search this blog

Saturday, August 26, 2017

The pseudo-steppe theory: last line of defense against the inevitable


A popular tactic used by those none too pleased with the presence of Bronze Age steppe or Yamnaya-related ancestry in South Asians is to claim that this ancestry isn't actually from the steppe. Thus, they call it "pseudo-steppe" ancestry.

This theory is based on the assumption that a population very similar to Yamnaya formed independently in South Asia, or at least as part of a cline of such ancestry running from South Asia to the Eastern European steppe.

At best, this is highly speculative and, at worst, insane; a last ditch attempt to counter mounting genetic evidence backing the Aryan Invasion Theory (AIT), or, if you're politically correct, Aryan Migration Theory (AMT). But it gets traction for the time being, including, unfortunately, among some Indian scientists (see here), simply because of the lack of ancient DNA data points for Central and South Asia.

However, we do have Iran_HotuIIIb, a Mesolithic or Neolithic forager from the Hotu Cave near the southeastern coast of the Caspian Sea in modern-day Iran. This location is not in Central Asia per se, but pretty damn close by anyone's standards, because it's next door to the Turkmenistan border.

Below is a map that I put together to illustrate the absurdity of the pseudo-steppe theory. Note that Iran_HotuIIIb is located between Yamnaya and present-day Brahmins from northern India, so if you're a proponent of the pseudo-steppe theory you'd expect him to pack quite a bit of Yamnaya-like ancestry, right? Well, he shows a slither, but much less than the Brahmins. How is this possible geographically and temporally, without a migration from the steppe to India some time after Iran_HotuIIIb was alive? Realistically, it's not.


The ancestry proportions on the map are based on an easily reproducible ADMIXTURE analysis. The test was run in unsupervised mode, but I designed the dataset to help the algorithm flesh out the so called Early Bronze Age steppe (Steppe_EBA) component. The full Q output is available in this spreadsheet.


In fact, there are several ways to show that present-day South Asians share relatively recent and direct ancestry with Bronze Age Eastern Europeans, including, for example, with a haplotype test (see here). But I'd say that a well designed ADMIXTURE analysis is an especially effective way of doing it, because both a strength and weakness of ADMIXTURE is that it's sensitive to ethnic-specific genetic drift. Thus, the so called ancestral populations that it infers are often just exaggerated signals of relative inbreeding, isolation, and/or rapid expansions experienced by founders of ethnic groups.

Clearly, my Steppe_EBA component is a signal of a relatively small, young founder population expanding rapidly across and out of the Eurasian steppe during the Bronze Age. That's because it peaks at extreme levels in genetically and archaeologically closely related Bronze Age steppe populations, such as Afanasievo, Poltavka and Yamnaya, and their recent descendants, such as Andronovo, Corded Ware and Srubnaya. In other words, ADMIXTURE has probably managed to pick up the genetic signal of a Bronze Age ethnic group.

This signal also shows up at relatively high frequencies in many geographically disparate Indo-European speakers, such as the Indian Brahmins and Icelanders. That's obviously good news for those of us who favor the Eastern European steppe as the Proto-Indo-European homeland.

Among the other ancients, those that plausibly can't have ancestry from the EBA steppe because they're too old, it's only seen at high levels in such populations as Eastern European Hunter-Gatherers (EHG), Caucasus Hunter-Gatherers (CHG), and Ancient North Eurasians (ANE). In other words, populations that contributed in a big way to the formation of the Bronze Age peoples of the Eastern European steppe.

It's important to understand also that my ADMIXTURE analysis isn't anything particularly original. Iran_HotuIIIb shows essentially the same behavior in this standard Principal Component Analysis (PCA) of West Eurasian populations, sitting just south of present-day South Central Asians (SC_Asia). This, of course, is out of whack with geography, and, again, can only be explained by significant gene flow from the north into South Central Asia after Iran_HotuIIIb's time.


See also...

The Out-of-India Theory (OIT) challenge: can we hear a viable argument for once?

Descendants of ancient European (fair?) maidens in Central Asia's highlands

Late PIE ground zero now obvious; location of PIE homeland still uncertain, but...

112 comments:

postneo said...

Iran hotu and EHG(purple) are mesolithic.
EBA yamnaya/afnasievo looks shifted towards hotu vs pure EHG.
modern SC asians are not that far from hotu despite a larger time gap.

The plot is not conclusive

Davidski said...

It's all very conclusive. You just need to pull your head out of your ass.

Davidski said...

By the way, Yamnaya/Afanasievo are shifted towards CHG. That's what the ADMIXTURE analysis shows as well, but in a more cryptic way.

And SC Asians aren't all that far from Iran_Hotu on the plot because they're pushed down by ASI ancestry, which Hotu lacks, as well as more basal Iran_N ancestry. Have a look at the ADMIXTURE spreadsheet.

Anonymous said...

I didn't know there was this pseudo-Steppe thing, where they propose an Yamnaya-like people forming close to their borders.
So, if I understand correctly, they don't even deny that the ancestry must come to an Yamnaya-like people, they just don't want them to be from Europe? Well, this really confirms what I've been thinking about this whole AIT thing for a long time: After the British left India, Indians took the cool-aid of Humanism and mixed it with Nationalism to create a deep hatred of Europeans.
I can see many Indians when I travel down south here in Italy, farmers have to hire them because the youth is either non-existent or is going to the cities, and they're really very Hinducentric even living in other people's country, they're disrespectful.
This whole thing just makes me more and more biased, specially now that I discovered their main case for being against AIT is precisely because of Europe.

There's no other explanation for people having EHG aDNA in India, there's no other explanation for them having a R1a yDNA derived from Europe, there's no other explanation for them speaking an Indo-European language and having an Indo-European religion (very changed now from Vedic times).
It was was infused there from the outside, AIT is a reality.

truth said...

Dave,

Could you include north-africans on the spreadsheet ?

Jijnasu said...

@JohnP
Perhaps you should educate your compatriots who propose an Out-of-Italy theory first especially the ones who are active commentators on this blog. Out-Of-India or Out-of-East Europe Hindus are one of the few groups who can claim a direct culture connection with their Early Indo-European ancestors. The very hymns sung over three millenia ago continue to be chanted with little change.

Anonymous said...

@JohnP

I think David's post is in response to Rob's recent comment, as David said he'd do a post on "pseudo-steppe". No one else seems to have mentioned anything like it on that page, though I can't be sure I didn't overlook anyone's comment. As far as I can tell, David's current post seems to be in response to Rob.

However, Rob only mentioned something EHG-like, not Yamnaya like. On the other hand, he had earlier suggested the possibility of something Yamnaya like in South Central Asia since the Mesolithic:

Rob said...
...
On the other hand the steppe in Indo-Aryans looks like Yamnaya stuff, but could be from much older Mesolithic populations in south-central Asia.
December 17, 2016 at 6:53 PM


Maybe he's not the only one who has made such suggestions, but I can't remember any one else.

Guess you don't read the comments, John? Rob, Alberto and David tend to be good to read.

Anonymous said...

@David

And SC Asians aren't all that far from Iran_Hotu on the plot because they're pushed down by ASI ancestry, which Hotu lacks, as well as more basal Iran_N ancestry.


On that. I want to ask about its relation, if there is any, to your earlier comment below,

"According to the Basal-rich K7, Iran_Hotu packs more ANE and more of some other, perhaps Central Asian, stuff that often looks like Andamanese-related admixture.
...
This might be some sort of ancient South Asian-derived ancestry that has been floating around Iran and Central Asia since the Mesolithic.

My new test might be able to spot more differences between Iran_Hotu and Iran_Neolithic. We shall see soon.
November 22, 2016 at 7:08 PM"


I might just have misunderstood what you meant, but back then I simply assumed it to mean there was something "ASI-like" in Hotu.

However, is there some intended distinction between ASI and Andamanese-related admixture that's some sort of South Asian-derived Mesolithic era ancestry in Central Asia, so that Hotu has the latter kind but not ASI? Or are the two referring to similar things, but it's that further analysis of the Hotu sample since the earlier comment has ruled out anything ASI like after all?

bellbeakerblogger said...

Anyone have access to this?
"Drilled teeth and shell artefacts from a graveat Prague-Březiněves and a review of decorative artefacts made from animal material from Corded Ware culture in the Czech Republic"

Anonymous said...

@Jijnasu
If you're talking about the whole R1b thing, it's up to you to find a sample older than Villabruna, until then I have nothing to say.
About Indians being the last bastion of Indo-European religion, I really appreciate that (Tengrism is too, but it's Turkic now and it's better to not dwell on it. Politically Corrects at Wikipedia locked the Tengrism article and removed every reference about its Indo-European roots. They did that to the Iranic Peoples as well, I think the next line of attack of the PC crew will be shutting down every article related to the new Indo-European questions and discoveries, they don't like it).

@ak2014b
>>Guess you don't read the comments, John?<<
I read comments, since a long time ago (specifically, I've been browsing these sites since the whole Diniekes vs Davidski thing), I just almost never post, so it may seem I'm new or anything.
But in Davidski's body of article, he talks about "Indian scientists" so I just made the connection on point. And I remember he discussing with Rob and saying he would create a post.

Rob said...

@ AK
To clarify, by "local" I mean hypothetically anything between Hindu Kush and south Siberia, and I don't think there's anything "pseudo" about it
As I said, the main reason is the lack of the KA / Iran Chalcolithic admixture which we know spilled over Iranian plateau and probably Turkmen in 4th- 3rd millennium

Rob said...

