232 thoughts on “Of horses, camels and extinct lineages”

1. See also, as amendment to my previous post:

   https://www.quora.com/What-is-the-consensus-on-Bangani-as-a-centum-language-How-did-it-develop-centum-features-and-be-devoid-of-the-RUKI-rule-What-is-its-implication-for-Indo-Iranian-linguistics

   Linguistic aspects are further elaborated upon by Zoller 2016. He concludes (emphasis is mine):

   “I have shown that at the time of Old Indo-Aryan there must have existed a linkage of lects, with Vedic just one of them. These lectal differentiations seem to suggest that the standard model of the three branches of Indo-Iranian is in need of a revision. Their existence also supports the idea of the earlier immigration of the ancestor(s) of the Outer Language which led to a strong encounter with Munda/Austro-Asiatic languages (but to a weak encounter in case of Vedic and Classical Sanskrit) which must have dominated the prehistoric linguistic area of northern India. This dominance must have extended far into prehistory because of the many parallels in the language isolate Burushaski. ”

   https://www.hf.uio.no/ikos/forskning/publikasjoner/tidsskrifter/acta/volum_77/ao_2016_cpz.pdf

2. @ Alberto

   “Any new paper on that subject that deals with the Anatolian influence in Thrace and the demise your refer to? This is pretty much a work in progress (for what I know) so still many details are poorly known. Maybe some aDNA could help too…”

   Indeed; but as mentioned; they are clearer
   I’ve sent you the info (listed & outlined also on the ‘Bronze age’ thread previously).

3. @Rob

   Got it, thanks.

   @Frank

   Yes, it would still be great to see that 3rd part of your CHG on the steppe series. After your HD crash I could never get to you by mail, but you still have access to continue with the post. If you have some new mail contact me when you’d like to write it or publish it.

4. @Alberto
   I do agree with the spirit that you convey, but think language = culture (and vice-versa) inevitably.

   In this vein, Mallory’s attempt to define I-E and Swadesh lists are effective somewhat.

   The verboten topic in genetics is whether culture derives from genes. No use poking that bear for the moment.

   Personally, as I’ve shared with Rob before, I think it is more effective to think in terms of Complexes when discussing a subject as expansive as this one.

   @FrankN
   Regarding: hata
   Don’t think it’s obvious that centum. s -> h more likely as noticed in other Iranic languages.

5. @Frank

   In your theory what would be the actual PIE/Indo-Hittite culture? Hajji Firuz?

   “and I find it very hard to imagine another place in time and space where/when IIA and Proto-Uralic could have interacted.”

   The thing is we do not know the actual PU homeland to pin point the actual relation between PU and II. Currently there are many theories on the PU homeland that, in my opinion, it’s very easy to imagine many other scenarios.

   A few examples:

   1. BMAC- Seima Turbino interactions
   2. BMAC-Andronovo interactions (if CW is Uralic as some claim)
   3. Issedones and other Scythian groups

6. @Alberto

   I do think that there is a PIE culture. We cannot fully construct it but we there are a few core ideas in all known IE cultures and religions. Eg. The stories of Kingship, the holy abode, the three brothers…etc. Some of them are unique to Indo-Europeans. Sure the stories of the thunder god change over time but at their core it’s a story of a god/king challenged by a serpent, and it is an okay representation of the ideology of those people.

7. @FrankN “My understanding is there is some of that, especially in Dardic. Kashmiri hata “hundred”, e.g., is obviously not Satem, but Centum, as is Kashmiri hun “dog”. Similarly, Khowar kuy “where?”, kyobachen ” for what?” seem to have conserved non-satemised traces of PIE “kw””

   kyobachen, kuy of khowar is seen as preserving kw is debatable. If so there are more mainstream words:

   when(eng), kabhI(hindi), keMvhA(marathi), kabE(bengali) etc where the NIA forms cannot be directly derived from kadA(sanskrit) which is strangely closer to latin derived forms (quand, quando) and slavic (kada, kedy)

   Even the often used kahAN (where in Hindi) is an odd one out vs other NIA (kitthE, kothay, kuThE) and Sanskrit(kutra) English (whither, where). None of these is conclusive by itself but I suspect that a structured study of such phenomenon does not exist.

   I was told that the English pair whither/tither is not related to kuThe, tithe (Marathi), kothay, tathAy(bengali), kutra, tatra(Sanskrit).

   don’t fully buy this … seems like inadequate research. As far as other reflexes of *kw in NIA we see a backed vowel after k in some cases but not after t and so forth. Is this significant or just random traits?… I don’t know yet and won’t have bandwidth to explore for a long time

8. @FrankN, if you imagine a PIE migration from Iran Chalc to Central Asia, then parts of South Asia are on the way and probably would not be circumvented.

9. In regard to Franks model

   * I have been interested in long duree models, however I presently agree with Kristiina in that languages are more likely to have been burst phenomena,.

   * The concern is that the social-demographic phenomena pointed to in the model seem too disparate to have carried strong language-cohering forces. Even more problematically, it would mean that half of Asia was Indo-European speaking, until the Bronze Age when it all suddenly reversed

   * It is likely that CHG ”arrived in the steppe” via the Caucasus, because at c. 4500 BC it peaks in Progress Eneolithic, not the Samara bend. But in any case, as Alberto pointed out, it really doesn’t matter which exact direction they arrived from, as they were 6th millenium forager-hunters (even if Kelteminar had some goats, etc)

10. https://www.cnn.com/2019/08/06/asia/turkmenistan-president-gateway-to-hell-intl/index.html

    I’m surprised by the cadence and meter of the language.

    https://www.cnn.com/2019/08/06/asia/turkmenistan-president-gateway-to-hell-intl/index.html

    It is strongly moratimed.
    resembling both dravidian languages with frequent gemination as well as classical Sanskrit because of frequent compounds.

