search this blog

Tuesday, March 6, 2018

Main candidates for the precursors of the proto-Greeks in the ancient DNA record to date


Thanks to the recent release of the Mathieson et al. 2018 dataset (see here), I've been able to spot a very interesting northwest to southeast genetic cline running from the oldest Peloponnese Neolithic (Peloponnese_N) individuals to the Bronze Age Anatolians (Anatolia_BA). Here it is, highlighted in my Principal Component Analysis (PCA) of ancient West Eurasian variation. The relevant datasheet is available here.


I don't think it's a stretch to assume that this cline represents, more or less, the genetic diversity that existed in the Aegean region during the early Helladic period, just prior to the incursions of Bronze Age steppe or steppe-derived peoples who, according to the current academic consensus, probably gave rise to the proto-Greeks and Mycenaeans (see here).

There are three main reasons for this: 1) the Peloponnese_N samples show a very deliberate "pull" towards Anatolia_BA, suggesting that the Peloponnese population experienced admixture from a source similar to Anatolia_BA prior to the Bronze Age, 2) the cline cuts right through the middle of an "Old European" cluster made up of Minoans, who lived on Crete and other Aegean islands on the eve of the aforementioned steppe-derived incursions, and 3) both the Mycenaeans and Minoans can be modeled in large part as Anatolia_BA and Peloponnese_N.

The identification of this genetic cline, and what it likely stands for, is important, because it should allow us to plausibly point to the source of foreign input that created the Mycenaeans, and thus the Proto-Greeks. And clearly, the trajectory of the Mycenaean "pull" away from this cline is towards most of the samples marked as "Eneolithic and Bronze Age steppe".

However, this doesn't mean that it's necessary, or even sensible, to look for the precursors of the Proto-Greeks amongst these samples. That's because there might be much more proximate options based on, say, geography, archeology, chronology and mixture modeling. Indeed, using various criteria, I've chosen three individuals who sit along the Mycenaean to Eneolithic/Bronze Age steppe cline in the above PCA and might plausibly represent the precursors of the Proto-Greeks, or close relatives thereof. The first two are from Mathieson et al. 2018 and the third from Olalde et al. 2018.

- if, as most academics posit, the people who were to become the Proto-Greeks came from the Early Bronze Age (EBA) Yamnaya horizon on the Pontic-Caspian steppe, then it's possible that they were similar in terms of genome-wide genetic structure to the only Bulgarian Yamnaya sampled to date: Yamnaya_Bulgaria Bul4

- on the other hand, if, as has also been postulated in academic literature, they derived from the Middle Bronze Age (MBE) chariot warrior groups of the post-Yamnaya Pontic-Caspian steppe, then they may have been similar to Balkans_BA I2163, who is also from Bulgaria, but dated to more than a thousand years later than Bul4, and clusters strongly with the said chariot warriors, such as the Sintashta people, and even belongs to the same Y-haplogroup: R1a-Z93

- but if they came from the Yamnaya horizon via the Carpathian Basin, which, I'm told in the comments here, is also a serious option, although admittedly I've missed it in my reading, then they may have been similar to Proto-Nagyrév individual Hungary_BA I7043, who belongs to Western European-specific Y-haplogroup R1b-L51, a marker fairly common amongst modern-day Greeks.

And here's a mixture model for the Mycenaeans, using the Global25/nMonte method (see here and here), and the above trio as potential reference samples, alongside Anatolia_BA and Peloponnese_N.

[1] distance%=1.9802

Mycenaean

Peloponnese_N,45.2
Anatolia_BA,35.8
Yamnaya_Bulgaria,16.4
Balkans_BA,2.6
Hungary_BA,0

Thus, it seems that the precursors of the Proto-Greeks came from Bulgarian Yamnaya. However, they, or the Mycenaeans, may also have had minor ancestry from the chariot warriors of the MBA Pontic-Caspian steppe. Yes, I'm probably reading far too much into these results, but I can't help it, because they appear so logical. Indeed, check this out:

[1] distance%=4.209

Mycenaean:I9033 (elite burial)

Peloponnese_N,38.8
Yamnaya_Bulgaria,33.4
Anatolia_BA,27.8
Balkans_BA,0
Hungary_BA,0

If this is just an artifact of the method, then it's a really nice one. But who are your main candidates for the precursors of the Proto-Greeks in the ancient DNA record to date? Feel free to let me know in the comments.

See also...

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

125 comments:

zulla said...

I have an amateur question unrelated to this post.

How can geneticists model current population as a mixture of 2 ancient populations, but not know the component haplogroups of the 2 ancient populations, eg in case of ANI ASI?

