Admixture (Ref3 K=11) HRP0171-HRP0180

Here are the admixture results using Reference 3 for Harappa participants HRP0171 to HRP0180.

You can see the participant results in a spreadsheet as well as their ethnic breakdowns and the reference population results.

Here's our bar chart and table. Remember you can click on the legend or the table headers to sort.

If the above interactive charts are not working, here's a static bar graph.

HRP0171 is our 2nd Tamil Vellalar from Sri Lanka and the results are similar to HRP0169.

HRP0172 has 1/16 Romani ancestry. The Onge component is about 0.4% which could be noise or possibly evidence of a South Asian connection via the Romany.

HRP0174 and HRP0176 are Kerala Nairs.

HRP0175 is a Georgian Svan and pretty similar to HRP0138 (who is Georgian but not sure which local ethnic group).

HRP0177 (Azeri) is a bit more northern European than HRP0083.

HRP0178, our first Punjabi Khatri, has admixture results more like the Punjabi Jatts than Punjabi Brahmins.

HRP0179, who is 7/8 Turkish and 1/8 Kurd, has the highest Siberian component (5%) other than the Kazakh participant.

HRP0180 is our first Pashtun even if he's only half-Pathan (the other half being English). I have heard grumblings on the net about the HGDP Pathans not being representative of the Pashtun tribes. If we use the HGDP Pathans and 1000genomes British averages to estimate HRP0180's recent ancestry, we get 45.5% Pashtun and 54.5% British. So it seems that the HGDP Pathan samples are reasonable for at least this individual.


  1. - The Nairs from Kerala are practically identical to South Indian Brahmins in terms of admixture proportions. A reflection of the fact is that HRP0176, who is half Nair and half Namboothiri Brahmin is indistinguishable from the other SIBs. At least in terms of HAP Ref.3 K=11; the two groups seem very much alike. Since the European component here is not Northern European-specific (it seems to have a gradual cline, maximal in the Finns/Lithuanians and gradually decreasing till the Tuscans/Sardinians a.k.a Southern Europeans. I wonder if there would be a difference between the two groups as far as the Euro fraction is concerned, if NEU and SEU were clearly differentiated. In Dienekes initial K=10 analysis, DOD720(=HRP0124) had 2% Southern European but practically no NEU, unlike the South Indian Brahmins, who instead had Northern European admixture(f.ex :- DOD395; DOD327; DOD331; DOD451; DOD464; DOD527).

    - Finally, a Khatri participant. While caste does indeed seem to be somewhat 'ancestrally arbitrary' in Northwest India (no particular correlation with ethnic group or caste), I predict that if we had a substantial number of Punjabi Brahmin samples, HRP019 would be an outlier. Just speculating, of course. Another interesting aspect about NW Indian admixture scores is that while South Indians tend to have more East Asian, NW Indians, if they have any East Eurasian admixture, is usually Siberian or American (seemingly more 'Northerly').

    Apart from all this, wow, submissions have indeed slowed down. The last batch was practically a month ago!

  2. still the most asian of them all :->

  3. I calculate ANI of 49% for HRP0171 close to the 51% for the other Vellalar HRP0169. The two new Nairs HRP0174 and HRP0176 each have ANI of 58% about the same as for all Brahmins south of the Vindhyas. The Punjabi Khatri HRP0178 has ANI of 74% in the same range as that of the Jatts.

    I don't agree with AV that HRP019 is an outlier. His ANI of 68% is close to the 70% value for the other Punjabi Brahmin HRP004 and the 71% for the Rajasthani Brahmin HRP0033. Reich et al. have reported ANI of 71% for Kashmiri Pundits. So we can say with some confidence that the ANI for North-West Indian Brahmins is about 70%. This is somewhat lower than the value for some Kshatriyas and Shudras from the same region and the explanation is mixture with Brahmins from the East and South.

    • We had a lengthy discussion about the validity of caste status as a measure of predicting ancestry in terms of population genetics in North-West India, here. Given the tribe-like workings of Punjabi society, I don't think Jatts can be deemed clear-cut Shudras especially given the variety of occupations they took up traditionally. Also, the reason HRP0019 might be less ANI (it's only minor - 2%, but FWIW) than the other Punjabi Brahmin is because he mentions he has some Muslim ancestry, and it has been found that Muslims have slightly more cosmopolitan ancestry than their Hindu counterparts in the subcontinent. Compare the reference Sindhi samples, who exhibit African admixture, with the Sindhi Hindu participants of the HAP (HRP62; HRP085; HRP108), for instance. Also, in what historical context does Punjabi Brahmins mixing with Brahmins from the East and South fit into?