But I'm not sure of any coherent synthesis for such a hypothesis
The specific dating of Kelteminar etc remains work to be done, but it seems to have been "swept away" (?) by Andronovo
Of course, some early Andronovo dates are coming from the East too, which might mean we're not looking at one-way traffic

Unknown said...

Hey Davidski, I've heard (idk?) you support a northeast Africa (Sudanese) Nubian proto-Afroasiatic homeland.

Is it possible you can make your opinion on proto-aa homeland knowing the Nubian E-M78 looks like the "main" proto-aa marker alongside a less important A-M13, with T1a1a being brought or having been present at an early date of proto-aa or post. I personally believe that the intially proto-aa pop was mix 50-50 East African/North African, with the North African population having T1 and some A-M13 and the East African carrying E-M215 (E1b1b) pre-M78. The North Africans already were 20% or so East African and carried A-M13 and L3 and L0 lineages likely.

However, the confusing thing for me lies in the mtdna and arena of East Africans. The seem 40-60% Eurasian in mtdna (consider some 5-10% of the L3 and such is likely Eurasian, probably 50-70%) why is there more Eurasian mtdna than ydna (which peaks in Somalis at 10-15% T1a1a, a clan-based founder effect). This is present in the long-severed proto-South Cusjites (split from East Cushites 6000+ years at minimum), who still were seemingly 90% E-M215 (with a founder effect of non-M78 lineages) and 10% T1a and 50% Eurasian mtdna.

Since bloggers like Mathalida (a prolific Eurocentric, less science only ideology) proposed that M78 arose in a mixed population (as I long believed) and that the Nubia was always mostly East African pre-Nilo Saharan migration (with a higher Eurasian ancestry pre-Nubian 40-60).

Anyways certain E-M215 lineages seem non or maybe para-aa like E-M81 and the Natufian E-M215 lineages. Certain Ethiopian Highland E-215 lineages are also local assimilated lineages. I'm torn on E-V6 however.

It's interesting to me how much proto aa ancestry populations like Berbers have (20% with earlier 10% pure East African ancestry - pre-Slave trade Niger Kordafian and Nilotic ancestry). The constant erosion of East African in Middle Eastern populations can be constructed to a pre-Bronze age Berber porportions of proto-aa.

Natufian (5-10% -debated). I believe these folks were para-aa possibly or influenced and not proto-Semitic related.

Pre-Iran Neolithic (30%) pre proto-Semitic admix - European Farmer and ancestor Anatoalian farmer were influenced by this population, since they carried E-V12 and M78 and very rare A-M13 lineages post-Natufian Nubian migration to Levant:

15% East African, 30% proto-aa. Mostly E-M215 and T1a with minor G and R1b. Significant L3.

Proto-Semitic Iran-Neolithic admixed - near-current levels of pre-slave trade pure East African ancestry in the Levant and Arabia

10-6% East African 20-12% proto-aa. Founder effects of Iran Neolithic lineages of J1e and influx of West Eurasian lineages. Minor L as well. Reduction of E-M215 and L3 lineages.

Askhenazi Jews (50% Southern European) are 1-3% East African, derived from a 5-6% population of exile Jews from pre-Islamic Levant and probably minor admix from Philistines and other non-aa pops with lesser to no East African.

This could be reconstructed almost exactly similarly in Egyptians, who became less EA with Iran Neolithic and later Amru (Asiatic) migrations and invasions. South Became increasingly South Sudan influenced and early Nilo-Saharan influence on proto-Egyptians in southern Egypt brought B and other non-aa A-M13 lineages. Also very rare E1b1a in Ramses was also brought with proto Egyptians since these early Nilo-Saharans were Kordafian-influenced maybe. Early Cushitic tribes had wholesale southward migrations to Kenya-Tanzania from southern Somalia and Ethiopia in this period and very minor incursions northwards into Levant (bringing minor E-V32 and E-V6).



Chad said...

There's no Sub-Saharan ancestry in Natufians.

Chad said...

Or Levant Bronze Age

Unknown said...

Which is what you say. Others (Awale) say otherwise.

I find both of you knowledgable which is why I said it was debateable. I'm leaning towards them having very insignificant (1-3%) East African. Which is a middle path. However in a situation of absolutes, yours seems to me more likely (admix calc), even though it would make little sense outside of a founder effect of probably a few mostly-Eurasian Egyptian individuals in a otherwise West Eurasian population. This doesn't explain the likely heavy cultural influence however (from the Harifians of northeastern Egypt), which supports Awale and the minor EA camp. As such, you can see why I'm very torn.

Anthro Survey said...

@Lathan

I also support the notion of insignificant SSA ancestry in the region during Bronze Age times, an increase from the 0% of the Natufian and possibly Neolithic eras. This would be consistent with and a marker of the spread of AA in the region from a north Sudanese urheimat. These early pastoralist speakers must have been a mix of Natufian-like and SSA, perhaps in a ~75-25 proportion(respectively) which would of course have been diluted upon their northward trek.

Chad said...

There was no Sub-Saharan in any Levantine from the epi-Paleolthic through the Bronze Age. Time to swallow that pill.

blogspot said...

today racists and fascists apparently rely on outdated European race theories of the 19th century, Fascist hungarians do that, turks too, and of course hindi racists and theit western useful idiots rely on this outdated ideologies believing in the absurdity of an indian origin of the so-called aryans.

Samuel Andrews said...

IMO, mtDNA haplogroup M1 and Y DNA haplogroup E1b1b1 are definitely Eurasian. How yHG E became popular in both Africa and southwest Asia is really interesting question. It it Eurasian admixture in Africa? Is it African admixture in Asia? Is it from something intermediate between Eurasian and African, something that contributed to both Natufians and Africans?

Anthro Survey said...

@Samuel Andrews

The thing is, lineage splits are not always sharp, instantaneous process as a cladogram would suggest. In reality, there are always some double-sided arrows at the bottom of each node. In other words, perhaps there was a time when limited back-and-forth admixture was taking place between proto-OAA and SSAs before the gate was sealed for good, so to speak.

This might account for what you're referring to.

Another possibility is limited admixture between Basal Eurasians and Africans at a later----but still quite ancient---date. Perhaps 30KYA or so.

Anonymous said...

Davidski, I have Polish ancestry, around 12% that didn't show up at all in my 23andme results. Instead I got a tiny Jewish percentage that went down to 0.2% at 90% confidence. Can you give me a calculator that can help find my ancestry and determine if the AJ was noise or not?

Unknown said...

It is likely proto-Ancestral East African and arose in Ethiopia.

The migration of East Africans to West Africa brought E1b1a and l3 to prominence and wiped out A00 and B lineages. The pre-East Africa. West African.Basal Human population was carrying A00, L0 and similar lineages. If we find Eurasian atdna in several ancient ~50-40k west African Basal Humans with East African ancestry sample then we can be suspicisous. However, I believe E and E1b1b to have began in a Ancestral East African population that no longer existed by the earliest Neolithic.

Basal Human - A00

San - A1 and A-M13

Pygmy - B

OOA/para-OOA Ancestral East African - E and CT

E became East African and CT Eurasian. A clear split, like A00 in Basal Humans and A1 in San-Everyoneelse

Eurasian and Ancestral East African are sibling clade's like Western non-Africans and Eastern Non-Africans .

M1a, N and U6 are the North African, L3 and L0 are L1 were from the East Africans. E-M215 is verified to have Ethiopian origins.

It's like saying R1a-zahatever is Indian and migrated to steppe. E1b1b is Ethiopian and migrated and mixed with mostly-Eurasian North Africans in Nubia to produce mixed E-M78 and T1a1a1a with M1a and L3 and U6 and L0 folks.

Unknown said...

I saw 2%. And stop saying SSA, were talking about East African not San or something else. That's not even a genetic clade.

This is why I am very critical of your posts, it's all blacks vs the enlightened eurasians. Constant anti-African slants in all African-related things.

You should stop lying so blatantly as well. Many people take your word on these posts. I would have taken this as fact, yet I saw EA admix. So sorry, no.

Davidski said...

@ak2014b

Individuals with substantial (>10%) East Eurasian, which includes ASI, ancestry are pushed right (or south geographically) on those sorts of West Eurasian plots. Iran_Hotu doesn't qualify. He probably shows minor Southeast Asian signals because of the lack of pre-Bronze Age samples from Central Asia, which will likely form their own cluster, and this cluster will probably show up at a substantial fraction in Iran_Hotu and Iran_N. But I doubt this will affect the Yamnaya-related ancestry estimates in South Asians. Rather, it'll change the Iran_N and ASI estimates for them.

@Lathan The Great

I don't have any opinion on the whereabouts of the proto-Afroasiatic homeland. Keep in mind that my views are often twisted on many forums, and I can't take part in all discussions across the net to correct these things. But I will reiterate what Chad said, which is that the minor Sub-Saharan ancestry showing up in prehistoric Near Easterners in ADMIXTURE runs like the Steppe_EBA K7 is an illusion. It's there probably because West Africans and Bantus form a very strong ethic-specific Sub-Saharan cluster that isn't homogeneous but includes Near Eastern-related input. And then there's also post-mortem damage on the DNA, which can cause minor basal, including faux Sub-Saharan, signals to show up in many ancient samples.

@truth

This analysis is specifically designed to show the Bronze Age steppe component. It's probably not very useful for studying African genetic substructures. Adding more samples might be interesting, but probably not very useful.