Thanks.

Davidski said...

@zulla

The origins of what were once known as ANI and ASI isn't much of a mystery. Have a look at the graph here...

https://eurogenes.blogspot.com/2017/06/ancient-herders-from-pontic-caspian.html

If you're referring to Y-chromosome haplogroups, then for ANI they are mainly R1a (steppe-derived) and J2 (Iran Neolithic farmer-derived).

I'm not quite sure about the ASI Y-haplogroups though. We might need a few ASI genomes from Mesolithic India to work that out, simply because of the massive male-biased genetic turnover in South Asia since the Mesolithic.

zulla said...

But surely the scientists who modeled the modern populations as mixture of ANI ASI know the Y haplogroups of ASI? Didnt their model tell them? Is it a case of them not giving out this data? Otherwise how does the model work?

Davidski said...

@zulla

The ANI/ASI model was based on genome-wide DNA from the 22 autosomal chromosomes, not Y-DNA or mtDNA, so it couldn't say anything directly about the ANI and ASI Y-DNA and mtDNA

But it is possible to make some inferences about the ANI and ASI Y-DNA and mtDNA from the model, because autosomal structure often shows strong correlations with Y-DNA and mtDNA markers. That's what I just did in my post above in fact.

zulla said...

@Davidski.

Thanks for the explanation.

Critias said...

Hi David,
How are you sure that including in the possible source populations of your mixture models 3 populations which are close to each other won't have problématic effects on your results ?
Critias

Folker said...

Only my guess, but I think that Yamnaya Bulgaria is a good proxi for the Steppe source of proto-Greeks, if not the source itself (not impossible, but not in the classical theory).
Have you been able to modelise Bul4?

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

You should have used Varna. They had EHG.

truth said...

The Hungary_BA I7043 is Iberian-like , which means too Western, I don't think is a good proxy for the proto-Greek source.

Synome said...

Very interesting.

Regarding the model for the Mycenaean elite, I would expect Balkans_BA to be more concentrated in the elites if there was actually such ancestry present. One might think chariot riding warriors would predominate over earlier "wagoneer" steppe herders. Curious that the Balkans_BA signal disappears.

Anonymous said...

These Mycenaeans were not the Achaeans.
Before the Achaeans in Greece, it is likely there were Luvics.
What does qpAdm show?

Anonymous said...

Great stuff.
In the Minoan/Mycenaean paper, they had to use Armenia to account for the Northeast PCA shift, but now that we know about those CHG-rich Neolithic peoples, the models really show a dramatic increase of Steppe ancestry (~1.5~2 times for Commons and ~3 times for Elites) and seem to be much more right on the money due to geographical proximity.
But neither Mycenaean shows Minoan ancestry, did you use any? Or nMonte decided not to use them of its own?
Also, is there any archaeological record, name or recognition for this 2nd Wave Anatolian Incursion or is this a novel mystery finding? I'll reread the paper to see if there's any mention to that that escaped the eyes.

Steven said...

Why doesn't the Greek language show any resemblance to other Indo-European languages that were spoken by carriers of I2a2 and R1b-L-51 if theses are the haplogroups that brought Yamnaya DNA to Greece?

Ryan said...

@David - "I'm not quite sure about the ASI Y-haplogroups though. We might need a few ASI genomes from Mesolithic India to work that out, simply because of the massive male-biased genetic turnover in South Asia since the Mesolithic."

I think H is a petty safe bet for one of them at least.

Matt said...

Re: I9033, and the Global25 model, I would say retest with simple CHG:Barcin_N:EHG:WHG or simple D stats models, comparing her to other Mycenaeans, see if there's a significant contrast.

I think the G25 is great, but I still struggle to understand how she can obtain the lowest AG3/EHG percentages in Lazaridis's formal stat driven models and have the most ancestry from Yamnaya.

Lee said...

@Davidiski

I am not sure why you are surprised by the cline from Peloponnese Neolithic (Peloponnese_N) individuals to the Bronze Age Anatolians (Anatolia_BA). The recent Mathieson paper in Nature talks about the similarity of these groups. Especially in the supplementary Table 2.


"Krepost Neolithic and Peloponnese Neolithic are both shifted towards CHG and away from WHG, relative to Anatolia Neolithic - a similar pattern to that seen in Minoans"

"The Anatolia Neolithic to Anatolia Bronze Age shift is driven by changes in CHG and Iran Neolithc ancestry not by migration from Steppe populations that have EHG ancestry"

Timing wise it would make sense that a population with CHG like genetics moved out of the Black sea area into the Balkans then later into Crete and Anatolia.