      • My hypothesis for Panjabi brahmins having slightly less European + SW Asian than Jatts and now the one Khatri is that the genetic contribution of later, historic intrusions into the subcontinent from the northwest (Parthians, Scythians, Kushans, Hephthalites) went preferentially to non-brahmin groups as brahmins considered them outside the pale of Vedic orthopraxy and would be less likely to intermarry with them. Thus, Panjabi brahmins likely preserve the genetic results of the original admixture between Indo-Aryan migrants and post-IVC remnants.

        • How about this proposition: Brahmans are from Madhyadesh (essentially modern UP) and wherever they went they became influenced by the local population - in Kashmir more like Kashmiris, in the Punjab more like Punjabies, in Sindh like Sindhis, in Bengal like Bengalies, and in the peninsula more like Dravidians, etc.

          • Since the direction of Indo-Aryan cultural expansion has been from the northwest to the southeast, I would say that Panjabi brahmins likely represent the remnants of the ur-brahmanic population of the Kuru kingdom that expanded from Panjab proper to the upper Ganga-Yamuna doab circa 1000 BCE (historic Madhyadesha). From this group, there was mostly male-mediated brahminic migration to other parts of the subcontinent like eastern UP, Bihar, Bengal, Orissa and the Dravidian lands. Though there may have been some back migration of brahmins into Panjab, I doubt that this was the founding population of present-day Panjabi brahmins.

        • If you are going argue or hypothesize about genetic contributions by 'intrusion', its not as if brahmin women or any other caste women would have a choice whether to get raped or not (unless they decided to burn themselves). As far as marrying within your 'caste' or tribe, jatt traditions are about as rigid or strong as any in the subcontinent. The so called 'honor' killings today are more likely to happen in jatt households.

          • I actually did not specify whether the genetic contribution of 'intrusive' historic populations (i.e. Parthians, Scythians, Kushans, Hephthalites) was male- and/or female-mediated and I certainly did not mean to imply a military, imperial policy of rape being the primary modality for the introduction of European and SW Asian genetic elements into the middle castes of the Panjab. Rather, I think there was a level of folk migration that accompanied these consecutive Central Asian dynasties that were centered in the Panjab in the early centuries of the common era. These groups were subsequently incorporated preferentially into mid-level castes rather than into brahminic groups due to certain prejudices that prevailed among the latter. I really cannot speak to how present-day caste-specific practices relate to population genetics that have their roots in events that occurred two millenia ago.

          • Interestingly enough many of the Jatt clans have names that reminiscent of some tribes mentioned in some ancient texts. The Sura tribe is mentioned in the Mahabharata Bhisma Parva. There happens to be a Jatt clan called Sura. The Paradas find a mention as one of the many mleccha tribes that fought in the Mahabharata (7:118). There are again two Jatt clans with similar names - the Paroda and Parodiya.

            Gnaeus Pompeius Trogus' Prologue, Book 41:
            "The report on the history of the Baktrians first speaks of king Diodotos by whom this realm was founded. Next, under which Scythian tribes' rulers – namely the Saraucae and the Asiani – Baktra and the country of the Sogdians was occupied. Next, under which [Greek] ruler Scythian tribes, namely the Saraucae and Asiani, occupied Bactria and the land of the Sogdians"

            A Jatt clan with a similar name - Asii/Asiagh

            Vayu Purana (47, 44)-
            "The Oxus river goes to the sea after irrigating the lands of the Sandhrans, Tukharas, Lampakas, Pahlavas, Daradas and Chhakans."

            Jatt clans Sandhran; Darada; Paroda respectively again are very similar.

            It looks as if many of the mlecchas of the classical period have been absorbed into the greater Jatt identity and have survived into modern times.

      • Adi Shankara was born in Kerala and traveled all the way to Kashmir as well as all over India. Of course Adi Shankara was a monk and celibate.

        The temple priest of the Shankaracharya temple in Srinagar even today is from Kerala.