Unknown said...

I meant CF and de. My apologies.

I don't believe such a Natufian-like population could existed as south as Nubia in this period. I could be wrong, I find it unlikely.

Pure East Africans were probably becoming extinct rather quick, mixing with Pygmies in the Sudan, SAN in east Africa, and Eurasians in the northern Peripheral of Nubia.

The E1b1b East Africans seem to be to proto-aa what R1a Ancient the Proto-IE.

Davidski said...

@MIND Domakr

Your Polish ancestry probably isn't showing up at 23andMe because the test at 23andMe relies on highly differentiated haplotypes to estimate ancestry proportions. And the thing about Polish haplotypes is that once they break up, they just look standard North European or European.

It's impossible to say whether your minor Jewish signal is real because it's so small. Keep in mind that East European Jews have some East European ancestry, so what you might be seeing is a signal of shared ancestry with Jews via Eastern European ancestry.

I'd just let this go, unless, for instance, you test your grandparents and one or more of them show higher Jewish ancestry than you.

Steven said...

Will the steppe calculator ever be in Gedmatch or available for download?

Davidski said...

Don't see how that's possible, because this test shows a huge calculator effect when not run from scratch with ADMIXTURE. That's due to the fact that some of the clusters, like Steppe_EBA and Iran_N, show very similar allele frequencies at many sites.

Davidski said...

By the way, anyone notice the difference between the Minoans and Mycenaeans in this test? Pretty awesome how it picked up the steppe admixture in Mycenaeans, and differentiated it from their Iran_N ancestry, which the Minoans have aplenty too.

https://docs.google.com/spreadsheets/d/1qYz-SsRykJEnlynp4MnQpjDETr4a0iZAZcAUJ8X2oT4/edit?usp=sharing

Gioiello said...

@ Jijnasu
"@JohnP
Perhaps you should educate your compatriots who propose an Out-of-Italy theory first especially the ones who are active commentators on this blog. Out-Of-India or Out-of-East Europe Hindus are one of the few groups who can claim a direct culture connection with their Early Indo-European ancestors. The very hymns sung over three millenia ago continue to be chanted with little change".

I let that JohnP answers you:
@Jijnasu
If you're talking about the whole R1b thing, it's up to you to find a sample older than Villabruna, until then I have nothing to say.
About Davidski I have to say that he didn't agree with me about my theory of an "Italian Refugium". Is his guilty to permit me to expose my opinions, what other fora don't and didn't? My theory is detailed in many thousands of letters in many blogs and fora, and, as you rightly are waiting for aDNA from Southern Asia, I am waiting that Italy (above all the Tyrrhenian one) is tested beyond the six samples tested so far against the hundreds of samples elsewhere. After the proof will be in the facts and no more in the theories.

P Piranha said...

Reiterating what David and Chad said, there is no evidence whatsoever of African ancestry in Natufian down to Jordan EBA and the Sidonians. It is not even clear that African ancestry of any type exists in the ancient Egyptians, as they show only 2% non-Eurasian ancestry when compared with Iran_N and no comparison with Natufian was done. African drift paths are extremely long, up to 10X the distance of drift paths between Eurasians, and very small fractions of African ancestry strongly influence formal stats. If Natufians had any African ancestry at all, even from Mota, the perfect symmetry that they show with Africans when compared with Eurasians will be gone.

The fact that Natufians are further from Eurasians as a whole but no closer to Africans probably means there is some para-Eurasian ancestry from North Africa that influenced them, and possibly Mota as well. We have some ancient mtDNA from Paleolithic North Africa, which is resolutely Eurasian. Berbers show a local maximum of Natufian ancestry, which is very difficult to explain other than a hypothesis that this may be a proxy for some sort of local Hunter-gatherer substrate that exists in mixed form in Natufians, and is therefore lumped with them in algorithms.

Right now, the genetic history of the Levant and Egypt is one of constant input from the North and East from sources in Asia, moving West and South, with no sign whatsoever of reverse movements until well into the historical period. The only possible movement from the South is from the transition between Levant_N and Jordan_EBA, for which Natufian ancestry increases. Unless Natufian-like populations existed in Sudan or the Beja area, the genetic and linguistic evidence has started to diverge for the AA, in contrast with the situation for the Indo-Europeans.

Anthro Survey said...

@P Piranha

The stats are significant when (Jordan_EBA, y, various SSA, Chimp) is run and y=Lebanese Christian or 3DRIF-27. How do we explain this?

Also, I had a few technical questions for you regarding drift path lengths since you seem to have a solid grasp of the concept. As I understood a while back, the length explains why even N. Euros share significantly more with Ancient Egyptians in f4 than modern Egyptians do, right?

Do you have an Anthrogenica account by chance?

Rami said...

Your previous models had pretty bad fits slightly better than before because you were using ENA shifted Scythians from Antiquity, almost 2000 years later after the first Indo Iranian nomads pushed through . Also this new chart you put up makes no sense you are modelling some populations as even much more Iran_N than they really are. Trans Caucasus and Plateau Iranians are mainly CHG not Iran_N and even the Steppe component is at most 50% CHG . So with the rationale you are utilizing then ,even Steppe can be considered part Iran_N once u take out the extra EHG, VHG lol. I would say Rob's assessment is probably the most sensible, because there is no way any Indian population is 40-50% Steppe. As for Iran Hotu ,its definitely ENA shifted compared to Iran_N and CHG which lack it completely.

Davidski said...

We'll soon see what estimates the Harvard crew come up with using their new Central and South Asian samples.

But don't expect anything less than significant steppe admixture in South Asia. This fantasy that South Asia is a magical place where anything is possible, including by sheer coincidence, the existence of a native Yamnaya people there, will come to an end.

Chad said...

You need to look at Neandertal levels. Lower archaic puts one closer to Africans. Look at Chimp/Neandertal, Mbuti; Iran_N, Natufian, for example. There's no Sub-Saharan.

Chad said...

They all have ENA since they have ANE.

Seinundzeit said...

David,

"But don't expect anything less than significant steppe admixture in South Asia. This fantasy that South Asia is a magical place where anything is possible, including by sheer coincidence, the existence of a native Yamnaya people there, will come to an end."

That pretty much sums things up.

Speaking of which, once the Central/South Asian aDNA paper is released, the reactions here at your blog will be totally popcorn-worthy. It'll be fun.

Folker said...

@Lathan the Great
History of Africa is very hard to draw given than much of the continent didn't use any form of writing systems, so was still in Prehistory a century ago, with other difficulties as climates (with rapid degradation of remains), big climatic changes (Green Sahara) and density.
From what modern and ancient DNA can tell us, there were limited and recent SSA introgression in Northern Africa and Eurasia, but with Eurasian introgression in Africa, heavy in Northern Africa (Egypt included), lower in SSA (but not limited to East Africa, given the presence of R1b V88 in Cameroon). It is a surprise by itself, as it was postulated that ancient Egyptians were a mixed population, resulting from migration from Sahara when desertification occurred.
Nubia had a complex history, given that Egypt tried to securise its southern border by "emptying" Nubia. So we have huge discontinuities there. Nubia seems to have been repopulated by foragers, but given they adopted pastoralism, a Eurasian admixture is likely.
I can't follow you about your deductions from modern Y haplogroups distribution. By example, A00 could be a relict from archaic admixture from a side branch of Homos. In any case, given the limited knowledge we have from Africa history, it is highly speculative.

Unknown said...

>Berbers show a local maximum of Natufian ancestry, which is very difficult to explain other than a hypothesis that this may be a proxy for some sort of local Hunter-gatherer substrate that exists in mixed form in Natufians, and is therefore lumped with them in algorithms.

It's possible Natufians were a mix of a less basal Berber-like population from North Africa and more basal one from for example Arabia who met in Levant. That's why some calculators detect Natufians as a mix of Mozabites and Southwest Asians. I think Arabians were more Basal than North Africans because Copts look much more South European-like than some Saudis and Bedouins with elevated Natufian ancestry despite having seemingly very similar ratio of ancestral components but that may be due to drift or selection. So it's possible North Africans have little to no real Natufian admixture but just ancestry from one of populations that contributed also to Natufians.

>Also, I had a few technical questions for you regarding drift path lengths since you seem to have a solid grasp of the concept. As I understood a while back, the length explains why even N. Euros share significantly more with Ancient Egyptians in f4 than modern Egyptians do, right?

That's probably due to significant East and West African admixtures in modern Egyptians. Africans and especially Pygmies and West/Central Africans with "Basal Human" admixture are highly divergent from all Eurasians. For similar reasons Sardinians have longer drift path with Natufians(and if I remember right Ancient Egyptians) than modern Saudis who've significant Iran_Chl/CHG and to lower extent African and South Asian admixtures.

Unknown said...


>IMO, mtDNA haplogroup M1 and Y DNA haplogroup E1b1b1 are definitely Eurasian.

Not sure about M1 but E Y-DNA may be Basal Eurasian's Y-DNA.

Samuel Andrews said...

@Unknown,

There's two reasons I think mHG M1 is Eurasian.

First mHG M itself is rooted in Asia, mHG M1 is the only branch that exists in Africa.
Second the ancient Egyptians had lots of mHG M1 but hardly any mHG L and hardly any African ancestry.

P Piranha said...

