search this blog

Monday, November 16, 2015

The "fourth strand" of European ancestry came from the Caucasus


From a news feature about a forthcoming palaeogenomics paper:

"The question of where the Yamnaya come from has been something of a mystery up to now," said one of the lead senior authors Dr Andrea Manica, from Cambridge's Department of Zoology.

"We can now answer that as we've found that their genetic make-up is a mix of Eastern European hunter-gatherers and a population from this pocket of Caucasus hunter-gatherers who weathered much of the last Ice Age in apparent isolation. This Caucasus pocket is the fourth major strand of ancient European ancestry, one that we were unaware of until now," he said

Read more at: 'Fourth strand' of European ancestry originated with hunter-gatherers isolated by Ice Age

Update: the paper is now out and open access at Nature Communications.

The two ancient Georgian genomes belong to Y-chromosome haplogroup J. So it looks like I was right when I said that this type of ancestry mostly entered the European steppe from the Caucasus via female mediated gene flow during the Bronze Age (see here and here).

However, one of the Eastern European Hunter-Gatherers (EHG) from Mathieson et al. 2015 also belonged to haplogroup J. This suggests that there was intermittent gene flow, including some paternal gene flow, between the Caucasus and the steppe well before the Bronze Age.

Nevertheless, it's now even more difficult to accept that Y-haplogroup R1 and the Proto-Indo-Europeans might have originated south of the steppe. Clearly, R1 appears to be a steppe marker from way back, and I seriously doubt that Indo-European languages were introduced into highly patriarchal steppe societies by female migrants from the Caucasus.


Image credit: Nature Communications, dx.doi.org/10.1038/ncomms9912

94 comments:

Rob said...

Massive

Karl_K said...

Very interesting. So these Georgia people had early mixture with the ancestors of middle eastern farmers. Can't wait to see the real article. This clarifies a lot.

Alberto said...

Wow, amazing! Great news.

So the "teal" component already appeared before the LGM? So this means it's not a pure artifact of ANE+ENF, but rather has its own drift and is a real component on its own right.

Also interesting:

"India is a complete mix of Asian and European genetic components. The Caucasus hunter-gatherer ancestry is the best match we've found for the European genetic component found right across modern Indian populations"

Which basically means that the Caucasus was the westernmost place where this population existed, but the area probably expanded east through the South Caspian to at least Afghanistan, and by the Neolithic it was probably all around Ariana.

Can't wait to read more about it. I hope they got some Y-DNA too!

Rob said...

I curious if this was the "R1 refuge zone", but I suspect it's too far west for that(?)

I wonder also of pockets of similar groups existed along the caspian littoral and belt zone during the LGM

Alberto said...

Yes, it's not clear from the article why they point to the Caucasus itself as the origin and the refuge during the LGM (the samples are from 13.000 and 10.000 y.a). The best refuge in the area was the South Caspian shores, and evidence of human occupation is earlier to the east than in the Caucasus as far as I know.

Anyway the fact that these are the "Teal people" and are the ones who make most of the West Eurasian ancestry in India (as expected) really point to them being all around to the east of the Fertile Crescent during the Neolithic. A migration in the Bronze Age of these people from the Caucasus to India is not attested nor it makes sense (apart from the fact that the Indus Valley had an estimated population of 5 million people already).

Davidski said...

The samples are from late Upper Paleolithic northwestern Georgia. These people went straight to the steppe after the LGM and mixed with EHG.

And there were two migrations to South Asia carrying this type of ancestry; one during the Neolithic from Iran, and the other during the late Bronze Age from the steppe bringing R1a-Z94.

Karl_K said...

Paper is out now

rozenblatt said...

Both of Caucasian hunter-gatherers - J haplogroup

Nirjhar007 said...

//Both Georgian hunter-gatherer samples were assigned to haplogroup J with Kotias belonging to
the subhaplogroup J2a (see methods). Haplogroup J is estimated to have arisen 31.7 kya in the
Middle East and is widely distributed in Eurasia, the Middle East and North Africa48,49. Patterns
of haplogroup frequency are consistent with an expansion from the Middle East towards Europe
which has been suggested to have accompanied the Neolithic transition in Europe48,50,51. In a
study exploring J haplogroups in 445 individuals from Eurasia, J2a was found at highest
frequency in Georgia and Iraq48. It is intriguing that both the mitochondrial and Y chromosome
haplogroups of our ancient Georgian samples have been associated with the Neolithization of
Europe. This tentatively suggests a genetic link between Georgian hunter-gatherers and early
European migrants from the Near East.
Bichon belongs to Y haplogroup I2a (see methods). Haplogroup I has been found at high
frequencies in Europe but is virtually absent elsewhere52. This haplogroup is suggested to have
a European pre-LGM origin53 and has been found in ancient samples with hunter-gatherer
backgrounds from central and northern Europe//

Nirjhar007 said...

Early European farmers
as well as modern European and West Asian groups additionally harbour a component
dominant in the Middle East (“magenta”), appearing at K=9, in agreement with a Middle Eastern
source of early Neolithic farmers. European groups from the late Neolithic onwards and many
West Asian groups also possess a component prevalent in South Asia, as soon as this appears
at K=7. This component is at first the same one that is prevalent in India (“dark purple”), but is
later replaced by the Caucasus-related component (“lime green”)
.

Davidski said...

No Y-DNA R1. So like I said, female mediated gene flow from the western Caucasus.

Caucasus farmers' daughters. Aha.

Nirjhar007 said...

You have to wait still to say that, Iran and C asian aDNA are coming :)....
And more sampling from Caucasus etc is also a must.

Nirjhar007 said...