    These traits are long lived and not easily transmitted. They transcend language families.

11. Rob:
    “The concern is that the social-demographic phenomena pointed to in the model seem too disparate to have carried strong language-cohering forces. ”

    Well, I didn’t yet elaborate on social-demographic features. But some are obvious:

    – Spread of Copper metalurgy,
    – Long-distance trade in precious stones (Carneol, Lapis Lazuli) and metals (gold, tin),
    – Wool sheep as key agricultural innovation (plus, in general, (south-)eastward spread of Iranian domesticates,
    – Non-light sensitive barley, originating on the Iranian Plateau, better adapted to moderate/ continental/ mountain climates than Fertile Crescent/Mediterranean varieties, and less draught-sensitive than wheat, allowing for effective exploitation of the Caucasus uplands (Sioni, KA), the Steppe, plus low-precipitation parts of Central Europe such as Kujawy, Bohemia and the Saale area after the end of the (humid) Holocene Climate Optimum.

    “Even more problematically, it would mean that half of Asia was Indo-European speaking, until the Bronze Age when it all suddenly reversed.”

    Well, for once, a good part of Asia, including Tajikistan, is still IE speaking. The reversal occured rather during the late IA to the Medieval (in Anatolia as late as 1923) via the Turkic/Mongolic expansions. That reversal equally affected the Pontic-Caspian Steppe, pre-dominantly Turkic (Khazars, Tartars) until Catherine the Great’s times, so I don’t really get your point.

    Otherwise, the village of Jastorf (the eponymous culture is commonly regarded as West Germanic homeland) lies just on the border between predominantly Germanic and Slavic toponymy. Austronesian is well alive and kicking, except for its Taiwanese homeland, where it has been relegated to endangered minority language status. Algic had virtually disappeared from the presumed homeland around Kennewick already in pre-European times, to instead stretch all the way from the Rockies to Newfoundland, and so on. Such geographic shifts were apparently anything but uncommon…

12. Frank.
    Cool. Looking forward to part III

13. @tim
    I think Shakas were there by the time Alexander came in. Definitely in greater Punjab if Greek and Persian records are to be believed.

    Not sure what you were meaning by stating you are Indian but I gather you mean it’s of particular interest to you.

14. A group of Early Medieval South Asians from Uttarakhand had J2, H, R2 and E1b (n=8 for the males). Interesting.

    https://www.nature.com/articles/s41467-019-11357-9#Sec16

15. @Marko
    The data is interesting.
    Their suppositions are idiotic.

    “Four individuals are very tall, with height in the range 184.41-187.55 cm.” Very tall? Plenty of people this tall in India.
    “it is tempting to hypothesize that the tall and robust individuals correspond to the Roopkund_B cluster of Mediterranean origin.”
    “It is tempting to hypothesize that the Roopkund_B individuals descend from Indo-Greek populations established after the time of Alexander the Great…”

    They should use their research money and travel to Crete and the Mediterranean to see all those exceptionally tall people. A cheaper option is to read historical records of the height of Alexander’s troops compared to the found locals in the Punjab.

    Also includes R1a, G2a, and T1. None of the haplogroups are surprising. Ignoring my above gripes, I’ve long suspected E1b (which is found in NW populations) is of Greek origin.

    This data will be useful to everyone other than the people on this paper. But still very good to have this data.

16. @Atri∂r

    The R1a, G2 etc. are from 1800 C. E, the R1a being typically Slavic. . The Roopkund_A group on the other hand is dated to ca 800 C.E. . So for Medieval Uttarakhand we have:

    J2a1
    H1a1d2
    H3b
    J
    R2a3a2b2c
    E1b1b1
    H1a1d2
    H1a2a1

    If the hypotheses regarding Proto-Bangani are correct, these samples are right from the area with a complex and supposedly very old presence of IE languages. See Franks comments as well.

    And yeah, Cretans probably aren’t particularly tall – weird conclusions. Seems very unlikely that they were Indo-Greeks, though. Maybe Ottoman merchants.

17. @Marko

    Which haplogroup group do you find intriguing? Which do you find rare? All of these are present in India in fairly old communities.

    As for languages, it’s useless to try and equalize language and haplgroups, especially from 800 c.e. The only time I see that as useful, in respect to early Indo-European languages, is the Bronze Age. Before and after that, useless. This in part, is how I predicted J2 in Mycenaea and Hittite ahead of the samples. The power of prediction is what confirms the strength of a model. Unfortunately, modern-day academia does not work like this so instead, most are forced to plod and waddle towards a proper model.

    I saw FrankN’s comments. I disagree with his premises; mainly because I see many of his starting premises as incorrect.

18. I find it interesting that the individuals from Group A (800 CE), and specifically those in cluster 1 within the group, are quite similar to the Swat Valley samples that go from 1200 BC to 1 CE. Now we see that population (with similar genetic structure and still no R1a) in a different location, 2000 years later than the earliest ones from Swat.

    One wonders if the R1a-rich groups were in very specific areas until after 800 CE and then started to become prevalent. The diversity of the haplogroup suggests a different story, so it might be just for random reasons that there’s still no R1a-rich samples from ancient India.