@ Anthro Survey,

Very, very unlikely that Jordan EBA have any ancestry from Mota or more basal than Mota. You can look at Lazaridis et al paper on Middle Eastern farmers supplementary information pg 41-43. Jordan_EBA/Levant_BA falls within rank = 2 (i.e. only two ancestry streams from "Basal Eurasian ancestry" which is less basal than Mota plus Eurasian ancestry proper) with p-value 7.90E-01, which is in fact one of the better p-values.

Also table S7.3: "Ancient West Eurasian populations other than the “Four” do not require an additional stream of ancestry from the outgroups", Levant_BA modeled as (0.583 : 0.301 : 0.134 : -0.018) (Levant_N : Iran_N : WHG : EHG) with p-value Levant_BA 7.96E-01, which is again one of the better ones.

Its possible the SSA shift when compared to Lebanese Christian is due to Basal Eurasian ancestry in Africa, which may exist at low levels throughout the continent.

The Sidonians cluster more or less exactly with Jordan_EBA ("Levant_BA" in Lazaridis) and like them carry Y chromosomal clades strongly associated with Semitic expansions. The fact that there's no evidence of Mota or others in them is quite significant.

Unknown explained the drift paths well, nothing much to add there.

P Piranha said...

Just realised the hypothesis of Natufian having para-Eurasian ancestry is explicitly contradicted by Lazaridis Supplementary Info pg. 41-43, where Basal Eurasian in Natufian and Iran_N is proven to come from a single population, and Natufian+Iran_N and all other ancients are consistent with rank = 2. Hmm.

P Piranha said...

Moving back to the original topic about Pseudo-steppe, this is really pure BS. The genetic data we have right now already rule this out more or less conclusively. Haplotypes with Eastern Europe render that admixture as recent, stronger with post Metal Age samples, stronger towards IA than BA, and indicate that that for some IA Steppics or E European cultures genealogical relation between them and Indians is as strong as that between them and Slavs. If it was purely some mesolithic phenomena from Central Asia, the pattern should by like that of Dravidians with West Asians, where the autosomal affinity in algorithms analyzing non-linked variation is high (ADMIXTURE, formal stats etc) but the affinity in segment-type analysis is low, which is what you can see in the segment matrix posted a few weeks back. There is some possibility that the "Steppic" contribution in Indians is partially due to Central Asian foragers, but to say that 100% of it is due to Central Asian foragers is simply false.

@ Salden

Here: http://www.tandfonline.com/doi/full/10.1080/24701394.2016.1258406

Note that this paper is very low-quality, the finer haplogroup determinations are almost certainly wrong, but one thing for sure is that most of the haplogroups are in Macro-haplogroup R and therefore Eurasian.

Ebizur said...

Samuel Andrews wrote,

"There's two reasons I think mHG M1 is Eurasian.

First mHG M itself is rooted in Asia..."

Granted that there is a relatively high probability of IBS and therefore reticulation in the inferred phylogenetic tree in the case of mtDNA because of its small number of base pairs, it does appear that mtDNA haplogroup M1 has derived almost directly from the root of the M clade. In contrast to this, most subclades of haplogroup M that occur frequently among modern South Asians (while also being found occasionally among West Asians, Central Asians, etc.) appear to be derived from a node that is significantly more recent than the initial spread of haplogroup M and the divergence of the M1 lineage. However, there is another subclade of haplogroup M, M2, which is also found among modern South Asians, and which appears to have a separate origin from the majority of South Asian haplogroup M.

Besides Americans, whose haplogroup M lineages are clearly derivative of those of East Asians, it looks difficult to determine the relative antiquity of haplogroup M in any region in which it is found today. Theoretically, haplogroup M1 should be considered equally relevant to the origin of haplogroup M as any of the other subclades that have derived directly from the root of M.

Anthro Survey said...

@P Piranha:

I think you misunderstood me. In my comment addressed to you, I was not implying that Jordan_EBA had detectable SSA ancestry compared to later Levant, but that the Levant of Late Antiquity had real SSA admixture compared to Bronze-Age times. This could have been mediated by migrations from South Arabia. Phillip the Arab, Ghassanids, and other proper nouns attest to this.
But, yes, indeed, the statistics (Natufian, Jordan_EBA; SSA, Mbuti) are not that significant. Not much quarrel there.

Anthro Survey said...

@Unknown

Yes, that is how I understood it! Basically, it would be the Muslim Egyptians who have more of the West/Central SSA admixture w/Basal Human DNA acquired through concubinage. The single digit SSA admixture in Ancient Egyptians(and Copts?)---if it lacked Basal Eurasian admixture---was most likely East African w/minimal Basal Human.

"For similar reasons Sardinians have longer drift path with Natufians(and if I remember right Ancient Egyptians) than modern Saudis who've significant Iran_Chl/CHG"

You mean shorter drift path, right?---since ANE-rich Iran/CHG and SSA lengthens it with Saudis.

Samuel Andrews said...

@Ebizur,
"In contrast to this, most subclades of haplogroup M that occur frequently among modern South Asians (while also being found occasionally among West Asians, Central Asians, etc.) appear to be derived from a node that is significantly more recent than the initial spread of haplogroup M and the divergence of the M1 lineage."

I was not aware of this. That South Asian M clade couldn't be much younger than M itself. It was quick break from mHG M and probably began expanding around the time mHG M itself did. It's sort of like the relationship between HV and H.

" it looks difficult to determine the relative antiquity of haplogroup M in any region in which it is found today."

I don't think so. Many distinct and very basal forms of M exist in Asia. It's safe to assume based on modern mtDNA that many of Asia's earliest humans carried mHG M. The the high mHG M1 frequencies in ancient Egyptians without much African ancestry is the nail on the coffin.

Ebizur said...

Samuel Andrews wrote,

"That South Asian M clade couldn't be much younger than M itself. It was quick break from mHG M and probably began expanding around the time mHG M itself did. It's sort of like the relationship between HV and H."

Haplogroup HV seems to have curiously low diversity compared to other subclades of haplogroup R. Likewise, South Asian M(xM2) seems to have low diversity compared to other subclades of haplogroup M.

I am unsure whether mtDNA diversity can be safely used as a basis for estimating the age of a population, but it appears on the face of it that both HV and South Asian M(xM2) have expanded at some time more recent than the initial AMH colonization of Western Eurasia or South Asia, respectively. I suppose haplogroup D4 could be considered as an analogue in northern East Asia, though it is not quite so hegemonic as M(xM2) in modern South Asia or HV in modern Europe.

In any case, I think that at least 15,000 years (perhaps more) have passed between the initial diversification of haplogroup M and the subsequent diversification of South Asian M(xM2).

"Many distinct and very basal forms of M exist in Asia. It's safe to assume based on modern mtDNA that many of Asia's earliest humans carried mHG M. The the high mHG M1 frequencies in ancient Egyptians without much African ancestry is the nail on the coffin."

The point is that haplogroup M1 is not a derived subclade, but rather a branch from the root of extant haplogroup M. I think very few people would argue for a sub-Saharan African (in the genetic sense) origin of mtDNA haplogroup M1, but an African origin in the geographical sense (e.g. Paleolithic Egyptian) is quite plausible.

Haven't ancient specimens from Paleolithic Europe been found to carry proto-M lineages basal to any M lineage known from extant humans? Does this mean that nearly all extant humans (but perhaps especially extant non-Europeans) have descended from inhabitants of Paleolithic Europe? I think that would be a bit of a stretch. It is rather more likely in my opinion that those Paleolithic European proto-M lineages have gone extinct (through genetic drift or otherwise) and that extant M lineages have descended from a population that lived somewhere in Asia (or perhaps even in some part of Africa, such as Egypt, where they were isolated from LO-, L1-, or L2-bearing populations who also inhabited parts of that continent).

Anthro Survey said...

@Piranha, Unknown and others:

This cladogram is not to scale w/regards to any branch lengths, but is my paradigm on these African groups correct?

I've revised the initial one I posted(if you've seen that one) for the technicalities.

https://justpaste.it/1al0a

Alexandros said...

Very useful analysis David. I have some observations/queries:

I did not expect such high Steppe admixture in Sardinians. Is there any chance this is inflated upwards for some reason? If not, what would you predict is the source population for this?

The Steppe admixture among modern Greeks is much higher than among Mycenaeans (that was already clear from the Lazaridis paper), but I can see that Cypriots also have slightly higher Steppe admixture than Mycenaeans do, which is quite interesting. Is there any chance to include Crete_Armenoi in your analysis?

Modern Lebanese have significant levels of Steppe admixture, almost as high as Cypriots. In fact, your estimations of steppe-derived ancestry among modern Lebanese (12-16%) is double that estimated in the Haber et al (2017) paper (7%). Btw, now that the Haber paper is officially published, is there any chance to get hold of the Sidon BA samples?

I have also noted the pseudo-SSA admixture among Natufians and other ancient samples, but that has been already picked up by Chad and justified by yourself. Just wanted to ask if there is a way to run some post-hoc analysis for 'correcting' this. The analysis results are great and it is a pity to have this small anomaly..

Davidski said...

@Alex

This test was designed to differentiate Yamnaya-related and Iran_N streams of ancestry in Iran_Hotu and South Asians.

I think the problem with the Yamnaya-related ancestry proportions for South Europeans and nearby Near Eastern groups is that they're slightly inflated due to some sort of more westerly southern input into the Yamnaya cluster, for which Anatolia_ChL and Hungary_N were good proxies in some of my recent qpAdm runs.

I'm not sure that I can fix this issue, or the faux Sub-Saharan admix in the Natufians, to build on this analysis and create a more complete test. If I fix one thing, something else will probably pop up as a result. It's sort of like trying to plug several holes in a leaking boat with one sock.