Here another :
The components characteristic of Native Americans populations (“dark green”, “light brown”,
“dark gray” and “dark blue”) are also worth noting. These components are present in MA1 and
eastern hunter-gatherers, in agreement with previous studies5,7, but is at very low levels in the Yamnaya and virtually absent from our samples from the Caucasus.
These components may
point to the presence of the Ancient North Eurasian ancestry of both native Americans and
eastern hunter-gatherers. Their absence in Caucasus hunter-gatherers is in agreement with
their southern geographical position and their separation from northern Eurasia by the Caucasus mountain range.

Rob said...

Yep the absence of any R1 is a major implication

Nirjhar007 said...

Rob,
Yes I agree, further testing from Chalcolithic is required (Caucasus), but i'm not surprised.
I suggest now to look for upcoming Mesolithic and Neolithic aDNA of Iran, they will reveal many things...

rozenblatt said...

BTW, Supplementary Fig. 2 looks strange, as if CHG are descendants of Basal Eurasian. Do I understand in correctly?

Nirjhar007 said...

But also ''Georgia is only a small western region, we dont know about other areas of the Caucasus, the mountains create isolation. ''

Unknown said...

both of them are from imereti region. they are not from the most north western parts of georgia.

Davidski said...

It doesn't change anything. I laid this out weeks ago...

http://eurogenes.blogspot.com.au/2015/10/yamnayas-exotic-ancestry-kartvelian.html

Ryukendo K said...

@ Alberto

What a stunning confirmation. Looking at supplementary table 10, High-caste north Indians such as GujaratiA, GujaratiB and Tiwari(Uttar Pradesh Brahmin) favour Afanasievo/Onge in D-stats, while GujaratiC, GujaratiD and other more southerly South Indians favour Kotias/Onge.

George Okromchedlishvili said...

FYI Imereti is right next to Mingrelia and Svanetia. So while it's not exactly North-West Imeretians tend to cluster near Mingrealians compared to Eastern Georgians.

Pretty exciting news. Finally we get ancient genomes from Caucasus.

Nirjhar007 said...

//it's exceedingly difficult to accept now that Y-haplogroup R1 and the Proto-Indo-Europeans may have originated south of the steppe.//
For you it is always difficult ;).
we must wait for the samples from C Asia,Iran,SC Asia.
We must also look for samples from Chalcolithic and Neolithi from Caucasus,
It appears the modern concentration of J2 around that area is quite ancient, it can be just a coincidence though.

George Okromchedlishvili said...

Looking at PCA plot they're even more Paleo than modern North Caucasians.

Why then the latter give a worse signal for admixture than Georgians (who look like they are a mix of these guys and EEF)?

Ryukendo K said...

Yep, looks like the existence of 'Q' in Yamnaya, v Basal, is proven more or less. The ANE k8 might have underestimated the amount of Middle Eastern ancestry in Yamnaya after all.

Hmm, CHG and EEF are equally far from East Asian. This does not make sense if CHG is a 100% Basal Eurasian population, which is the authors' model in Supp Fig 2.

Supp Table 8 shows a consistently -ve but non-significant D(Yoruba, MA-1, CHG, EEF) stat.

There may be some gene flow into CHG from Crown Eurasian, hidden, which the authors did not test for, perhaps because the situation was a bit too complex for clear conclusions to emerge.

Looking at the behaviour of the ADMIXTURE, all the transitions between different K are pretty logical, and it seems like many questions raised by ADMIXTURE are now cleared; e.g. why SC Asian is shared between Euros, Indus and Caucasus (CHG+LNBA); why Mediterranean is shared between Bedouin and European (EEF); why N Euro is shared between Euros, Siberians, and SC Asians(EHG+ANE+LNBA), why S Asian is shared with Bedouin at low K (CHG).


@ Davidski
Looking at the ADMIXTURE, there seems to be a shocking surprise, at K=20 the Kalash break off to form a component that discriminates strikingly between two different groups of Yamnaya, as well as splitting the LNBA pops from each other. Smoking gun?

George Okromchedlishvili said...

CHG as far as I understand is a mixture of a separate group of Western Crown Eurasians and Basal Eurasians.

Ryukendo K said...

Hmm, and it seems like the main difference between the Middle east today and the middle east of the past is the massive rise in Kotias-like ancestry, "CHG". Perhaps this is consistent with the derivation of Semitic, another bronze-age language family, from northern Mesopotamia.

Krefter said...

Looking at modern diversity, I expected both R1a-M417 and R1b-M269 to be from south of the Caucasus. So this is surprising. However, we only have two Teal Y DNA samples.

Alberto said...

@RK

Yes, indeed. But this still leaves a lot of questions. One interesting one is what's the relationship between these CHG and ANE, because it seems none! And this is something you wrote about when Haak et al. was published, proving that Q was ENF, not ANE. But we still don't know why it seems to give an ANE-like signal in admixture and other methods?

I'm still just reading the paper (caught me with work, so I go slowly). But I think these findings are amazing and will be very useful. However, I think they're just the tip of the Iceberg. We have to see what was to the east. Where they CHG? A mix of CHG and ANE? Something else?

Unknown said...

For someone whose interests are not so parochially limited to IE languages and the origin of R1, I find this study incredibly exciting. It confirms my speculations that I/I2 are the Y chromosomes for European hunter gatherers while J/J2a are the corresponding Y's for now West Caucasus hunter-gathers. Bar-Yosef, who led the excavations for Upper Paleolithic/Mesolithic West Georgia is a superb archaeologist and now, I would wager that I tracks the expansion from a European refugium, J from a Black Sea/Caspian refugium and G from a Levantine/SE Anatolian refugium. Since the Neolithic originated in the latter region, the Is and Js later adopted farming. There are also linguistic implications: is J2a the keystone to Caucasian or Kartvelian languages? Fascinating stuff!

Ryukendo K said...
This comment has been removed by the author.
Krefter said...