19. @Alberto

    Good observations.

20. Re; R1a in India, it is often presumed that priestly groups in India have more R1a and steppe related ancestry because they ‘preserve’ to a stronger extent an ‘original’ Indo-Aryan group rich in both. (And this as evidence for an Indo-Aryan migration directly from groups very rich in steppe related ancestry).

    But I wonder if depending on when endogamous norms formed, and the norms of Hinduism may formed more recently than often thought*, it may be that these groups (and some military ones) were actually instead just more open and simply engaged in much more long distance exchange of migrants with Central Asia during the Iron Age to early Classical period, before endogamous norms took shape.

    *The “Hindu Synthesis” may have formed relatively late – “develop(ing) between 500 BCE and 300 CE” – and as Razib Khan states on a long period of development, “the reality that I have seen … is that Indian Buddhism and Indian Hinduism existed in dialectical tension for 1,500 years between 500 BCE and 1000 CE (when Indian Buddhism was intellectually and culturally exhausted, more or less)”.

21. Alberto, Rob & other friends,

    Recently I saw some parts of a David Reich lecture recorded more than 6 months.

    One thing that struck me which I hadn’t realised earlier was that out of a total of 14 Indus Periphery samples which are going to be published in the final version of Narasimhan et al, 10 are from the site of Shahr I Sokhta itself.

    Now, Shahr I Sokhta is a very peculiar site. It is a site with the earliest evidence of woollen textiles and also the Bactrian camel from what I can recall.

    But more relevant to our discussion is the fact that the site suddenly came into existence around 3200 BC and does not have any earlier phase.

    While there is a western input in its formation there is also a strong input from the east, more particularly from the Chalcolithic cultures of Baluchistan. There is a strong evidence of Baluchistani cultural influence on Shahr I Sokhta which had led scholars to argue even a probable migration to the site from the east.

    It is also fascinating to observe, out of the hundreds of cattle figurines from the site, the vast majority of them are of Zebu cattle – clearly an import from the east.

    So, IMHO, the Indus Periphery samples that we find in Shahr I Sokhta most definitely look like migrants from Baluchistan or atleast should have come to the site via Baluchistan.

    In this context, can we find how the modern Baluchis, Brahui, Sindhi compared with these Indus Periphery samples ?

    _____________

    We also have 4 Indus Periphery samples from the site of Gonur in BMAC.

    While their provenance appears less clear, they could have come from Shahr I Sokhta rather than from the Harappans proper, since the archaeologists discern some cultural influence of the Helmand civilization and more particularly from Shahr I Sokhta in the formation of BMAC.

    So the Indus Periphery samples are not representative of the human diversity of the Harappans, rather they may be part of the ancestral groups of the Baluchistan – Sindh region. I would like you folks to share your views on this – especially if we can muster some relevant genetic data on this subject.

    Thanks.

22. Jaydeep, I’m not 100% willing to comment on that precise statement of archaeology and modelling (will leave that to others), but it is certainly a bit of an assumption open to question in the paper to assume that these subjects are representative first generation migrants from Harappa/IVC.

    In my view they are likely to be locally admixed, and as you say, even if they are migrants, may be representative of cultures along a cline between Harappa/IVC and the sampled sites (such as the cultures in Balochistan at this time).

    To solidify their assumption that these people are first generation migrants actually from IVC, they would at least need careful biomineral oxygen isotope data and probably strengthen that by analysis of grave goods. Maybe this will come in their finished paper, but it does not seem to be there at present.

    (Though there’s nothing distinctive about grave goods in existing outliers as described by supplement, however, other than that I8726 has an alabaster vessel which ” a distinctive pottery type in this furnishing belongs to a distinctive cluster of graves and is, “possibly local or northeast oriented (Kandahar area)” and same vessel seems to be present in I8728’s grave.

    Note that I8728 is the sample which is much richer in ASI and is later in the Bronze Age; I8726 is at 3200-3000 BCE, I8728 at 2550-2450 BCE.

    At Gonur, I2123 outlier is dated 2452-2140 BCE, and is an infant, suggesting being locally born.).

    The following relates to discussion here, but also that I have recently been having at Davidski’s place, and I am aware there is a bit of cross traffic on discussions here and there:

    I mean, thinking about what Narasimhan 2018 does, in hindsight what they does seem kinda strange, and maybe this has held up the paper in peer review:

    1) Narasimhan preprint identify outliers at Gonur and Shahr-i-Sohkta with South Asian (AASI) ancestry.

    2) Rather than assume that these samples are outliers admixed with locals at Gonur and SiS, they seem to assume that their average (which is in first preprint dependent on 3 varying samples!) is representative of a population hundreds of miles away in South Asia.

    3) Because the present-day populations of India are obviously not on the same cline as the outliers, they then assemble a model where a population represented by the outliers fused with separate AASI and Steppe_MLBA populations to create two ANI and ASI population, which then actually do (mostly) explain the cline.

    This seems like a pretty questionable chain of assumptions, looking at it with a bit of distance of time.

    The simpler explanation would probably be that the outliers are themselves admixed, that ASI already existed (SiS3 is close to that point anyway) and only one new population formed and entered South Asia, a population that exists beyond the present day North-South Asia cline and exists only today in form diluted by ASI.

    Hence I’d imagine them going back to the lab again to try and find more outliers and more forcefully make their claim, or to remove it in favour of remaining agnostic on the actual autosomal identity of IVC (which I would imagine they do not want to do).