This early movement clearly shows up starting about ~5.5-6 BCE. The timing of this movement remarkably aligns with the climatic cooling that happened around 6200 BCE. A very severe Mini-Ice age This is also co-incident with the flooding of neolithic villages in the sea of azov. Presuming people wer leaving a flooded land and moving to warmer climes it would make sense for them to move away from north of the black sea.

It is nice to see your analysis matching with Mathieson, definitely supports the strengths of the methodology you use.

Lee

Davidski said...

@Critias

How are you sure that including in the possible source populations of your mixture models 3 populations which are close to each other won't have problématic effects on your results?

This is always a risk in these sorts of fine scale analyses, but it doesn't seem to be a problem in my models, because Yamnaya_Bulgaria is chosen as by far the best steppe-derived source, while Hungary_BA totally ignored.

@LeeAlbee

I am not sure why you are surprised by the cline from Peloponnese Neolithic (Peloponnese_N) individuals to the Bronze Age Anatolians (Anatolia_BA). The recent Mathieson paper in Nature talks about the similarity of these groups. Especially in the supplementary Table 2.

Yes, but they talk about CHG influence causing this, and if you look at my PCA, which is somewhat different from the Mathieson et al. PCA, the cline I'm talking about is a very neat one from Peloponnese_N to Anatolia_BA, and doesn't look to be derived from CHG influence per se. Also, only after the dataset was released was I able to confirm that this cline made sense chronologically for the Peloponnese_N samples.

@Romulus

You should have used Varna. They had EHG.

Only Varna_outlier is relevant, and that's basically a steppe sample, and not representative of Varna, who are in the Old Europe cluster in my PCA.

@Steven

Why doesn't the Greek language show any resemblance to other Indo-European languages that were spoken by carriers of I2a2 and R1b-L-51 if theses are the haplogroups that brought Yamnaya DNA to Greece?

The Hungary_BA sample is my third choice, mainly just to make up the numbers. As for Yamnaya_Bulgaria, I'm expecting R1b-Z2103 in such Balkan Yamnaya when more are sampled, aren't you?

@Matt

Re: I9033, and the Global25 model, I would say retest with simple CHG:Barcin_N:EHG:WHG or simple D stats models, comparing her to other Mycenaeans, see if there's a significant contrast.

I think the G25 is great, but I still struggle to understand how she can obtain the lowest AG3/EHG percentages in Lazaridis's formal stat driven models and have the most ancestry from Yamnaya.


I'll have a look later today, though it's possible that this discrepancy is caused by the difference between Yamnaya_Samara and Yamnaya_Bulgaria/Steppe_MLBA.

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

It would certainly be nicest if Proto-Greek were spoken by a later steppe population (Sintashta, etc).

This would make it easier to posit a discrete, later Graeco-Aryan clade or sprachbund, adjacent to pre-proto-Balto-Slavic. Sintashta (etc) = Graeco-Aryan.

If Proto-Greek was instead spoken by much earlier migrants directly to Greece, though, that would mean that Graeco-Aryan has to be projected much further back, right back into the immediate aftermath of PIE, which is geographically and conceptually much messier (for a start: why did such a clear sprachbund develop that exluded the northwestern languages, if all the languages were still being spoken in more or less the same place?). And if - as IIRC is the theory? - we're tracing the Aryan invasions to an originally Corded Ware culture back-migrating to the steppe, but Proto-Greek is directly from the steppe, that make Graeco-Aryan really confusing as an entity.

The earlier date would also, ironically, push us closer to the 'traditional' reconstruction of PIE, based on Ancient Greek and Sanskrit. The later date would allow us to put more of that reconstruction into the age of 'Graeco-Aryan' (/'chariot warriors'), allowing different versions of PIE itself.

The main reason this matters is probably what it says for the stop series reconstructions. Traditionally, PIE is reconstructed with a bizarre voiced aspirate series (but no voiceless aspirates), but the direct evidence for this is basically all from Greek, Armenian and Indo-Iranian (specifically Indo-Aryan). The chariot theory for Greece would let us limit the voiced aspirates to a temporary phase in the Graeco-Aryan sprachbund on the steppe, and would let us have something more like the glottalic theory for PIE itsef; the straight-from-the-steppe theory, on the other hand, makes it much harder to do that, which would make the putative glottalic phase only PRE-pie, if it existed at all.


But of course, what's neatest linguistically isn't always what actually happened...
---------



Could someone clarify something for me? An earlier Mathieson paper talked about the neolithic migration to the peloponnese being distinct from the main migration to the balkans that late produced both the southern/maritime Cardial and northern/continental LBK cultures - Cardial/LBK came from west anatolia, while the peloponnese migration seemed to come from the Levant via Cyprus.