Switzerland 13,700 BP: mtDNA U5b1h, Y DNA I2a.

This is very consistent. I just can't see much non-U5 having a pre-Neolithic origin.

Ryukendo K said...

@ George

The authors' model did not confirm this. It will be up to us, once the genomes are released.

For CHG to be the same distance from East Asian as EEF, the level of Basal vs Crown has to be the same, however it does not prefer either one of MA-1/WHG/EHG over the other (though the EHG/WHG comparison approaches significance). Perhaps it has some undifferentiated West Eurasian ancestry that is close to neither of those three.

George Okromchedlishvili said...

@ ryukendo kendow

yeah, seems like formal test don't quite support my idea.
On the other hand simply eyeballing the plot these guys sit roughly on the same "latitude" with EEFs. And the Y-axis seems to at least partially reflect Basal/Crown ancestry proportions.

Nirjhar007 said...

Krefter,
I expected both R1a-M417 and R1b-M269 to be from south of the Caucasus
Uh well the Georgian samples are from 13000 YBP and 10,000 YBP, i don't think M-417 and M-269 existed then :), but yeah more sampling is the word, we still have miles to go.

Anonymous said...

Was someone able to extract Y-DNA data from the Fastq files (http://www.ebi.ac.uk/ena/data/view/ERP012743)?

Alberto said...

@RK

Where are you seeing the admixture results? I can't seem to find any other than the one in the supplementary information PDF, which is very low resolution to see anything. Is there a high resolution image somewhere else?

@others

Re: R1, these samples do give a better possibility of R1a and R1b in Yamnaya/CW being from EHG mixed with this population. But genetically possible alone is not enough to answer the open questions. Other factors still have enough weight to expect a southern origin of modern R1, so the question is still open, IMO.

For now I'm more interested in seeing how this leaves the WHG-EEF-ANE model. It seems that there is a 4th component, "Teal" (or CHG now), that behaves like a mix of ENF and ANE, but that it seems to be neither (if ENF is taken as Anatolian Farmer kind). This will affect Europeans, but even more Near Easterners and Iranians/S-C Asians. I hope the genomes become available soon so we can learn more about this.

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

what is the mtdna of the caucausian HGs?

bellbeakerblogger said...

Really interesting to have a profile concretely attached to someone of the Azilian Industry. (I said Aterian on Maju's blog several weeks ago and meant Azilian)

In any case, clearly the Swiderians and Azilians were this I/U5 group of people. This means we can probably expect Mesolithic Britian to look similar in the Creswellian stage (which we already have U5 mtdna haplotypes) And this will persist as Britain thaws until the first farmers relatively late.

And again, I point to the early mtdna studies in Portugal and suggest that every one of those samples is contaminated or Neolithic admixture.

Matt said...

"West Asian" is real! So cool that so much has been resolved in almost only one year.

I think I'd probably agree with Davidski on this one, at first you might think "Aha, these are the guys who brought Indo-European languages to India", but it seems very good evidence still that the Indo-European languages dispersed from Yamnaya and later cultures.

Whether this population (CHG) brought pre-proto Indo-European languages to the steppe is a hard question. It seems like proto-Indo European itself is likely to have expanded from the steppe. It's not impossible for them to brought the languages there, then for y replacement to have happened. It seems like this has happened in other IE cultures (e.g. R1b being largely replaced by R1a).

What this does mean for sure is that we can get more sensible and workable models for South Asia that don't try to use Yamnaya or Sintashta and present day West Asian inputs that don't really make sense (as much as maybe the qpAdm did seem to fit). We can finally understand much better the peopling of the subcontinent (it'll still be pretty hard without proper adna from India to confirm whether it's CHG ancestry or something close, which mixed with relatively similar to, but distinct from, Onge.)

I think it'll be hard to know if all the CHG related ancestry went to South Asia from the Neolithic on. It may not have done.

From the trees, Kotias (CHG) looks like the most European HG unadmixed Middle Eastern population. We will have to see if Early Neolithic Anatolia changes this though.

Interesting that EF is the only population to have maintained high population size. Exploring this will be cool, because Yamnaya do not show unusually high FSTs compared to European MN or EEF. So that seems to raise a question mark from me. Maybe we will find a less inbred version of this population elsewhere (or maybe not).

For models of population this also means that models with free floating EHG, CHG, WHG and Anatolia_EN can be used for Europe, so we could end up seeing better and more informative fits, than we got just through WHG, European EN, Yamnaya/EHG, particularly if there is some degree of variance in relatedness to CHG relative to EHG that is not quite captured by using Yamnaya (which has a fixed ratio).

Dude Manbro said...

@ Romulus,

K3 and H13c

Matt said...

Grab bag of thoughts in no particular order:

- CHG and EF do not form a clade with respect to WHG (or thus modern Europeans) but do form a clade with respect to Eastern Non-Africans and MA1. This would support how qpAdm can't really tell them apart so well using only Yoruba, Native Americans and East Asians, and using modern West Asians as candidate donors.

- Looking at the shared drifts in the supplement. There's quite a bit of variation in Neolithic farmers in relatedness to CHG (Kotias and Satsurbalia). Bichon OTOH seems to really be a signal of "WHG" or European HG ancestry (perhaps more of one than Loschbour or La Brana).

- Strange that there's still the strongest f3 signal for some populations with ANE, despite the presence of EHG... I suppose this is a peculiarity of the f3 statistic.

- Also in Europe, from Fig4a seems slightly more contribution from Kotias in NW relative to NE, even where we have previously found slightly less from EHG...

- Re: Basal Eurasian, with Kostenki, CHG and EEF all having the same shift from East Asians, despite all having different relatedness to WHG, seems strange. How can they have all got the same Basal Eurasian effect, separately and independently to the same degree?