    Not that I’m quite sure how more outliers at sites well outside South Asia would strengthen their case (a larger set of admixed outliers to use as a basis for their model, it is still based on a set of outliers from well outside South Asia)…

    (Jaydeep, you may differ on the model I propose for of the genesis of the ANI, but I think in any case you will find the other points stand?).

    To add, one implication of Narasimhan’s model of ASI being created late from Indus_Diaspora+AASI is that it would tend to suggest constraining the entry of Austroasiatic languages to India until after late IVC, since those populations represent an end-point of ASI+Austroasiatic ancestry. This contrasts with Witzel’s view that widespread Munda+para-Munda was already present in South Asia, and with analysis of split times within Austroasiatic (https://www.eva.mpg.de/fileadmin/content_files/linguistics/conferences/2015-diversity-linguistics/Sidwell_slides.pdf). That’s not necessary if you do away with Narasimhan 2018’s assumption of a late genesis of ASI….

23. Others may have noted all the above, and I may be a latecomer in these comments. I had a look over Alberto’s treatment of the Steppe MLBA / CHG / West Siberia admixed BMAC outliers in his posts from July last year, when having a recent look again myself, and it seems like that analysis was pretty good and quite insightful.

    Effectively as I see it, no indication of a sustained enrichment of Steppe_MLBA ancestry in BMAC (or really even that late outliers presented so far preferentially are Steppe_MLBA admixed, ancestry seeming being most plausibly in models for other sources for essentially all outliers apart from two at early Gonur, which we can’t test with G25).

    This obliges that models suggesting mediation of BMAC to Indo-Aryan are correct and proposing an early “takeover” of BMAC (circa 2000-1900 BCE), there is a substantial genetic as well as archaeological ‘kulturkugel’ effect required. But genetic and material ‘kulturkugel’ proposals in a general sense are something Reich lab seems to be trying to dodge… (Because if one genetic ‘kulturkugel’ is admitted, much of their wider framework falls into question. E.g. in this instance why a North->South language dispersal without genetic change, rather than the converse? Etc. for other proposed instances of language change).

    Alternatively there is the direct migration model across the IAMC directly into Swat, which the paper seems to have plumped for. But this will not work if Indus_Diaspora is not actually representative of IVC and the ground truth is more heavily ASI like.

24. @Matt

    Re: R1a in India, it’s still a very open question. It’s true that high castes are more likely to engage in long distance exogamy (army men probably too, for mobility reasons), so that could somewhat explain the higher steppe ancestry and higher R1a frequencies, yes.

    Another point to take into account is that there may be two different stories about R1a in India. The one about R1a-L657 (parallel branch to the steppe R1a-Z2124) so far looks like some sort of founder effect, give it’s never been found on the steppe in spite of the very large number of samples carrying R1a-Z93+ (from Sintashta, Andronovo, Srubnaya, Scythians, para-Scythians, post-Scythians…). It’s story could be not too historically relevant, and this is in fact the main clade in India and the one prevalent in Brahmins, for example.

    More important historically ma be the presence of R1a-Z2124, but this clade is relatively minor in modern India, with its frequencies rising in some groups from Pakistan and further north (Pashtun, Tajik) which have a more obvious relationship to the steppe. Though this clade could have expanded in SC Asia quite later than the 2nd mill. BC fo the little we know so far.

    No way to really know until we get aDNA from the relevant time and place(s). For me, the presence of R1a-Z93 in Sintashta had always been the single strongest point for a steppe to India migration related to IE languages. But the data from Narasimhan et al. preprint somehow diluted the strength of it, and these latest peripheral samples from Roopkund, whoever they were, but from 800 AD and some of them with 20%+ steppe levels but no R1a looks surprising again for what we would expect given modern frequencies.

    I agree too that it’s difficult to make much out of the 3 Indus periphery samples. We can be quite sure that they were migrants from India, but we can’t know if mixed or unadmixed, if representative of the main population of the IVC or not.

    @Jaydeep

    Continuing with the Indus periphery samples and SiS origins, I don’t think I can say much about that subject. Both modern populations available from Balochistan are outliers both genetically and linguistically. Genetically they have more Iranian ancestry (Chalcolithic Iranian type, containing ANF) than the rest of the populations from South Asia, and therefor no special affinity to the Indus Periphery samples. But both of them may be rather recent migrants to the area, so who knows.

    The samples we have from the BA which is not labelled as Indus periphery are very similar to the populations from BMAC area, so if they came from India, it must have been at a time where there was hardly any AASI in India. Or else, they are native to the area or migrants from the North East.

25. What would you guys say would be a good representative sample of the early Indo-Aryans? It is my belief that with the exception of the Dardic groups, they did not extend much beyond the greater Punjab in the early Vedic period, so I would be very surprised if we found anything substantially different from the Gandhara grave population.

    Granted it is a bit more complicated than Greece and Anatolia, but if in addition to the Gandhara samples we had a few samples from farther south in the Punjab, that would settle the case IMHO.

    The basal diversity of R1a-L657 seems to be concentrated in the east and the south of the subcontinent in any case, that’s why I suspect that in the original Vedic regions there might have been some turnover.

26. @Marko

    While it’s hard to say the extent of early Indo-Aryans as a whole, at least the early Vedic ones should indeed be in the greater Punjab region, estimated around the first half of the 2nd mill. So samples from that time and place would indeed clarify the situation greatly.