So, is your Peloponnese Neolithic sample, with its CHG, a representative of that southern migration, or are we here talking about a second, LATER migration from the near east into the region?


Also, just to point out the practical use of some of these investigations: if we could establish a CHG-shifted aegean culture, that might be really important for our knowledge of Etruscan. If Etruscan remains showed the same CHG-shift, it would be a huge boost for the theory of an Etruscan migration from the Aegean, and would probably suggest that Minoan, in turn, was related to Etruscan. If, on the other hand, Etruscan remains did not show the same CHG-shift, it would suggest that Etruscans were native to Italy or the Alps (or, a personal pet speculation, Dalmatia?), and that Lemnian represented a later migration eastward - and consequently that Minoan was probably not related to Etruscan.


Am I also right in thinking we don't actually have clearl Hittite/Luwian genetic data yet? That would seem to be the holy grail. If we could have Hittite, Hattian and Hurrian genetic results laid out, that would be so valuable in piecing together the complicated map of who was moving where, and when. [like: did Hittites pick up CHG from Hattians, or were Hittites more CHG-rich than Hattians? the former would suggest an Anatolian migration route from the west, while the latter would suggest a migration route from the east, and probably an earlier date, and would make the idea of PIE being ultimately Caucasian much more attractive...]

Interesting times...

Ariel said...

So I just want to quickly point out that I9033 has virtual SSA caused by post mortem damage. No Minoans or Mycenaeans have that much SSA if any.

Mycenaean:I9033
"Peloponnese_N" 50.9
"Yamnaya_Bulgaria:Bul4" 24.3
"Armenia_ChL" 21.6
"Yoruba:NA18489" 3.2

That said, I tested Minoans and they don't get any Yamnaya Bulgaria, and I think that's really telling.

Minoan_Lasithi
"Peloponnese_N" 68.95
"Anatolia_BA" 24.05
"Armenia_ChL" 7
"Yamnaya_Bulgaria" 0

Also, I tried a lot of combinations (like 50), and Minoans always get less Armenia CHL, Armenia MLBA, CHG than Mycenaeans. I understand that the fact that Mycenaeans have extra steppe could influence the results giving a more marked "caucasian shift". But Mycenaeans are absolutely packed with ancient Armenian admixture, so much so that Yamna and EHG dissapear if you also had Amernia MLBA and Armenia CHL.

[1] "distance%=4.5218 / distance=0.045218"
Mycenaean:I9033
"Peloponnese_N" 40.05
"Armenia_ChL" 31.5
"Hungary_BA" 19.55
"Armenia_MLBA:I1656" 8.9
"EHG" 0
"Yamnaya_Samara" 0

So I do think that Mycenaeans have some significant amount of EHG (5-10%) and steppe admixture that came via the balkans, but I fell like there is also additional influence from the "east" when you compared them to Minoans. Also Yamnayas from Bulgaria have their fair share of Armenia CHL, that's probably why are such a good fit for Mycenaeans.

[1] "distance%=3.3991 / distance=0.033991"
Yamnaya_Bulgaria
"Hungary_BA" 35.95
"Yamnaya_Samara:I0231" 33
"Armenia_ChL" 27.75
"Narva_Estonia" 3.3

Rob said...

@ Ariel

Yes, Yamnaya Bulgaria has itself significant west Asian admixture; of a different character to Minoans
Not sure what it represents entirely, apart from an obvious migration to Greece and Thrace , but drops precipitously in West balkans

Ariel said...

*I removed Hungary BA and Mycenaeans show some Yamnaya even with Armenia CHL and Armenia MLBA.

[1] "distance%=4.6337 / distance=0.046337"

Mycenaean:I9033
"Peloponnese_N" 53.75
"Armenia_ChL" 32.5
"Yamnaya_Samara:I0231" 5.4
"Armenia_MLBA:I1656" 5.2
"Narva_Estonia" 3.15

Davidski said...

@Rob

Yes, Yamnaya Bulgaria has itself significant west Asian admixture; of a different character to Minoans

Yeah, it's called Yamnaya ancestry from the steppe.

Ron said...

Do people from India have Anatolian Neolithic admixture besides Iranian Neolithic and steppe?

Lee said...

@Davidiski

"Yes, but they talk about CHG influence causing this, and if you look at my PCA, which is somewhat different from the Mathieson et al. PCA, the cline I'm talking about is a very neat one from Peloponnese_N to Anatolia_BA, and doesn't look to be derived from CHG influence per se. Also, only after the dataset was released was I able to confirm that this cline made sense chronologically for the Peloponnese_N samples."