- One more interesting thing here, if I'm reading this correctly: Bichon, the European Upper Paleolithic Hunter Gatherer is here *derived* for SLC45A2 (light skinned variant), as is CHG. So what is going on here, with Loschbour and La Brana being the odd ones out? Did selective pressures shift during the LGM? (It seems unlikely that it was giving WHG Cantabrian Refugia guys a fitness hit, then not people in Georgia and Anatolia).

Rami said...

You don't have genome data from Bronze Age Central Asia or Iran, so you cannot make conclusions already. I agree with Nirjhar on this.

Also the article insinuates ANI arrived in South Asia with Indo Iranians, this is WRONG, when it clearly arrived there in the late Ice Age. There seems to be 2 types of CHG, one which is archaic and more localized (because its so old) to S/SC Asia and one which was brought to South Asia with Indo Aryans in the Bronze age. Though it gets more complicated as the Indo Aryans who arrived in South Asia had very likely mixed with BMAC populations already.

Alberto said...

@Matt

Great to read your comments and see that we agree about many things (the IE question I'd personally leave it for later, though. This is not any definitive answer to that question).

"What this does mean for sure is that we can get more sensible and workable models for South Asia that don't try to use Yamnaya or Sintashta and present day West Asian inputs that don't really make sense (as much as maybe the qpAdm did seem to fit)."

Yes, indeed. This is going to be a great step for better models. For Europeans too, but more so for Near Easterners and S-C Asians. I'm wondering how will this impact ANE, since up till now this "West Asian/Teal" seemed to behave like a mix of ANE/ENF, but now it will be a component on its own, probably reducing ANE in Asia? But this has caveats (more later).

"nteresting that EF is the only population to have maintained high population size. Exploring this will be cool, because Yamnaya do not show unusually high FSTs compared to European MN or EEF. So that seems to raise a question mark from me. Maybe we will find a less inbred version of this population elsewhere (or maybe not)."

Yes, my first thought is that these samples are just a small part of what we'll find further east. If you look at the map (page 6 of SI) it's clear why they were isolated from the north and from the south (mountains) and from the west (Black Sea), but they were connected to the east through the South Caspian shores. I think the biggest part of this population will be from the South Caspian to the east, and I'd think they expanded throughout he Neolithic through all of Iran and S-C Asia. I also doubt the population that mixed in Yamnaya is this very same one. In the Caucasus nothing much was happening before 4000 BC. while in North Iran, Turkmenistan and SE Uzbekistan they had Neolithic since around 6500 BC (and earlier all the way down to the Indus Valley). So I suspect that the population that brought domestic animals to the steppe was not this one from the Caucasus, but rather another one entering from the east Caspian.

More importantly, would that population be just CHG-like? Or would it have mixed with ANE? CHG doesn't seem to have much affinity to MA1, while for example the Kalash do. And the Kalash don't seem to have any Anatolian farmer ancestry (so no European ancestry either), so they can't be just CHG + Kharia (or Onge-like). They could have Afanasievo on top of that to bring them ANE affinity, but their ANE (MA1) affinity seems to exceed what would be expected from Afanasievo admixture alone (plus no evidence of Afanasievo migration to South Asia). So what I'm thinking is about the possibility of Central Asia as a place where ANE and CHG mixed, creating an Afanasievo-like population (but lower in WHG).

Interesting times. Let's see if before the year end we have the Indian DNA and really get the full picture.

Krefter said...

I've insisted ever since Laz 2013 Basal Eurasian didn't originate with farming. It didn't make sense to me that Teal was a mixture of EEF's brother and native "ANE"-like Caucasus folk. The mtDNA separation between Yamnaya's Near Eastern side and EEF is from Palaeilthic times.

There were lots of people in West Asia before farming began, and these Hunter gatherer Caucasus genomes prove they have hunter gatherer ancestry. EEF could have had a lot of Anatolian hunter gatherer ancestry.

mtDNA diversity in West Asia suggests a more complicated history than population-replacement by early farmers in the Levant.

Aniasi said...

Would you be able to run an analysis that looks at the percentage of CHG in South Indian populations, or South Asian if you lack data, and see how much CHG is found in each one? I would be very interested in seeing how this compares to generic Central Asian/Kalash/West Asian.

Krefter said...

The paper states it as fact that the Indo Aryans came from a CHG-type population from the Caucasus, because S/C Asians have a lot of CHG-relates ancestry.

"CHG left their imprint on modern populations from the Caucasus and also central and south Asia possibly marking the arrival of Indo-Aryan languages."

They don't understand language can spread without a huge impact in genes. S/C Asians have CHG-type stuff from North European-like early Indo Iranians and actual CHG ancestors from West Asia.

Davidski said...

Nope, they're not that stupid. Page 5:

Finally, we found that CHG ancestry was also carried east to become a major contributor to the Ancestral North Indian component found in the Indian subcontinent. Exactly when the eastwards movement occurred is unknown, but it likely included migration around the same time as their contribution to the western European gene pool and may be linked with the spread of Indo-European languages. However, earlier movements associated with other developments such as that of cereal farming and herding are also plausible.

Davidski said...

By the way, the genome files are getting processed, but might take a while, because they're pretty big.

Matt said...

@ Alberto: I've split a response to you into a couple parts.

(the IE question I'd personally leave it for later, though. This is not any definitive answer to that question) .... This is going to be a great step for better models. For Europeans too, but more so for Near Easterners and S-C Asians.

With IE, I don't know if it's necessarily a shut book, though, for me, to try and make the case as it seems to me now, when it comes to the language dispersal, I don't really too strongly hold to the steppe model of dispersal, but it seems to be favoured by linguists and there are some linguistic facts which I believe make some sense in terms of Indo-Aryan languages being a late dispersal from the steppe, rather than from CHG alone.