    Here is the talk by David Reich I mentioned above (from the point it starts talking about South Asia). He talks about he formation of the clines that Matt mentioned above (indeed a bit complicated scenario, but I became sceptic about modern clines – Europe’s one does not reflect its genetic history very well, going from modern Crete to around SHG, which is pretty meaningless). But here’s the sentence I was referring to:

    And then in individuals from Pakistan from 3000 y.a., from Northern Pakistan, in Swat Valley, we see through these chunks of DNA that have fragmented, that the steppe ancestry has been there for at least 500 years. So we now can limit when this steppe ancestry got injected into South Asia to a relatively narrow window between 4000 to 3500 y.a.

    So indeed it seems that they are going for that scenario where the Sintashta people expanded simultaneously throughout Kazakhstan (Andronovo Culture) and North India (Cemetery H, OCP) c. 1900-1400 BC. Which is indeed what would be required for the Steppe Hypothesis to work, but that doesn’t seem the most likely scenario with the (admittedly sparse) current genetic data or the archaeological data (we would expect clear Sintashta parallels if such thing happened in North India).

    Let’s see if now that the summer is ending they finally decide to publish the final version and we can see exactly what they have and what they’re arguing for.

27. @Alberto, cheers, the structural difference in R1a-Z93 subclades is interesting to me, (though I don’t know quite how to take it at present – the split is not as deep as the split within R1b-M269 that is still a bit mysterious in the somewhat analogous case in Europe, for ex, and there’s not any sudden explosion in the adna record, so in this case it seems easier to explain by sampling gaps. For the time being, at least…).

    Yes, certainly there is ultimately some broader South Asian migrant background in the outliers, though whether from places that are within Pakistan / India today or Afghanistan, and how through many generations of residence elsewhere, we don’t know.

    I’m really leveling my objection here at the assumption that they are representative first generation migrants, then a complicated model based on this.
    In no other case where outliers have been found well outside a given region under study have outliers been treated as likely to be anything other than locally admixed and representative of a population from another region.

    (Imagine if we had a different history of sampling, where steppe admixed outliers in SE Europe had come to light before Yamnaya samples, and been treated as representative of an unadmixed steppe population which then must have later admixed with EHG and CHG to form Steppe EMBA! Seems similar to what Reich lab is doing with the Indus_Diaspora samples and ASI… Perhaps the model works, but it departs strongly from parsimony and you really need a transect, other lines of evidence that subjects are direct migrants, etc to make it viable, IMO.).

28. While Alberto is right to be somewhat skeptical about estimating ancient structure from present day clines*, I do think they can drop hints, and I was having a look at the G25 reprocessed PCA of South Asia data today, and it seems to me like there may be a subtle South Indian tribal population cline that is distinct from the both the Austroasiatic cline and the general cline of increasing AASI within most of India.

    See: https://imgur.com/a/Q7GvIxG

    Essentially the above graphics show that it seems that a few South Indian populations in G25 (including the Irula, Paniya and Pulliyars) seem to be enriched in “Eastern Non-African”, but are neither a linear extrapolation of the general South Asian cline of increasing ASI, nor are they shifted towards East Asia (as Austroasiatics).

    Interestingly, this cline to me seems to suggest that it may represent the product of a population fairly close to the Shahr-I-Sohkta BA3 sample with a distinct population from South Asia which is both more similar to the source of most of Shahr-I-Sohkta BA3’s ancestry than to the bulk of ancestry in Austroasiatic groups, but somewhat distinct as well.

    That may reflect a model where Harappa / IVC was represented by Shahr-I-Sohkta BA3, and there really was further introgression as a population . This is similar to Reich lab’s model of ASI (as in the graphic from “Caravan” magazine included in the above gallery), but would have involved far less “new” introgression of AASI and AASI of a distinct type, and been more regionally circumscribed within India.

    *Reich lab might argue that castes in India means structure is much more preserved and clines much more information, but I’m not so sure about this!

29. @Matt @Alberto
    I’ll just point out that Reich Lab (incl. Narasimhan), Davidski, Razib Khan, and most commenters do not understand caste; or rather all understand it in the same erroneous way. Because of this, significant bias and errors are taking place. Ergo, none of the above are intellectual authorities on this subject and should not be quoted on the subject.

    Most understandings of “Brahmins” specifically, including those here, are flawed.

    This is one example: “It’s true that high castes are more likely to engage in long distance exogamy (army men probably too, for mobility reasons), so that could somewhat explain the higher steppe ancestry and higher R1a frequencies, yes.”

    Just look at data and forget current misunderstandings; just observe. The data then leads to proper conclusions imo.

30. Matt,

    Let me start from your last point addressed to me. Witzel’s para-Munda hypothesis does not have any leg to stand upon. It is pure guesswork with very little evidence to support it. It is nothing more than nonsense. The problem the likes of Witzel and other linguists are bedeviled with is this – if Indo-Aryan is not native to South Asia, and if Dravidian too is not native to North India atleast, what were the Harappans speaking. So they try to come up with weird theories to shore up their untenable positions.

    Austro-asiatic ancestry in South Asia is recent and is defined by y-dna O2a-M95 which is only present in some Central Eastern Indian tribals like the Munda, and nowhere else. They could have entered South Asia as recently as 4000 YBP. So their ancestry type is far from being representative of the native South Asian ASI ancestry.

    https://www.nature.com/articles/s41598-019-40399-8

    As far as the ASI ancestry or AASI is concerned, what no one so far has thought about is this – if as per Narasimhan et al, AASI split up from Onge, Papuan and other ENA ancestries since 47 kya, could it have managed to remain monolithic all these years ? I mean surely, if it was spread across the vastness of South Asia, it could not have managed to remain as a single population but rather got diversified into several different population groups by the time the Neolithic began. And in the meantime, would there have been no genetic interactions between South Asia and West Eurasia ?