Fair enough. Mind you looking deeper at the data-a direct link from the Peloponnese_N to Anatolia_BA seems unlikely. The CHG content in Anatolia_BA is much higher than that in the Peloponnese_N. More likely the CHG content comes from a common or closely related source.

Ariel said...

Davidsky

Look at your PCA, Yamna Bulgaria is clearly CHG shifted. It's a huge outlier...

Folker said...

@vacuouswastrlel
Yes, if Yamnaya from Bulgaria (and Thrace, same population) is ancestral to proto-greek, and since it appears archeologically before 3000BC, it means a revision of the IE tree, as Greek would be one of the 1st branch which separated to PIE. That's mainly why I'm favoring instead an identification with the Anatolian branch. But we will see.
We don't have any sample for Hittite, Palian or Luwian. Yet. Neither a Hattian one. But we have some samples from BA Anatolia which could be good proxi for Hattian. So probably already CHG rich.

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

https://ibb.co/duWSdS

Look at how the Myceneans compare to Varna on genetiker's k14. They have less EHG than Varna. Even the Myceneans with the most EHG has less than Varna. As well it's all around 5%.

Davidski said...

@Ariel

Look at your PCA, Yamna Bulgaria is clearly CHG shifted. It's a huge outlier...

Well that's an interesting interpretation of reality, but alas, it's wrong. See here for a visual explanation of what is really going on...

https://drive.google.com/file/d/1q414FA9hrRECfO33dXd3qEKXc8eUZIve/view?usp=sharing

Davidski said...

@Matt

Here are a couple of qpAdm models comparing I9033 to the other Mycenaeans, with as many of the usual outgroups as possible while retaining over 100K SNPs.

https://drive.google.com/file/d/1tu8T76sSKXKZ9c4IoRBsEOaIvBd5k0T9/view?usp=sharing

https://drive.google.com/file/d/1S1t9RiSNpRnz9FhCfKfdsoUkl_5QMYXp/view?usp=sharing

The standard errors cover the difference, but I9033 always shows this sort of shift towards steppe and steppe-related samples, both in my PCA (I9033 is the most steppe-shifted Mycenaean in my PCA above), and in mixture models using formal stats and PCA data as input.

So something's going on, even if the difference is, strictly speaking, not easy to discern.

Michalis Moriopoulos said...

I favor the first option: an EBA origin for the Proto-Greeks. It's nice to see it supported in Dave's model. I'd like to see remains from throughout the Early Helladic period tested as I think that will yield the most promising fruit.

If we found, for instance, that remains from Early Helladic I resembled Neolithic Greeks and Minoans, but remains from EHII or EHIII had significantly more steppe inpuT (presumably more than the Mycenaeans), that would be a major discovery.

Dr. Lazaridis is surely aware of how important both the early Greeks and Hittites are to understanding the spread of IE. If we're lucky maybe we'll get a follow-up paper soon with more samples from both Greece and Anatolia.

Ariel said...

I don't get any Armenia (0%) with central Euro Beakers and I get (a lot) Armenia with Yamna Bulgaria, also that strait line would imply that Yamna flew from the steppe to Bulgaria without mixing with any of the people in between (WHG/SHG shifted pops that will make the position in the PCA just impossible). But I challenge you to model Yamna Bulgaria as just Balkan CHL and Yamna, you just can't do it .

Yamnaya_Bulgaria
"Yamnaya_Samara:I0231" 43.6
"Balkans_ChL:I2425" 27.7
"Armenia_ChL" 24.4
"Loschbour:Loschbour" 4.3

Beaker_Central_Europe
"Yamnaya_Samara:I0231" 46.25
"Balkans_ChL:I2425" 30.75
"Iberia_ChL:I1277" 17.7
"Loschbour:Loschbour" 5.3

Also if I remove all the armenians I still get 10% extra CHG that I don't get with Beakers (0%).

Yamnaya_Bulgaria
"Yamnaya_Samara:I0231" 46.75
"Balkans_ChL:I2425" 40.4
"CHG" 9.8
"Loschbour:Loschbour" 3.05

Beaker_Central_Europe
"Yamnaya_Samara:I0231" 46.25
"Balkans_ChL:I2425" 30.75
"Iberia_ChL:I1277" 17.7
"Loschbour:Loschbour" 5.3

Ariel said...

Even better example

Yamnaya_Bulgaria
"CWC_Baltic" 45.05
"Balkans_ChL:I2425" 22.35
"CHG" 14.2
"Yamnaya_Samara" 13.3
"Anatolia_BA:I2683" 5.1

Beaker_Central_Europe
"CWC_Baltic" 30.5
"Yamnaya_Samara" 28.75
"Iberia_ChL:I1277" 21.35
"Balkans_ChL:I2425" 19.4
"CHG" 0!

