1. I think the more strictly caste/community based samples are more accurate. The broader ones probably are too much of a hodge podge to really get a good idea.

      But this is just an assumption based on intuition, so I am not really sure if it merits any substantial meaning. It would make sense that it would be harder to get a big enough sample to adequately and proportionally represent all groups, say in Punjab, to get an accurate set of numbers for Punjabis. However, it would be easier to get a bunch of people from the same caste/community (ie. UP Yadavs) and run their samples to get a relatively more accurate read for that caste/community.

    2. the method is the same across the samples. i didn’t put the sample sizes in there. might effect. also, it is not unheard of mislabeling to occur.

      Lohana having more Steppe than Sikh Jatts?

      this makes sense tho. if you look even without ancient DNA, ‘south pakistanis’ are more ‘west asian’ than ‘north pakistanis’, and north pakistans are more ‘european’ shifted than ‘south pakistanis.’

      this is pretty clearly due to what we now call steppe vs. iranian farmer/indpe differences. the lohanna are pretty obv. an extension of the sindhi distribution.

      1. “north pakistans are more ‘european’ shifted than ‘south pakistanis”

        Right, so why does a Northern Punjabi (Sikh Jatt) have a lower European Steppe percentage than a Southern Sindhi Lohana?

        If I’m missing something obvious, sorry.

        1. you need to standardize to onge fraction. not the absolute. no idea why, but it does look like there is more AASI in northern NW s asia than southern NW s asia. e.g., baloch basically are not on the ASI-ANI cline. pathan kind of are.

          my explanation is simple: like into swat, there was massive reflux migration up through the indo-gangetic plain into the upper indus valley region. sindh and western gujurat were shielded from that probably (in any case, there is some stuff that guju lohannas come from sindh).

          1. p.s. there is some variance in ‘jatt’ communities of AASI % too. the jatt might be a bit more heterogeneous than we think and sampling is going to cause major sensitivity.

          2. “northern NW s asia”
            “southern NW s asia”

            wow. some geographical gymnastic. i almost reached out for a compass to figure out what regions were being talked about. 🙂

            can we at least use india as a synonym for s asia? i know it will cause some heartburn to our pakistani friends, but at least it will be one less directional dimension to process in the head.

      2. Sample size is in the 43 for Jatts. Yeah Jatts are diverse, over one hundred million strong in population.

        But many of these groups are quite big. I think there are like 65 million banias.

        1. I think around 25-30 million is more accurate for Banias. Still big, but smaller than you’d think. ~2% of the pop

  4. My community is Punjabi Arain for which the mythology is that they came over from Iran with Muhammad bin Qasim in the 700s. I assumed that was BS and the origins were conversion from a lower-ranked Hindu caste. That being said a number of my immediate and extended family could pass as Iranians so maybe there is something to it.

      1. I’ve seen two decent sized Arain sets, one from Xing and one collated from Anthrogenica. The former I think is what you are referring too, as the Arain look similar to Gujjars there.

        The latter one was a bit different, as these Arain’s had the largest Caucasus % of any Indic-speaking group, the largest SW Asian % of any group besides Muslim Sindhis, but without the high Baloch % of the Sindhis. Which lends credence that Arains could have some connection to the West-Asian Muslim incursions (though their specific folk story is certainly made up).

        On the other hand, my father’s own Arain sample looked like a classic Sikh Jat, so I’m not sure if there’s just sampling issues going on between different sub-tribes or what.

        1. Did the whole Arian caste convert to Islam? That’s quite strange , i haven’t met or heard about any Hindu/Sikh Arian caste. I guess it had a different caste name before Islam (like the caste “Mughal”)

          1. Yeah this is another reason I think there could be a kernel of truth to the “we wuz Arabs/Persians” theory for Arains.

            They were the largest group of Punjabis after the Jatts and Rajputs per the 1931 census, and virtually all Muslim (99.99%). Which is weird, as almost all significant Punjabi tribes fall within the 25-75% range. You only get that 99.99% among Pathans and specific groups like Syeds/Qureishis who were designated “upper Muslim-caste” names for Brahman/Rajput converts. There’s no real solid mythos about where the Arains came from (not breakaways from older Hindu tribes), and the British really only noted them for their Islamic orthodoxy (unusual among Punjabis at the time of their arrival).

            As I hinted above, some Arains don’t even look like Punjabis genetically, but somewhat like Muslim Sindhis. Relatively low Euro%, high Caucasus/SW Asian%, but without the high Baloch% seen in Sindhis. Meaning their possible West-Asian admixture isn’t from centuries of mixing with the Baloch (like Sindhis), but straight from Western-Iranians or Levantine/Iraqi Arabs.