From what's new from this paper, when it comes to y chromosomes haplogroups, the complete isolation of CHG from ANE / EHG seems like it argues against any haplogroup R being likely to be found in CHG, and that would argue for at least some y dna switching in steppe IE and then a wide dispersal from that, to spread R to Europe and India, ideally at a later stage when R1a seems to supplants R1b on the steppe (at the same time as the steppe changes to a structure with more Early European ancestry).

What I really found questionable about the spread of Indo-European was mainly the idea that all the autosomal relatedness to Yamnaya in South Asia could've come through either a Yamnaya / Sintashta->South Asian path, and this seems to show promise in sort that out to a large degree.

One thing with South Asia which CHG does raise is investigating how CHG and MA1 relate to the Kalash. It may not be so crazy for them to be modeled as a CHG like sister clade after all, with other South Asian input (despite how we were all like "Oh, of course they're just a Bronze Age mixing of populations, plus drift"), although I am wary of predictions at the moment.

In time I guess someone may be able to look at Central Asia to see if it looks like CHG has been moving independently, or if all populations there model with mixes of Yamnaya+South Central Asians+appropriate East Asians (although those all seem quite hard to test together, because of the array of different combinations even within West-Central and South-Central Asia, very complex).

Matt said...

@Alberto: I'm wondering how will this impact ANE, since up till now this "West Asian/Teal" seemed to behave like a mix of ANE/ENF, but now it will be a component on its own, probably reducing ANE in Asia?

Yeah, one of the interesting things here is that, although it is found in close correlation with ANE influence, CHG doesn't seem to be any more related to MA1 than ENF is.

Looking back at what we were talking about before, there tended to be two views on here:

- "Teal" (CHG) is a mix of Yamnaya / EHG with an Anatolian EN like component (ENF), created spuriously by ADMIXTURE.
- "Teal" was a mix of ENF and ANE.

As you say, this is what the K8 analysis of teal seemed to show. But both of these now seem to be us getting it utterly, totally wrong, if CHG is teal, as this paper suggests.

Together with its lack of relatedness to MA1, CHG also seems far too anciently divergent from ENF to be ENF plus anything (unless we find something way, way, more divergent than CHG from ENF, with exactly the right properties).

So trying to explain this: It seems like it must've at least partly been the consequence of using ADMIXTURE methods that forced the teal / West Asian component within supervised ANE and ENF clusters, and the fact CHG and EHG tend to be found in close correlation in many populations. ADMIXTURE must to some extent have been making a bad compromise to find "teal" as ANE plus ENF, but really doing the best it could with the options we gave it.

Then maybe that was compounded by using the qpAdm methods that did not have adequate outgroups to distinguish between CHG and ENF (like trying to find the difference between Onge and Han using only the CHG and ENF groups - totally useless, as CHG and ENF both form a clade together to Onge and Han).

It seems if that's correct, ANE will go down across the board. Then again, IIRC, absolute affinity (shared drift) to MA1 was never that high in South Central Asia. It was only relatively high there to MA1 compared to the WHG / ENF affinity there (some populations having an almost even degree of relatedness to MA1 and WHG, unusual when the rest of West Eurasia is on the whole more WHG related).

But that makes sense, since CHG is supposed not to have any special relationship to WHG or ENF I think. So *relatively* high relationship to MA1, but not *actually* high, just an even degree of relatedness to WHG and ENF relative to MA1, which is unusual for West Eurasia, where people tend to fair amounts of WHG / ENF (and particular South Asia seems to lack ENF and WHG, probably as less of this was brought over time, by later migrations).

We tended to interpret this in terms of MA1 affinity being high there, then at the same time some mysterious spooky X factor has decreased it, but it seems to me now like there was never any need for it to have ever been unusually high to begin with, and something related to CHG plus something ENA related may be sufficient.

I can't dismiss your idea that the real "teal" is ANE-CHG mix, CHG itself, though. It seems plausible that such populations could exist, and I haven't totally read through the paper (lot of information to digest and I've only had a quick skim through, and am a bit busy so don't have much time to read today), so there may be some details that might indicate that something like that was possible.

Matt said...

Btw, I got it wrong about Bichon's pigmentation genetics "We found that
Kotias and Satsurblia have the ancestral version of the SLC45A2 (rs16891982) variant but both CHG have the selected version of the SLC24A5 (rs1426654) gene (Supplementary Table 21) encompassed by the most commonly associated haplotype (C11) found in modern populations. Bichon on the other hand has the ancestral version of both genes suggesting that our Caucasus hunter-gatherers may have had lighter skin than our western hunter-gatherer, Bichon (Supplementary Table 20).

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

About Bichon: "The human skeleton was determined to be of a young male, 20-23 years of age, of the Cro-Magnon
type. "

There we go, Haplogroup I confirmed for Cro-Magnon, so we should see it in Aurignacian and Gravettian as well.

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

@Matt

he was heterozygous for the Blue eye gene as well, which is consistent with La Brana and Loschbour.

rozenblatt said...

I wonder whether we will see any new papers from this team soon. If you look at the list of samples: https://sites.google.com/site/pinhasierc/home/samples , there are still a lot of unpublished samples.

Matt said...

Btw, just a quick one, assuming the proportions from Haak et al's models and that Yamnaya=50:50 CHG:EHG, while European EEF=90:10 Antolia_EN:WHG, I think this is what the proportions for some populations may look like - http://i.imgur.com/YyJhqL6.png. I reckon the real thing will change from that, where CHG and EHG can vary freely from one another.

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

@Matt

I think the Haak estimates will look a lot different once the new data is accounted for, they were very simplistic and I see a lot of contradictions.

Alberto said...