However, I will try to add a cluster or two when the Mathieson et al. 2017 data is released, and we'll see what happens.

Ryan said...

I have my own OIT - Out of Inuit Theory.

I feel it has about as much support from DNA and linguistics as the Out of India Theory. Maybe more so - the Inuit only have to lose East Asian ancestry through some sort of magic or alien intervention, whereas Out of India requires aliens to magically delete several layers of South and East Asian ancestry.

Linguistics disfavour Out of Inuit of course, but less than Out of India actually. Instead of having to conjure up imagined IE substrates and just neglect any explanation for how those families got to Europe/Anatolia/China/West Asia from India, nearby Chukotko-Kamchatkan has actually been linked to PIE. So at least we're starting from the right spot in the tree so to speak - at the base.

I mean, aliens manipulating DNA isn't very plausible, but you need more of that for Out of India Theory too, so less aliens makes my OIT better, right?

Ryan said...

I think we need a post debating the shape of the Earth too. It's a cylinder I tell you!

Archaelog said...

I just hope political correctness doesn't drive any of the researchers or labs to suppress or water down the hard truth. Eagerly awaiting the SA results

Atriðr said...

This post is conflation and deflection.

OIT is a dead theory. No use to conflate it with the model which makes the most sense since at least the beginning of the year.

jv said...

If you find this, could you please let me know? Online I find illustrations of Bell Beaker Culture attire and jewelry( or at least an artist's interpretation) Do you know of any websites or books, articles that would illustrate a Corded Ware Culture woman's attire? Thank you

Samuel Andrews said...

@Ebizur,
"In any case, I think that at least 15,000 years (perhaps more) have passed between the initial diversification of haplogroup M and the subsequent diversification of South Asian M(xM2)."

That's fair.

"Haven't ancient specimens from Paleolithic Europe been found to carry proto-M lineages basal to any M lineage known from extant humans?"

No, they belonged fully to M and to a subclade that no longer exists.

" I think very few people would argue for a sub-Saharan African (in the genetic sense) origin of mtDNA haplogroup M1, but an African origin in the geographical sense (e.g. Paleolithic Egyptian) is quite plausible."

Yes, I guess so. North African mtDNA fits within West Eurasian mtDNA diversity meaning their Eurasian ancestors may have arrived in Africa "recently." I wouldn't bet on Natufian or Anatolia Neo-like people in Africa 20,000 years ago but I may be wrong. Of course, it's entirely possible M1 derives from a totally different type of people that did, in fact, live in Africa 20,000 years ago.

Davidski said...

@Atriðr

This post is conflation and deflection.

OIT is a dead theory. No use to conflate it with the model which makes the most sense since at least the beginning of the year.


This post isn't about OIT. It's about the fact that there was a big migration from the Pontic-Caspian steppe to India during the Bronze Age.

But it also shows that your Out-of-BMAC and/or Out-of-Siberia theory is nonsense, hence your hissy fit.

Atriðr said...

@Davidski
hissy fit

Why do you get so emotional?

Davidski said...

@BAtriðr

Why do you get so emotional?

I don't, you do. That's what I was pointing out.

For months now you've been babbling about something remarkable that the new data are showing about the PIE homeland, failing to say anything close to coherent on the topic.

To me it's rather obvious that you don't know how to read the data, and just hoping that it's showing something different than what it is actually showing, which is that the PIE homeland was on the western steppe.

Atriðr said...

@Davidski
I don't, you do.

I graduated from kindergarten with flying colors. Let me know if you need to see my report card.

Till then, keep patching your ship. Or better yet, you can re-adjust geographic terminology.

data
This word means nothing to you. Don't kid yourself.

Late PIE is still on the Pontic Steppe and that's your consolation prize.

Balaji said...

@Davidski

You probably realize a serious problem with the notion that the “Aryans” of the AIT came from the steppe. Archaeologists had identified the Andronovo as the ancestors of the “Indo-Iranians” and that seemed plausible until aDNA became available which indicated too much EEF in the Andronovo for them to have been ancestral to Indians. You therefore suggested that only the elite Andronovo had much EEF ancestry and that the commoners were more like the Poltavka outlier.

But now you find that Poltavka outlier does not work well and have instead used Sarmatian_Pokrovka for your modeling. But iron age Scythians are too late to have been the “Aryans” who were supposed to have invaded in 1500 B.C. It will be too much of a stretch to hypothesize that the Andronovo commoners were like Sarmatian_Pokrovka.

Bogdan said...

jv:

Not much example that far back specific to CW culture as not much textile has survived. Egtved Girl attire might be an example, but that is more recent 1370 BCE. I'd say flax linen bleached with urine and milk, nettle yarn, basic woolen cloakes, drapes and hides closed by bone fabula/broach, fur linings. Maybe some decorated fringes, rows of knots and amber bead adornments.

A good reference is "Survey of Historic Costume", fourth edition, by Tortora/Eubank. Most of the ancient world focus circa 3000 BCE is Mediterranean and most of the book covers much more recent than Bronze Age though..

Davidski said...

@Balaji

It will be too much of a stretch to hypothesize that the Andronovo commoners were like Sarmatian_Pokrovka.

Why's that?

They were Indo-Iranian speakers and obviously their ancestors didn't pick up much EEF from the Sintashta and Srubnaya elites that moved into and via Samara to Central Asia. So what's the problem?

P Piranha said...

@ Atriðr

You have an incredibly supercilious and irritating tone, and present no evidence at all to back up what you are saying. Not a good combination, even for someone who graduated kindergarten.

Have you forgotten Kumtepe4 with ~30% Steppe ancestry in Anatolia from the vicinity of Troy?

Please present what makes you think Central Asia is more favorable. Or, if you think Yamnaya lookalikes also existed in Central Asia, please explain why Iran_hotu right next to the plains was so different.

Even if you are right, people like you deserve no credit.

Bob Floy said...


@ Atriðr

David and P Piranha are right, you know.

P Piranha said...

@ Anthro survey

Yep, I misunderstood you there, sorry for that.

Nope, I don't post on Anthrogenica.

That cladogram seems identical to the one in Schlebusch on South African genomes. It may work, but as usual with these things each phylogenetic configuration is only stable until the next set of ancient genomes appear. For example, the dichotomy between WHG and ANE was a parsimonious explanation of population relationships until Kostenki was sequenced, and the picture is complexified further by the other para-WHG Paleolithic Europeans, revealing both ANE-like and ENA-like inputs into what we know now as "WHG".

Ancient genomes from Tropical Forest Africa will be a treat. It may be the case that BasalHuman is not basal enough and Eurasian input is discovered in Africans. Gurdasani et al, for example, found a signal of Eurasian input into even Yoruba at 7.5-10.5 millenia ago, and it looks like there is non-zero Neanderthal in even West and Central Africa.

Bronze said...

Nope, east africans are overwhelmingly eurasian on the paternal side. Y-dna E is confirmed to have origins in non-african/non-black/non-SSA people, since natufians where confirmed to not have any SSA admixture. So these natufians/basal eurasians migrated into africa and mixed with native SSA women and brought y-dna E haplogroups. E-m215 is not verified to have ethiopians origins, many e-clades in east africa also has an origin in north africa, but ultimately it seems those lineages also came from the middle east.

Bogdan said...

Cattle is currency, then and now. Horses are currency, then and now. (especially the larger horses from the Steppe, relative to everywhere else at the time). Dope (opium and ganga) is currency, then and now, and the Steppe people's appear to be the worlds first dope dealers as wel..

So it is rather easy to understand how Steppe peoples could spread around far and wide, given their superior mobility and huge amounts of tradable wealth, knowledge and dare I say military skills, to thoroughly dominate, obtain/buy many wives, take or barter for more pasture land, take or trade for more metals and ore and generally simply dominate trade and become head honchos. It's basic common sense.

Appears S.Asia-Indus was no exception. That R1 paternal seed is planted far and wide for a very good reason. Follow the money and who has it.. It was even more true then, than it is now...

postneo said...

The pca shows the large impact the "Armenian chalc" type pop has on wide ranging populations.

modern, BA, chalc Armenians plot in the same place. but in contrast

1) EHG->yamnaya->modern Volga Ural swerves rapidly then drifts towards Armenia
2) Levan-neol->Jordan EBA -> near eastern pops also drift towards Armenia
3) Iran Neol->chalc->IA also show large movement towards Armenia thru time.

Armenia chalc is a strange attractor.

Iran Hotu clearly demonstrates a "psedo steppe" like signal 8-9k years ago.

Bob Floy said...

@postneo

"Iran Hotu clearly demonstrates a "psedo steppe" like signal 8-9k years ago."

No he dosen't.

Davidski said...

@postneo

Iran Hotu clearly demonstrates a "psedo steppe" like signal 8-9k years ago.

No he doesn't and this is clearly seen in the ADMIXTURE analysis.

2.34% noise membership in the Steppe_EBA cluster due to being less basal than Iran_N. The PCA confirms the result.

The pseudo-steppe theory is nonsense.

Ryan said...

What's pseudo-steppe-like anyways? Wouldn't that be pseudo-pseudo steppe? How many pseudos do we add before it just becomes not-steppe-like.

Alberto said...

@Davidski

If the Khvalynsk Q1a sample is related to the Kelteminar culture, as you suggested (and that's plausible), that's already like saying that a population 90% Steppe_EMBA lived in Uzbekistan 5000 BC. So unless you now think that was a brain fart of yourself, I don't see what this all means.