            [one thing, if the ancestry is western persian/arab, it should be detectable. the problem would be if it was ‘eastern persian’/afghan, which is much more like NW indian, or NW indian is already much more like that -razib]

            This is only brainstorming however, I’m skeptical of this theory without more data.

          2. For instance very few Punjabi castes Islamicise their names; the Jats, Gujjars & Rajputs remains as such.

            I’m patrilineally Qaqazai but they are a Pathan tribes who settled Hoshiapur.

            They don’t speak Pashto nor observe Pashtuns but they are a bit like the Kashmiri Punjabis (Kabir’s people).

          3. Off-topic / Deep has pre-existing author privilege.

            BP always encourage multiple voices & perspectives; I know it’s quite a surprise but I don’t quite know everything 🙂

          4. Xerxes,

            Unlikely. The Arora are related to the Khatri (both cluster together genetically, were mercantile tribes), but the Arain are distinct genetically and occupationally. Most Khatri/Arora converts to Islam also took the name Kojah/Sheikh.

          5. Yeah Arora are mercantile , most mercantile castes remained Hindu/Sikh. Also this sudden rise of a caste out of nowhere is similar to Ahoms situation . Today they are classified as OBC , even though we know they came from outside.

            Just a health warning on the rajput thing though. Pakistan is filled with these fake rajput claims.Nowhere in any Indian state do the rajputs are over 10 percent. Not even in their home state Rajasthan. In Rajastahn even jats and gujjars outnumber them in most places. To think of them being the 2nd biggest caste in Pakistan is seriously laughable.

          6. Xerses,

            Quraishys, Sheikhs, Syeds, etc are all titles adopted by mid and high caste converts to Islam (Brahmans, mercantile, administrators, soldiers, etc). Hardly any come from agricultural backgrounds.


            I’m not sure how you can call Paki-Rajputs fake when Rajput itself is a fake caste/tribe. A Rajput is literally just someone from a middle-high caste who was able to gain some measure of political power (via either administration or military service). Rajputs across all South-Asian ethnicities cluster with their mid-high caste neighbors, not fellow Rajputs.

          7. Xerses

            Why would Rajput make sense in Punjab? Perhaps few families might have settled during Mughals time and all. Its different then sindhi rajputs considering i dont think they are that big pops in Sindh. Which is accurate considering that you have rajputs ranging upto 5 percent in all neighboring states like Gujarat( Jadeja) and UP and Bihar (Purbias) . Punjab is an outlier then where somehow Rajputs outnumber even their own home state. LOL


            i dont know from where you got that anyone can be a rajput from a middle/high caste though power. This perhaps is the reason why Pakistan has so many “Rajputs”. Rajputs are few castes in India which have their lineage clearly drawn out listing out which castes are and clearly aren’t rajputs. I would go further and argue that rather than “clustering” they are actually the most endogamous of all high caste where they dont even intermarry to local Kshartiyas of their area. In Bihar they dont consider Bhumihar as their own, just to give you an example.

          8. Saurav,

            Rajputs were literally random Hindus from a variety of middling castes that were able to seize some type of power in parts of North/West India during the Medieval period. They didn’t coalesce into an actual endogamous tribe/caste until the 1600’s (largely due to Mughal incentives), though a significant number ended up converting to Islam.

            After 1700 Rajput membership had become largely restricted to descent in most of India, with the exception of the Punjab region, where high-achieving Hindus/Muslims continued to adopt the title prior to partition.

            I suppose one could say these aren’t “real” Rajputs since they entered after the bulk had decided to close membership, but Punjabis would (accurately) note that the entire Rajput construct is artificial anyway, so why should they care if older Rajputs don’t accept them.

          9. “I suppose one could say these aren’t “real” Rajputs since they entered after the bulk had decided to close membership,”

            I dont know about Pakistan, but this is how caste works in India. You dont “get” to be a Brahmin, just because you felt so.

      2. Some of the generalist sources like wikipedia posit a connection to the rayeen community of western UP. They are also small landowner/horticulturalists and sunni muslims.

        1. I haven’t seen genetic studies on Rayeen, but if anything they are a tiny subclan from the much larger Arain tribe that broke-off when they both inhabited east-Punjab, going east to settle in UP.

          I think people underestimate how large the Arain are. Arains in British Punjab (East/West Punjab, Haryana, Himachal Pradesh) made up the same percentage as Rajputs did in Rajasthan.

          1. Yes, i know quite a few arains, they seem quite prosperous and fairly well established in the state apparatus as well. Anecdotally, their appearances also seem a touch iran-shifted compared to other groups. The numbers of west asian immigrants to india was not trivial by most accounts, shouldn’t surprise us if communities absorbed some of them.
            Regarding arain not having a cognate in other hindu castes, its possible that it was the name that many new muslims took to when they reconstituted as a horticultural caste. Historically, there have been so many cycles of caste dissolution and reconstitution because of political and economic exigencies. Horticulture is one of those things that urban artisans/tradesman shift to when there is deindustrialization and de-urbanisation. This happened fairly frequently and has analogs all over the subcontinent. They may have been descendants of a more urban population that went back to the land, which would also explain a slightly more cosmopolitan lineage.

  5. What is roughly the AASI percentage in the InPe (in the genomic formation preprint, it seems to be somewhere between 14 and 42% – did you just take an average?).

    Another (naive) question: why do you have to use Onge? Shouldn’t it be possible to create an artifact better approximating AASI than Onge, and use that?

  6. A kind of facetious exercise; is it possible to, using the data above, set up an oppression model, where Onge, IP and Steppe %s are inputs?

    Something like G = Ax +By + Cz; and we select for A B C such that, for example high Steppe ancestry would show up greater privilege and higher Onge would show up greater oppression (roughly).

    I’m sure the Arundhati Roys would buy into this 🙂

  7. Hi Razib,
    Very interested to know the steppe, InPE and AASI/Onge of the kerala populations.
    If it’s not too much trouble, would you please include in the doc Kerala Nasranis, Nairs, and if you have it Ezhava or Nambudiri (brahmin).
    Would be very curious and appreciate this info.

  8. I usually refrain from commenting on genetics posts ( it agitates against my lefty-liberal sensibilities), but I will indulge myself this time.

    Some of the data points should be discarded, or at least put aside till more information can be obtained on the samples used. For e.g.

    1. Chaudhary is a common surname, used from W Punjab to E Bengal. Unless we know who this sample belongs to, it is meaningless to draw any conclusion. This particular sample used in the dataset almost looks like a S Indian population.
    2. Punjabi. There are many kinds of Punjabis. Masihi/Mazhabi Sikhs to high class Muslim Rajputs and Jatt Sikhs. The variation in ancestral components can be huge. Unless there is more information, no conclusions should be drawn.
    3. Coorghis seem to have very low steppe. Coorghis are a distinct ethnic group from Karnataka, who are somewhat famous for being very caucasoid looking among a see of dravidian people (they actually show some attitude because of this). Their ratios seem incongruous.
    4. Lingayath_TN numbers don’t seem right either. The numbers indicate a very AASI shifted community, almost like S Indian tribals. Linagayaths are actually quite a N Indian looking people.

    1. Does genotype always have to equate with phenotype though?

      I mean inbreeding can have some pretty profound effects. Not saying you are wrong about the Chaudharys or Punjabis, but I am not sure about the other two.

      Just anecdote again btw, but I have noticed that people that actually submit kits and post their pics on anthrofora tend to be among the more atypical phenotypic members of their communities. While phenotype doesn’t always equal genotype, insofar as the prior is generally just eyeballed and therefore rather subjective, I think there is still some merit.

      I think intuitively, these people would be more interested in finding out more details of their ancestry, given the relative incongruency of their looks with others of the same community around them. Perhaps, a random sample of that group would be better than what is posted with some of the kits on different anthrofora.

    2. Regarding coorgis, doesn’t surprise me that steppe is not enriched with them. They are an elite tribe of sorts, but historically somewhat antagonistic to brahmins, and quite un-sanskritized. The kodava language is a great example of a dravidian language with less prakritic/sanskritic influence. Although their distinct skin tone is exaggerated a bit, it may be a general feature of western ghats dwelling communities that have lower onge proportions due to isolation, like “malnad” gowdas and badagas. Wouldn’t be surprised if bunts also had lower onge, and if there were an east-west aasi cline in the deccan.
      Lingayats are more of a sect/religion (depending of where one stands in that debate) than a jati. That some adivasi profile tamil communities became lingayat shouldn’t surprise us. Even in the lingayat heartland of north karnataka and south maharashtra, there are many distinct endogamous sections. Much like sikhs, they would need to be segmented along those lines.