    Infact as I see it, South Asian populations were not simply all AASI clones before the onset of Neolithic. Rather, there was likely composed of 3 ancestral components, one being AASI, the other Iran_N/CHG like and the last being ANE-like. There is also some deep connection between Iran_N & AASI on the one hand and AASI-ANE on the other. The Dzudzuana paper clearly showed that there is some ENA admixture in Iran_N & CHG, and going by admixture graphs, this ENA mostly likely corresponds to AASI. Similarly, MA-1 clearly has some ENA admixture which could come from some AASI like population. And the fact that the ancestor of y-dna R & Q, P1 comes from SE Asia (where AASI ancestry is likely rooted) is also quite revealing.

    Once, one is willing to acknowledge this deep complexity of South Asian genetic history, only then can one come around to understand more recent population mixtures.

    ————————–

    As far as the Indus Periphery samples are concerned, if I am not wrong, just like Indus Periphery, the Baluchis and Brahui are also off-cline from the modern South Asian cline. This is, correct me if I am wrong, because of less steppe and less AASI ancestry and excess Iran_N. So we need to check if there is some special closeness of these Western populations like Baluchis to Indus Periphery samples.

    The origin of as many as 10 of these samples from Shahr-i-Sokhta, a site which most likely received human migration from Baluchistan in its formative phase starting around 3200 BC (the SiS2 incidently dates to this very period), makes this a very valid avenue of inquiry.

    https://www.sciencedirect.com/science/article/pii/S2352409X18304875

    Rather interestingly, Shahr-i-Sokhta not only has ceramic parallels with Baluchistan and Mundigak (Afghanistan) but also with Namazga in Central Asia. To complete the picture, even Sarazm had ceramic parallels both with Namazga and with Afghanistan, Baluchistan and even the Greater Indus Valley.

    https://www.researchgate.net/publication/308795205_2014-Contacts_across_the_Hindu_Kush_in_the_Early_Bronze_Age_Additional_Insights_from_Sarazm_-_Soundings_11-11A_Tajikistan

    So there was quite early a large interaction zone already in place, which stretched from NW South Asia to Eastern Iran and into Southern Central Asia, right from the SE Caspian (Kopet Degh mountains) to the Zeravshan valley in Tajikistan. How people moved about in this sphere and what was its genetic legacy we do not know much but it should be an interesting area of focus.

    Coming back to Indus Periphery samples, the fact that :-

    1. They do not have ANF ancestry,
    2. They are the closest to the Indian tribal Birhors among all ancient Central Asian samples ( as shown in the Narasimhan et al Supplement),
    3. They have significantly higher AASI ancestry than the other Iranian and Central Asian samples,
    4. They are quite close to the Iron Age Swat samples

    make their origins in South Asia, a very tempting proposition.

    At any rate, I do not accept the idea that any population native to South Asia should necessarily have had very high AASI ancestry by default and that therefore SiS3 is more South Asian and others less so. Thats an unreasonable proposition.

    However, as I said, their geographic origins within South Asia could just as well be restricted to the Baluchistan and regions close to it. The fact that most of the Indus Periphery samples come from Shahr-i-Sokhta where a lot of Baluchistani people might have migrated and settled right from its formative phase i.e. right at its beginning, also means that those Indus Periphery samples could perfectly well have received no admixture, settling as they were in virgin territory. But ofcourse we cannot be 100 % sure. But any admixture they received surely could not have come from a population already admixed with ANF – which was well spread otherwise in Central Asia & even present in SiS1 samples.

    The Indus Periphery samples from Gonur in Central Asia are a different matter. The 1 sample we have in the preprint dates to around 2500-2400 BCE, which is right when the BMAC begins to take shape. And around this period, they was already an Indus civilization colony at Shortughai in North Afghanistan, close to Sarazm, which lasted for 3 to 4 centuries. Gonur itself, shows significant influence from Indus civilization. But it appears to me that there were greater parallels between BMAC and the slightly older cultures from Eastern Iran (the Helmand & Jiroft civilizations) especially when it came to ideology and religion. This inclines me to think that perhaps the Gonur_BA2 sample may have its origins in Eastern Iran from where most of the Indus Periphery samples originate.

    ———————–

    Overall, I think that the Indus Periphery samples are definitely not a good proxy for the main population group of the Indus civilization which centred around the greater Punjab region (stretching from western UP & Haryana to the margins of Afghanistan and northern Baluchistan). Let us wait for the Rakhigarhi dna. The samples, if in decent coverage, will be better proxies forBronze Age South Asian populations without a doubt. However there is one caveat, we should not expect that the populations across the vast spread of the Indus civilization were all homogenous. There is every possibility that there were regional population structures already in place. On the other hand, at cosmopolitan sites like Rakhigarhi, Harappa or Mohenjodaro, we can expect groups from different geographic origins and therefore different ancestry compositions coming and interacting. So we need decent sample sizes from such locations.

31. Jaydeep, I would suggest that SIS2 and Gonur2, samples *could* represent unadmixed individuals from Harappa, they also can be modelled perfectly well with SIS / Gonur and SIS3. There is nothing evident in my view in the Global25 data for them that rejects such a model, and nothing off cline about them.