Look at the distance between the Zagros and the South Caspian, and then look at the distance to, say, Tajikistan. If Iran_Hotu has an extra ~10% EHG over Iran_N, it's a real possibility that those samples from Kyrgyzstan that are said to be coming will look much closer to EHG/ANE in that cline.

We also have the Srubnaya_outlier sample which doesn't match with other contemporary steppe samples by a large margin. She couldn't have come from further east than the Altai, since that was already high East Asian territory as shown by Okunevo et al. So the only plausible place for her to have come from is further south, from the inner Asia mountain corridor. And she's still the best proxy for "steppe" admixture in SC Asians.

But in any case, from a purely AIT point of view you have 2 options here:

- An Afanasievo-like population (which branched out from pre-Yamnaya groups and went east, carrying R1a-Z93 -and L657 specifically- migrated to South Asia before the arrival of Sintashta/Andronovo types, going completely unnoticed in BMAC or elsewhere.

- Or it was a later invasion by Sintashta/Andronovo types, and the excess of ANE in SC Asia was already there (so a model with Iran_N + AG3 + Sintashta + ASI, like Chad has shown).

I honestly find the first option extremely unlikely. No evidence for it at all. So the real question is not if there figure we get from Yamnaya in SC Asia are inflated. The question is how much they are inflated. Ad that's something that only aDNA can answer.

There's third possibility that western Scythians/Sarmatians brought R1a and excess of ANE to South Asia, but that would be linguistically and culturally irrelevant.

(BTW, glad to see that with qpAdm using more ingroups in the pright pops instead of just outgroups you get more accurate results, closer to what Dstat/4mix-nMonte were showing (and even the PCA/nMonte). This should be quite a satisfaction for Matt and yourself, who made that method possible long before they figured it out with qpAdm).

Davidski said...

@Alberto

If the Khvalynsk Q1a sample is related to the Kelteminar culture, as you suggested (and that's plausible), that's already like saying that a population 90% Steppe_EMBA lived in Uzbekistan 5000 BC. So unless you now think that was a brain fart of yourself, I don't see what this all means.

I don't know what the Q1a sample represents exactly. There are a number of options, but you have to be honest that there's no evidence whatsoever that a 90% Steppe_EMBA population lived in Uzbekistan around 5000 BC.

Look at the distance between the Zagros and the South Caspian, and then look at the distance to, say, Tajikistan. If Iran_Hotu has an extra ~10% EHG over Iran_N, it's a real possibility that those samples from Kyrgyzstan that are said to be coming will look much closer to EHG/ANE in that cline.

But look at the distance between Iran_Hotu and the Central Asian desert steppe. He's right there on the edge of it, and people like him probably colonized the major river valleys of the region, like that of the Amu Darya. So I'd say expect Iran_Hotu clones all over pre-Indo-Iranian Central Asia, but maybe with a fair whack of something like ASI?

And I have no idea why you're positing the existence of EHG in Kyrgyzstan? How are you making this link? Even ANE is a stretch, because of the topography (huge mountains) and totally different ecosystem from where the two ANE samples were found in Siberia.

We also have the Srubnaya_outlier sample which doesn't match with other contemporary steppe samples by a large margin. She couldn't have come from further east than the Altai, since that was already high East Asian territory as shown by Okunevo et al. So the only plausible place for her to have come from is further south, from the inner Asia mountain corridor. And she's still the best proxy for "steppe" admixture in SC Asians.

Srubnaya_outlier is just Srubnaya with more ANE and less basal admixture. She probably has ancestry from western Siberia. Forget South Central Asia, or even Central Asia.

I honestly find the first option extremely unlikely. No evidence for it at all. So the real question is not if there figure we get from Yamnaya in SC Asia are inflated. The question is how much they are inflated. Ad that's something that only aDNA can answer.

You ignored the most plausible scenario, which is that the early Indo-Iranians were the descendants of Abashevo and Sintashta who moved east and, in varying degrees, mixed with the remnants of the earlier Afanasievo and related groups, producing people similar to the Pokrovka Sarmatians, some with more EEF and others with as little as Afanasievo.

There's third possibility that western Scythians/Sarmatians brought R1a and excess of ANE to South Asia, but that would be linguistically and culturally irrelevant.

Not possible, because the main clade of Z93 in South Asia is L657. This is not a Scythian clade for several obvious reasons.

Ebizur said...

All right. I went back and checked Posth et al. 2016. They placed the mtDNA of GoyetQ116-1 (35,000 years old), GoyetQ376-3 (34,000 years old), and LaRochette (28,000 years old) in a subclade of haplogroup M separate from one subclade that consists of two Japanese, a Siberian Eskimo, and a Guarani (I suppose this should be haplogroup D) loosely linked with the mtDNA of an Indian and a Chinese, a second subclade that loosely links the mtDNA of an Australian aborigine and the mtDNA of an individual from coastal Papua New Guinea, and a third subclade of relatively low diversity that includes two Warao, an Evenk, a Buryat, and a Kyrgyz (perhaps haplogroup C).

It looks like you were right, Samuel, that the mtDNA of the Goyet and La Rochette individuals in question appears to be derived from haplogroup M just as any extant subclades of that haplogroup rather than descended from a basal "pre-M." I'm not sure about the lineage of the specimen from southern Italy examined by Fu et al. 2016. Does anyone here know whether the Ostuni specimen's mtDNA is also descended from the common root of extant haplogroup M, and, assuming that it is, whether it belongs to the same subclade as the two Goyet specimens and the La Rochette specimen?

Furthermore, has it been demonstrated that the GoyetQ116-1, GoyetQ376-3, and LaRochette specimens do not belong to any known subclade of haplogroup M? Figure 2 ("Maximum Parsimony Tree of Present-Day Human and 55 Pre-Neolithic mtDNA Genomes") of Posth et al. suggests that the mtDNA of those specimens must belong to a subclade of M other than haplogroup C, haplogroup D, or the subclade to which the Australian aborigine and the coastal Papua New Guinean belong (perhaps haplogroup M29'Q). In the passage "Phylogenetic analyses" under "Supplemental Experimental Procedures," the authors have written:

"Finally, three pre-LGM individuals (LaRochette, GoyetQ116-1 and GoyetQ363-3) carry mtDNA hg M defined by four positions from the L3 split (T489C, C10400T, T14783C, G15043A). Looking more closely at the intragroup M phylogeny, these three pre-LGM individuals are all placed on a branch not yet seen in modern-day individuals, either in Europe or elsewhere. In particular GoyetQ116-1 and GoyetQ363-3 from Belgium dated ~34-35 ka, show 6 (G207A, C1556T, C6045T, C8619a, A11084G, T16297) and 8 (G207A, C1556T, A6040G, C6041T, C6045T, C8619a, A11084G, T16297) derived mutations from the M root, respectively, whereas LaRochette from South France dated ~27.5 ka presents 7 (G207A, C1556T, C6045T, C6164T, C8619a, A11084G, C12816T) derived mutations from the M root. LaRochette carries a possible additional back mutation at position C16297T! (16Ts and 4Cs), compared to the two Belgian sequences both carrying the T16297C mutation. The four Cs at this position cannot be explained by damage and are likely the result of contamination or heteroplasmy, therefore an unassigned base (N) was placed at that position in the final consensus sequence (see above). Despite their geographic and temporal separation all three pre-LGM individuals share five mutations (G207A, C1556T, C6045T, C8619a, A11084) from the MRCA of the basal M lineage. This finding suggests that this M branch arrived in Europe before 35 ka and survived at least until 27.5 ka, before the LGM started (~25 ka)."

The authors have stated that "these three pre-LGM individuals are all placed on a branch not yet seen in modern-day individuals, either in Europe or elsewhere," but does the data set against which they have compared the mtDNA of those specimens really contain representatives of every known subclade of haplogroup M?

Alberto said...

@Davidski

I don't know what the Q1a sample represents exactly. There are a number of options, but you have to be honest that there's no evidence whatsoever that a 90% Steppe_EMBA population lived in Uzbekistan around 5000 BC.

Agreed. Difficult to say what that Q1a sample represents. But it was your own suggestion that it could be related to Kelteminar, which is admitting the possibility of a Yamnaya-like population being in Uzbekistan. And yes, we have no evidence of it. None at all. We have no evidence of the contrary either. Basically because we have zero samples from south of Kazakhstan.

And I have no idea why you're positing the existence of EHG in Kyrgyzstan? How are you making this link? Even ANE is a stretch, because of the topography (huge mountains) and totally different ecosystem from where the two ANE samples were found in Siberia.

The Altai is a mountain range that extends south through what's been called the Inner Asia Mountain Corridor. It's far easier that people adapted to the mountains could move along that range than they ventured into the arid steppe. That's why they favoured the forest steppe. But still, Karelia is much further away from the Altai than Kyrgyzstan is. To expect Paleo/Meso samples from Kyrgyzstan to be clones of Iran_Hotu is quite more of a stretch than expecting something closer to Srubnaya_outlier. But we'll wait for the results to see.

Srubnaya_outlier is just Srubnaya with more ANE and less basal admixture. She probably has ancestry from western Siberia. Forget South Central Asia, or even Central Asia.

No, Srubnaya_outlier has no admixture from Srubnaya people. Zero. She has ANE and maybe some mix of Afanasievo and Iran_N.