Rob said...

Dave can you list all the sources you included in the above result ?

Anonymous said...

Modelling Yamna_Bulgaria as Yamna_Ukraine + Balkans_Chl + IronGates_HG yields the best results for me:
"distance%=4.3806"
Yamnaya_Bulgaria
Yamnaya_Ukraine,62.8
Balkans_ChL,33.4
Iron_Gates_HG,3.8
I tried adding Anatolia_BA, Anatolia_Chl and various CHG enriched European samples like Peloponnese_N_Outlier and Krepost_N and they don't work very well.
So I agree with David, there doesn't seem to be additional CHG in Yamna_Bulgaria.

Ariel said...

Lauχum

Try to put an actual CHG population or armenia, why not CHG themself, come on..

Ariel said...

Yamnaya_Bulgaria
"Barcin_N" 38.55
"EHG:I0061" 30.75
"CHG" 25.75
"WHG:I1875" 4.95
"Natufian" 0
"Iran_N" 0

Beaker_Central_Europe
"Barcin_N" 45.5
"EHG:I0061" 30.4
"CHG" 12.45
"WHG:I1875" 11.65
"Natufian" 0
"Iran_N" 0

Davidski said...

@Ariel

Quit being so naive.

Here's a formal model of Yamnaya_Bulgaria as a two-way mix between Balkans_ChL and Yamnaya_Samara. Tail = 0.89! Low std. errors. So it works just fine and makes perfect sense.

https://drive.google.com/file/d/1da2Uxt_rsjrfyY7vYdLhr2xKGmI94WWD/view?usp=sharing

Like I showed on the PCA plot, this is what Yamnaya_Bulgaria basically is.

@Rob

I did list all of the sources. I was testing a very specific model from the PCA, which is actually supported by formal stats modeling, so I don't need to use more sources to complicate the model.

Ariel said...

Wrong. Why use Iron gates,Yamnaya Ukraine and Balkan BA? ALl those sample have some EHG and sucks from Yamna, is not armenia CHL fault, try again. Use Loschbour, Yamnaya SAMARA and Balkan CHL, and don't forget CHG or armenia CHL.

Ariel said...

Davidski

Ok, it works. But not convinced yet. The question here if additional CHG will improve the fit.

Anonymous said...

@Ariel
whoops, misread and accidentally added BA instead of Chl. Now with Balkans_Chl it doesn't work.
I used Yamna Ukraine, Iron Gates and Balkans Chl because they are more relevant populations. Yamna_Bulgaria is most likely derived from Yamna Ukraine not Samara. Iron Gates is a more regional WHG pop as well.

Anonymous said...

So Using Armenian Copper/Bronze Age and Anatolia Copper/Bronze Age populations significantly worsens the fit. Whilst the model of Yamna + Balkans_Chl works rather well.

Davidski said...

@Ariel

I don't need to improve the statistical fit, because I'm getting basically the same successful model with Global25/nMonte and qpAdm.

[1] distance%=4.008

Yamnaya_Bulgaria

Yamnaya_Samara,58.4
Balkans_ChL,41.6

...

Yamnaya_Bulgaria
Yamnaya_Samara 0.573
Balkans_ChL 0.427

chisq 8.784 tail 0.888551

Davidski said...

But this is pretty funny. Adding Armenia_EBA to the qpAdm model just blows it up.

https://drive.google.com/file/d/1MZaildIC0ZBozse2azJ4pIKNnIm2gTVe/view?usp=sharing

Anonymous said...

@Ariel
This here you made caught my attention:
Yamnaya_Bulgaria
"Barcin_N" 38.55
"EHG:I0061" 30.75
"CHG" 25.75
"WHG:I1875" 4.95
"Natufian" 0
"Iran_N" 0

You see, you have to look beyond. What this model is saying to me is:
1. There's an EEF (Barcin_N 38.55% + WHG 4.95%)
2. There's an Yamnaya (CHG 25.75% + EHG 25.75%)
3. There's extra EHG, probably Narva (EHG 5%)

Of course, generalizations, but you got the idea.
Also, as many here are reporting, I've been running loads of models with these new CHG-rich peoples, and the Steppe dwellers always reject it (Yamnaya and CW).
The dates are vastly different as well. This CHG-rich people incursion must have been very small and diluted with time when they settled in Europe (we can see older samples with this CHG ancestry, but very reduced, almost to the point of being noise).

Ariel said...