    3. lingayats is a religious sect. Its not true they all look north indians. karnataka lingayats look pukka south indian. These are Tamil nadu lingayaths.

  9. Panta Kapu have 0.645 of IP whoa! Surely something amiss here? For comparison, Muslim kashmiris have 0.605 only

    1. again you have to look at the ratios. these southern non-dalit groups have always been enriched for west asian stuff. even before ancient DNA this was obv.

    2. Indophilus

      not really, considering they would have migrated south from indus valley with other dravidian speaking upper castes before large scale admixture with aryans happened (which would have diluted indus component in kashmiris)

      1. They need not necessarily have come from the Indus valley. Some kind of a similar ancestral profile in upper Deccan and indeed the southern Deccan from an appropriate time period contemporary with the Indus civilisation could also do. In fact, I have seen on Anthrogenica where they model the Panta Kapu, et al with SiS BA3 as opposed to the averaged Indus_Periphery and not surprisingly get 80%+ percentages of SiS BA3 for these groups (with not-so-bad fits or even better fits or things like that; saying these things makes me cringe so hard because I don’t know what I am talking about at all lol) and it is possible SiS BA3 with his 42% AASI may have been a migrant to the relevant Shahr-i Sokhta village from somewhere in the more easterly parts of the Indus civilisation or more southerly regions in inner India with Indus influence (Maharashtra? even south India (Karnataka)?) (since Sindh seems to have had very less AASI) (actually, it is possible that the SiS BA3 guy actually spoke some kind of Dravidian language as a native language lol).

        1. “Some kind of a similar ancestral profile in upper Deccan and indeed the southern Deccan from an appropriate time period contemporary with the Indus civilisation could also do.”

          But ultimately the main iranian farmer-AASI admixture happened in the IVC region. Those theoretical populations in the Deccan are ultimately derived from further north at some point in the remote past.

          1. Okay, I don’t understand this very much but is it not possible that some relatively unmixed Iran_N pastoralists from Balochistan or something bypassed the developing Indus civilisation and entered the Gujarat-north Deccan area (and even more south possibly) where their primary admixture with the local AASI would happen? But yeah, I do see your point. You seem to say that no much admixture took place for some south Indian groups in south India after they migrated here at all? Like an evidence for the presence of caste-like endogamy at a much older period than normally thought for south India? I also believe in something like this because the 80%+ percentages of SiS BA3 for some of these south Indian groups like Panta Kapu, etc. is really crazy. I sincerely apologise if I have thoroughly misrepresented you there though.

  10. One small thing: at least the first two south Indian groups listed in the illustrative table of 12 rows in the body of this post (Irula (who speak Irula which is an ancient language branched off before the separation of Tamil and Malayalam and very close to them) and Ulladan (who likely speak a Malayalam language (there are several Malayalam languages in Kerala spoken by small tribal communities like Paniya, Ulladan, etc.) based on their location in the Idukki district of Kerala)) are not SC or other historical lower caste south Indian groups; they are ST groups. Palliyar also may be ST though the reader Vijay might know this better.

    Also, as Vijay noted here once before long ago, the differences between at least the historically lower end of the non-Brahmin Dravidian-speaking castes and at least some SC groups is quite less (I can also see this for the groups of AP and TS; the situation maybe the same for K’taka and TN too). Some castes and some SC groups are both like 51-49, 49-51; those kinds of percentages for Indus_Periphery and AASI. But it may be the case that those in the lower end of the historical SC ladder have very high AASI like the ST groups.

  11. My first thought on the high caste Onge % was that there was probably less AASI admixture in the south Indus Valley than the north, because the Thar desert would have separated them from the big AASI reservoir to the east. Whereas in the north, the Indus plain would have bordered directly on the Gangetic jungles, and more mixing could occur.
    How does the other existing data support/refute this idea?

  12. The Jatt samples seems to be clearly mislabeled. According to the paper they were collected in a hospital in Bathinda, which is a very vague description. It’s known in Punjab Mazhbis (valmikis) and Ravidassi (chamars) use Jatt surnames openly. So it feels like a good chunk of those samples are not jatts.

    We’ve a decent number of Jatt users with their data on tool called nmonte where we’re able to replicate the results from this research, and they’re nowhere close to this in Iran N, steppe and AASI %age. It’s laziness at best, makes me question the sampling of all the research they’ve done.

    1. If I had to guess, the 43 Jatt Sikh samples are probably extremely heterogeneous and probably have a wide range of Steppe, Iran N and AASI scores like the PJL (Punjabi Lahore) samples. Personally, I don’t think this means Jatt Sikhs themselves are extremely heterogeneous (I’ve seen the GedMatch results for 100 different Jatt Sikhs and they clearly form a cluster between Punjabi Khatris & Hindu Jats) but rather there are probably “chamar” or partially mixed individuals included. If the samples were publicly available, running them individually through nMonte would clarify the situation.

  13. When the first time comment was posted, the story of higher Onge in lower castes doesn’t seemed to make sense when I ranked the caste hierarchy in AP. It seemed like InPe had stronger correlation to hierarchy than Onge. Now that Razib consolidated it into percentages, I plotted steppe/onge vs. (InPe + Steppe)/Onge and ranked castes according to forward (high, excludes brahmins), OBC (medium) and SC (low) status in AP.
    Too lazy to upload plot somewhere to share (and can’t figure out Korean webpages now), but outcome:
    *High castes have high (InPe+Steppe)/onge ratios but indistinguishable with low and medium castes on Steppe/Onge scale. (e.g., ediga and budagajangam has higher Steppe/onge than Naidu or Vysya)
    *Within group differences in OBC doesn’t track with Steppe/onge scale either.
    *Within group differences in SC seem to be tracking more along Steppe/onge scale but still Mala and yerukali show up as anomalous.
    Doing caste ranking split by regions, so as to not to mix up geographical cline with local hierarchies, would be more useful to tease out the three-way mixing. It may be that mixing events occurred in difference sequences (and different proportions) in different regions.
    But InPe is not completely independent of onge, and so I don’t know how that would fall out if split in to true AASI vs. onge vs. Iran_neo components.

  14. Aren’t Rors similar to Jatts in Haryana? Could that mean Haryana Jatts are more like the 30% Steppe, as seen in Kalash than the 25% in Sikh Jatts? How come phenotype wise the Kalash look pretty different than Haryana Jatts, even despite Haryana Jatts looking relatively more West shifted? Did Sikh Jatts mix more because of relatively anti caste nature of Sikhism?

    “Our study, based on DNA profiling of over 200 persons in Haryana and Pakistan, reveals that Rors have strong genetic similarities with the Pathan and Kalash communities in Pakistan. The study indicates that Rors may have moved from the Swat Valley, a peripheral region of the Indus Valley region, about 1,500 years ago,”

    1. Rors are children of the Scythians?

      It may be the largest extent of Scythian population is now in the north-west Subcontinent. It would stand to reason that they would Hinducise pre-Islam; it’s fascinating though they weren’t granted particularly high caste status (Brahmin status is jealously guarded).

      1. Maybe they were refugees that left, when Islam came? They might have come after caste hierarchy established so they were relegated to shudra function of cattle herding.

        Article with quote from chief author

        “Our study, based on DNA profiling of over 200 persons in Haryana and Pakistan, reveals that Rors have strong genetic similarities with the Pathan and Kalash communities in Pakistan. The study indicates that Rors may have moved from the Swat Valley, a peripheral region of the Indus Valley region, about 1,500 years ago,” said Chaubey, senior author of the report.

        Light skin colour and average height of 6 feet were distinct characteristics of Rors, he said. “Saliva and blood samples of 154 Rors from 11 villages, including Amin, Katalheri, Sataundi and Dhatrath, were collected in 2012. Later, more samples were taken from Pakistan by our collaborators and the analysis were conducted in a state-of-the-art laboratory in Estonia,” he added.

        Rors are predominantly an agrarian community in Haryana. They are present in various parts of the state and have a dominant presence in several villages of Kurukshetra, Kaithal and Karnal. Kadian, a native of Karnal, said he wanted to end widespread confusion about the Ror ancestry.”

        I want to see these Rors and Jats of Haryana when I visit India one day. If what the author is saying is correct and they are 6’0 tall and fair skinned on average, they just might look like the Kalash. 6’0 average would make them as tall as Scandinavians. Internet images I found of Haryana Jatt Agitation (some sort of riot it looks like?) don’t show anyone that really looks Kalash. Perhaps, those images are not representative? They look like other Indians to me, maybe a bit bigger?

      2. I’m not entirely sure, but I think Razib might have be able to point it out if there were any clearly Scythian descended groups in this set. Doesn’t Scythian ancestry tend to have a few distinct things associated with it? E.g. a different kind of steppe, some elevated East Asian?