You ignored the most plausible scenario, which is that the early Indo-Iranians were the descendants of Abashevo and Sintashta who moved east and, in varying degrees, mixed with the remnants of the earlier Afanasievo and related groups, producing people similar to the Pokrovka Sarmatians, some with more EEF and others with as little as Afanasievo.

We have samples fro the descendants of Abashevo/Sintashta who moved as far east as the Altai. Their very similar to Sintashta itself. And carry R1a-Z2124.

Not possible, because the main clade of Z93 in South Asia is L657. This is not a Scythian clade for several obvious reasons.

Exactly. This was just to remark that the models with Sarmatians were irrelevant for Indo-Aryans.

Ebizur said...

Actually, the names for the samples in the phylogenetic trees published by Posth et al. 2016 are a bit ridiculous. One of the samples that appears to belong to mtDNA haplogroup D has been labeled "AsiaJapanese," whereas another that appears to belong to the same haplogroup has been labeled "EurJapanese." The Guarani individual whose mtDNA appears to belong to haplogroup D has been labeled "SamIndianGuarani," the two Warao whose mtDNA appears to belong to haplogroup C have been labeled "SamIndianWarao" and "SamIndianWarao2," a Pima whose mtDNA appears to belong to haplogroup B2 has been labeled "NamIndianPiman," and the arctic individual whose mtDNA appears to belong to haplogroup D has been labeled "NamSiberianInuit." "Siberian Inuit" would be an obvious misnomer, so I assumed it to be a tacky PC attempt to avoid using the word "Eskimo," but upon further consideration, I have realized that the authors may have used this label to imply association of this sample's autosomal DNA with a cluster of Siberians as opposed to two other clusters of "NamIndian" (North American Indians, such as the Pima) and "SamIndian" (South American Indians, such as the Warao or the Guarani). However, "North American Siberian" is as silly of a label as "Siberian Inuit," so it is poor labeling either way.

In conclusion, I am unsure whether the "NamSiberianInuit" individual is an Inuit from North America or a Yupik from Siberia, though the distinction is irrelevant to my previous comment.

Davidski said...

@Alberto

So a Yamnaya-like population first formed in the desert steppe of Uzbekistan, and then it formed separately in the very different steppe of Eastern Europe? That sounds like the pseudo-steppe theory, just a slightly less magical version of it.

And Eastern European Hunter-Gatherers (EHG) lived in Kyrgyzstan did they, while Srubnaya_outlier has ancestry from Neolithic farmers in Iran? :)

Please take a look at some maps, especially the topographical ones, so you can see distances, deserts and mountains. I think you'll get a nice reality check.

Srubnaya_outlier is obviously of forest steppe origin from somewhere; clearly more northern not southern than regular Bronze Age Eastern European steppe samples.

And obviously we do have an ancient sample from south of Kazakhstan: Iran_Hotu. As I said, he's sitting on the very edge of the Central Asian desert steppe that you want to be home to EHG.

Your theories won't work out. This is already obvious.

Matt said...

Only one point comes to mind I would slightly question is the idea here that the ADMIXTURE is picking up the signal of a "Bronze Age ethnic group" from an "exaggerated signals of relative inbreeding, isolation, and/or rapid expansions experienced by founders of ethnic groups."

If we take the proportions for EHG and CHG from the test, and then overlay on Lazaridis's PCA from 2017 - http://i.imgur.com/7VSh9Nj.png

The proportions for the EHG and CHG just seem to fit where CHG and EHG are as close as is possible given the constraints. Iran_Hotu wouldn't get much Steppe_EMBA component, just because it is quite close to the Iran_N cluster (and ADMIXTURE cluster are fuzzy approximations at the margins).

Yes, 70% Steppe_EMBA, 27% Euro_HG, 3% North Asian isn't the great model for EHG (and neither is the equivalent CHG model), but this is ADMIXTURE doing the best it can with its method and the Supervised clusters.

Anyway, just my take on it. Even still, bottom line, totally as you say, the Iran_Hotu sample doesn't really look much like a Iran_N+EHG = pseudo-Steppe or pseudo-CHG here, or hint too much at this as a possibility. However it is that Iran_Hotu is variant from Iran_N, it looks like it is in a different way and not the same genetic direction.

Made a set of bar plots for % Steppe_EMBA averages in the spreadsheet for South and Central Asian populations in the panel:
http://i.imgur.com/yp4aEhl.png / http://i.imgur.com/SkU8s1M.png.

Seems believable in the Steppe_EBA component. Long term I don't really believe the pseudo-steppe thing (as explanation for everything) but still want to understand why ADMIXTURE generally prefers to create South Asian components to other models.

Matt said...

For anyone interested, locations of samples from Lazaridis 2016 dropped onto an equal area topographical map: http://imgur.com/a/GCt5i

(Of course there are more samples since then...)

Davidski said...

@Matt

This test wasn't supervised in the normal way. I just dropped the populations that were making their own clusters ahead of the ones that I wanted to create clusters.

That's why there aren't any South Asian groups in the dataset apart from the Brahmins and Makranis. The latter are helpful in creating the Iran_N cluster, which I think demonstrates how much direct ancestry they have from ancient groups closely related to Iran_N.

Considering the age and location of Iran_Hotu, I don't buy the suggestion that Central Asia was home to EHG, or anything even similar.

The Eurasian steppe is blocked off from the more fertile areas of Central Asia by the Karakum and Kyzylkum deserts. So we can't expect that EHG moved south across these ultra dry landscapes.

On the other hand, Iran_Hotu is at a location from which it should be easy to move east through fairly high rainfall mountains to reach the big river valleys in Uzbekistan and other fertile areas of Afghanistan and Tajikistan. So the foragers and different types of farmers that colonized Central Asia were probably very similar to the samples we already have from Iran.

Rob said...

The Iran Hotu sample and other sites around South Caspian are thought to be related to & largely descended from the Trialetian industries of the southern Caucasus, although locally adapted

(Eg Iran_Hotu
"Satsurblia" 83.35
"Ust_Ishim" 13.6
"Yoruba" 2.55
"AfontovaGora3:I9050.damage" 0.5)


So it'll be interesting to see what foragers from further east/ northeast look.

Rob said...

@ Dave

Were you able to analyse Srub-Outlier formally ?
I would tend to agree it's something western steppe admixed with local HGs, but PCA gives inconsistent results (i.e. they flip n flop)

Davidski said...

Srubnaya_outlier comes out like this with qpAdm, but the number of markers is low.

right pops:
Mota
Ust_Ishim
Kostenki14
GoyetQ116-1
Vestonice16
MA1
ElMiron
Villabruna
CHG
EHG
WHG
Iran_N
Iran_ChL
Natufian
Levant_N
Barcin_N

left pops:
Srubnaya_outlier
AfontovaGora3
Andronovo

best coefficients: 0.090 0.910

std. errors: 0.043 0.043

taildiff: 0.220481292

Davidski said...

Ah, crap, I ran two Srubnaya_outliers instead of the one of interest. I'll check this again.

Davidski said...

Here we go...

right pops:
Mota
Ust_Ishim
Kostenki14
GoyetQ116-1
Vestonice16
MA1
ElMiron
Villabruna
CHG
EHG
WHG
Iran_N
Iran_ChL
Natufian
Levant_N
Barcin_N

left pops:
Srubnaya_outlier1
AfontovaGora3
Yamnaya_Samara

best coefficients: 0.180 0.820

std. errors: 0.067 0.067

taildiff: 0.161760786

Balaji said...

@Davidski

Your hypothesis of an EEF-rich elite among the Andronovo along with EEF-poor commoners, has no support from aDNA or archeology. Also it is in contradiction to what we know about the social organization of highly mobile nomadic people. They tend to be relatively egalitarian with high social mobility. A further difficulty for you is that Scythians are known to have extra-Iran_Neolothic related ancestry but no EEF related ancestry. They could not have got this from Iran since by the iron age the populations of Iran, the Levant and Anatolia had been thoroughly mixed up.

Alexandros said...

Thanks for the justification David. Makes sense. If possible, it would be great if you could also include Crete_Armenoi, as well as the new Levant BA samples in any subsequent analyses you may run.

Ryan said...

@Alberto - "I honestly find the first option extremely unlikely. No evidence for it at all. So the real question is not if there figure we get from Yamnaya in SC Asia are inflated. The question is how much they are inflated. Ad that's something that only aDNA can answer."

I'd just point out that related populations getting swept up into the Proto-IE expansion from Yamnaya doesn't detract from AIT - it supports it.

Matt said...

@ Davidski, interesting stuff. So it's that you got these K populations because of the sample and population selection (e.g. not too many South Asian samples to trigger a South Asian cluster, good amount of Yamnaya related Bronze Age Europeans and Steppe populations and so on) and not by using any supervised mode.

Btw, can you put up the Fst scores if possible as I'm always interested in those (whether or not they're always the most meaningful...?

Anonymous said...

@Alberto

The Altai is a mountain range that extends south through what's been called the Inner Asia Mountain Corridor. It's far easier that people adapted to the mountains could move along that range than they ventured into the arid steppe. That's why they favoured the forest steppe. But still, Karelia is much further away from the Altai than Kyrgyzstan is. To expect Paleo/Meso samples from Kyrgyzstan to be clones of Iran_Hotu is quite more of a stretch than expecting something closer to Srubnaya_outlier. But we'll wait for the results to see.