    That is, there is no reason they *have* to be from Harappa, in a composition of ancestry that could not be explained by Harappa being like. (In the case of Gonur2, it is particularly unlikely that this infant was not born locally to Gonur).

    I also only say that it is less likely that outlier samples we find in a particular place are unadmixed locally than that they are locally admixed, as it is predominantly the experience in all other locations. That a sample is unadmixed and representative of another region requires more general evidence than the contrary, particularly if they are variable outliers and not one of a homogenous cluster.

    On the other points those seem like rehashes of the “pseudo-steppe ancestry from population structure in South Asia which is vastly complex and huge” argument which has been discussed before and doesn’t need any fresh comment (and still isn’t believable IMO). I would note also I don’t cite Witzel as being correct that Harappa spoke Austroasiatic / “para-Munda”, but that evidence points to an early entry of Austroasiatic to India, earlier than seems to me is suggested by a late formation of “ASI” as per Reich’s model.

32. “Ignoring my above gripes, I’ve long suspected E1b (which is found in NW populations) is of Greek origin.” — @ Atri∂r, iirc, there were a few E1b males in Narsimhan’s pre-print in South Asian IA samples. If i am not wrong then Greek incursions into NW India started from around 400 BCE. Thus, i am not sure if all the E1b in NW populations are of Greek origin.

33. Expanding on Alberto’s comments, I would be very interested to know when the herders entered the arid plains of southern Afghanistan and Baluchistan. This might then have significant implications for the population history of the subcontinent as well, especially Gujarat and more southerly regions south of the Thar via the Bolan pass.

    R1a-L657 seems to have some presence on both sides of the Persian Gulf in any case, and I suspect this is the result of a somewhat older presence there, seeing the distribution of the basal haplotypes. A somewhat older Iranian study supposedly found Iron Age Elamites to be strongly dominated by R1a-M417. Never published in western journals, so to be taken with a grain of salt, but I don’t see any reason to doubt it, either.

34. “Most understandings of “Brahmins” specifically, including those here, are flawed.
    I’ll just point out that Reich Lab (incl. Narasimhan), Davidski, Razib Khan, and most commenters do not understand caste; ” —- @Atri∂r, very true, imo , jaati is a constantly evolving structure( i can witness it’s ‘evolution’ even today). Few of the well-known jaatis were still in their formative phase in the early 20th century (e.g Yadavs of North india).
    Similarly, Rajputs are quite a heterogenous bunch themselves.

35. @Tim
    Yeah, some good examples.

    @Marko
    The answer you’re looking for is in your own comment. 😉

36. @Alberto, FrankN, Rob or anyone,

    Concerning my previous reference here to some online discussion of genetic associations between Dzungharia and Tibet, Kristiina has since detailed some specific indications of Tibetan connections to the ancient Shirenzigou samples over at anthrogenica, in posts 2797 and 2799.

    From this, the pre-Buddhist contact between Tibet and the region is firstly confirmed, going back to at least the period of the sampled ancient Shirenzigou inhabitants, and is possibly to be tied to the shamanistic forms that linked the two regions before Buddhism. But secondly, it also indicates that the eventual and general Tibetan Buddhist persecution of Mongolian and related shamanism, was not accompanied by largescale genocide of the population or took other forms, at least in this region.

    However, do we have genetic samples from the modern inhabitants of the Dzunghar Basin, to confirm the reported scale of the more recent Qing era genocide and Turkic replacement of the native Dzunghar population that was at the time significantly Tibetan Buddhist? Any (un)published samples you know of that can be compared to the Shirenzigou ancients?

37. @ak2014b

    I’m not aware of any modern population from the Dzungar Basin that is labelled as such. The area, as far as I know, is currently populated by Uyghurs, Xibo or Han people (which are not ethnic Dzungars), but even among those ones it’s not specified which samples may be from Dzungaria and which ones not so we could check for any genetic admixture from the original Dzungars (and we don’t have ancient DNA from this population either to know exactly how they were genetically).

    I guess we’ll have to wait for a specific study about this. But given the political implications (even if not so relevant today) is may not happen anytime soon.

38. Thanks as always, Alberto.

    I’ve been unable to find any published modern samples from that part of Asia myself. You’re right, it’s possible it may be considered a sensitive subject despite the time that’s elapsed.

    Discussion at Anthrogenica has it that the Indus Valley paper will appear between today and tomorrow. This may be the paper that was expected two or three years ago around this time, now appearing quite suddenly after anticipation dropped ever since samples were eventually said to be too low quality to provide meaningful genetic data. Data from an unsampled part of Asia, should be fascinating.

39. The rakhigarhi paper is online folks
    https://www.cell.com/cell/fulltext/S0092-8674(19)30967-5

40. Yes, also Narasimhan et al. 2019:

    https://science.sciencemag.org/content/365/6457/eaat7487

    I’ll try to have some post about them soon.

41. It looks pretty bad, atleast as far as them speculating about how IE got to India, but I’ve only read a short snippet.

42. The Rakhigarhi paper has been delayed for years due to quality issues, and never because of political ones as many people speculated. I’m glad to see it finally published, though it’s just one sample they could salvage.

    I’m not too sure about the quality of the sample given some contradictory results we can see in the graphs. However, it seems to be within the IVC cline, which gives more confidence on the outliers from other sites published in Narasimhan et al. as being representative of the IVC population (though we have to cautious still, given it’s one sample and not very clear if it’s high or low AASI one).