        Brahmins from U.P. seeking employment witn a royal or wealthy business patron could have migrated to Punjab in significant numbers and if they became successful, they would have become integrated into the Punjabi brahmin community. The population of U.P. is much larger than that of Punjab. And Malthus teaches us that a surplus population is quickly generated!

    • How do you calculate ANI and ASI?

      • The terms ANI and ASI were created by Reich et al.

        They calculated ANI values for various Indian and Pakistani populations using statistical methods that they invented and described in their paper.

        Zack has obtained the genome data for the populations that Reich et al. used. He has published the admixture analysis for these on the HRP website.

        We can get a good estimate for ANI using ANI = 32% + SW Asian + European + 0.18 * S Asian.

        ASI + E Asian + African = 100% - ANI

        African is negligible except for a few individuals. I believe E Asian is small but significant but it is hard with these data to estimate.

  4. Hi Zack. My comment, for whatsoever reason, is awaiting moderation (never seen this before). Kindly accept :).

  5. @ RGM Using your hypothesis, you would see a similar elevated genetic component of 'intrusive' in other non brahmins besides the jatts and one khatri. Rest of punjabi non-jatt/khatri are pretty much in line with punjabi brahmins. You do not see that. My hypothesis is that there was a minor geographic shift in hindu population due to Islamic onslaught. Prior to partition it was pretty common for people from Peshawar and Kabul to move around in current India. People shifted and caste system kept their original genetic imprint in place.

    • Vivek, You make an interesting point: perhaps there exists a west-to-east genetic gradient even within the erstwhile undivided Panjab and the Panjabi brahmins, Ramgarhias, etc. sampled on this site represent an eastern Panjabi sub-population while the Jatts and the Khatri represent a western Panjabi sub-population. FYI, HRP0178 (Panjabi Khatri) happens to be my wife and her family fled to Indian Panjab from the western portion during Partition. To test your hypothesis, we would require a more granular specification of geographic origins.

      • From what I observe as far as the ADMIXTURE results and the attempts at inferring West Eurasian-ASI results from the same; it's completely the opposite. It's more of a East-West cline as far as slightly higher amounts of ANI/SW Asian/European are concerned. The Pakistani Punjabis are some 2-5% less West Eurasian than the Punjabis on the Indian side, especially the Punjabi Jatts. All the unspecified Punjabis are all likely from Pakistan. Even in the case of the Kshatriyas, compare HRP106, 107, 111 and 135, all Punjabi Muslim Rajputs* to the Punjabi Khatri. They both belong to the same formal caste but there are some slight differences between them in a genetic context, at least insofar. We need more Khatris and Rajputs to form anything conclusive, though. Also, thanks for divulging that bit of information regarding your wife. Are you also a Khatri? Have you tested with 23andMe?

        * Zack - could you also add Rajput in addition to Punjabi to their ethnicity in the appropriate spreadsheet? I've had exchanges with the individual who manages all three profiles and this is what they have told me (i.e they are Punjabi Rajputs from Lahore).

        • You are conflating religion and region.

          Punjab is a variegated region and the majority of Punjabis are Muslims on the order of 70%.

          East Punjab has a significant Dalit population. The Khatri castes of both Sindh and Punjabi (the forward Sindhi castes are Punjabi immigrants in the medieval ages) were basically Kshatriyas (Kayasthas like Todar Mal) who occupied the niche moneylending zone (though not landowning/feudal) that was semi-encouraged by the Muslims.

          Comparing Punjabi Rajputs (whose locus is South Punjab, Malwa, Rajasthan, Sindh, Gujarat) with Khatris is not like for like comparison. There are certain Khatri castes that converted to Islam (Qureshis, Shaikhs particularly) that would hold up as good examples.

          You are leaping to conclusions; also the statement that "Muslim ancestry is more cosmopolitan than Hindu ancestry hence lower ANI" flies in the face of the historical narrative (and also UP data and other data throughout the subcontinent).. The idea would be that if anything the post-Islamic Turanian influx would have wholly assimilated into the Muslim populations; whereas the Jats may have had discernable influence (also in the West Punjab Jats wholly converted to Islam, only in Majha Punjab where Jats had not pledged to Islam they reverted to Sikhism whereby then Khatris for the large part remained Hindu but continued the tradition of raising one son Sikh etc etc..)