@Matt

Thanks for all the insightful comments. It all makes sense to me. I'd still like to explore that possibility of S-C Asians having some real ANE and not just CHG (but that should be easy to find out with these genomes and get a clear answer), and the IE question I'm also really quite neutral about it in spite of what it might seem, but I honestly have doubts about the current model, or part of it at least (but let's see how much these new findings help us in that regard).

@Davidski

So you have the genomes already?? Great news! I can't wait to see some tests with them. Thanks once more for all the time you put into this. I've seen you really busy running many requests and helping out everyone these days, and now these new genomes come in and will require a lot of time too. All your work is much appreciated!

Joshua Lipson said...

ryukendo: on the contrary, this is yet another piece of evidence that proto-Semitic has nothing to do with the north or haplogroup J. Its relationship to Egyptian, Berber, and Cushitic places it squarely within an African phylogeny. The only alternative is much less parsimonious: to suggest that all of the Afro-Asiatic languages, from Semitic out to Chadic in northern Nigeria and South Cushitic in Tanzania have Mesopotamian roots.

If proto-Semitic itself is to be located outside of Africa, the place to look is the Levant. And despite the preponderance of J there now (it's not even that overwhelming -- look at the non-Islamized populations of the Levant and you'll find an uneven mix of patrilineages that's <50% J), a late arrival in the Levant is suggested by the pre-Neolithic presence of J far away in the Caucasus. Existing phylogenies of both J1 and J2 already pointed to a northeastern origin.

However J-P58 came to be associated with Semitic populations, it seems much more likely that Semitic was initially spread by E-bearing populations (specifically E-M34, E-V22).

Tobus said...

The placing of CHG on the "Basal Eurasian" branch seems a little premature to me. Reading the paper it seems to be based primarily on the zero DStats for EF/CHG vs both ANE and ENA, but I don't think that's necessarily conclusive. I'd like to see some direct comparisons against UI like D(Chimp, UI; X, CHG) where X is ANE, ENA, WHG, K14 etc. to see if there really is something non-UI in CHG.

Grizzlor said...

I wonder, how much Neanderthal admixture these CHG had? I've always pondered whether the myth of hairy armenoids having significant neanderthal admixture was true.

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

@Grizzlor

I am also disappointed we haven't seen Neanderthal estimates for a long time in any of these ancient samples. I've asked David and others to try but they decline. Last time we got Neanderthal estimates was in Laz

Table S6.1. Estimates of Neanderthal ancestry for ancient samples
Estimate Std. Err. from a Block Jackknife
AG2 1.7% 0.5%
MA1 1.6% 0.3%
LaBrana 2.2% 0.3%
Loschbour 2.1% 0.3%
Motala1 2.2% 0.5%
Motala2 1.8% 0.6%
Motala3 1.9% 0.4%
Motala4 1.2% 0.7%
Motala6 2.1% 1.1%
Motala9 0.8% 1.8%
Motala12 1.9% 0.3%
Motala_merge 2.0% 0.3%
Skoglund_hunter 3.8% 1.0%
Skoglund_farmer 3.8% 1.8%
Stuttgart 1.8% 0.3%

It seems most ancient WHG samples were below the European average of 2.7% with the exception of the Skoglund samples which were closer to the East Asian average of 3-4%. That might just be a function of the higher error though.

Rob said...

BTW can anyone make out Supp Fig 5 (all the modern ADMIXTURE graphs) better ?

Chad said...

Please, let's retire the talk of teal and ANE. Clearly, other stuff is going on. Teal was never an ancient pop to begin with.

On another note, looking at some f3 stats, I find it interesting that the Baden sample shows a more significant fstat with the Caucasus samples than Beaker and Hungary BA. I got something similar on Admixture, but was wondering if it was a quality issue. Perhaps not. I'm interesting in checking this out. Continuous flow from Anatolia makes me wonder if this was harbored in Hungarian Neolithic samples after 5000BCE. I can't wait to run these in Admixture. Funnelbeaker sharing wagons, burial mounds and copper, could be a link with them and pre-Maykop back towards the SW Caucasus and NE Anatolia.

Rob said...

What a great paper.

I'd like to share a few comments.

@ Davidski

" Clearly, R1 appears to be a steppe marker from way back, and I personally find it very difficult to entertain the notion that Indo-European languages were introduced into highly patriarchal steppe societies by female migrants from the Caucasus."

-> Yes, fair point. Although, language inheritance works in more complex ways than that of Y haploid markers. In fact, ethnographic evidence has often pointed to maternal language inheritance, even in 'patriarchal' societies, especially if the exogamous females were from 'prestigious' regions. The shift from Pictish to Scots Gaelic comes to mind as a broad analogy. But at the moment a more northern origins is difficult to argue against; but I don't think its specific local need have been the steppe. I note Pinhasi's article very wisely describes Yamnaya as a "vector" for spread from Caucasus to Europe. They must have been reading this blog ha ha


@ Nirjhar

"further testing from Chalcolithic is required (Caucasus), but i'm not surprised.
I suggest now to look for upcoming Mesolithic and Neolithic aDNA of Iran.."

-> Of course. Cant' wait.

@ Roy King

" I find this study incredibly exciting. It confirms my speculations that I/I2 are the Y chromosomes for European hunter gatherers while J/J2a are the corresponding Y's for now West Caucasus hunter-gathers. Bar-Yosef, who led the excavations for Upper Paleolithic/Mesolithic West Georgia is a superb archaeologist and now, I would wager that I tracks the expansion from a European refugium, J from a Black Sea/Caspian refugium and G from a Levantine/SE Anatolian refugium"

-> Couldn't agree more ! But i'd stipulate that the J refugium was very much in the south Ponto-Caspian region. ie right here in the Caucasus where found. I suspect other refugium existed further east.

I'm still not satisfied that we have found the R/ ANE refugium.

@ Bell Beaker

'In any case, clearly the Swiderians and Azilians were this I/U5 group of people"

-> Basically, these people are all descendants of Magdalenian re-populants of Europe. They'll all be U5b and I2.

@ Matt

"Whether this population (CHG) brought pre-proto Indo-European languages to the steppe is a hard question. It seems like proto-Indo European itself is likely to have expanded from the steppe. It's not impossible for them to brought the languages there, then for y replacement to have happened. It seems like this has happened in other IE cultures (e.g. R1b being largely replaced by R1a).'

-> Quite plausible if not parasiminous. But language expansion if more difficult than a matter of tracing lineal genetic descent. For now, although the picture looks more or less clear, I want some genomes from India and Greece to seal the deal :)

Whatever the case, the simple Kurgan out-of-Yamnaya scenario is wrong on multiple levels- which is what I always maintained.

NB thanks for the new "splits". Any values for central & south Asian populations ?

@ Alberto

"I also doubt the population that mixed in Yamnaya is this very same one. In the Caucasus nothing much was happening before 4000 BC. while in North Iran, Turkmenistan and SE Uzbekistan they had Neolithic since around 6500 BC"

-> Very true. But I want to see Majkop - and even pre-Maykop ("Meshoko" phase) genomes. The people that moved into north Caucasus just as easily could have been a mixed EHG/ WHG groups from the northwest.

Unknown said...

Admin, could you please run me on Eurogenes K8, my code is M535853

thank you :)

Davidski said...

Errm...I don't have access to your genome via GEDmatch.

Also, the K8 is no longer available. There's a new test on the way. More info on that soon.

Krefter said...

"@Davidski

So you have the genomes already?? Great news! I can't wait to see some tests with them. Thanks once more for all the time you put into this. I've seen you really busy running many requests and helping out everyone these days, and now these new genomes come in and will require a lot of time too. All your work is much appreciated!"

I second that.

Unknown said...

The autosomal component behind Y-Dna can change rapidly (I2a went from WHG to EEF quickly, we have J* EHG), so this doesn't prove anything other than the "teal" comes from CHG. R1b was ubiquitous all around the Caspian Sea, while R1a occupied Russia. A south Caspian R1b population (M269* is most popular in Northern Iran) acquiring CHG admixture and moving north is a lot more realistic than R1b1* mutating to M269 -> L23 -> Z2103 -> 05 -> 09 in just 2 thousand years in the same spot.

The one question this study brings up is which clades of J in Europe today are IE, EEF, or Etruscan-like.

Davidski said...

There's no evidence that R1b-M269 was ubiquitous all around the Caspian Sea before the Middle Bronze Age.

Unknown said...

Can someone check whether CHG mtDna matches up with Yamna mtDna?

Davidski said...

From the supp info...

"Sub-haplogroup H13 is most common in the Near East and Caucasus reaching highest frequencies in Georgia and Daghestan 39. Interestingly this sub-haplogroup has been found in individuals from the Late Neolithic Bell Beaker culture in Germany and the Early Bronze Age Yamnaya culture from the Pontic Steppe 7."

Krefter said...

Several Steppe mtDNAs are H13. But none are H13c. I don't much about it.

Unknown said...

Maciamo's comment "What surprises me is that neither the Y-DNA nor the mtDNA lineages show an association with Yamna or any Indo-European culture. MtDNA K3 is found almost exclusively in Georgia nowadays, which means it wasn't part of the maternal lineages that mixed with Proto-Indo-Europeans. Likewise H13c is essentially found in the Caucasus and Middle East. Although H13a was found among Yamna and Bell Beakers, all H13 subclades are relatively rare in Europe today, except in Sardinia (8%), which is the only part of Europe with virtually no Steppe ancestry. When we see that the paternal lines belong to J* and J2a, two lineages also more common in the Middle East than Europe, it makes me wonder whether Mesolithic Caucasian really are ancestral to Yamna and other IE people."

Davidski said...

Maciamo can wonder all he likes, but the descendants of CHG obviously contributed ~50% of the genome-wide structure of Yamnaya and almost as much of most Corded Ware.

It's also hard to ignore that western Georgia is very close to the steppe. So why look elsewhere unless for some reason you're not happy with these results?

Krefter said...

Maciamo is not up to date. He doesn't have the time to pay attention. 5 years ago haplogroups were the focus and so he still focuses on them.

Krefter said...

The ADMIXTURE results are impossible to read in Supp info. But the paper gives a good description of the results in Supp. Note 9. In Summary what they say is...

>A Middle Eastern component appears in K=9.
>From K=9-K=15 CHG is a mixture of it and a South Asian centered component.
>The South Asian component first appears in late Neolithic Europeans, and today most Europeans and West Asians have membership in it. The rest of West Eurasians score in SHG/EHG/WHG centered component and Middle Eastern component.

In K=15 the older CHG scores 100% in a new Caucasus-centered component but the younger one scores just under 10% in a Middle Eastern component. Caucasus today score almost 50% in Middle Eastern component and over 50% in CHG. From K=15-K=20 nothing really changes.
South Asians are a mixture of CHG component and another South Asian component.

>Europeans are mostly a mixture of SHG/WHG/EHG+CHG+MIddle Eastern. West Asians are mostly a mixture of CHG+Middle Eastern.

>By K=18 it looks like Middle Easterns splits into typical Mediterranean and SW Asian.

Kapak said...

Davidsky, funny how long you been entertaining this ANE hostile takeover of Caucasus women like some Tolkein-style fantasy. I'm no psychologist but it's almost as if you have a ego-investment in all this. I hate to break it to you, but even the Ruskies got bloodied up badly when they struggled with little Chechnya, let alone Poles. Leave Caucasus women out of it, sir. There are better ways to convey what you're trying to say rather than using women as some kind of commodity.