    Regarding the no migration from the early West Iranian farmers, it was more or less known that you can’t model neither the Indus_periphery samples, nor modern populations from the area as Zagros_Neolithic + AASI (or adding Sintashta for modern pops), since it requires the extra ANE seen in East Iranian farmers. And no one would expect that the Mesolithic IV area was a mix of ANE and AASI.

    That said, the analysis is not very nuanced. It’s not an all or nothing kind of thing. One can still place a decent amount of West Iranian farmer ancestry + East Iranian ancestry + AASI. It all depends on how eastern or western was the original Iranian-related ancestry. I guess I’ll have to write a post about this to explain it in more detail.

    The other thing I’ll need to explore in more detail is the IVC cline and its possible meanings. It’s probably the most interesting part but none of the papers cared to even look into it. We can still only speculate about it, but at least it will be interesting to explore the possible scenarios and their possible meaning.

43. Alberto and Atrior talked about it here, the very close relationship between BS and IA, and they do as well.Unfortunately the authors got their vectors wrong or maybe reversed.Maybe the third time is s charm?

44. @Al Bundy

    Well, I didn’t have high hopes about the interpretation of the data after what I was reading lately from the people involved. So that was more or less expected. Still, about Narasimhan et al. there are a few improvements, but we should expect better. I’ll mention the good and the bad as soon as I can read it in more detail.

45. I am somewhat disappointed, the samples I had hoped forr aren’t really there, just more of the same.

    One of the interesting bits is the addition of Late Medieval samples from Swat – by then R1a had become the dominant lineage. In the metal ages E-Z827 stands out as perhaps the most frequent haplogroup. Does anyone know if these calls are reliable?

    Also interesting: the presence of Q1 of both northern and southern variety.

46. @Marko

    Yes, after one and a half years it’s a bit of a shame that the new samples in Narasimhan et al. only add quantity, with very few exceptions.

    I noticed that interesting trend of R1a frequencies rising in the middle ages in samples from Swat Valley. It’s good that they added that the steppe admixture in the Swat IA samples is female biased (a suspicious omission in the preprint, that made no mention to the surprising lack of R1a samples), but then they had to spoil it with a speculative assertion that in other parts of India it is the other way around based on modern frequencies of R1a. It seems they didn’t even pay attention to the graph they cared to make about it, showing very poor correlation between R1a frequencies and steppe admixture. Pity they didn’t do the same with mtDNA (steppe mtDNA vs. steppe admixture) to check if that correlation is significantly better or not.

47. I only had a brief look at the haplogroup data and the paper.

    A couple of things I noticed:

    “Kushan empire context individuals from Ksirov, Tajikistan (n=5) ”
    3/5 Kushan samples are mtDNA U2e1e. Maybe Kushan movements explain the source of this steppe mtDNA in modern South Asia? One of the remaining 2 Kushan samples was R1a1a1b2a2a, which is R-Z2123.

    An (eastern) Scythian from a much earlier paper was also U2e1e and R1aZ93. Maybe related to the Kushan people?

    The second item is that the paper says “the R1a Y chromosome associated with Central_Steppe_MLBA ancestry in South Asia is also present in the Swat District Late Bronze and Iron Age individuals (two copies)”

    So both copies found in Swat are of the R1aZ93 subclades characterising Steppe MLBA, meaning R-Z2124? I wonder when the other R1aZ93, the one said to be more common in South Asians, will be found.

48. Alberto and others, how solid do you think their autosomal models are? I don’t see very strong evidence of a steppe introgression in the mtDNA (no U5 in particular). To be honest, I don’t find their clinal models very helpful, and I think that if we ever get pre-Neolithic South and Central Asian DNA they’ll be shown inaccurate.

    As for mtDNA and steppe admixture, I strongly suspect the latter was mediated through a third population like the Burusho, hence the paucity of northern uniparentals.

    I guess without DNA from PGWC and such these questions won’t be resolved.

49. “(though we have to cautious still, given it’s one sample and not very clear if it’s high or low AASI one).” — @Alberto, the paper says that the only fitting two-way
    models were mixtures of a group related to herders from the
    western Zagros mountains of Iran and also to either Andamanese hunter-gatherers (73% ± 6% Iranian-related ancestry; p = 0.103 for overall model fit) or East Siberian hunter-gatherers (63% ± 6% Iranian-related ancestry; p = 0.24). They also say that the estimated proportion of ancestry related to tribal groups in southern India in I6113 is smaller than in present day groups, suggesting that since the time of the IVC there has been gene flow into the part of South Asia where Rakhigarhi lies from both the northwest (bringing more Steppe ancestry)
    and southeast (bringing more ancestry related to tribal groups
    in southern India).

50. To come back to a point that I had made earlier, if the Painted Grey Ware horizon turns out to be similar to Gandhara, it might be possible that groups more similar to contemporary Brahmins (i. e. rich in R1a) have entered the subcontinent from the Bolan pass, for example with the horse-breeding Pirak culture.

    I suppose it will be years until we have samples from those sites at the current pacee, unfortunately.

    Of the groups in the subcontinent, Brahui/Baloch/Makrani still diverge most strongly towards West Eurasia, so something is going on there for what it’s worth:

    http://4.bp.blogspot.com/-qivBhEiq6gE/TuE4e8064eI/AAAAAAAAArw/HhNq-O6YP9Y/s1600/Metspalu2011PCA.png

West Iranian vs. East Iranian ancestry (with Vahaduo’s tool tutorial) »

---

Comments are closed.