  15. Article about the study I linked

    “Our study, based on DNA profiling of over 200 persons in Haryana and Pakistan, reveals that Rors have strong genetic similarities with the Pathan and Kalash communities in Pakistan. The study indicates that Rors may have moved from the Swat Valley, a peripheral region of the Indus Valley region, about 1,500 years ago,” said Chaubey, senior author of the report.

    Light skin colour and average height of 6 feet were distinct characteristics of Rors, he said. “Saliva and blood samples of 154 Rors from 11 villages, including Amin, Katalheri, Sataundi and Dhatrath, were collected in 2012. Later, more samples were taken from Pakistan by our collaborators and the analysis were conducted in a state-of-the-art laboratory in Estonia,” he added.

    Rors are predominantly an agrarian community in Haryana. They are present in various parts of the state and have a dominant presence in several villages of Kurukshetra, Kaithal and Karnal. Kadian, a native of Karnal, said he wanted to end widespread confusion about the Ror ancestry.”

    Are Rors and therefore Haryana Jatts really on average 6’0 tall and fair skin? I looked up images of Jatt Agitation in Haryana and they look like regular Indians. Now, I really want to visit India and see this community. This is very intriguing. If what the author is saying is true, I can certainly believe they look identical to the Kalash. The google search images I found of agitated Jatts in Haryana must not be representative. I couldn’t really see anyone in those crowds resembling Kalash.

    It is interesting that they are cattle herders. I am still confused why they aren’t Brahmins. Maybe, they were shepards in the region of the Kalash and fled when Muslim invasion came? They could be refugees that came after the caste system was already established and thus relegated to shudra societal functions. Could someone with more knowledge clarify?

    1. Skin pigmentation on Rors vs other Indians given by:, where μ is the Melanin Index and N is sample count

      Population – Latitude – N – μ
      European – 48.14 – 135 – 28.32
      Ror – 29.69 – 44 – 42.16
      Nihali – 20.94 – 33 – 58.01
      Korku – 20.94 – 35 – 52.46
      Kapu – 17.39 – 272 – 43.41
      Naidu – 17.39 – 112 – 43.57
      Reddy – 17.05 – 844 – 43.72
      Toda – 11.41 – 11 – 42.96
      Kurumba – 11.41 – 24 – 56.41
      Badaga – 11.41 – 31 – 46.18
      Kota – 11.41 – 27 – 45.13
      Saurashtrian – 9.92 – 25 – 42.36
      Brahmin – 9.92 – 20 – 41.07
      Yadava – 9.92 – 27 – 58.21

      On a scale where the average for South Asian population is 1 and European 0:
      Population – M Scale
      European – 0.00
      Brahmin – 0.67
      Ror – 0.73
      Saurashtrian – 0.74
      Toda – 0.77
      Kapu – 0.79
      Naidu – 0.80
      Reddy – 0.81
      Kota – 0.88
      Badaga – 0.94
      Korku – 1.27
      Kurumba – 1.48
      Nihali – 1.56
      Yadava – 1.57
      Rors are still mostly pretty close to other South Asians in M index.

      Based on dna Rors probably fit as about 40% Steppe_MLBA, 60% Indus_Periphery, where a group like Kalash is probaby 26% Steppe_MLBA, 17% Bactria Margiana pre-Steppe, 40% Indus_Periphery. Brahui probably about 60% Bactria Margiana pre-Steppe, 15% Steppe_MLBA 25% Indus_Periphery.

      There looks like a relationship of West Eurasian ancestry with skin colour, but not steppe ancestry beyond that. Groups like Naidu and Reddy who have quite little steppe ancestry have more or less as light skin as Rors.

  16. Regarding Rors/Jatts being on average 6ft, its certainly false. Eastern European populations with more Steppe ancestry and better nutrition aren’t even 6ft on average.

    Jatts/Rors are originally from Punjab, and traditionally outside the Brahmanical-Hindu system of India. During the middle ages some migrated into NW India (Haryana, Rajasthan, UP), and became Hindus, but were still considered caste-outsiders.

    Regarding how they look, I wonder if sun exposure has something to do with it. I know Kashmiris are known for being quite fair-skinned even relative to Punjabis, despite many Punjabis actually having more Steppe ancestry. Perhaps mountain dwellers (Kalash/Kashmiri) have more cloud cover, compared to their tanned low-land Punjabi neighbors.

  17. I see that many talk about Scythians. Can someone say who were they, what was their language, what were their genes, what’s happened with them considering that on the map occupied almost whole continent (optionally – mythology, any their toponym in SA or anywhere, any relationship with Sanskrit and vedic literature)? Thanks.


Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.