          Also your Sindhi examples don't fly because half of Sindhis are forward caste Kshtrayias (amil bhaibands, of possible Punjabi descent and the other half are Haris.

          Sindh's ethnicity did change (many Hindu castes in nearby Gujarat, Haryana seem to be emigres from Islamic invasions) but there are certain Sindhi communities that need to be examined.

          The idea is that genetic data is being used without context to make certain inferences that don't really hold water. For instance East Punjab was settled along West Punjab (which was settled pre-historically) so unless ANIS were already there or there was a genocide event it doesn't make sense..

          It sort of reminds me when genetic studies were proclaiming that Palestinians were not the autochthonous inhabitants.

          I do think that all of this myriad fascination with origins (which are very irrelevant); it can lead to some spurious and counter-intuitive analysis.

          • We do need more data but from what we have it appears that Muslim Jatts and Khatris are less west eurasian than Sikh and Hindu Jatts and Khatris. As you say the "the post-Islamic Turanian influx" should have resulted in Muslim Jatts and Khatris becoming more West Eurasian.

            This paradox can be explained by the practice of Islamic elites importing large number of concubines from all over the world. Sharia is usually interpreted to mean that a man can have up to four wives and as many concubines ("captive that your right hands possess") as he can afford and wishes to have. A perusal of the Reference 3 admixture data will show that many Muslim groups have significant African admixture. For example Iranians and Lebanese have 3% and 6% African admixture respectively. In Pakistan Brahui, Makrani and Sindhi have 3%, 6% and 2% African admixture respectively.


            More concubines would have surely been imported from India to Pakistan than were imported from Africa. It is this practice that has resulted in the lower west eurasian percentage in Pakistani Jatts and Khatris.

            A recent paper on the ancestry of the marsh Arabs of southern Iraq is most interesting.


            They find that 10.4% of the maternal ancestry of the marsh Arabs comes from the Indian Subcontinent and 7.7% from Africa.

            Also interesting is this story.


  6. I have heard grumblings on the net about the HGDP Pathans not being representative of the Pashtun tribes.

    There seem to be quite a few Pashtun data-sets here? -'PO000355N

  7. { Brown Pundits } » From Harappa Dna to Britain’s social system - pingback on October 12, 2011 at 9:53 am
  8. I recently got the time to read through old posts on this site. I am curious about the SW-Asian component in South Indians. I would assume for the Keralite muslims, this is probably Arab contribution from traders(but I could be wrong), but for other populations, where does this come from? Are Persian/arab/levantine SW-Asian components different?The dendograms show difference, but it is not clear where the difference comes from. If so, is it possible to figure out which of those are closer to the SW component of South Indians? An obvious conclusion would be that it is Persian, but is that correct?

  9. It would be interesting to see the origin of the sw asian ancestries that seems prevalent in a lot of south indians. Also, Kerala represenative populations in particular, seem to have one similarity in their proportion of ancestries. The Kerala Syrian Christians, Muslims, Nairs, and Brahmins all have the same proportion of south asian (58%). Could that point to a small founder population for keralites overall? I would be curious how close it could be traced the language split from tamils....

  10. We do need more data but from what we have it appears that Muslim Jatts and Khatris are less west eurasian than Sikh and Hindu Jatts and Khatris. As you say the "the post-Islamic Turanian influx" should have resulted in Muslim Jatts and Khatris becoming more West Eurasian.

    Also becoming more East Eurasian. But from the available results it seems that Muslim Jatts and Khatris are not more Mongoloid-admixed than Hindu and Sikh Jatts and Khatris.

    So, it seems that the post-Islamic Central Asian input was so small that it is genetically undetectable in all parts of South Asia (India + Pakistan + Bangladesh + Sri Lanka + Nepal).

    • Excellent point, Onur! There is also a nice discussion of this in Razib's GNXP blog.

      The population of the Indian Subcontinent has always been so enormous that invaders could have had only a minor genetic effect. And how many Turks, Afghans and Iranians would want to make the hard trek through the deserts of Afghanistan and through the Khyber pass to earn a living as a peasant in Pakistan or India? After all there are only a limited number of jobs as emperors or high military or civil officers!

      However it is much easier to transport slaves and concubines within the Subcontient.

Trackbacks and Pingbacks: