search this blog

Monday, July 27, 2020

Ancient ancestry proportions in present-day Europeans (to be continued)


This year has already been massive in all sorts of ways, including for new data and software releases. So I'm thinking it might be time to update many of the analyses that were featured at this blog a while ago.

Let's start with the classic hunter vs farmer vs herder mixture model for present-day European populations. The rules of the game are as follows:


- run the latest version of qpAdm using qpfstats output

- use transversion sites and 1240K capture data

- pick a set of diverse and chronologically sound outgroups

- for a model to be successful the p-value must reach 0.01

- tweak the left pops in models that are clearly underperforming

- follow high end scientific literature, logic and common sense


Obviously, the reason that I decided to limit my analysis to markers from transversion sites is to mitigate problems associated with modeling the ancestry of modern, high quality samples with relatively low quality ancients. One of these problems appears to be qpAdm assigning faux East Asian/Siberian admixture to present-day Europeans (for instance, see figure 4 here).

My starting reference populations and outgroups are listed below. In qpAdm terminology the former are known as the "left pops", while the latter as the "right pops". Most of these samples are freely available at the David Reich Lab website here.

left pops:
HUN_Koros_N_HG
TUR_Barcin_N
UKR_Yamnaya

right pops:
CMR_Shum_Laka_8000BP
MAR_Taforalt
Levant_Natufian
IRN_Ganj_Dareh_N
Levant_PPNB
CZE_Vestonice16
BEL_GoyetQ116-1
Iberia_ElMiron
RUS_Karelia_HG
RUS_West_Siberia_HG
MNG_North_N
RUS_Ust_Kyakhta

As you can see, I picked a wide variety of right pops. But I chose most of them specifically to be able to differentiate the three streams of ancestry - from ancient hunters, farmers and herders - that are the focus of my analysis. I also intentionally avoided using samples in the right pops that may have experienced gene flow, including cryptic gene flow, from the populations in the left pops.

I somewhat speculatively earmarked HUN_Koros_N_HG, from the Early Neolithic Carpathian Basin, and UKR_Yamnaya, from the Early Bronze Age North Pontic steppe in what is now Ukraine, to represent the hunter-gatherer and pastoralist streams of ancestry, respectively.

That's because I expected HUN_Koros_N_HG to be the best proxy for the hunter-gatherer ancestry that was initially absorbed by the early farmers who fanned out from the Aegean region across much of the European continent, and of course it made sense to choose a steppe pastoralist population that was located close to Central Europe where such groups first made the biggest impact outside of the steppe.

Interestingly, HUN_Koros_N_HG and UKR_Yamnaya did prove to be among most effective choices for the types of ancestries that they represented. For instance, UKR_Yamnaya generally produced much stronger statistical fits than a very similar set of Yamnaya samples from the Caspian steppe (more precisely, from the Samara region in Russia). However, this might well be an artifact, due to very specific characteristics of these few ancient individuals. Larger sample sets would be welcome, especially from Yamnaya sites in Ukraine.

Below, dear audience, is a spreadsheet featuring the preliminary results. Click on the image to view and/or download the spreadsheet. The general rule is that the higher the tail prob, or p-value, the more likely it is that the ancestry proportions are close to the truth (a tail prob of well below 0.05 is usually a strong indication that something isn't right). For a detailed look at each of the qpAdm runs, feel free to consult the zip file here.


Note, however, that many of the European groups in my burgeoning genotype dataset are yet to make an appearance in the spreadsheet. That's because their models with the standard left pops showed p-values well under 0.01, which essentially meant that they failed, and I'm still trying to make them work.

But round one has certainly revealed some fascinating stuff. For instance, except for Hungarians and Estonians, none of the Uralic-speaking groups can be modeled successfully in the standard three-way model.

However, I managed to significantly improve the statistical fits in their models by adding a Siberian population, RUS_Baikal_BA, to the left pops. This is unlikely to be a coincidence, because the Proto-Uralic homeland was almost certainly located in or very near Siberia. Iain Mathieson please take note.

Saami
HUN_Koros_N_HG 0.134±0.043
RUS_Baikal_BA 0.270±0.015
TUR_Barcin_N 0.081±0.026
UKR_Yamnaya 0.515±0.058
chisq 19.865
tail prob 0.0108571

See also...


536 comments:

1 – 200 of 536   Newer›   Newest»
Davidski said...

There's someone trolling at a couple of places online about Poles having a lot more Karelia_HG ancestry than Northwestern Europeans.

This person appears to be seriously mentally unstable, which isn't unusual in this scene, but in this case the trolling is very persistent.

Of course, Poles don't have any more Karelia_HG-related ancestry than what we have as part of our ~50% of Yamnaya-related ancestry. This is easy to demonstrate in any number of ways, including a variety of formal statistics-based analyses. Here's a qpAdm/qpfstats mixture model.

Polish
HUN_Koros_N_HG 0.127
RUS_Karelia_HG 0.000
TUR_Barcin_N 0.321
UKR_Yamnaya 0.552
chisq 7.554
tail prob 0.478197

https://drive.google.com/file/d/1fKs65Q4K3Tc1wBOUVf6jKH2Hnz4weMYt/view?usp=sharing

Poles certainly do have more indigenous European hunter-gatherer ancestry than most Western Europeans (not including Scandinavians), but this is derived from Central Europe and the south Baltic region, and the foragers in these places weren't similar to Karelia_HG.

ANI EXCAVATOR said...

Why the shoutout to Mathieson?

Matt said...

The results look intuitive. The caveat would be that this model has to be kind of stylized at its limits to be applicable across Europe; we know that in NE Europe, some of the flow that formed BA-IA populations probably from HG groups with some level of ANE/EHG related ancestry (Narva/Motala like groups or straight EHG groups).

The 3-way WHG:Anatolian:Yamnaya model probably only really obtains at the fringe of NW Europe. But those are unlikely to do too much more for most populations than shift things at the edges by a few %.

That said, some cross checks with G25+Vahaduo for consistency: https://imgur.com/a/pYknRom

There's a slight difference where Vahaduo data follows roughly the same cline as qpAdm, but is shifted about -10% away from Yamnaya along it. Within that, there is a slight difference that there is a bit more geographical patterning in WHG related ancestry to NE Europe in Vahaduo run, and that there is a bit more regional stability in Yamnaya/Barcin related ancestry.

(Here's another quick experiment, using the proportions from the qpAdm, you can use multivariate equation to produce ghosts based on those proportions. Pastebin of results of that: https://pastebin.com/CXeb5YHT, then these ghosts dropped onto a Global25 Vahaduo West Eurasia - https://imgur.com/a/V9hTxrm. These ghosts have similar positions to the real ones, but some shift - the "ghost" Anatolian is shifted away from Europe, as forms loewr proportions, while ghost Yamnaya is shifted north relative to the real one, and the HG ghost is shifted towards populations, as less consistently affiliated in qpAdm towards populations which are more northern on G25).

Tigran said...

@Davidski

There's also a few morons arguing that northern EHGs were actually Proto Uralic speakers and that Proto Uralic is not related to heavily East Asian populations from the Baikal region carrying ydna N.

Copper Axe said...

What is the Uralic homeland according to Mathieson? Or rather, which ancient populations does he think were Uralic speaking. Does it start with an Eastern?

Penkki said...

Could some of the statistical fits be made better with the addition of a West Asian population?

weure said...

thanks, useful!

pnuadha said...

There's someone trolling at a couple of places online about Poles having a lot more Karelia_HG ancestry than Northwestern Europeans.

was that the motivation for this post? jk

Can we take any of these results as precise and accurate, or even in terms of the "rankings". How much do these 3 components shift when you use different variations and ideally the actual ones that contributed to a population. I find it fishy that the Irish have an extra 4% steppe (0.572 - 0.529) than the scottish. Maybe thats how it really is but how could we truly verify this?

Why does Norweigan score relatively low on this model but on the Haak model it was the highest?

Also, shouldn't there be an EHG category separate from steppe? We know EHG survived separate from the 50/50 yamnaya type.

Samuel Andrews said...

It is good to see an updated effort to get ancient ancestry percentages for modern populations.

The Yamnaya scores are about 10% points higher than we are used to. The WHG scores are lower than we are used to.

I'm not saying "This is WRONG". I guess different methods get slightly scores. I agree with the older scores: 45-50% Yamnaya at the highest not 55-60%.

Davidski said...

@Samuel Andrews

I guess you didn't notice that I used UKR_Yamnaya instead of RUS_Yamnaya_Samara.

RUS_Yamnaya_Samara produces lower Yamnaya estimates for these Europeans, generally in the 50% range, but the fits are much worse.

So it seems like the Yamnaya or Yamnaya-related group that founded the Corded Ware population came from further west than Samara and already had some extra forager and even farmer ancestry.

Davidski said...

@Matt

The caveat would be...

I investigated this caveat. It doesn't appear to be a significant factor. For example:

Polish
HUN_Koros_N_HG 0.127
RUS_Karelia_HG 0.000
TUR_Barcin_N 0.321
UKR_Yamnaya 0.552
chisq 7.554
tail prob 0.478197

https://drive.google.com/file/d/1fKs65Q4K3Tc1wBOUVf6jKH2Hnz4weMYt/view?usp=sharing

Also see figure 4 here.

https://www.nature.com/articles/s41467-018-07483-5

It really does seem that the extra hunter-gatherer ancestry that exists in Eastern and especially Northeastern Europe is overwhelmingly of the Western European hunter-gatherer (WHG) type.

Davidski said...

@pnuadha

Also, shouldn't there be an EHG category separate from steppe? We know EHG survived separate from the 50/50 yamnaya type.

But did it really? Where's the evidence for this?

Look at this recent paper. EHGs were basically totally wiped out by migrants from the south who were rich in WHG ancestry.

https://eurogenes.blogspot.com/2020/07/fatyanovo-males-were-rich-in-y.html

Also, see my reply to Matt above.

Samuel Andrews said...

@Davidski,

Last I checked, in G25 PCA Yamnaya Ukraine is the same as Yamnaya Samara. I am surprised using Yamnaya Ukraine, qpADM scores are so high using Yamnaya Ukraine.

Yamnaya Ukraine is swallowing a lot of what was before counted as hunter gatherer ancestry. Early Corded Ware genomes cluster slightly north of Yamnaya in G25 PCA, but i don't think they have they can swallow up this much hunter gatherer.

Samuel Andrews said...

Um......We need high coverage Early Corded Ware and Western Yamnaya/Sredny Stog I guess.

"RUS_Yamnaya_Samara produces lower Yamnaya estimates for these Europeans, generally in the 50% range, but the fits are much worse.

So it seems like the Yamnaya or Yamnaya-related group that founded the Corded Ware population came from further west than Samara and already had some extra forager and even farmer ancestry."

Gabriel said...

Is it accurate for Hungarians, Croats and Slovenes? They show up as having more steppe and less HG than Germans.

Davidski said...

@Samuel Andrews

The differences that I saw between UKR_Yamnaya and RUS_Yamnaya_Samara are interesting, and I don't know how to explain them.

It's possible that they're largely an artifact of using only 2 UKR_Yamnaya samples vs many more RUS_Yamnaya_Samara samples.

So it may well be a technical thing that affects qpAdm output.

But, in any case, that's what I got, and you need look at the standard errors as well when comparing models. They're not there just for the hell of it.

Davidski said...

@Gabriel

Is it accurate for Hungarians, Croats and Slovenes? They show up as having more steppe and less HG than Germans.

Germans should have more HG ancestry than these more southern populations. But you need to look at the models in more detail, like the tail probs and standard errors, to see what's going on.

If a tail prob is well below 0.05 than something might be missing and being covered by one of the three components. Indeed, it seems that the pops from the Balkans and nearby do need something else, like maybe extra CHG-related ancestry that arrived from somewhere else than the steppe?

Davidski said...

@ANI EXCAVATOR

Why the shoutout to Mathieson?

He once featured a very strange qpAdm run in one of his videos that showed the Saami as ~100% Yamnaya.

I e-mailed him about it, and he basically said the run wasn't supposed to be anything comprehensive or even all that accurate, but still, come on.

I'm pretty sure there are some loonies out there who now believe that Yamnaya ancestry peaks in the Saami.

Davidski said...

@Tigran

There's also a few morons arguing that northern EHGs were actually Proto Uralic speakers and that Proto Uralic is not related to heavily East Asian populations from the Baikal region carrying ydna N.

Yeah, I think Nick Patterson and David Reich did a talk at Harvard about 6 years ago in which they, or one of them, claimed that Karelia_HG may have represented the Proto-Uralic population, while Maykop may have been Proto-Indo-European.

OK, they were taking a fresh look at things based on very new data, speculating and no doubt having a bit of fun.

But those ideas have now taken on lives of their own online, and there are practically sects of loonies out there who religiously believe that Karelia_HG was a Uralic speaker and R1a is a Uralic marker.

And how do you argue with a crazy person who believes that a Harvard professor is always right, no matter how speculative and at odds with the facts their theory might be? You can't.

Bob Floy said...

Swedes have more Yamnaya than Lithuanians and Ukrainians?
Huh.

Davidski said...

@Bob Floy

Mind those standard errors.

There are no significant differences between the Swedes, Lithuanians and Ukrainians in terms of their Yamnaya ancestry proportions.

Bob Floy said...


@Davidski

"Mind those standard errors"

Right, makes more sense. Also would explain why the Scots seemed to have less Yamnaya than one would expect, I guess.

Ric Hern said...

I wonder if the big problem is largely due to Linguistics. Trying to fit Uralic or Proto-Uralic into the Proto-Indo-European story...? What was loanwords and when and for this purpose R1a as a whole got dragged into the story due to its current proximity to countries where Uralic languages are spoken ? Now it looks more like what was borrowed in Siberia and what was borrowed after the expansion of Uralic into Europe....

Slumbery said...

@pnuadha
Also, shouldn't there be an EHG category separate from steppe? We know EHG survived separate from the 50/50 yamnaya type.

I looked at this question with G25 nMontes multiple times with a lot of test runs with different assumptions. I settled with the conclusion that the only modern European population that has EHG-related ancestry more than noise-level above of what delivered by "steppe" is Saami and even that is more SHG than EHG. The assumption that EHG is survived in north and got incorporated into Uralic speakers (then subsequently Northern Russians) is logical, but does not appear to be true. At least not above an 1-3% level that is hard to identify in the strong presence of similar ancestries.

Davidski said...

Yeah, the idea that significant EHG survived in the genomes of Eastern Europeans was a very plausible one, but it just hasn't been backed up by any of the data.

In fact, it's been contradicted by both ancient and modern data.

But because it's such a plausible theory, it's assumed to be fact, and gets repeated over and over in these sorts of discussions.

It's time to move on now. The really interesting thing is the fact that most Uralic speakers don't show any significant EHG or Western Siberian HG (WSHG), but they do show significant eastern Siberian ancestry.

How did that happen exactly?

Matt said...

@Davidski, I did see that model; fundamentally I do think it is still ultimately going to be more logical that the extra contributor of HG ancestor was like all the other such populations in the region who show some EHG related ancestry, than that the contributor was a ghost population that showed none. We'll see how the models evolve over time. This is a valiant attempt at modelling here. G25 results do tend to be different, which may be because of some different influence of compression towards modern samples in ancients that behaves differently.

weure said...

Nice work David, not able to check check double check it.

Many use it as a model in G25 now:

any comments to that? Do's and don'ts?

Davidski said...

@Matt

The G25 doesn't really show anything different, apart from some noise due to recent drift. Refer to Slumbery's comment above, which is based on the G25.

Hunter-gatherer ancestry in Europeans is indeed of the WHG type, and it spread across Eastern Europe very recently, along with Corded Ware and closely related populations.

That is, it's not local hunter-gatherer ancestry. The local hunter-gatherers, like the R1b-rich Volosovo people, were wiped out.

Davidski said...

@weure

I don't know if it's sensible to try and copy this model with the G25, because of the overall low quality of the reference populations.

We really need more steppe samples from Ukraine, and then I think the qpAdm/G25 models will converge.

Shaikorth said...

@Matt

There are several options for building a sensible phylogeny, but it takes proximate LBA and IA samples to find out which populations were actually involved in the formation of modern genetic landscape. Even the ones used in below fit are just inadequate substitutes.

Estonia_IA_OLS10 (N1c1 carrier, isotopic analysis suggest he wasn't a local Estonian)

Estonia_BA_d 0.807
Sintashta_MLBA_o 0.193

chisq 7.126
tail prob 0.895538

right pops:
Ust_Ishim
Kostenki
MA1
WHG
EHG
CHG
Levant_N
Ganj_Dareh_N
Villabruna
LapaDoSanto
Barcin
Han
Mbuti
Mota
West_Siberia_N

Davidski said...

@Shaikorth

I'm guessing that model was run with an older version of qpAdm and based on very few markers, because the Estonia_BA_d is a damage only sequence.

That's why the fit looks so good probably.

Slumbery said...

@Davidski

G25 nMontes show excess EHG ancestry virtually in everybody in Europe with the same references, but this something I noticed in the past. That is: G25 nMontes inflates EHG ancestry in deeper ancestry tests.

"sample": "Hungarian:Average",
"distance": 3.9964,
"Barcin_N": 41,
"Koros_N_HG": 12.5,
"Yamnaya_UKR": 39,
"Karelia_HG": 7.5

"sample": "Polish:Average",
"distance": 5.5979,
"Barcin_N": 33,
"Koros_N_HG": 17.5,
"Yamnaya_UKR": 41,
"Karelia_HG": 8.5

"sample": "Estonian:Average",
"distance": 6.4951,
"Barcin_N": 25.5,
"Koros_N_HG": 20,
"Yamnaya_UKR": 37,
"Karelia_HG": 17.5

As you can see as far as Gn25 nMontes concerned even Hungarians have significant EHG ancestry above their "steppe". Not very realistic.

However including either Ukraine Mesolithic or Ukraine Neolithic completely eliminates both Körös_HG and Karelia_HG from the mix. And that steppe migration came from Ukraine.

"sample": "Hungarian:Average",
"distance": 3.7017,
"UKR_N": 24,
"Yamnaya_UKR": 33.5,
"Barcin_N": 42.5,
"Koros_N_HG": 0,
"Karelia_HG": 0

"sample": "Polish:Average",
"distance": 5.2136,
"UKR_N": 33.5,
"Yamnaya_UKR": 31,
"Barcin_N": 35.5,
"Koros_N_HG": 0,
"Karelia_HG": 0

"sample": "Estonian:Average",
"distance": 6.0686,
"UKR_N": 42.5,
"Yamnaya_UKR": 31,
"Barcin_N": 26.5,
"Koros_N_HG": 0,
"Karelia_HG": 0

These fits are pretty bad though, so something is missing anyway. Plus the Yamnaya ancestry is estimated to be much lower in nMontes.

Shaikorth said...

@Davidski
numsnps used was 226196 which I suppose is decent considering how many you can get out of OLS10. Mezhovskaya instead of Sintashta_o with similar amount of SNP's also had a workable but much lower tail prob.
But someone can probably try to repeat these kind of models

Queequeg said...

Even taking into account ancient Avar and Yakut type of eastern cases, anything younger than N-L708 TMRCA 7400 years does not in my understanding really support an Eastern Siberian origin of Uralic, in terms of geography and provided that we assume that paternal N is somehow related to that. Ob-Irtysh, maybe, but even that remains to be seen. For instance mostly Volga-Kama based N-Y9022 is, as a sublineage, older than say the Yakut specific sublineage. That being said, maybe there once was an Eastern Siberian looking ancient population in Ural area, including West Siberia. We'll see about that.

Matt said...

@Davidski, we'll see how these things evolve over time; even with the best models, these qpAdm have standards errors which are significant relative to the size of the differences we are talking about (a few percent).

(As last points, I don't think that an internal reconstruction of the cline of Lithuanian->Baltic_LTU->Baltic_LVA on G25 suggests a WHG like source without any EHG related ancestry. That could also be analysed formally using some of the methods Lazaridis has used to extend clines within populations using formal stats.

And a purely drift based effect in G25 would not be expected to make an unrelated HG population Baltic_Narva_LTU preferred to Koros_HG in Vahaduo - it would simply make both more distant.

And yet that seems to be the case in all models... I seemed to find that Baltic_LTU_Narva was preferred to Koros_HG, SRB_Iron_Gates, and Western HG groups, in Vahaduo+G25. There also seem to be very *clean* cleavages between present day Western and Eastern European samples in how Vahaduo selects POL_GAC vs Narva+Koros, or DEU_HG/Narva (e.g https://imgur.com/a/iCnM7PU / https://pastebin.com/3hUyAwMJ / https://pastebin.com/Gn1yY86n).

That is not strong evidence of anything but it seems strange to me that something so geographically plausible - indeed geographically expected - should be due to random drift effects, which somehow place an unrelated population relatively closer while placing unrelated ones relatively further away. Rather than displacing all populations, true ancestors and non-ancestors, an additional equal extra distance away.)

This isn't to say I'm confident Global25 is a superior or better method to the qpAdm here clearly! Not obviously so at all. There could be severe compression and "constraints of modern diversity" issues, where Global25 tells us less about "true" position of ancients relative to moderns and ancients to each other than we think. Ultimately all these models have limitations at dealing withe low %s of ancestry and with complex models. This looks to be an improvement over last previous published models in Haak's paper and probably last ones on here, anyway. It's useful to get an update to see what the more recent qpAdm methods using only ancients with similar capture and age in pright and pleft do when using moderns as targets. Anyway, that's about all worth saying on the topic for now (since it seems to have become emotivized for some reason.)

Slumbery said...

Additional note for G25 nMonte simulation: Dereivka (combined) works even better, but eliminates ancestry from Yamnaya proper even further. Not necessarily true, but a fit is significantly better (but still not stellar good) than the model with just Yamnaya and WHG + EEF + EHG.

And again, from the Y-Hg-s we can at least suspect that the source cannot be Yamnaya proper (or we have to assume that something big within Yamnaya went under the radar). So I think Ukrainan non-Yamnaya ancestry is plausible / withstand reality test.

As for why G25 nMontes gives such different with the same reference pops, I do not know and it is above my pay-grade to speculate about it in a forum full of people who understand the mechanics of these testes much better than me.

Davidski said...

@Matt

It's become emotivized because there's been a lot of trolling about this issue. It looks like a lot of people find it offensive to see that Eastern Europeans have as much steppe ancestry as Western Europeans, if not more.

But I already have a good idea about what happened in Eastern Europe.

Take it or leave it for now, but at some point you'll see a lot more data from there and from Siberia, and the recurring theme will be population replacements, over and over again.

Samuel Andrews said...

I compared Steppe ancestry scores in G25 using Yamnaya Samara compared to Corded Ware early from Poland and Baltic.

When, using CWC_Early, Steppe scores are 4%+ higher across the board.

So G25 supports that Steppe ancestry is actually higher in Europe than we thought, even we already knew it was high.

Samuel Andrews said...

Highest G25 Steppe scores using Corded Ware Early Poland and Baltic (who had no more farmer ancestry than Yamnaya).

Lithuania_RA/VZ: 52-53% (ranges)
Lombard: 52%
Saxons (England): 52%
Norway: 52%

Ukrainain: 51%
Polish: 51%
Icelandic: 51%

Irish: 50%
Swedish: 50%
English: 48%

Samuel Andrews said...

Something which seems off about these results is that East/West Slavs don't score extra hunter gatherer.

They score only 13-15% hunter gatherer, which would suggest their non-Steppe ancestry was on average 35% hunter gatherer which is not much higher than Globular Amphora's 30%.

Polish & Ukraine scores are very similar to Norweigan and Swedish scores. But, they should have more hunter gatherer.

un said...

The UKR Yamnnaya belongs to the Bronze Age, unlike others. What are the proportions of EHG CHG ANF WHG in it?

Samuel Andrews said...

Looking at G25, it looks like in Iron age Europe that Germanic and Balto-Slavic groups had highest Steppe ancestry at 53%ish. *When Ignoring Scythians who had more.

Iron age Europe......
Germania: 53%
"Venedi", Ukraine Belarus Baltic States: 53%ish
Britain Ireland: 50%
"La Tene Celts": 43%
Central-South France: 35-40%
Spain: 30%
Italy: 27%
Western Balkans: 27%
Eastern Balkans: 20%

Penkki said...

@Davidski

As for Balkans seeming to lack something in the models, that's what I was referring to with the West Asian population.

Davidski said...

@Penkki

I'm working on it.

But it might take me a while to come up with a successful format for all of Europe.

Slumbery said...

@Davidski

In G25 nMontes I usually use Kura-Araxes Kaps for modern Balkan populations. It works reasonably well.

An example (just to see the difference between Balkan and nearby non-Balkan):

"sample": "Hungarian:Average",
"distance": 2.9421,
"Dereivka_I_En2": 30,
"Dereivka_I_En1": 2.5,
"Yamnaya_RUS_Kalmykia": 31,
"Baden_LCA": 36.5,
"Kura-Araxes_ARM_Kaps": 0

"sample": "Bulgarian:Average",
"distance": 2.8355,
"Dereivka_I_En2": 14,
"Dereivka_I_En1": 2.5,
"Yamnaya_RUS_Kalmykia": 24,
"Baden_LCA": 43.5,
"Kura-Araxes_ARM_Kaps": 16

The qpAdm application probably need more modification. I am not sure Natufian and Ganj_Dareh_N are good outgroups if we drop KAC into the mix.


BTW, if I model "steppe" as Yamnaya + Dereivka (might be a wrong assumption) then G25 nMontes gives very high steppe ancestries too.

Samuel Andrews said...

We don't need to use the mysterious CHG name anymore to describe post-Neolithic Near East admix in Europe. There are many new ancient genomes from Near East that we can use more recent proxies.

Kura-Araxes is good proxy for early waves in Eneolithic, Bronze age which impacted Greece and Sicily.

Then the next wave of Near East admix into Europe came in Roman imperial era, and it is best proxied by Bronze age Syria and Southern Turkey, who are significantly different from Kura Araxes. It mainly impacted Italy but small amounts probably also exist in Southeast Europe.

Gabriel said...

@Samuel Andrews

It is possible that La Tene Celts had less steppe than that, given some of the Central European Iron Age and Roman era samples that have popped up. Gauls could have simply been rich in earlier Beaker ancestry.

Samuel Andrews said...

@Gabriel,

The Iron age French samples from Brunel 2020 overlap with modern North French in G25 PCA. A Hallstatt sample from Czech republic, also clusters squarely with modern North French. So, La Tene Celts probably had the same Steppe ratios as modern French.

The France Bronze age Iron age samples are not in David's G25 PCA right now. I can post the coordinates, I forgot where I got them from, maybe David knows.

Samuel Andrews said...

" given some of the Central European Iron Age and Roman era samples that have popped up"

Swiss_IA_Wartau:SX18 is from Italy. She clusters perfectly with Iron age Latin samples.

Germany_Roman soldier, is from Spain or maybe Southern France, he not a local Gallic person in Roman army. He clusters with closest to modern Spain.

ambron said...

David, you wrote that the Eastern European HG is from CWC. And what about the Baltic HG, which determines the Balto-Slavic drift? The early CWC didn't have of it. He appears only in the later CWC (N47 and N49).

Davidski said...

@ambron

The earliest sample with obvious Balto-Slavic drift is Spiginas2. This thing just suddenly appears out of nowhere, because the earlier Baltic and Polish Corded Ware individuals don't really show it. They're more similar to Scandinavians in terms of fine scale drift.

None of the hunter-gatherer samples show it either, although some are better fits than others for modeling Balto-Slavic ancestry. LTU_Narva is probably the best, but I'm guessing that it's only a stand in for the real thing from further south somewhere.

So what does Spiginas2 represent exactly? A migrant or a new population moving into the Baltic region? If so, from where exactly?

vAsiSTha said...

I haven major qualms about the modeling done by davidski here.
One has to bear in mind that sources are distal in nature only, and fine tuning of each population cannot be done when running same models on large number of populations.

It is difficult to get good p values when moderns are modelled with ancient sources, so if you are getting 0.3, 0.40, 0.50 etc it is to be considered pretty good.

The addition of barcin N as source makes the job of selecting an older right pop harder. I wish that davidski had selected 2 more proximal sources such as globular amphora and some other with a diff mix of whg:anatolia_n as source instead of whg and anatolia_n.

So that whg, anatolia_n, chg etc could be put in right pops and natufian etc could be removed. In that case he could also do away with vestonice goyet etc so that all right pops would be of similar ages.

vAsiSTha said...

The line should start with
"I don't have major qualms..."

Rob said...

@ Davidski
Rzucewo culture ( syncretic features of CWC, GAC , late Narva) could be a candidate

Davidski said...

@Shaikorth

New model for OLS10.

https://drive.google.com/file/d/1-w7vTipek-JVENssF9os8ETzLZz-46Hj/view?usp=sharing

Considering the fit, it looks like something can be done better, but clearly the eastern ancestry is more like Baikal_BA than the Sintashta outliers.

When_in_Rome said...

Would these be good models for measuring Near Eastern DNA in Sicilians / South Italians?

Using:
HUN_Koros_N_HG (for WHG)
TUR_Barcin_N (for EEF)
Yamnaya_UKR (for WSH)
CMR_Shum_Laka_8000BP (for African)
MAR_Taforalt (for North African)
Levant_Natufian (for Southern Middle Eastern)
IRN_Ganj_Dareh_N (for Iranian-related)
Levant_PPNB (for Nothern Middle Eastern)
CZE_Vestonice16 (for UP-EUHG)
BEL_GoyetQ116-1 (for UP-EUHG)
Iberia_ElMiron (for UP-EUHG)
RUS_Karelia_HG (for EHG)
MNG_North_N (for East Asian)
RUS_Baikal_BA (for Siberian)
RUS_Tyumen_HG (for WSHG)
LAO_Hoabinhian (for AASI)
RUS_Devils_Gate_Cave_N (for East Asian)
JPN_Jomon (for Southeast Asian)
SYR_Ebla_EMBA (for Middle Eastern)

Target: Sicilian_West
Distance: 2.1526% / 0.02152592
35.8 TUR_Barcin_N
32.8 SYR_Ebla_EMBA
20.8 Yamnaya_UKR
7.2 HUN_Koros_N_HG
2.8 MAR_Taforalt
0.4 IRN_Ganj_Dareh_N
0.2 CMR_Shum_Laka_8000BP

Target: Sicilian_East
Distance: 2.0713% / 0.02071333
37.8 TUR_Barcin_N
34.8 SYR_Ebla_EMBA
22.0 Yamnaya_UKR
3.6 HUN_Koros_N_HG
1.2 MAR_Taforalt
0.2 CMR_Shum_Laka_8000BP
0.2 CZE_Vestonice16
0.2 LAO_Hoabinhian

If so, can we assume that the TUR_Barcin_N is the EEF DNA not shared in Middle Easterners and therefore received from the European side?

Shaikorth said...

@Davidski
The 2-way fit for OLS10 never worked with Sintashta_o3, it has too much direct EHG. Try o1, or Mereke_MBA which hasn't been tested yet but should be less like o3. The highest P-value fit used all Sintashta outliers together but that kind of population probably didn't exist.

epoch said...

@Slumbery

With regard to the affinity of ANE to Kostenki14: This seems to be a pretty good fit.

[1] "distance%=4.8547"

RUS_Sidelkino_HG

RUS_Kostenki14,77.2
RUS_AfontovaGora3,22.8

Maybe we miss something...

Anonymous said...

@epoch

This is a failure.

Target: RUS_Sidelkino_HG:Sidelkino
Distance: 4.4274% / 0.04427439
56.2 RUS_AfontovaGora3
43.8 RUS_Kostenki14

Target: RUS_Sidelkino_HG:Sidelkino
Distance: 4.0067% / 0.04006704
55.2 RUS_AfontovaGora3
44.8 RUS_Sunghir
0.0 RUS_Kostenki14

But this two way-way modeling is all bad, too, because it's far away.

Target: RUS_Karelia_HG:UzOO77
Distance: 4.2723% / 0.04272276
65.8 RUS_AfontovaGora3
34.2 RUS_Kostenki14

Target: RUS_Karelia_HG:UzOO77
Distance: 3.9704% / 0.03970436
65.4 RUS_AfontovaGora3
34.6 RUS_Sunghir
0.0 RUS_Kostenki14

Vladimir said...

@ Davidski

So what does Spiginas2 represent exactly? A migrant or a new population moving into the Baltic region? If so, from where exactly?

As I have already said, the main and probably the first population of the region from the Black sea to the Baltic sea was the population of haplogroup I2a. In the East it reached Moscow. Applied to the Baltic region, this is the Kunda culture. In Russia, this is the Valdai culture and, apparently, the Upper Volga culture. The I2a found in Volosovo is this particular population. All this post Svidersky population. Where the border of this population in the South is not yet clear. But if the results from the don show the presence of I2a in them, then this population reached the don. In the Narva culture, this population was already diluted by R1b-M73 coming from the South-East, bringing EHG, but further South, for example, in Belarus, this population could retain the original WHG in full, as it was in the Kunda culture.

Slumbery said...

@epoch

I don't know your exact test setup, but in G25 nMontes this is not really a "pretty good" fit.

"sample": "Sidelkino HG:Sidelkino",
"distance": 13.6783,
"RUS_AfontovaGora3": 67,
"RUS_Kostenki14": 33

Now, qpAdm is more reliable when we go after deep ancestries like this, but comparing sources in G25 nMontes with otherwise the same setup should be telling. And:


"sample": "Sidelkino HG:Sidelkino",
"distance": 7.2242,
"RUS_AfontovaGora3": 66,
"Grotta_Continenza_Meso": 29,
"RUS_Kostenki14": 5

Even more, we should ask the question if the remaining KO14 here represents itself or represents something that is missing?

"sample": "Sidelkino HG:Sidelkino",
"distance": 6.7625,
"RUS_AfontovaGora3": 63,
"Grotta_Continenza_Meso": 29.5,
"GEO_CHG": 7.5
"RUS_Kostenki14": 0

Well, CHG had KO14-related ancestry after all...

And then even Goyet is faring a tad better than KO14 (interestingly eliminating even CHG from the picture):

"sample": "Sidelkino HG:Sidelkino",
"distance": 12.9466,
"RUS_AfontovaGora3": 65,
"BEL_GoyetQ116-1": 35,
"RUS_Kostenki14": 0,
"GEO_CHG": 0

So we can model EHG as KO14 + AG3 sure, because KO14 is roughly in the direction of WHG from AG3, but that won't be a good model.

This disagree with my earlier statement that KO14 is closer to ANE than to WHG-s:

"sample": "Sidelkino HG:Sidelkino",
"distance": 21.4451,
"RUS_Kostenki14": 70,
"Grotta_Continenza_Meso": 30

But that question is a bit out of the reach of nMontes anyway, even by an indirect/interpolating route.

Note, the tests made on Genoplot, penalty off.

Anthony Hanken said...

@Davidski

How similar are Baikal_BA and the Sinashta outliars? Both have Q1a, with PU speakers presumably living somewhere between Sintashta and Baikal_BA.

The true source of eastern ancestry in OLS10 was probably from an N1c rich population. It seems to have replaced some of these Q1a lineages but may have existed on a cline with the outliers and Baikal_BA?

mzp1 said...

I'm not sure all the Yamnaya samples are uniform in terms of their EHG affinity. It looks like some of them may have more EHG than others.

1 (Afanasievo, Chimp) (Bulgaria_EBA_Yamnaya_o, Sidelkino) -0.0011 0.0102 -0.108 3261 3268 46360
3 (Ukraine_Ozero_Yamnaya, Chimp) (Bulgaria_EBA_Yamnaya_o, Sidelkino) 0.0254 0.0150 1.69 2424 2304 33540

With Chimp as an outgroup, Afanasievo is closer to Sidelkino than to Bulgaria_Yamnaya_o and Yamnaya Ozero is the opposite.


J.S. said...

@Samuel Andrews and David

Actually, Bronze and Iron French samples overlap all the modern French on Brunel's PCA .

Thus, can you explain the f3 statistics, f3(Mbuti, Ancient BA/IA French, Modern French)?

https://imgur.com/a/nLbyPN1

Simon_W said...

@Samuel Andrews

"The Iron age French samples from Brunel 2020 overlap with modern North French in G25 PCA.(...) The France Bronze age Iron age samples are not in David's G25 PCA right now. I can post the coordinates, I forgot where I got them from, maybe David knows."

Seriously? There are G25 coordinates for them?? Afaik only David and no one else has the data to calculate G25 coords, so if true, he must have inofficially posted them somewhere. Or maybe you confused them with the Eurogenes K15 data that is available on Vahaduo? In any case, please post the coordinates, if you have them.

Shaikorth said...

@Anthony Hanken
This should give an idea about the ancient clines and the relations between those samples.

https://i.imgur.com/h9kOVAf.jpg

Note the EHG shift in o3. 0LS10 worked with o1, Potapovka outlier and Srubnaya outlier (low quality) so Mereke_MBA might also work.

gamerz_J said...

@Davidski

Do all of the Balkans require this CHG-like ancestry? I thought only Greece and south Italy had it.

gamerz_J said...

@Samuel Andrews

"It mainly impacted Italy but small amounts probably also exist in Southeast Europe."

I am not aware of any Syrian-like admixture in Southeast Europe. Serbs, Croatians and Slovenians don't really need it. Nor do northern Greeks apparently thought Greek islanders might have it.

Samuel Andrews said...

@Simon_W,

I can email you the French Iron, brone age coordinates if you give me your email.

Simon_W said...

@Samuel Andrews

"Swiss_IA_Wartau:SX18 is from Italy. She clusters perfectly with Iron age Latin samples.

Germany_Roman soldier, is from Spain or maybe Southern France, he not a local Gallic person in Roman army. He clusters with closest to modern Spain."

The C14 date of SX18 doesn't gel well with an Italian or Roman/Latin origin. The area where she was buried was first conquered by the Romans in the year 15 BC. Her C14 date is 178 BC - 2 AD. So a Roman origin would necessitate an arrival and death of her during the first 17 years of Roman rule. What would a woman do there, virtually following in the footsteps of the Roman legionaries, and far from a Roman colony? Moreover the predominant Roman burial rite at that time, during the early imperial age, was cremation. So an inhumation, even more so in a natural cave, would be unusual for a Roman. As for the distances in G25, among all ancient pops she is closest to DEU_Roman; Bell Beaker ITA comes next, and then the Prenestini, who are the only Southern French-like ones among the Latin samples. Among the modern samples she is closest to Spanish_Castello, which is northern Catalan, right behind the Spanish/French border. But indeed, the Latins are also mostly Catalan-like, with the Romans more resembling La Rioja and the Prenestini the southern French. Conversely the DEU_Roman is closest to CHE_IA, among all ancient pops, and also Catalan-like among the modern pops. So a Spanish or South French origin isn't proven, as the Latins and CHE_IA are also Catalan-like.

Tomenable from Anthrogenica has made this map showing the affinity of CHE_IA SX18 to modern pops: https://justpaste.it/51k1p

According to this, SX18 has the closest affinity with modern Southern Swiss, followed by south-central French. It would be interesting to know what exactly he measured and compared there, but he didn't yet have the time to reply. Obviously it's not based on G25 data.

Rob said...

@ Vlad
Kunda is an eastern tradition, coming from Butovo and other groups to the east. It’s important bearers would’ve been R1b-P297 and Q1. The I2a present in the east Baltic are from post-Ahrensburgian, swiderians arriving concurrently from west
And these probably account for the ~ 30% WHG found in Russian HGs

gL said...

@Huck Finn said...

@Davidski said...
"most Uralic speakers don't show any significant EHG or Western Siberian HG (WSHG), but they do show significant eastern Siberian ancestry."

I am slightly confused on what is considered Eastern Siberian in this context.

modern Siberia can be divided in 2 parts - Western Siberia:
https://upload.wikimedia.org/wikipedia/commons/a/ab/%D0%97%D0%B0%D0%BF%D0%B0%D0%B4%D0%BD%D0%B0%D1%8F_%D0%A1%D0%B8%D0%B1%D0%B8%D1%80%D1%8C.jpg
and Eastern Siberia.

Anyway, I'm hoping, that @Davidsky will expand on this more.





Though, even with Eastern and Western Siberia division it is really hard to see how Eastern Siberian ancestry is showing up for most modern Uralic ancestry, unless in this context Eastern Siberian = North Eastern Asian ancestry, that was located in Amur river basin and Liao region.

Uralic speaker history is a bit more complicated as they arrived in Northern Asia in many waves, which split afterwards. Most modern people with y-dna N in Europe, including Poland probably missed Eastern Siberia entirely and went through Western Siberia(along Irtysh river) and then to Volga basin, before expanding in Europe.

https://www.researchgate.net/profile/Monika_Karmin/publication/305036948/figure/fig3/AS:669301560328217@1536585335957/Frequency-Distribution-Maps-of-Individual-Sub-clades-of-hg-N3-Data-points-from-Table-S2.jpg

In this map(N3 = N-TAT) the only Eastern Siberian populations are N3a2, N3a5 and N3a6 and they all are later expansion branches of the same N-Tat branches that are mostly found in Europe/Russia. N3a1 does not show any expansion happening to/from Eastern Siberia - Baikal area - quite opposite N3a2(Yakut/Yukagir) overlaps with all other N3a on this map in Southern Ural area - Southern Irtysh river basin and Altai region. Volga basin probably is not the source, but very early expansion from Irtysh river. The fact is that Uralics became Uralics only on Volga river basin, so what languages they spoke before we can't call Uralic, as clearly Yukagir language is so different, that it is not considered Uralic.

https://www.cell.com/ajhg/pdfExtended/S0002-9297(16)30160-4
It is known, that Samoyed people are descendants from Sayan mountain area and they indeed were coming from Baikal area - probably pushed out of there by N-Tat spread to Baikal area and then later further to north and east. N-P63(N2a1) spread to the north in much later times and Nenets people reached Arctic Ocean only 1000 years ago. All the Southern Samoyed languages in Sayan mountains have died out about 200 years ago, but N-P43 is still present there.

There was some back and forth movement happening for Hungarians, but they were not Eastern Siberians, but Western Siberians with North Eastern Asian ancestry.

Same is for Polish y-dna N, as it is not from Eastern Siberia. Eastern Siberian Uralic branched much later than Uralic was present in Volga basin.

In short - origins:

M46 - Korea/NE China
which spread as two branches:
M178 - Southern Siberia(Irtysh basin)
L666 - Baikal Shamanka and also to Southern China
Both of M178 and L666 also spread in China, Japan and Korea and probably more to other places.
The truly northern branch of M178 is F1419, that interest us - European-Siberians, that might have originated in southern part of Western Siberia.
The same for L666, where P43 originated in Baikal area.

Then there exists Y6503, which spread from Southern Siberia(did it originated from there?) among Turkish and Bosnian people. Y6503 was found in ancient samples of Baikal area, but most probably it was completely replaced/pushed aside by later migrations of previously mentioned.

So, there are at least 3 different sources of European N y-dna, where only P43, that reached Finland originated in Eastern Siberia - rest of them(and for majority of European Uralic) originated in plains of southern Siberia and probably their common point of origin goes so far back in history, that it was located in southern China/SE Asia.

EastPole said...

@Davidski

“So what does Spiginas2 represent exactly? A migrant or a new population moving into the Baltic region? If so, from where exactly?”

I have one of the highest amounts of Spiginas2 among all Poles:

Target: EastPole
Distance: 4.4507% / 0.04450683
51.2 Corded_Ware_Baltic:Spiginas2
20.8 POL_TRB:N19
13.6 Corded_Ware_DEU:I1538
13.2 Corded_Ware_Baltic_early:Gyvakarai1_10bp
1.2 Corded_Ware_Baltic:Kunila2

It is characteristic of Eastern Poles. It is also high in Balts, Estonians, Belarusians, Smolensk Russians. Lower in other Poles, Ukrainians, and Finns.

It probably is the result of mixing Corded Ware groups migrating from Poland east and mixing with local substratum:

Target: Corded_Ware_Baltic:Spiginas2
Distance: 3.9695% / 0.03969475
51.2 Corded_Ware_POL:N49
16.0 UKR_Dereivka_I_En2:I4110
14.6 UKR_Meso:I5885
11.2 Baltic_LTU_Narva:Donkalnis7
4.4 RUS_Karelia_HG:I0211
1.0 HUN_ALPc_Tiszadob_MN:I2375
0.8 HUN_Sopot_LN:I4184
0.6 BGR_EBA:I2176
0.2 TUR_Ikiztepe_LC:IKI037


Target: Corded_Ware_Baltic:Spiginas2
Distance: 4.0063% / 0.04006274
51.2 Corded_Ware_POL:N49
15.6 UKR_Meso:I5885
14.2 Baltic_LTU_Narva:Donkalnis7
5.4 RUS_Karelia_HG:I0211
5.0 Corded_Ware_DEU:I1534
4.4 BGR_EBA:I2176
1.4 Corded_Ware_DEU:I1539
1.4 HUN_ALPc_Tiszadob_MN:I2375
1.0 HUN_Sopot_LN:I4184
0.4 TUR_Ikiztepe_LC:IKI037


Interestinf Spiginas2 is also high in some Welzin samples:

Target: DEU_Welzin_BA_outlier2:WEZ56
Distance: 5.3890% / 0.05388955
82.6 Corded_Ware_Baltic:Spiginas2
8.0 POL_TRB:N19
5.8 Corded_Ware_DEU:I1539
3.6 Corded_Ware_CHE:MX190

Target: DEU_Welzin_BA:WEZ59
Distance: 4.4717% / 0.04471654
50.0 Corded_Ware_Baltic:Spiginas2
23.6 POL_TRB:N20
9.4 Corded_Ware_CHE:MX190
8.8 Corded_Ware_CHE:MX188
5.0 POL_Globular_Amphora:RISE1163
3.2 Corded_Ware_DEU:I1544

Target: DEU_Welzin_BA:WEZ15
Distance: 3.3114% / 0.03311357
45.6 Corded_Ware_Baltic:Spiginas2
15.8 Corded_Ware_POL:N47
12.8 POL_TRB:N19
12.0 POL_Globular_Amphora:I2403
8.2 Corded_Ware_DEU:I1538
2.8 Corded_Ware_CHE:MX190
2.8 POL_Globular_Amphora:RISE1246

Davidski said...

@Shaikorth

https://drive.google.com/file/d/1KCLRpfoVlYNRMzzVqbnVFPR3sNkx2mxw/view?usp=sharing

https://drive.google.com/file/d/1xUMOjT8uFZwXfpXhLCEOLKrW1ZRQZhWa/view?usp=sharing

Anonymous said...

@gL

"The fact is that Uralics became Uralics only on Volga river basin, so what languages they spoke before we can't call Uralic, as clearly Yukagir language is so different, that it is not considered Uralic."

What a nonsense. Name Uralic is an outdated traditional name, they certainly began to speak Uralic, or more precisely Finno-Ugric, not on any Volga river. It is firmly established that Finno-Ugric languages come from Taiga Siberia, see Samoyedic languages, it is a part of Uralic languages. The Yukaghir language is the closest relative to Uralic languages, it forms a clade with it, approximately as Hittite with all other Indo-European languages.

Tigran said...

@Davidski

What tools do you use for your analysis? Is there a good way for one to learn to start understanding these stats? Thanks.

Davidski said...

@gL

The fact is that Uralics became Uralics only on Volga river basin, so what languages they spoke before we can't call Uralic, as clearly Yukagir language is so different, that it is not considered Uralic.

This is not a fact, it's just your speculation.

And I'm not sure why you keep mentioning Poland in the context of Y-hg N, because it's not a very common marker in Poland and most of it comes from contacts with Balts.

Davidski said...

@Tigran

What tools do you use for your analysis? Is there a good way for one to learn to start understanding these stats?

Mainly ADMIXTOOLS.

There are papers and websites freely available online about it.

Shaikorth said...

@Davidski
So fit with Mereke is the best so far by a small margin, how's Kumsay_EBA, Potapovka_o or Chemurchek_EBA?

mzp1 said...

I was able to get a 'working' qpadm run with EHG, WHG, Yamnaya and Anatolian_N for euro populations. I'm not sure if it's correct but it looks ok to me. Don't mind suggestions for improvement..

target Anatolia_N WHG Yamnaya EHG
4 French 0.348 0.083 0.491 0.078
5 Engli… 0.232 0.084 0.609 0.074
6 Czech 0.227 0.098 0.564 0.111
7 Hunga… 0.281 0.062 0.569 0.087
8 Ukrai… 0.296 0.067 0.544 0.094
9 Polis… 0.268 0.062 0.513 0.157
10 Eston… 0.145 0.111 0.623 0.12
11 Lithu… 0.237 0.123 0.516 0.124
12 Russi… 0.191 0.086 0.599 0.123

Source pops EHG, WHG and Yamnaya are combinations of related samples. EHG is Karelia. Samara and Sidelkino, WHG is Loschbour and Italy_N_oWHG, Yamnaya is Ozero and Bulgaria_EBA_Yamnaya_o (those with least EHG).

Not exactly sure where the p value is though...

outgroups=c( "Russia_Ust_Ishim.DG", "Russia_MA1_HG.SG", "Russia_AfontovaGora3","Iran_GanjDareh_N", "Georgia_Kotias.SG", "BIR.SG", "Anatolia_Epipaleolithic", "Russia_Steppe_Eneolithic", "Italy_N_oWHG.SG", "Italy_North_Villabruna_HG")

mzp1 said...

Southern Europe didnt run with EHG, so replacing EHG with HajiFiruz

target WHG Anatolia_N Iran_C_HajjiFir… Yamnaya

1 Turki… -0.001 -0.004 0.713 0.292
2 Greek 0.052 0.294 0.292 0.362
3 Ital S 0.067 0.239 0.406 0.289

Tested the above for French

1 French 0.117 0.24 0.076 0.566

Only showing 7% HajiFiruz for French so looks ok. Still quite high for Italy and Greek so maybe I'll try with different Iranian sources.

mzp1 said...

Only using Yamnaya Ozera for the Yamnaya group (removing Bulgaria_EBA_o) seems to give more realistic results

target Anatolia_N WHG Yamnaya EHG
1 Engli… 0.289 0.135 0.443 0.132
2 French 0.381 0.12 0.376 0.123
3 Polis… 0.316 0.08 0.404 0.201
4 Ukrai… 0.337 0.104 0.411 0.149
5 Lithu… 0.268 0.161 0.404 0.168
6 Russi… 0.24 0.135 0.457 0.168

Davidski said...

@mpz1

You're hopeless.

Yamnaya Ozera is an outlier with significant Maykop ancestry and irrelevant to European population prehistory.

Samuel Andrews said...

@Davidski, You don't have an EEF or WHG in right pops. That seems problematic.

Davidski said...

@Samuel

How am I supposed to put EEF and WHG in the right pops when I have Barcin_N and WHG in the left pops?

Putting the same or derived populations in the right pops screws up models. They have to be ancestral.

Davidski said...

@Shaikorth

Note that only Chemurchek_EBA produces a decent model without Baikal_BA helping out.

https://drive.google.com/file/d/1lDiV1n0hd9KcDAP9NMxfoLNrSKluafsT/view?usp=sharing

https://drive.google.com/file/d/1MhFQdUQDEUrQ5UyzA5n71D048hDua1V9/view?usp=sharing

https://drive.google.com/file/d/1pErWQ_W9wF8BzJTTYpc-DulguQtdAmVE/view?usp=sharing

Slumbery said...

@gamerzJ
"Do all of the Balkans require this CHG-like ancestry? I thought only Greece and south Italy had it."

In the above G25 nMonte setup all of them take KAC ancestry in the model. The order of amount is Greeks (with strong regional differences) > Albanians > Bulgarians > Macedonians > Romanians (including Moldovans) and Serbians > Bosnians and Montenegrin > Croatians.

It is pretty much down to near noise level in the Croatians and further populations (Slovenians, Hungarians, Ukrainians) do not take it.

Davidski said...

@All

I'm going to try modeling with different types of EEF pops and put Barcin_N in the right pops, because nothing is working for at least half of Europe.

If I can't get that to work for most European nations, then I reckon I'll have to leave things for now and try again when more relevant samples become available.

Slumbery said...

@Davidski

I think you cannot find a significantly better EEF reference than Barcin_N. In fact I found that some of them are worse than Barcin_N. The only way to improve things from that side is to give up on the separation of EEF and WHG and move to later farmer populations that are already admixed.

Above I used Baden. It was not a random choice, for Central Europe and the Balkan it gave just as good results as the old classic GAC (by and large) and has the advantage of being older than the steppe migrations.

Slumbery said...

@Davidski

In addition the pre-steppe Europe was not monolithic. You won't find a model that works for all Europeans. I can see that by putting WHG and ANF against the much later Yamnaya (that had complex recent ancestry) you tried something universal. But the cost is that they are not actually the counter-part of Yamnaya. And if you move to later populations, you have to use multiple of them. Setting aside the second wave (CHG rich) Anatolian ancestry in Italy-Balkan, Italy and and Iberia surely has strong local farmer ancestry that is not brought there from Central or Eastern Europe.

Shaikorth said...

@Davidski
The single decent Potapovka fit also has clear majority coming from Potapovka_o, so looks like this direction is improving the fits for whatever reason. If you have time, might also try redoing these fits with WHG and EEF in rightpops since there's no issue with putting them there in this particular case.

ambron said...

David, are you 100% sure that the N47 and N49 did not have the Baltic HG but had something similar to the SHG? I find it hard to believe that Arza missed this difference. In his models for the N47 and N49, the Baltic BA gives about 20%.

Queequeg said...

I personally don't have any difficulties in accepting Baikal_BA biased Mereke_BA type of groups being behind OLS_10, in terms of his non-Baltic features. Actually, taking into account fex Seyma Turbino phenomenom is might even be expected. OLS_10 himself in Estonia represents a herding community and even if Mereke is located by one of the tributaries of Ural river, groups residing in and around Mereke, i.e. in Western Kazakhstan, may very well have contacts with much more eastern herding communities. That being said, the paternal N lineage of OLS_10 apparently still represents the Mereke or Greater Ural type of features of OLS_10.

EastPole said...

Corded_Ware_POL:N49 and Corded_Ware_POL:N47 can be modeled as steppe, HG, EEF:

Target: Corded_Ware_POL:N49
Distance: 1.9056% / 0.01905585
29.8 Yamnaya_RUS_Kalmykia
18.2 UKR_Globular_Amphora
16.2 UKR_Meso
11.8 RUS_Poltavka
10.4 POL_Globular_Amphora
9.8 Baltic_LTU_Narva
3.8 UKR_Trypillia

Target: Corded_Ware_POL:N47
Distance: 1.8288% / 0.01828767
28.0 POL_Globular_Amphora
15.8 RUS_Poltavka
15.6 UKR_Meso
14.0 UKR_Trypillia
13.6 Yamnaya_RUS_Kalmykia
13.0 Baltic_LTU_Narva

Corded_Ware_Baltic:Spiginas2 can be modeled as steppe, HG and EEF too.

Target: Corded_Ware_Baltic:Spiginas2
Distance: 4.6748% / 0.04674802
35.0 UKR_Meso
25.6 RUS_Poltavka
13.8 POL_Globular_Amphora
10.8 Baltic_LTU_Narva
8.8 UKR_Globular_Amphora
4.4 Yamnaya_RUS_Kalmykia
1.6 UKR_Trypillia

But the model with Corded_Ware as the source is better:

Target: Corded_Ware_Baltic:Spiginas2
Distance: 4.0760% / 0.04076025
57.6 Corded_Ware_POL:N49
17.8 UKR_Meso:I5885
12.6 Baltic_LTU_Narva:Donkalnis7
6.6 Corded_Ware_DEU:I1534
3.2 Corded_Ware_DEU:I1539
2.2 POL_Globular_Amphora:RISE1252

The origin of various CWC groups may be complex and we can’t exclude that Ukrainian HG language was the source of PIE.

Samuel Andrews said...

@Davidski,
"Putting the same or derived populations in the right pops screws up models. They have to be ancestral."

I did not know this about qpADM. It explains why right pops in studies always use distantly related pops.

@Slumbery,

There are a handful of Early Neolithic Europe samples who have no WHG. They do best for modelling Neolithic Europe in G25. In fact, Neolithic Europeans score higher WHG when those EEFs are used than when Barcin is used, so they are best reference we have for Anatolian ancestor of Neolithic Europeans.

But I very much doubt they are high enough coverage to use in qpADM.

Samuel Andrews said...

@Simon_W and others,

Here's G25 coordinates of France Bronze, Iron age. You can find the location and age of the samples in Brunel 2020.

France BA, IA
https://script.google.com/d/139aBUIXVmxExUPEzSbrm1-BgxX_caqmuA7Ol_Vvfy6xRFSt9ZMhadY1M/edit?usp=drive_web&folder=0AO4--OcB_rMfUk9PVA&splash=yes

There's quite a bit of variation in the samples btw. It is pointless modelling them with nMmonte. They are low coverage so they never get good fits. But their position in West Eurasia PCA is informative.

When I put them on West Eurasia PCA, I see that French BA forms a cline north of modern Western Europe meaning they had more WHG. While, France IA clusters in modern Western Europe and most samples fit in Northern France (Alsace, Nord).

But several France IA are more northern than modern North France, one sample is probably a German.

Hallstatt sample DA11 perfectly fits in Northern France. I think we are seeing a pattern.

old europe said...

@ eastpole

We can't exclude???

Debunking the Ukraine HG ( WHG+EHG) origin of PIE is now quite impossible IMHO

Linguistic archeology and now genetics point in this direction.

Archaic PIE: Ukraine Meso and Neo
Full PIE: Ukraine Meso and Neo + Balkan Carpathian EEF ( predominantly GAC)

Davidski said...

@Shaikorth

https://drive.google.com/file/d/1quwI3KT4t0xRykV1wP5KAvZTUc1YOFVk/view?usp=sharing

Shaikorth said...

@Davidski
Comparing all these fits it looks like El_Miron might be causing problems in right pops, commonly generating Z scores of >2. If there's a WHG reference in Pright it shouldn't be needed in any case, and you could try switching it with Barcin_N.

Vladimir said...

@old europe said...
Archaic PIE: Ukraine Meso and Neo
Full PIE: Ukraine Meso and Neo + Balkan Carpathian EEF ( predominantly GAC)

CHG?

Anonymous said...

old europe said...
" We can't exclude???
Debunking the Ukraine HG ( WHG+EHG) origin of PIE is now quite impossible IMHO
Linguistic archeology and now genetics point in this direction.
Archaic PIE: Ukraine Meso and Neo
Full PIE: Ukraine Meso and Neo + Balkan Carpathian EEF ( predominantly GAC)"

Everything you wrote is nonsense and completely untrue. I've debunked that mistake, I've given you this data, all your claims, I'm a categorical error. You're not modeling anything at all, you're just making it up. There's no Meso and Neo in Yamnaya or CWC. Linguistics categorically disagrees with you, so there is nothing to make up and write a lie.



gamerz_J said...

@Davidski

Greek farmers are more CHG-shifted than Barcin_N IIRC. It may be that southern Europeans have more ancestry from these farmers than Barcin_N. I don't know why or how Barcin_N avoided this admixture but maybe that's an issue. There could be 2 farmer groups?

@Slumberry

Global25 does not give any CHG really to Serbs that I know, will check again though. Greeks have it but it drops to really low in Greek_Macedonia and Bulgarians.

"And if you move to later populations, you have to use multiple of them. Setting aside the second wave (CHG rich) Anatolian ancestry in Italy-Balkan, Italy and and Iberia surely has strong local farmer ancestry that is not brought there from Central or Eastern Europe"

Local farmer ancestry as in differentiated by drift?

Slumbery said...

@Samuel Andrews

There are a handful of Early Neolithic Europe samples who have no WHG. They do best for modelling Neolithic Europe in G25. In fact, Neolithic Europeans score higher WHG when those EEFs are used than when Barcin is used, so they are best reference we have for Anatolian ancestor of Neolithic Europeans.

I know there are EEF populations, but I am pretty sure none of them will solve Davidski's problem. Mediterranean and Mediterranean impacted populations cannot be modeled with just EEF + WHG + Yamnaya, because they have non-Yamnaya CHG ancestry, Natufian ancestry, Iran_N ancestry and whatnot. (Not necessarily all of them in the same population of course.)
This problem cannot be solved by looking for alternatives for Barcin_N.

If you look at the list in the article you can see that exactly Mediterranean and Mediterranean impacted populations that are missing, because EEF + WHG + Yamanaya fails for them. It is not the fault of using Barcin as EEF, it would be the same with all EEF.

That why I said (maybe not in this form) that he can address this either by
- accounting for the later Middle Eastern migrations (in case of the Balkan with KAC)
- or usage of regional late-neolithic / Chalcolithic populations and give up on the unified test.

Davidski said...

@Shaikorth

The WSHG-rich pops like Kumsay and Mereke BA aren't relevant to the westward spread of N-L1026 and Uralians.

You have to wait for samples from Seima Turbino and the groups that were moving west across the forest zone during the Iron Age.

vAsiSTha said...

@davidski

"
I'm going to try modeling with different types of EEF pops and put Barcin_N in the right pops, because nothing is working for at least half of Europe."

Do the same for whg also by putting it in right pops and an admixed whg pop in left.

CrM said...

@Davidski

Your qpAdm runs are more akin to G25 Unscaled runs. I've noticed Unscaled gives more Steppe ancestry compared to Scaled.
But the difference between Scaled and Unscaled when you model Dagestanis is very drastic, latter gives like 5-8% more Steppe. Which one is more accurate in your opinion? And if possible, can you do a run for Avars or Dargins on qpAdm?

https://i.imgur.com/kPbTAKw.png
https://i.imgur.com/aqezaQI.png

Anonymous said...

@Vladimir
"@old europe
CHG?"

In fact, what he writes is exceptionally untrue, the main thing is that it is instantly refuted. It is not even clear how such a lie can be made up. You can see that he doesn't think anything at all, he just imagines it.

Target Distance RUS_Karelia_HG GEO_CHG RUS_Sidelkino_HG UKR_Meso RUS_Samara_HG UKR_N
Yamnaya_RUS_Samara:I0429 0.04343664 65.8 34.2 0.0 0.0 0.0 0.0
Yamnaya_RUS_Samara:I0439 0.04669944 63.4 36.6 0.0 0.0 0.0 0.0
Yamnaya_RUS_Samara:I0357 0.03764562 61.0 35.6 1.8 1.6 0.0 0.0
Yamnaya_RUS_Samara:I0438 0.04064214 60.6 39.4 0.0 0.0 0.0 0.0
Yamnaya_RUS_Samara:I0444 0.04307701 56.8 29.8 13.4 0.0 0.0 0.0
Yamnaya_RUS_Samara:I7489 0.04858407 54.6 39.8 3.8 0.0 1.8 0.0
Yamnaya_RUS_Samara:I0231 0.03767896 42.2 34.2 23.6 0.0 0.0 0.0
Corded_Ware_POL_early:poz81 0.04641657 35.2 32.8 27.6 4.4 0.0 0.0
Yamnaya_RUS_Samara:I0370 0.04489146 33.4 38.0 28.6 0.0 0.0 0.0
Yamnaya_RUS_Samara:I0443 0.04491385 29.8 37.4 32.8 0.0 0.0 0.0
Corded_Ware_Baltic_early:Plinkaigalis242 0.04900116 28.2 30.0 39.4 2.4 0.0 0.0
Corded_Ware_Baltic_early:I4629 0.05521564 21.2 31.0 47.8 0.0 0.0 0.0
Corded_Ware_Baltic_early:Gyvakarai1_10bp 0.04781119 18.8 31.6 39.0 10.6 0.0 0.0
Average 0.04507798 43.9 34.6 19.8 1.5 0.1 0.0

CHG can be divided into basal part, then its part goes to EEF.

Target Distance RUS_Karelia_HG GEO_CHG RUS_Sidelkino_HG POL_TRB POL_Globular_Amphora RUS_Samara_HG UKR_Meso UKR_N
Yamnaya_RUS_Samara:I0429 0.04339271 65.2 33.4 0.0 1.4 0.0 0.0 0.0 0.0
Yamnaya_RUS_Samara:I0444 0.04237409 64.8 27.6 1.4 2.2 4.0 0.0 0.0 0.0
Yamnaya_RUS_Samara:I0438 0.04064214 60.6 39.4 0.0 0.0 0.0 0.0 0.0 0.0
Yamnaya_RUS_Samara:I0357 0.03471129 57.6 31.2 0.0 0.0 11.2 0.0 0.0 0.0
Yamnaya_RUS_Samara:I0439 0.04462325 57.4 32.2 0.0 0.0 10.4 0.0 0.0 0.0
Yamnaya_RUS_Samara:I7489 0.04822516 56.2 38.2 0.0 0.0 4.6 1.0 0.0 0.0
Corded_Ware_Baltic_early:Plinkaigalis242 0.04495869 51.0 25.2 7.8 0.6 15.4 0.0 0.0 0.0
Corded_Ware_POL_early:poz81 0.04465082 50.4 29.8 9.2 0.0 10.6 0.0 0.0 0.0
Yamnaya_RUS_Samara:I0231 0.03645868 49.0 31.2 12.6 7.2 0.0 0.0 0.0 0.0
Yamnaya_RUS_Samara:I0370 0.04447466 40.6 36.4 18.0 0.0 5.0 0.0 0.0 0.0
Corded_Ware_Baltic_early:Gyvakarai1_10bp 0.04182719 37.0 24.2 20.4 13.6 4.8 0.0 0.0 0.0
Yamnaya_RUS_Samara:I0443 0.04470670 35.6 36.4 24.4 0.0 3.6 0.0 0.0 0.0
Corded_Ware_Baltic_early:I4629 0.05398186 35.6 28.2 26.6 0.0 9.6 0.0 0.0 0.0
Average 0.04346363 50.8 31.8 9.3 1.9 6.1 0.1 0.0 0.0

UKR_Meso = 0, UKR_N = 0 strictly. However, it is a part of EHG shifted towards WHG, so it can appear in minuscule amounts of late samples having WHG as a proxy for WHG.

Slumbery said...

@gamerz_J


"sample": "Serbian:Average",
"distance": 3.3227,
"Baden_LCA": 43.5,
"Yamnaya_RUS_Kalmykia": 27.5,
"Dereivka_I_En2": 20.5,
"Kura-Araxes_ARM_Kaps": 8.5

This should translate to about 3-6% excess CHG, depending on the other references (or even zero, if something eats it up to balance out other references). Plus there is also Levant and Iranian ancestry. Not very high, but plenty enough to screw up the model if you won't account for it.

I just tried it and when I use CHG + Natufian + Barcin_N instead of KAC, the extra CHG being absorbed into Dereivka_En2 and when I kick out Dereivka, its gets eaten by Yamnaya (resulting in rather high Yamnaya ancestry).

Local farmer ancestry as in differentiated by drift?

That too, but what more important is differential admixtures that are not EEF or WHG.

old europe said...


@archie

You contradict yourself by saying that Yamnaya lacks Ukraine HG, it has but indirectly, through Sredni Stog that is ancestral to Yamnaya but then in other posts you root for the theory that R1b is not IE in origin. Yamnaya is mostly R1b.
So where you stand exactly on the PIE question?

Shaikorth said...

@Davidski
I'm not suggesting anything from Bronze Age Kazakhstan was ancestral to Estonia_IA, but looking for proper proxies. Seima-Turbino is at least chronologically in the right place, the spread of L1026* happened during the Bronze Age but we knew this already from Bolshoy,which culturally doesn't seem to be Seima-Turbino-related though.

IA forest zone groups from Siberia and the Urals are too recent to be relevant to Estonian/Baltic_IA transition or the initial spread of Uralic branches of L1026 (CTS10760 and Z1936), though all sorts of Volga groups certainly have significant direct ancestry (verifiable with IBD and rarecoal) from them.

gamerz_J said...

@Slumberry

Would it be still the same if you use Peloponnese_N and Yamnaya_Samara or Yamnaya-Ukraine to model Serbs? I am not close to my PC atm, so I can't try it myself.


That's interesting, it seems all of southern Europe has later "non-European" admixture, tbh I did not expect this.

"That too, but what more important is differential admixtures that are not EEF or WHG."

Which admixtures in farmers? The CHG stuff?

Anonymous said...

@old europe
"You contradict yourself by saying that Yamnaya lacks Ukraine HG, it has but indirectly, through Sredni Stog that is ancestral to Yamnaya but then in other posts you root for the theory that R1b is not IE in origin."

You're just talking amateur nonsense. Who told you that Sredniy Stog has anything to do with the Ukrainian Mesolithic and Neolithic? You could only imagine such nonsense yourself. Sredniy Stog has nothing to do with the Ukrainian Mesolithic or Neolithic. Strictly speaking, Sredniy Stog is not the ancestor of Yamnaya culture, but its predecessor a thousand years before it. Whether the Yamnaya culture originates from the Sredniy Stog culture is unknown, but nothing that I wrote does not contradict it, but contradicts your statements.

Everything you write is unscientific nonsense, contrary to all linguistics, archaeology and genetics in general. And you're writing this untruth purely in a categorical manner, because you have no proof, you know you're just cheating.

old europe said...
This comment has been removed by the author.
epoch said...

@Slumbery

I must have done something wrong. When setting up again I get the same as you.

[1] "distance%=13.6783"

RUS_Sidelkino_HG

RUS_AfontovaGora3,67

Strange..
RUS_Kostenki14,33

Anonymous said...

old europe said...
"ancestor and predecessor are the same thing, they have the same meaning."

It's not the same, your ancestor is your ancestor, your daddy is your ancestor, you had a predecessor at work, but he is not your ancestor.

old europe said...

@archie

Where is the PIE homeland according to you?
Which was the genome wide ancestry of the populations involved in the PIE ethnogenesis?

Slumbery said...

@gamerz_J

Please be more specific if you would like me to run an nMonte for you. But I try. I included modern Peloponesse for comparison.


"sample": "Greek Peloponnese:Average",
"distance": 2.2705,
"Peloponnese_N": 58,
"Yamnaya_UKR": 23.5,
"Kura-Araxes_ARM_Kaps": 14,
"Koros_N_HG": 4.5

"sample": "Serbian:Average",
"distance": 3.9598,
"Peloponnese_N": 50,
"Yamnaya_UKR": 39,
"Kura-Araxes_ARM_Kaps": 0,
"Koros_N_HG": 11

As you can see this set-up decreases the ancestry drawn from KAC significantly. But the fit is worse.

Also:

"sample": "Peloponnese N:Average",
"distance": 1.5511,
"Barcin_N": 94,
"GEO_CHG": 6

So the reason Peloponnese Neolithic kicks out KAC in Serbians is exactly because the (small) extra CHG can be supplied from it.

However it is not kicked out from Southern Greeks, because they have Levant ancestry too. (And then I gave you an example for the second question.)

"sample": "Greek Peloponnese:Average",
"distance": 1.8931,
"Peloponnese_N": 46.5,
"Yamnaya_UKR": 22,
"Levant_Sidon_MBA": 15,
"Dereivka_I_En2": 7,
"Kura-Araxes_ARM_Kaps": 5,
"Baden_LCA": 3,
"Koros_N_HG": 1.5

"sample": "Serbian:Average",
"distance": 3.2048,
"Peloponnese_N": 38.5,
"Yamnaya_UKR": 19,
"Levant_Sidon_MBA": 3.5,
"Dereivka_I_En2": 39,
"Kura-Araxes_ARM_Kaps": 0,
"Baden_LCA": 0,
"Koros_N_HG": 0

Rob said...
This comment has been removed by the author.
Anonymous said...

@old europe
"Which was the genome wide ancestry of the populations involved in the PIE ethnogenesis?"

There were two types of EHGs: Northern EHG represented by Karelia, Onega and Veretye, and Southern EHG represented by Sidelkino.
The Ukrainian Mesolithic origin seems to be entirely from the Southern EHG, but it disappears completely in the Ukrainian Neolithic and does not continue anywhere or participate in anything. Yamnaya Samara originates apparently entirely from the Northern EHG, maybe with little involvement from the Southern EHG. But the PIEs are located between the North EHG and the South EHG, very approximately 2/3 of the North EHG + 1/3 of the South EHG, because of the great kinship of the North and South EHG proportions can not be accurately specified.
All this has been added to 30%-40% CHG, it has already happened strictly in the Volga-Don region.
And EEF (=ANF+WHG) was added further in Central Europe.

mzp1 said...

@Davidksi,

If we look at the f3s below comparing shared drift of different Yamna samples with reference populations

5 Steppe_Maikop_o Anatolia_Epipaleo… Chimp.… 9.55e14
6 Steppe_Maikop_o Iran_GanjDareh_N Chimp.… 1.11e15
7 Steppe_Maikop_o Russia_Sidelkino_… Chimp.… 9.60e14
8 Steppe_Maikop_o Luxembourg_Loschb… Chimp.… 1.08e15

17 Ukraine_Ozera_EBA_Y… Anatolia_Epipaleo… Chimp.… 1.11e15
18 Ukraine_Ozera_EBA_Y… Iran_GanjDareh_N Chimp.… 1.31e15
19 Ukraine_Ozera_EBA_Y… Russia_Sidelkino_… Chimp.… 1.13e15
20 Ukraine_Ozera_EBA_Y… Luxembourg_Loschb… Chimp.… 1.28e15

9 Yamnaya_Samara Anatolia_Epipaleo… Chimp.… 1.83e15
10 Yamnaya_Samara Iran_GanjDareh_N Chimp.… 2.24e15
11 Yamnaya_Samara Russia_Sidelkino_… Chimp.… 2.04e15
12 Yamnaya_Samara Luxembourg_Loschb… Chimp.… 2.28e15

13 Russia_Afanasievo Anatolia_Epipaleo… Chimp.… 1.84e15
14 Russia_Afanasievo Iran_GanjDareh_N Chimp.… 2.26e15
15 Russia_Afanasievo Russia_Sidelkino_… Chimp.… 2.05e15
16 Russia_Afanasievo Luxembourg_Loschb… Chimp.… 2.29e15

Maykop_o and Ozera are furthest away from all Anatolian_HG, Ganj_Dareh, Sidelkino, and Loschbour. It means those pops were the most isolated ones and have the least external admixture for sure, and all possible sources of admixture are covered. In PCA they are close and the other Steppe samples are all closer to EHG. We would think Ozera has more Anatolian/Iranian but not so from the f3.

So what is Steppe_Maykop_o and Yamnaya Ozera, because it doesnt look like external admixture pulling them away from the other samples?

gamerz_J said...

@Slumberry

Sorry, I should have indeed been more specific, I mentioned Peloponnese_N to replace Barcin_N precisely because it has some CHG. I seem to remember Mathieson et al (2018) found old layers of CHG ancestry in Southeastern Europe, thus some of it could be old and not from Kura Araxes.

The Levant_N is a different story. Most Greeks do have it indeed. Serbs apparently too albeit I find it suprising.

I think Serbs might also have some low EHG-like ancestry but Davidski showed it's not the case for Poles, and if they don't have it, I doubt Serbs will.

Regarding Dereivka_I_En2, where is that from?

Anyways, thanks for running these for me!

Davidski said...

@mzp1

I get a headache looking at the output you produce with ADMIXTOOLS.

But anyway, you can see at the link below how Yamnaya Ozera differs from the main Yamnaya cluster, and that it's not relevant to European population prehistory because it's an obvious outlier from the ancient Steppe > Europe Middle Neolithic genetic cline that produced most modern European groups.

https://vahaduo.github.io/g25views/#WestEurasia

Of course, I don't expect you to understand anything I just said. I actually said it for the benefit of everyone else reading this.

Anonymous said...

@mzp1
"If we look at the f3"

You don't seem to know how to use qp3Pop at all, you don't know what you count, you don't know how to count, what to input to the program, what to output as a result and what all this means.
If you don't know how to use then don't take it.

CrM said...

Matt noted previously that Yamnaya Ozera, aside from having Yamnaya and Maykop ancestry also has some EEF.


Target: Yamnaya_UKR_Ozera_o
Distance: 2.0838% / 0.02083788
53.8 Yamnaya_UKR
31.6 RUS_Maykop_Late
7.4 UKR_Trypillia
7.2 RUS_Maykop
0.0 RUS_Maykop_Novosvobodnaya
0.0 Yamnaya_BGR

Target: Yamnaya_Ozera_without50%YamnayaUkr
Distance: 4.1672% / 0.04167228
63.2 RUS_Maykop_Late
15.0 UKR_Trypillia
14.0 RUS_Maykop
7.8 Yamnaya_UKR


If that EEF came with Steppe ancestry, then the best match for it would be Corded_Ware_Baltic_early


Target: Yamnaya_Ozera_withoutMaykopLate
Distance: 3.7488% / 0.03748831
89.8 Yamnaya_UKR
10.2 UKR_Trypillia

Distance to: Yamnaya_Ozera_withoutMaykopLate
0.04010414 Corded_Ware_Baltic_early
0.04267295 RUS_Potapovka_MLBA
0.04596805 MNG_Afanasievo_1
0.04619164 TJK_Dashti_Kozy_BA
0.04634172 KAZ_Georgievsky_LBA
0.04670550 Yamnaya_UKR
0.04802342 UZB_Kokcha_BA
0.04814597 UKR_Catacomb
0.04957567 UKR_Cimmerian
0.04998284 KAZ_Dali_MLBA

Target: Yamnaya_UKR_Ozera_o
Distance: 1.8010% / 0.01800974
51.6 Corded_Ware_Baltic_early
25.0 RUS_Maykop_Late
23.4 RUS_Maykop
0.0 RUS_Maykop_Novosvobodnaya
0.0 UKR_Trypillia
0.0 Yamnaya_BGR

Mayuresh Madhav Kelkar said...

Archi

"Name Uralic is an outdated traditional name, they certainly began to speak Uralic, or more precisely Finno-Ugric, "

Seems relevant:



https://www.nature.com/articles/s41431-020-0699-4.epdf?sharing_token=O-fipjjPqnEtyR7yadlPctRgN0jAjWel9jnR3ZoTv0OIN6ZnOkUjTZGn8IWjCBW_YEC4Iprryqd2KEzHN63TyUOdtfcpp8ho1270BWHfkYiOPW8q3uq3PIQuxtmcv-IjN70DYO47F2vbnPR9js7pI4f4iNQB2qPMtcfOtlmmfyI%3D

ambron said...

EastPole, have you tried to use Spiginas2 on the N49 or N47 model?

Slumbery said...

@gamerz_J

Regarding Dereivka_I_En2, where is that from?

From Mathieson et al. The genomic history of southeastern Europe (2018). Sample number I4110, dated 3634-3377 calBCE, Dereivka site 1 (Ukraine). Davidski earlier speculated in a post that a similar population might be a partial ancestor for Yamnaya. But here I used it with the assumption that a similar population might mixed into Yamnaya during its westward expansion.
The site is by the Dnieper river about halfway between Kiev and the Black Sea.

Anthony Hanken said...

@Shaikorth

I appreciate the PCA. So in your opinion would a PU population look like something between the Steppe outliers and Okunevo or more Baikal_BA shifted?

In regards to the IA forest groups refrenced by Davidski, I believe the Akozino-Akmylovo culture which resided on the Volga is the best candidate for spreading the Tarand graves to Estonia.

Simon_W said...
This comment has been removed by the author.
Simon_W said...

@Samuel Andrews

"Here's G25 coordinates of France Bronze, Iron age. You can find the location and age of the samples in Brunel 2020.
France BA, IA
https://script.google.com/d/139aBUIXVmxExUPEzSbrm1-BgxX_caqmuA7Ol_Vvfy6xRFSt9ZMhadY1M/edit?usp=drive_web&folder=0AO4--OcB_rMfUk9PVA&splash=yes"

Thanks for sharing! I immediately ran the IA samples on Vahaduo, they make plenty of sense, more than all other analyses I had seen of them. I checked their distances to modern populations, and they get very decent fits, not bad and noisy ones.

I'm already seeing some patterns:

The Hallstatt samples from Alsace tend to come out quite southern. NOR4 resembles Spanish_Castello, Jeb8 and NOR3-15 resemble French_Occitanie, NOR2B6 resembles the Swiss_French, they are also not very northern. Only NOR3-6 comes out closest to French_Brittany. So we're seeing here, in the older Iron Age, a relatively southern-like population in the Alsace.

This picture changes with the second part of the Iron Age, the La Tene era. Although there are still southern-like individuals - ERS88 resembles Spanish_Terres_de_l'Ebre and COL153i comes out closest to French_Occitanie - the majority is now more northern, and even seems Germanic admixed to full blooded Germanic. ERS86 and COL153A are closest to French_Alsace and ERS1164 to Belgians. Both modern Belgium and Alsace are Germanic-admixed. And COL11 resembles the Orcadians, he may thus be a pure Germanic. This makes complete sense: The Triboci who settled in the Alsace were a Germanic tribe. And the southern-like individuals were probably remainders of the local Hallstatt population.

Now for the IA Picardie, of which we only have La Tene samples: ATT26 is closest to French_Paris, and BFM265 closest to English_Cornwall, thus the Celts there were northern French-like or even more insular Celtic-like.

From Southern France we have the La Tene samples PEY53, which is French_Brittany-like, and BES1248 which comes out closest to the English, but French_Brittany follows instantly on the second place. There was a third IA sample from Southern France which isn't in your table, it came out Catalan-like with Eurogenes K15 data. The Breton-like individuals in the very South of France are in all likelihood the result of Celtic population movements from more northerly locations of Gaul to the south.

The picture is clear: Hallstatt Alsace was quite southern-like, so it's very easily possible that in the mountain valleys of Switzerland and Bavaria similar southern-like Celts were preserved deep into the La Tene era and even into Roman times, alluding here to SX18 and the DEU_Roman individual.

IA Northern Gaul on the other hand has probably always been more northerly than that.

Fascinating stuff, thanks again for sharing.

Matt said...

As kind of on topic, I think we'd benefit on the history of Europe from a longer transect study of Hungary that combines existing samples with new samples and extends on from Lipson's 2017 study that closed out its period of study at 2900 BCE (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5973800/). Sort of a similar thing to Olalde's long term study of Spain.

Here are a few graphics of PCA position by time, and ancestry proportions by time, using G25 and West Eurasia PCA data - https://imgur.com/a/BSZDxcH

It seems like Hungary persists in having samples with fairly low, 10%ish additional HG beyond what Barcin has, even after the apparent arrival of steppe ancestry - first sample approx 2500 BCE, but there is a sampling gap between 3100 and then.

There seems to be potentially little ancestry in NE-Central Europe today from Globular Amphora related populations and related populations. That could be explicable as follows:

If you have Baltic_BA populations as in the Global 25 models where comes out with 66% CWC_Germany, 30% extra HG, and then 5% extra farmer, and CWC_Germany is about 20% GAC_Poland and rest steppe, then that implies about 13% GAC ancestry in Baltic_BA. Then if you follow on from that that present day NE-Central Europeans are 60-80% Baltic_BA, with the rest being from BA populations like those in Hungary who have little GAC related ancestry, then that could take that 13% further down to 8-11%. That could explain why there seems like there could be fairly little specific GAC related signal in G25 for these populations compared to what seems to be the case in Western Europe - the Middle Bronze Age CWC and post-CWC populations didn't have too much (20-25%) which then got somewhat overwritten by HG ancestry and ancestry from other cultures rich in Barcin related ancestry, which persisted better in some areas of Northeast-Central Europe than GAC did.

(Bonus point in favour of this would be that the first ancient sample we have which showed very strong autosomal haplotype matches with present day populations was from Hungary Bronze Age in Cassidy's 2016 paper).

Big study on Hungary that covers 3000-1500 BCE in detail would help to begin resolve all this.

Coldmountains said...

@Matt

Not really anything suggests that the Farmer substrate of Balto-Slavs is not GAC derived. What you see here is rather the overfit of nmonte what is the same reason why Balto-Slavs show here Karelia_HG.

Balto-Slavs share a lot of mtdna with GAC, formed in earlier core ares of GAC and are Corded Ware-derived. All late CWC groups show GAC-like admixture even far in the east and west so if Balto-Slavs lack GAC than they should form outside of CWC what is not very likely.

mzp1 said...

Yeah, I think Ozera might have some additional Anatolian N or something.

The qpadm setup I've got running looks pretty good, I can get quite a few sources in there and get decent results for different populations. Even run it with two Yamnaya sources side by with Anatolian_N, EHG and WHG.

That should allow for some really good analysis...

Rob said...
This comment has been removed by the author.
vAsiSTha said...


@crm i see you are not at all willing to give Ozera outlier any south of caucasus ancestry.
with your models.

Target: Yamnaya_UKR_Ozera_o:I1917
Distance: 1.4252% / 0.01425217
38.0 Corded_Ware_Baltic_early
22.8 RUS_Maykop_Late
21.2 Yamnaya_UKR
18.0 IRN_Hajji_Firuz_C

a much better fit

Tigran said...

@coldmountains

What's the history of R1a in Poland? Was there any R1a there during the CW Period or Bronze/Iron Age?


@Davidski
Can this analysis be broken down to the most atomic(or at least more atomic) components?
Like how much WHG, WHG like/UHG, how much from the western part of ANE, how much from Basal Eurasian, Ancient North Africans, East Asian?

Coldmountains said...

@Tigran

R1a was probably never absent in most parts of Poland prior to Slavs and after the first steppe groups arrived just the frequency of it drastically varied through the ages and regions. Early CWC,Late CWC/EBA and Unetice Poland all carried both R1a and R1b. But not sure how much of this R1a caldes survived among modern day Poles only future genetic studies can show that.

Rob said...

@ Tigran

“ What's the history of R1a in Poland? Was there any R1a there during the CW Period or Bronze/Iron Age?”

Why don’t you go read all the papers and make notes ? That’s the best way to learn instead of constantly asking others

Norfern-Ostrobothnian said...

Hey I have no idea how this works but does it have Mari samples?

Norfern-Ostrobothnian said...

Did we ever get that Volosovo-Danilovsky sample from a month ago?

Matt said...

@coldmountains, CWC cultures did not have *huge* amount of GAC like ancestry (25%) and plausible other ancestry contributions from other populations reduced further. Not saying it is lacked entirely but plausible only 10% or so while 50% ancestry Western European groups. I don't think the best model will ultimately be "CWC+more MN European GAC type farmers+a little HG"+drift for East-Central Europeans (e.g. East-Central Europeans have more of less same history as West European plus a little extra HG), probably more "CWC+HGs+Hungary like farmers with less HG than GAC". I don't think anyone had ever had much luck reconstructing autosomal (or any) population history from mtdna frequencies. Time will tell.

Rob said...

@ Old Europe

This is what is presently evident

https://imgur.com/JRYsVgZ

I have left out R1b-M269 because we still don't know where it was ~ pre-4000 BC

Coldmountains said...

@Matt


1. CWC in Slovakia/Czechia shows strong GAC-like admix just like other CWC groups and unlike earlier EEFs there. So even if Proto-Balto-Slavs came from there (Eastern Unetice) they would get their EEF mostly from GAC.

2. Modelling modern day Balto-Slavs as mix of Baltic_BA+ Balkan_IA is problematic because Baltic_BA is a peripheral Balto-Slavic group which likely picked extra HG ancestry further north and not migrated southwards.

3. Rather most Balto-Slavs including Balts like Lithuanians are derived from BA groups further southwest than the currently sampled Baltic_BA groups. We lack the right BA/IA references for most modern day Balto-Slavs

4. Using Baltic_BA unrealistically inflates Balkan_IA and LBK-like ancestry because it is more shifted towards HG than any modern what will be compensanted by nmonte with something very low in HG ancestry

5. Like mentioned before most Balto-Slavic mtdna clades (especially H clades) are not just shared with GAC but also with Scandinavians/NW Europeans, what points that both formed out of CWC+GAC-like groups.


Arza said...

@ Samuel Andrews

What can you tell us about such set of mtDNA haplogroups? What contemporary group can this set represent?

H1c
H27
H2a1
I1a1a
I4a
J1c
J1c
J2b1a1
K1c1
T1a1
T2b
T2b
T2b
U2e1a1
U2e2a1a2
U4b1b1
U5a2b1a
V3a

Samuel Andrews said...

Slavs, East/Central Europe. Is this ancient mtdna?

Vladimir said...

@Rob
In General, I agree, except for the date of the starting point on the Dnieper. There, this population was later approximately 3500 BCE, just at the time of the destruction of the Dnieper group Tripolye. And it came there from the don, most likely from the upper reaches of the don, at least from the section between the Don and the Dnieper. Because even the most recent Dereivka samples do not show the autosomal composition of pre CWC.
Do I understand you correctly, you consider I2a2-L701 Hittites?

pnuadha said...

The assumption that EHG is survived in north and got incorporated into Uralic speakers (then subsequently Northern Russians) is logical, but does not appear to be true. At least not above an 1-3% level that is hard to identify in the strong presence of similar ancestries. Slumbery

Ok, but what are North Russians, Finns, and Estonians composed of? According to Davidski's model they all have a lot of yamnaya like ancestry, which I find hard to believe.

Yamnaya like ancestry was formed on the PC steppe, and nearby, from around 7000 to 5000 years ago. Furthermore, we know that CHG became diluted everywhere around 5000 to 4000 years ago which means that yamnaya like ancestry could never again recreated from scratch. So unless I am mistaken, all yamnaya like heritage comes from the PC steppe and nearby from 7000 ybp to 5000 ybp.

From the creation of yamnaya like heritage we can trace most of the migrations, which are largely linking to indo europeans (yamnaya, corded ware, Afanasievo, ect...).

So how did the non IE speaking Finns, Estonians, and North Russians end up deriving about 1/2 from yamnaya like people?

Before you and Davidski spoke against the idea of EHG surviving apart from yamnaya like groups I figured that the yamnaya in these uralic groups was just a proxy for EHG/ANE and some farmer/southern stuff.

Its certainly possible that not all Yamnaya like groups in 3000 were IE. Its also possible that IE CW people took on Uralic at some point in the northeast. But still, 50% is a lot. I expect population replacement on the steppe but I wouldnt expect it in the far north...

CrM said...

@vAsiSTha

To be honest I didn't even think of this. But Hajji Firuz C (or something very similar to it) was important in the formation of various Caucasian cultures, especially KAC and most likely Leyla Tepe - the proposed ancestor of Maykop. AZE_Caucasus_lowlands_LC replaces Hajji Firuz C in my run if you include RUS_Maykop.

Target: Yamnaya_UKR_Ozera_o
Distance: 1.7449% / 0.01744921
42.2 Corded_Ware_Baltic_early
21.6 RUS_Maykop_Late
13.4 RUS_Maykop
12.6 Yamnaya_UKR
10.2 IRN_Hajji_Firuz_C
0.0 Kura-Araxes_ARM_Kaps
0.0 RUS_Maykop_Novosvobodnaya

Target: Yamnaya_UKR_Ozera_o
Distance: 1.7013% / 0.01701256
50.4 Corded_Ware_Baltic_early
26.6 RUS_Maykop_Late
19.6 RUS_Maykop
3.4 AZE_Caucasus_lowlands_LC
0.0 AZE_Caucasus_lowlands_LN
0.0 IRN_Hajji_Firuz_C
0.0 Kura-Araxes_ARM_Kaps
0.0 RUS_Maykop_Novosvobodnaya

Rob said...

@ Vladimir

Just because C-T was reaching the Dnieper; it doesn’t mean the resident hunter-gatherer groups which had been there since ~ 5000 BC disappeared. They were certainly there in the native habitats along the river
New data tells us there were eneolithic barrows as early as 4000 BC as far as Podolia; in between CT - like a chess board

“ Do I understand you correctly, you consider I2a2-L701 Hittites?”

No categorical hypothesis about Hittites specifically . Merely that I expect groups which moved toward anatolia via Bulgaria to consist of I2a2; given the data from EBA Bulgaria

CrM said...

Forgot to include this

Target: Yamnaya_UKR_Ozera_o
Distance: 1.6306% / 0.01630593
35.0 Corded_Ware_Baltic_early
22.2 RUS_Maykop_Late
19.8 Yamnaya_UKR
12.6 RUS_Maykop
10.4 AZE_Caucasus_lowlands_LC
0.0 AZE_Caucasus_lowlands_LN
0.0 IRN_Hajji_Firuz_C
0.0 Kura-Araxes_ARM_Kaps
0.0 RUS_Maykop_Novosvobodnaya

Arza said...

@ Samuel Andrews

Slavs, East/Central Europe. Is this ancient mtdna?

Thanks. Yes, it's aDNA from the Tollense Valley.

WEZ56 (Baltic_BA-like) has Y-DNA described as R1 despite that the quality of the sample seems to be sufficient to give him a much deeper assignment... just like in the case of MX265.

The rest is under I2a2 (most derived - I2a2a1a2a1a1) and R1b (most derived - R1b1a2a1a2).

https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3565013
Genomic Data from an Ancient European Battlefield Indicates On-Going Strong Selection on a Genomic Region Associated with Lactase Persistence Over the Last 3,000 Years, Burger et al.

Abstract

Lactase persistence (LP), the continued expression of lactase into adulthood, is the most strongly selected single gene trait over the last 10,000 years in multiple human populations. It has been posited that the primary allele causing LP among Eurasians, rs4988235*T (Enattah et al. 2008), only rose to appreciable frequencies during the Bronze and Iron Ages (Mathieson et al 2015; Olalde et al. 2018), long after humans started consuming milk from domesticated animals. This rapid rise has been attributed to an influx of peoples from the Pontic-Caspian steppe that began around 5,000 years ago (Allentoft et al. 2015; Furholt et al. 2016). We investigate the spatiotemporal spread of LP through an analysis of 14 warriors from the Tollense Bronze Age battlefield in northern Germany (~3,200 BP); the oldest large-scale conflict site north of the Alps. Genetic data indicate that these individuals represent a single unstructured Central/Northern European population. We complemented these data with genotypes of 18 individuals from the Bronze Age site Mokrin in Serbia (~4,100 to ~3,700 BP) and 37 individuals from eastern Europe and the Pontic-Caspian Steppe region, predating both Bronze Age sites (~5,980 to ~4,250 BP). We infer low LP in all three regions, i.e. in northern Germany, south- eastern, and eastern Europe, suggesting that the surge of rs4988235 in Central and Northern Europe was unlikely caused by Steppe expansions. We estimate a selection coefficient of 0.06, and conclude that the selection was on-going in various parts of Europe over the last 3,000 years.

Arza said...

https://publications.ub.uni-mainz.de/theses/frontdoor.php?source_opus=100003480&la=en
Genetic variation related to the adaptation of humans to an agriculturalist lifestyle
Jens Blöcher

This study investigates the influence of climate and subsistence on the genetic variation of prehistoric individuals of Europe. Partial genomes from more than 100 individuals were analyzed, originating from a region spanning from north-western Anatolia, over Germany to Lithuania. The data is spread over a period of more than 10,000 years and thus covers the phases before and after people started to become sedentary. To fill temporal and geographical gaps, 27 individuals were sequenced specifically for this study. For this purpose, a hybridization-based enrichment process was used in combination with next-generation sequencing to obtain predefined genomic segments from ancient DNA (aDNA). These regions consist of over 400 functional markers, 68 non-recombinan Y markers, and more than 5,000 neutral regions of 1kb in length. The data set was completed b genomes of three modern reference populations. These made it possible to calculate haplotype- and frequency-based neutrality tests such as the integrated haplotype score or Tajima’s D and thus to be able to detect signals of past selection events. Using the allele frequencies of the ancient genomes, the timing of possible selection processes could be narrowed down.
The analyzes focused on three main topics: 1.) the development of skin, eye and hair color phenotypes in Europe; 2.) The development of Europe’s predisposition to type 2 diabetes and the metabolic syndrome; and 3.) The development of the immune system with respect to autoimmune diseases. The historical focus was on the influence of subsistence and the transition to farming on the evolutionary processes.
The results suggest that a high proportion of the genetic variation in genes associated with the energy and fat metabolism has been affected by adaptations to the local climate in Central Europe. Furthermore, clear indications were found that due to the reduced UV radiation in Central and Northern Europe, depigmentation of the skin had already begun before the onset of the Mesolithic. On the basis of the available data, it can be assumed that the the shift in diet and lifestyle that accompanied the transition to farming partially reinforced already existing selection pressures. Thus, with the Neolithic, a significant increase in allele frequencies associated with lighter pigmentation phenotypes was found. This is probably due to the reduced vitamin D content in the diet. Furthermore, evidence was found that indicates a metabolic adaptation to a plant-rich diet. For the later Neolithic and Bronze Age periods, an increase in allele frequencies has been found in genes that enhance protection against infectious diseases. A connection with the increasing population density and a sedentary lifestyle is very likely here.

pnuadha said...

The really interesting thing is the fact that most Uralic speakers don't show any significant EHG or Western Siberian HG (WSHG), but they do show significant eastern Siberian ancestry.

Davidski

Has that been area been catalogued well? I thought we only knew that at one point they were EHG and next they were like modern Uralics. I didnt see the mixing of Siberians with whatever was in North Russia/North Baltic.

It looks like a lot of people find it offensive to see that Eastern Europeans have as much steppe ancestry as Western Europeans, if not more.

both your model and Haak's model says its just north vs south. You say Sweden (top), Iceland, and Ireland along with Lithuania and Ukraine, while Haak says Norway (top), Iceland, and Scottish along with Lithuanians and Estonians.

Just curious, do you think Haak's estimates are at all dated?

As for your estimates, I notice two things.

One is that components do not follow a strict cline. Remember in 2012 when you modeled Europeans with only modern populations. Something like "North East European" formed a perfect cline in the sense that if NEE peaked in Lithuania, then it would be higher in Belarus than it would be in Ukraine. But the yamnaya compenent in your current model sort of jumps around. Yamnaya peaks in Sweden, then dips in Norways and Scotland, then peaks again in Iceland and Ireland. (Also surprised that Ukraine has so much Steppe, and more than Poland.) Can you explain why your 2012 models (only moderns) followed a unipolar cline while this model does not and what that says about true ancestry?

Also, have you noticed that in your model all non southern europeans are extremely similar in overall composition, much more so than in the Haak model. The yamnaya and barcin components only vary by 10 and 20% in your model but they vary by 15 and 30+% in Haak's model. Any comment on which you think is more accurate?

ambron said...

Arza, did I understand you correctly? Is WEZ56 something close to the M458, like the MX265?

Davidski said...

@pnuadha

I ran many more samples than Haak et al. They had something like three Scots and a few more Ukrainians than that. I ran 10-15 individuals for most of these sets.

Apart from that, as I've said before, you have to look at the fits and standard errors. These weren't reported for the Haak et al. estimates.

And how the hell are those old estimates supposed to be super accurate anyway when they used modern populations as outgroups? Think about that.

As for late EHG survival, I'm not aware of anything like that in already published or soon to be published data.

The EHG and even EHG-heavy foragers got wiped out both in the steppe and forest zones of Eastern Europe. From the Early Bronze Age all you'll see in the forest zone are groups like Fatyanovo.

But eventually Fatyanovo and Abashevo disappeared somewhere, and groups from the west took over Eastern Europe. Of course, these more westerly groups were rich in extra WHG not EHG.

Then there's an expansion of Siberian-admixed Uralians across Northeastern Europe, and when they merged with the locals we got people like those early Finnics from Iron Age Estonia.

So if you're aware of a modern or ancient population that doesn't fit this pattern, then let us know what it is and how it formed. By the way, the Saami do seem to have quite a bit of Scandinavian hunter-gatherer ancestry, but that's much closer to WHG than EHG.


Davidski said...

Czesc Arza!

Up to no good, I see, as usual. ;)

Where did you get those Y-haps for the Tollense Valley warriors? The link you posted is dead.

Arza said...

@ ambron
It's R1.

ambron said...

Arza, mtDNA explains well Welzin's autosomal similarity to Poland.

Matt said...

@coldmountains: "Modelling modern day Balto-Slavs as mix of Baltic_BA+ Balkan_IA"

Not proposing Balkan groups (or "Slovakia/Czechia ") but Hungarian ones! A transect of Hungary through Bronze Age will tell us more about how viable this is.

I don't think that G25+Vahaduo picking Baltic_BA as a source for Slavic populations is an illusion, which you seem to be suggesting (that Baltic BA is chosen as an ancestor for Slavic populations, even though it wasn't, and because of this Hungarian Copper and Bronze Age populations are chosen as another population to balance out). I think a population like this really did move southwards (or on from wherever its origin was) and did contribute.

Arza said...

@ Davidski

Cześć!

Luckily I've downloaded the pdf before the maintenance has started. You have it in your mailbox already.

Samuel Andrews said...

@Azra, ok that makes sense. On first glance the tollense valley mtdna is also similar to already published bronze age Poland mtdna.

So I believe this WHG-rich Steppe-poor pop lived in Poland during Bronze age and contributed to maternal ancestry of Early Slavs.

Bronze age Baltic also shares mtdna with modern eastern Europeans but different shares different mtdna.l than what is in tollense valley. Maybe I'm wrong but that is my guess.

Norfern-Ostrobothnian said...

Davidski, Ananyino culture (Permians and Mari) doesn't overlap with Fatyanovo-Balanovites at all. As for HG in their area, SHG or Narva makes no sense. WSHG makes some sense in theory.
From what I have read, there was famine and desolation in the latter era of Fatyanovo-Balanovo and Volosovo. Then the Net-Ceramics arose in their region.

Davidski said...

@Matt

You've got things backwards, both in terms of genetics and linguistics.

Sure, Baltic_BA is related to modern Balto-Slavs, obviously with LTU_BA being more closely related than LVA_BA.

However, Baltic_BA is not ancestral to modern Balto-Slavs, who obviously came from somewhere nearby, and in all likelihood from somewhat different, more southern-shifted populations.

LTU_BA may have been in large part ancestral to West Balts, but forget about it being ancestral to East Balts or Slavs, who moved into that region much, much later.

LVA_BA is out of the ballpark even for West Balts.

Davidski said...

@Norfern-Ostrobothnian

WSHG is basically irrelevant to the genetic structure of modern Uralic speakers.

I don't know why and how this happened yet, but it's a fact.

Most Uralians in Europe closely resemble their non-Uralic speaking neighbors, except they usually have some extra Nganasan-related ancestry.

I'm not sure yet who the Uralians east of the Urals resemble exactly, but they don't carry much WSHG either. Their western ancestry seems to be derived from various steppe groups, while their eastern ancestry, which is considerable, is Nganasan-related as well.

Matt said...

@Davidski doesn't look to me like the Baltic_BA pops represent a composite of "drift that happened somewhere else in populations with more southern farmer ancestry" + "local extra HG ancestry" and then being picked up falsely as ancestor in models due to paucity of other samples.

Davidski said...

@Matt

This drift did happen somewhere else.

We know when it arrived rather suddenly in the East Baltic region, because we've got the smoking gun: Spiginas2.

More smoking guns are on the way, from nearer the Carpathians.

Matt said...

Great, we'll see if there is ever evidence of pops showing up with as much or more drift along a Balto-Slavic dimension, but less HG ancestry than Baltic. Spiginas2 isn't that (although IRC he doesn't show the full tendency of drift either). At the moment these are completely correlated, indicating a single cline...

Norfern-Ostrobothnian said...

Should Narva/Kunda admixture represent some of the Balto-Slavic HG admixture?

Davidski said...

@Norfern-Ostrobothnian

Balts and Slavs took with them a very western type of HG admixture, probably similar to Lithuanian Narva, deep into Russia.

So yes, we should expect this type of ancestry in Uralians to be largely of Balto-Slavic origin, but Fatyanovo and related groups also had very western HG ancestry (not including the EHG they got from their steppe ancestors of course).

Also, there's a lot of Germanic admix in Finland and even some in Estonia. It's not yet clear when it first spread into Fennoscandia, and if it had any impact on the Uralians in Russia.

It may have, because Iron Age samples from Karelia show Germanic-like fine scale ancestry.

Norfern-Ostrobothnian said...

Karelia republic or Karelia ancient? Because Ancient Karelia was big on trade because they held access to Neva, which was enroute to Volga. As for the republic, White Sea area wasn't settled by Finnics but used as foraging land by the Saami. So if it's from that area, they might have been explorers/hunters/taxers from the west.
As for other Uralic groups the Germanic ancestry probably is irrelevant. Mordovians are the only so called West Uralic group that didn't interact with the Nordics in any way.

Coldmountains said...

@Samuel Andrews

The mtdna of Welzin warriors looks mostly generic North-Euro/Central Euro with a slight shift towards modern day Balto-Slavs. Basically all of this clades are today found from Britain to the Urals but with a strong presence in Poland and Scandinavia/Finnland. For now it is very hard to say if they are just shared between modern day Slavs and Welzin people because of common GAC, Steppe and WHG ancestry or because of Welzin-like people contributing directly to early Slavs. All of this clades formed long before 2000 B.C so only deeper analyses can say that.

H1c: found in BA Germany, EBA Britain and BA Latvia -> today frequent in Poland, Scandinavia and Russia/East Europe.

H27: found in BA Germany and GAC -> today found in Scandinavia,Russia and East Europe

J1c: very broad/widespread clade but frequently found in GAC so likely GAC related

H2a1: found in Yamnaya, Iron Age Lithuania, BA Germany -> frequent in Russia, Scandinavia

I1a1a: -> frequent in modern day Russia, Finnland, Central Europe/Poland, Germany, West Europe

I4a: found in CWC Czechia, Bell Beakers, CWC Baltic (Spiginas2 with lot of Balto-Slavic drift) -> frequent in Scandinavia, Poland, Germany, Russia

J2b1a1: found among ancient Minoans -> found in Britain, Germany, Russia and Poland

K1c1: found among Bell Beakers, Fatyanovo and BA Hungary -> today frequent in Scandinavia, Russia, Poland, Ukraine, Britain, Germany

T1a1: found in CWC Poland, Fatyanovo and Unetice -> frequent in East Europe, Scandinavia and Central Asia.

T2b: very broad and widespread but found basically wverywhere in Europe.


U2e1a1: found in CWC Germany and Bell Beakers -> Today frequent in Scandinavia, Poland, East Europe and Central Europe.

U2e2a1a2 -> found in Germany and Scandinavia

U4b1b1 -> found in Mesolithic Serbia and Neolithic Ukraine -> found in Russia, Germany and Scandinavia


U5a2b1a -> found in modern day Scandinavia, Russia, Belarus and Poland

V3a -> found in modern day northwest Europe.

Samuel Andrews said...

@ColdMountains,

There is a direct link between a few of the mtDNA clades in Tollense Valley and modern Slavs. I labelled their mtDNA likely to be from East/Central Europe before I knew the mtDNA was ancient mtDNA from Tolenese Valley, so I mean that adds credibility.

U5a2b1a is pretty much exclusive to Eastern Europe. You might find one example in Scandinavia, but overall it is pretty much only in Eastern European.

I1a1a, also basically only Eastern Europe. mtDNA I, in other parts of Europe do not fall under I1a1a.

H1c is pan European but because it is alongside I1a1a, U5a2b1a that is more support to idea this was East European population.

vAsiSTha said...

@davidski

"WSHG is basically irrelevant to the genetic structure of modern Uralic speakers.

I don't know why and how this happened yet, but it's a fact."

Because genes do not carry information about language spoken, archaeology & literature does. Sometimes people on this blog need a reminder of the same.

Rob said...

I think the “archaeological TMRCA” (if you’re excuse my neologism) for Balto-Slavic could be the Milohrad culture
There’s almost no point trying to model a putative population given the data we have; all the more given the cultural & demographic tessalations and permutations which occurred before and after

Davidski said...

@vAsiSTha

No, that's definitely not the punchline.

You'll see what it is when more samples come in from Siberia, especially from Seima Turbino sites.

vAsiSTha said...

You are free to make up your mythology about what ancient people spoke what language based on purely their genes.

I look for archaeological and literary evidence to decide which ancient region spoke what kind of language.

Rob said...

@ Vasistha

people spoke languages. ADNA tells us how they moved; when ; and how they interacted with their neighbours; what power relations were set up
Texts 1000 years after the fact can’t give that insight

Norfern-Ostrobothnian said...

@ Davidski
Are we really expecting Seima-Turbino samples? What cultures? I'd want some Chirkov culture samples to be honest, maybe they could've put input to the Mari.
https://arheologija.ru/chirkovskaya-kultura/

EastPole said...

@Rob

“I think the “archaeological TMRCA” (if you’re excuse my neologism) for Balto-Slavic could be the Milohrad culture”

And how do you explain the correlation between R1a in Slavs and Vedic Aryans with the similarity in languages and cultures, in particular religion? Vedic Gods have Slavic origin. Milohrad culture existed thousands of years after Slavic and Indo-Iranian separation.

https://i.postimg.cc/BbByfGTw/Language-Tree.jpg

In my opinion, CWC was Indo-Slavic. CWC that went East became Indo-Iranians, CWC that stayed between in the Vistula-Dnieper area were proto-Slavs.

Many people here use the word Slavs without understanding what it means. Not even trying to define the word.

Slavs were the people who first used Slavic words!

Mouthful said...

"Vedic Gods have Slavic origin."

Why is such bullshit allowed to be posted on this blog?

Davidski said...

@Mouthful

His comment wasn't against any rules or laws, and it wasn't even rude, unlike yours.

You may well be correct about what he said, but the fact that it passed moderation doesn't mean that I endorse it in any way, and he's absolutely free to make these sorts of claims.

Mouthful said...

Maybe I've said it in a mean way, but I can see that a lot of hecklers (Won't name them as everyone knows them) get put in their place while posting obviously far from reality posts, while in this case and not once or twice he keeps repeating this over a large period of time.

Also, sorry for O.T, if needed delete this post.

Samuel Andrews said...

EastPole is the Polish Gaska.

Mouthful, I agree it is annoying to see him say Corded Ware was Slavic and to say the Vedic Aryans were Slavic. Like with Gaska, it is impossible to reason with him.

Davidski said...

@Norfern-Ostrobothnian

Are we really expecting Seima-Turbino samples?

Yes, lots of ancient samples are on the way from Siberia, especially from cultures potentially linked to the Uralic puzzle.

I have a feeling that something big is coming about the spread of Uralic languages and genes. Actually, it's more than a feeling...

Norfern-Ostrobothnian said...

@Davidski
So what cultures/samples are you supposing are linked to them? Krotov culture seems pretty good in my opinion with the aforementioned Chirkov material culture descending from them.
Do you think that we can finally get some good G25 samples for Mari? Currently there is no way to properly represent them.
Target: Mari
Distance: 8.1860% / 0.08186011
65.8 FIN_Levanluhta_IA
22.0 RUS_Bolshoy_Oleni_Ostrov_o
7.6 RUS_Maykop
2.8 DEU_LBK_KD
1.8 EGY_Hellenistic

Norfern-Ostrobothnian said...

I think Finns might end up with more "eastern" admixture than now if the samples are good. The Nganasans absorbed a lot of Paleo-Siberian admixture and loanwords so I don't think they work as the perfect component for the Uralic eastern ancestry. I'd maybe expect something along the lines of 10%?

EastPole said...

@Mouthful
“"Vedic Gods have Slavic origin."

Why is such bullshit allowed to be posted on this blog?”

For example:

Slavic ‘Ogni’ > Vedic ‘Agni’

If you don’t agree prove it.
My claim is ‘Ogni’ was used in CWC and from there it got into Indo-Iranian. So we have an example of Slavic words in CWC.

Cpk said...

I am really curious about the ancestry proportions of Proto Turks.

Anonymous said...

@EastPole
Slavic ‘Ogni’ > Vedic ‘Agni’

Why not Slavic ‘Ogni’ < Vedic ‘Agni’? Are Hittite agniÅ¡ and Latin ignis also from Slavic?
What does childish linguistic franticism have to do with this blog? Stop being ridiculous, you seriously don't know anything.



@Arza

You can put the Tollense paper in the public domain?

CHG Chad said...

Can someone give an example of how Proto-Uralics would have been like in terms of autosomal DNA?

vAsiSTha said...

@eastpole

I would like to know the date of first recorded history of the Slavic God Ogni.

Thanks

Davidski said...

@G2a-M406 Anatolian Farmer

Can someone give an example of how Proto-Uralics would have been like in terms of autosomal DNA?

Similar to modern Nganasans, but more western shifted.

There's nothing like this yet in the G25 datasheets.

Norfern-Ostrobothnian said...

@Davidski
So Khanty-like? Nganasans have a lot of Paleo-Siberian from Taimyr and later on absorbed Tungus groups. So all in all I think there's no good proxy at the moment.

Mr. J said...

A bit unrelated but since we are talking about Uralians do pretty well all Finns in Finland today even those with roots from the far West and South of Finland have some level of Saami ancestry and DNA?

EastPole said...

@Archi

“Slavic ‘Ogni’ > Vedic ‘Agni’

Why not Slavic ‘Ogni’ < Vedic ‘Agni’? Are Hittite agniÅ¡ and Latin ignis also from Slavic?
What does childish linguistic franticism have to do with this blog? Stop being ridiculous, you seriously don't know anything.”

Your stupid questions suggest that you don’t know Brugmann's law.

CHG Chad said...

@ David

Ty David.Yep,thats why i asked actually.. there is nothing yet in G25.I was reading some crazy shit about Uralics and Corded on Carlos forum yesterday and i was very confused.

Anonymous said...

@EastPole
"Your stupid questions suggest that you don’t know Brugmann's law."

Don't talk nonsense, don't involve irrelevant laws. My questions are the smartest one, but you can't understand theirs because you don't have a mind. Your statement is stupid, because IE [a] gave in Slavic [o].

You don't know anything about languages, you just flaming nonsense. You don't know anything in languages, it's a fact, so don't write in them at all. I'm tired of children pampering with languages on non core blogs.

vAsiSTha said...

bruggmans 'law' is not like Newton's law. Hardly anyone accepts it to be true.

All Bengalis pronounce Agni as Ogni, just as they pronounce most a's as o's. Does not mean that sanskrit is the daughter of Bengali. It's actually the other way round.

You did not answer my question.
1. When is the earliest attestation of the word in literature?
2. Why did cwc pass it on to slavic and not other European languages?

Mayuresh Madhav Kelkar said...

Russian огонь (ogon), Polish "ogień", Slovenian "ogenj", Serbian oganj, and Lithuanian ugnis


Also Germanic ignite, ignition all from Proto-Indo-European root *h₁n̥gÊ·nís

dsjm1 said...

Just looked at the Y-DNA for the 14 Tollense Battle warriors chose for the paper published about Lactase Persistence.

Am a bit disappointed that not one U106 appears in the table. 3 x P312 (WEZ35=R1b1a2a1a2, WEZ54=R1b1a2a1a2, WEZ59=R1b1a2a1a2), 9 x I2 (WEZ15 is listed as I2a2a1a2a1a1, the others less).

Ahh well - a long wait and a dissapointment. But it does tell us P312 & I2 were battling in the Sth Baltic near Wolin Island 1300 BCE.

«Oldest ‹Older   1 – 200 of 536   Newer› Newest»