Davidsky

We were talking about Armenia CHL, or CHG, you used armenia EBA

distance%=3.4477
Yamnaya_Bulgaria
"Yamnaya_Ukraine" 41.05
"Armenia_ChL" 27.65
"Balkans_ChL" 23.05
"Iron_Gates_HG" 8.25
"Barcin_N" 0

Ariel said...

Namedguest

Yes but BB Central Europe and CWC don't get much CHG in very same run

Davidski said...

I don't see what significance Armenia_ChL plausibly has for Yamnaya_Bulgaria? Just a whole bunch of indirectly shared ancestry.

Doesn't work anyway.

https://drive.google.com/file/d/1ng2JBC-SAY1ZvTc_1mM9KyL5mSrdQ7ku/view?usp=sharing

Davidski said...

@vacuouswastrel

So, is your Peloponnese Neolithic sample, with its CHG, a representative of that southern migration, or are we here talking about a second, LATER migration from the near east into the region?

Peloponnese_N is only partly of early Neolithic European origin, and the younger the samples, the more Caucasus-shifted they are. So we're definitely looking at fresh migrations from the Near East into Southeastern Europe, especially the Aegean, during the tail end of the Neolithic, and continuing well into the Bronze Age.

This is no doubt how the Minoans came to be.

If Etruscan remains showed the same CHG-shift, it would be a huge boost for the theory of an Etruscan migration from the Aegean, and would probably suggest that Minoan, in turn, was related to Etruscan.

I have a feeling that Etruscans will indeed show elevated CHG, and might actually look similar to Beaker_Sicily.

http://eurogenes.blogspot.com/2018/03/on-origin-of-steppe-ancestry-in-beaker.html

Am I also right in thinking we don't actually have clearl Hittite/Luwian genetic data yet? That would seem to be the holy grail. If we could have Hittite, Hattian and Hurrian genetic results laid out, that would be so valuable in piecing together the complicated map of who was moving where, and when. [like: did Hittites pick up CHG from Hattians, or were Hittites more CHG-rich than Hattians? the former would suggest an Anatolian migration route from the west, while the latter would suggest a migration route from the east, and probably an earlier date, and would make the idea of PIE being ultimately Caucasian much more attractive...]

No remains belonging to actual or even suspected Anatolian-speakers have been tested yet at any lab, as far as I know.

But I'm pretty sure that a paper on the Hittites and Hattians will basically be a replay of the paper on the Mycenaeans and Minoans, with a clear signal of steppe ancestry in the Hittites, but not the Hattians. And then there will be all sorts of debates about how much steppe is enough steppe.

Arza said...

@ Davidski
But this is pretty funny. Adding Armenia_EBA to the qpAdm model just blows it up.

Nope, nothing exploded here. Fit is better and it tries to say that Balkans_ChL is too much Armenia_EBA-shifted. Something from behind Balkans_ChL (going from Armenia_EBA direction) will work better.

Samuel Andrews said...

Rob, steppe admixture shows up big in iron age Bulgaria and the Mycenaean from Crete.

Chalcolithic, Eba Bulgaria had too little steppe on average to make Mycenaeans normal for Balkans going back thosends of years.

Also, a route with ehg from Armenia to Greece looks unnecessary considering some kind steppe movements hit the Balkans around the time of myceneans and of course the rest of Europe.

I expect more genomes from ancient Greeks, including myceneans, to show 20-30% steppe admixture. The Mycenaean from Crete with 30% steppe can't be explained away as a random immigrant or daughter of an immigrant. She's very significant to the kitten question. She indicates some early Greeks had significant steppe ancestry like the iron age, Thracian(?), individual from Bulgaria.

Davidski said...

Renfrew was wrong for so many reasons, that even 50% EEF admixture in western Yamnaya won't mean anything.

Folker said...

@Ariel
Speaking archeology, Yamnaya from Bulgaria and Thrace were culturally influenced by locals.

https://www.researchgate.net/publication/285673711_Pit_graves_in_Bulgaria_and_the_Yamnaya_Culture

Mind that the Turkey part of Thrace remains blank in this paper, but:

http://www.hurriyetdailynews.com/tomb-in-istanbuls-silivri-years-biggest-archaeological-discovery--99235

And this quote:
“Thrace received migrations from the north. This is a kurgan-style tomb and such tombs exist in my studies, too,” said Istanbul University Archaeology Department Prehistoric Department Professor Mehmet Özdoğan. “I know that lots of kurgan tombs have been destroyed in Thrace. We have rescued one of them from the digger. But this tomb is older and is from the Bronze Age. It is a very important discovery. I believe scientific examinations will lead to interesting results.”

Rob said...