Davidski said...

One day dear Kapak you'll comprehend such basic shit as female exogamy and female mobility, and then you'll understand.

Kapak said...

On a more positive note, I would like to say you are somewhat heading in the right direction, as Caucasus is a hot-bed for the haplogroup u5, and seems to be a source for it rather than a dumping ground. But, why must you rule out the possibility that J & R1a was already native to eastern Europe/steppe before being stamped out by incoming R1b folk? Sometimes it's hard to follow your logic.

Davidski said...

Lots of things are possible, but only some things are plausible.

Krefter said...

@Kapak,
"On a more positive note, I would like to say you are somewhat heading in the right direction, as Caucasus is a hot-bed for the haplogroup u5, and seems to be a source for it rather than a dumping ground. "

There's nothing suggesting Caucasus is a source for U5. It's more popular in ever part of Europe than anywhere outside of Europe. We have U5* in Czech Republic from over 25,000 years ago.

Kapak said...

Aye, well let me say my logic is razor sharp so heed my words. BTW I preached of the 50%-50% Caucasus-ANE composition of the Proto-Indo-Europeans long before any Reich et en al. and what have you even mentioned it. See here and check the date-stamp:

http://histogen.blogspot.com/2013/10/the-origin-of-scythians.html

I'm not just babbling shit when I say something sounds off about "they took our women!" scenario. I think you're on the right track but something, is off, can't put my finger on it exactly.

Seinundzeit said...

Amazing stuff. I think this provides strong genetic evidence for a Kartvelian-Dravidian connection (we might be able to posit a deep, distant rooting of those language families in the languages spoken by these hunter gatherers, just like how we might be able to see Indo-European and Uralic possibly having deep roots with EHG).

It is also of interest, as RK noted, that the highest d-stat for GujaratiA and GujaratiB involves Afanasievo (IE, steppe population), while GujaratiC and GujaratiD are closest to Kotias. I can't wait to see some qpAdm models with these samples.

pconroy said...

My wife and our kids are mtDNA H13, maternally she is Sicilian...

I'm Irish and show 4% Caucasian Lezgin ancestry on admix tests...

Fanty said...

"something sounds off about "they took our women!" scenario."

Isnt a "they took out women scenario" something "biblical"?

I cant tell exactly where it was written and I might have read that 20 years ago, but something in the back of my brain tells me, god said in that one that in a war against another people, you shall kill all what is male. Kill the men, keep the women.
Oddly I can recall it even was said about the livestock of the enemy. Kill all male animals, keep the female ones for your male livestock.

Question is, are those pure fantasies or does this really have an example in real life.

Fanty said...

"entertaining this ANE hostile takeover of Caucasus women like some Tolkein-style fantasy.

Not Tolkien. I am sure its the Old Testament.

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

Maternal gene-flow between neighboring communities is a fact of life in all human societies. They probably swapped goats & cattle too.

Anonymous said...

@Rob

According to this graphic from the paper 24.000 years ago was in the middle of the LGM. That is the age of Malta. So maybe Siberia was the R refuge?

http://www.nature.com/ncomms/2015/151116/ncomms9912/images/ncomms9912-f2.jpg

Rob said...

Hi Epoque.

I think you asked that same question while ago. The peak glacial began 22 KYa and ended at 18. So I think mal'ta boy was "the last of his kind" in Siberia, so to speak

Ryan said...

"Nevertheless, it's now even more difficult to accept that Y-haplogroup R1 and the Proto-Indo-Europeans might have originated south of the steppe. Clearly, R1 appears to be a steppe marker from way back, and I seriously doubt that Indo-European languages were introduced into highly patriarchal steppe societies by female migrants from the Caucasus."

Davidski - I realize you only mean specific R1 groups, but I wish you would be more specific, since we know for a fact that at least some R1b had spread well beyond the Steppe prior to Western Europe prior to the Indoeuropean expansion.

Rafs said...

"But, why must you rule out the possibility that J & R1a was already native to eastern Europe/steppe before being stamped out by incoming R1b"

R1b was already in the steppe before Yamnaya and CHG admixture. No evidence that it was in the Caucasus before that.

Alexandros said...

@Roy King + others
If J2 originated in the Caucasus, any ideas on how and when it penetrated the Eastern Mediterranean, dominating places like Lebanon, Cyprus and Crete? Neolithic, Bronze age, or more recent??

Anonymous said...

@Rob

Yes I know we quibbled about this before, so to speak. I merely wanted to point out the authors apparently do consider 24.000 ya at the heart of LGM. But frankly I think you are right on the time frame. This link provides some insight on Siberia during the LGM. Whatever it is, it must have not been a very hospitable place.

http://www.esd.ornl.gov/projects/qen/new_eurasia.html

It does provide *some* evidence of some remaining forestry, but mostly it seems arid tundra. It also states that some animals such as mammoths, horses and saiga antilopes remained.

So while I think generally this supports your point it still may allow for some pockets within the area.

Rob said...

Epoch

I'd call it an amicable debate;)
But the fact is even academics are locked in debate about the settlement history of Siberia during the LGM (even with radiocarbon dates, etc) ; so I don't think you and I at the moment can resolve it

But I will state some concluding remarks

1) the peak LGM was 22-18 according to Palaeolithic archaeologists (I know some genetics papers have a lightly "off" idea of when the peak LGM was)

2) it is of course possible that *some* groups survived the peak cold in Siberia . But what really matters is was it reproductively viable and did it actually contribute much to later groups; or was it simply swamped by newer albeit related arrivals ?

3) mal'ta boy and his gang are an extinct lineage. They contributed neither to EHG nor Amerindians. Rather, a related group from somehwere else- perhaps from somewhere close by- did. .