I have his personal theory that migration is something we should split between HG-migrations and farmer and pastoralist migrations. The latter can pick up he ecosystem they maintain and take it where the conditions are right. Seeds can be packed to travel over deserts and, more importantly, herds can be driven through hostile environments to new green pastures and fertile field.

HG's don't work that way - or at least that is what I think. They follow herds rather than drive them so their migrations must makes sense, must be following herds. The spread of the Magdalenians and the groups associaed with Hamnburg, Ahrensburg and Swiderian culures seems completely driven by the trek of the reindeers. They went north and hence a whole culture went north. The entry of humans in the America's can also be explained that way. See Davids mammoth topic a while ago.

With respect to ANE I think that would mean that they would follow he mammoths as long as they existed and thus their ancestry popping up in both Europe and American-Indians makes sense. However, IIRC the Altai mountains were under an ice sheet.

P Piranha said...

About hunter-gatherer population dynamics, an interesting difference between them and farmers/pastoralists is that stable and almost impenetrable genetic frontiers exist in between farmers/HGs or farmers/pastoralists, and you can see this for example in the GAC/HG or Balkan/HG frontier, or just the European Middle Neolithic/HG interpenetrating genetic boundary, all of which remained stable on the millenial timescale. For the limited gene flow that did exist to have taken place demands extremely low rates of cross-cultural intermarriages, virtually zero per generation. In contrast, we have multiple smooth admixture clines that appear among hunter-gatherers. Anatolian Hunter-gatherers appear to have smooth genetic gradation into CHG Natufians and WHG, CHG grades between EHG and Iranian Hunter-gatherers, SHG between EHG and WHG, and the mesolithic repopulation of the Scandinavian peninsula looks like a smooth admixture process as well. If Karasuk_Outlier is any indication gradations continued into Siberia.

It seems weird to claim that Samartian Pokrovka cannot represent Indo-Aryan input. These Samartians are basically Andronovo plus ~10% input from the Causcasus and ~15% Siberian input encompassing both ANE-EHG and East Asian ancestry. All three types of ancestry would be present in South Central Asia which the Andronovo would have passed through, no? We already have Iran_Hotu and judging from the Burusho the East Asian input is not a problem either. It was an already highly BMAC-admixed cultural group that was responsible for further movement south.

A further difficulty for you is that Scythians are known to have extra-Iran_Neolothic related ancestry but no EEF related ancestry.

This is just false. Scythians are clearly descended from Andronovo and Sintashta and in mtDNA are closest to Srubnaya. The only Steppic population that appears to have descended from Afanasievo types absent EEF is Okunevo.

The idea that the EEF elite concept is still needed falls by the wayside now we know input is best represented by the Pokrovka Samartians. Now it is possible that all steppic input likely arrived through a homogeneous population. Even then, we know that population structure among pastoralists is not a problem from aDNA, just that it probably does not track stratification, which contrary to what you said actually does exist among pastoralist cultures, just not in a stable dynastic manner, or at least not as stable as among farmers.

The Corded Ware culture already ranged from 0% to 25% EEF ancestry depending on region, so population-level differences can exist within cultures with the same material remains. Possible that they represented different tribal groupings.

It strikes me the archaeogenetic record overall seems to suggest that rapid changes in population affinities and high levels of population heterogeneity both within and between regional variants of a culture is more characteristic of pastoralists, or cultures undergoing interaction with pastoralists, than it is characteristic of farmers.

No, Srubnaya_outlier has no admixture from Srubnaya people. Zero. She has ANE and maybe some mix of Afanasievo and Iran_N.

This is the kind of thing that happens when you set too much store by unstable methods like nMonte.

Davidski said...

@Balaji

Uniparental markers suggest strongly that at least some Scythians derived in large part from EEF-admixed Srubnaya, Andronovo etc. See here...

http://eurogenes.blogspot.com.au/2017/03/north-pontic-steppe-scythians-heirs-of.html

Some of them did pick up new admixture in Central Asia, and I suspect that even the Pokrovka Sarmatians have Iran_N-related ancestry that Bronze Age Eastern Europeans (inc. Yamnaya) lacked. But clearly they, and the Samara Scythian (Samara_IA), on the plot below from the Scythian paper, show membership in the orange EEF cluster at almost all of the K at which it appears.

https://drive.google.com/file/d/0B9o3EYTdM8lQYW56OHJ4d0UwSG8/view?usp=sharing

So the conclusion that Sarmatians and Scythians don't harbor any EEF ancestry doesn't appear to be supported by all of the data, and it's probably the wrong conclusion.

@Matt

https://drive.google.com/file/d/0B9o3EYTdM8lQOW5ITlQ5ZVpWZTQ/view?usp=sharing

postneo said...

Iran hotu is a side show. its a lone sample of great age

But what about Armenia chalc? its more fleshed out.

its more a vortex than strange attractor.
its pulling steppe, levant, Iran and Greece over time. Particularly the steppe and Iran swerves very rapidly to it at an early stage.

Davidski said...

Thanks for the comic relief. But please learn some basics before dazzling us with your theories.

postneo said...

Why don't you explain:
if you draw contour lines connecting samples of the same age you get smaller circles converging on Armenia.

postneo said...

As the samples become younger

Ryan said...

@epoch - "With respect to ANE I think that would mean that they would follow he mammoths as long as they existed and thus their ancestry popping up in both Europe and American-Indians makes sense. However, IIRC the Altai mountains were under an ice sheet."

It's a good point, but one would expect the Altai mountains to be repopulated first by people living adjacent to the melting ice sheet.

Romulus the I2a L233+ Proto Balto-Slav, layer of Corded Ware Women said...

Is there any trace of WHG autosomal in India?

Jaap said...

Sorry for being stupid, but is anyone there willing to explain why zilch R1b made it to India?

Davidski said...

@Jaap

Sorry for being stupid, but is anyone there willing to explain why zilch R1b made it to India?

Same reason Corded Ware had zilch R1b.

See here

Unknown said...


The Iran Hotu sample and other sites around South Caspian are thought to be related to & largely descended from the Trialetian industries of the southern Caucasus, although locally adapted
(Eg Iran_Hotu
"Satsurblia" 83.35
"Ust_Ishim" 13.6
"Yoruba" 2.55
"AfontovaGora3:I9050.damage" 0.5)
So it'll be interesting to see what foragers from further east/ northeast look

Look pretty good but are you sure about Yoruba? ADMIXTURE is showing a small amount of Karitiana Amerindian related admixture in Hotu. Could you test that?

https://genetiker.wordpress.com/

>Is there any trace of WHG autosomal in India?

I think yes but only small amounts(like 1-5% maybe and less than 10% in all cases) in upper castes of some ethnic groups. I remember seeing somewhere some Jatts having some. It was probably mediated through their EHG admixture.

Davidski said...

Iran_Hotu doesn't have anything resembling Yoruba or Karitiana admixture, or even any real EHG.

He's basically like the Iran_N farmers, but less basal, and this is the my explanation for the non-basal ancestry both in Iran_Hotu and Iran_N.

http://eurogenes.blogspot.com.au/2017/07/working-topology-for-eurasian.html

Bogdan said...

For anyone not familiar with natural routes into Indus from the steppe regions, I advise studying topographical maps. There are only a few natural, relatively 'easy' routes prior to Silk Road. The best one is through modern day Turkmenistan, skirting the modern day west Afghanistan / east Iran border, to around modern day Herat. From there you follow south and head east skirting the Hindu Kush to area of modern day Quetta or a bit more south to Kandahar and follow the numerous mountain gaps on to Indus River.

Distance from Mary-Turkmenistan to Indus on this route is only approx 1000mi/1600km. To put that in context, an above average HG from 12000 years ago could cover that distance just naturally hunting at a rate outpacing a modern day Olympian the entire distance, in less than 2 months, like a walk in the park. The HG would be waiting quite a few days at the finish line for the modern day Olympian to show up, and then crush his skull and take his waiting wife at the finish line.... (HGs get no respect)

Using a Chisholm trail example, 10 men on horseback could cover that same distance / ground driving 3000 head of cattle in about 2 months as well, keeping stock well fed along the way...

Migration over vast distances in antiquity was no big deal...





Anonymous said...

@Ryan

I just wanted to raise the point, obviously you could be right. This is how mammoths were distributed over Eurasia over time.

https://3.bp.blogspot.com/-aXMOqM8W2to/VzQhRfVKUgI/AAAAAAAAEc0/WLJO_Di3DH8EESFzeAp7ocP9TzPci2OoACLcB/s879/woolly_mammoth_migration.png

It would have been nice if this had autosomal data:
http://www.sciencedirect.com/science/article/pii/S2352409X16306927

truth said...

David,

Could you include in the spreadsheet some balkans (Serbians, Macedonian, Montenegrin...) ??
Thanks

Davidski said...

The Balkan groups that are already there (Albanians, Croats, Romanians etc.) are the only Balkan groups in the two datasets that I used. I'd have to run a different test using a different dataset to add more Balkan samples. Can't really do that at the moment.

Rob said...

Pseudo African admixture just reflective of quality. So not real

puszta said...

Hello, I've got an unrelated (perhaps a bit stupid) question

Is the close proximity between the Volga-Ural and Yamnaya samples on the PCA a manifestation of the former being largely descended from the latter? I'm asking because as far as I know, North Slavs generally have Yamnaya ancestry comparable to other Central and Northern Europeans. The people from Volga-Ural region seem much closer to Yamnaya, however.


I'm sorry for OT but since this thread is already dead I thought it might be okay