@ Sam

“Chalcolithic, Eba Bulgaria had too little steppe on average to make Mycenaeans normal for Balkans going back thosends of years”

Well actually, the Varna Outlier has more steppe than any of the Myceneans , and where there’s smoke there’s fire
But I’m not suggesting there wasn’t further migration later ; but until we get more samples, the exact scenario is stil unbeknownst .


*”The Mycenaean from Crete with 30% steppe can't be explained away as a random immigrant or daughter of an immigrant. She's very significant “

Incorrect
She’s not random but of course she could be an immigrant - that’s what the data in fact shows.
Mobility intermarried other mobility far away

I’ll get back about the Thracian

Rob said...

^ nobility

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

The same cannot be said about steppe proponents, because R1a-M417 and R1b-M269 are from the steppe.

The details concerning female mediated admixture into the early Indo-Europeans, like the EEF in Sintashta, are interesting, but they can't overturn the narrative.

Alberto said...

I'm all for testing different options with the samples we have in addition to what was tested in the paper. But I also think we should point out what's very speculative from what's realistic. So keeping the paper as a baseline of reality might help here a bit.

Table S2.2, with Mycenaean individual and more distant sources:

https://drive.google.com/file/d/1sBUIpNDyizgWtCCJV95isRyPf28N9tAN/view?usp=sharing

Table S2.26 with Mycenaean average and closer sources:

https://drive.google.com/file/d/1r_ynn46zlDBbiTe-qPexD3NmPJ0Ea4LB/view?usp=sharing

This is from Global 25 adding enough realistic sources but avoiding the difficult to interpret Balkans_BA and Hungary_BA. Some might say that's overfitting, but for me it's more realistic with this late samples with complex admixture to give more choices than to force a specific model (more so when it agrees with the simple models from the paper):

https://drive.google.com/file/d/1wMaMV1XttNlVyFTyRbsvkSVnw9vfR392/view?usp=sharing

We can all agree that 4 samples are not enough and in no way definitive. And then we can agree to speculate to our heart's content about what future samples will show. But we should also agree on the simple reality that these 4 samples and the single Y-DNA we have for now clearly give a very weak genetic support for Mycenaean Greeks (or their language) coming from the steppe.

Davidski said...

@Alberto

Mycenaeans clearly have ancestry from the steppe, while Minoans don't. This was the nightmare scenario for you guys looking for an alternative, any alternative, to the steppe theory.

Well, it came true, so it's time now to shift the goal posts. But please keep in mind that you can only keep shifting the goal posts so much until you run out of space.

Rob said...

@ Dave

Your results show Myceneans are 15-30% Yamnaya Bulgaria which means 7.5-15 % Samara (and one is ~ zero); which is what what I was suggesting (and others) which you had issues with.
The details remain.

More generally, modelling Yamnaya Bulgaria with Yamnaya Samara or CWC shows shared ancestry, because they are contemporaneous. Its also more mixed: Hungarian bronze age is a checker bored - some are EEF, some very steppe - all in the same settlements. So let's look at theoretical exercise using these sources, c. 4000 BC, when blocks were more homogeneous:
Balkans Chalc, Boleraz, Tisza, Khvalysk, Ukraine Eneolithic (non-outlier), Arm Chalc, etc - ie groups in the immediate surrounds likely to have mixed, and existed c. 45 - 3800 BC, prior the Yamnaya horizon and contemporary groups in SEE. Narva were included (because they existed c. 4000 BC, and there for a source of late HG admixture. I did not include CHG on the theoretical premise that a 'pure' CHG population did not exist, even if there were CHG-rich groups.
https://imgur.com/a/BD25f



Yamnaya_Samara
Samara_Eneolithic:I0122 72.25 %
Armenia_ChL 27.75 %
Balkans_ChL 0 %
d 0.5%

Yamnaya_Ukraine
Samara_Eneolithic:I0122 69.75 %
Armenia_ChL 30.25 %
Balkans_ChL 0 %
d.05%

Yamnaya_Bulgaria
Ukraine_Eneolithic:I4110 41 %
Armenia_ChL 37.55 %
Samara_Eneolithic:I0122 11.15 %
Anatolia_ChL 10.3 %

Vucedol:I3499
Anatolia_ChL 35.5 %
Tiszapolgar_ECA 27.2 %
Ukraine_Eneolithic: 17.05 % + Samara_Eneolithic:I0122 11.35 %
Greece_Peloponnese_N 8.7 %


Remedello_BA
Tiszapolgar_ECA 90.65 %
Protoboleraz_LCA 8.85 %
Balkans_ChL 0 %


Sam,
Balkans_IA:I5769
Greece_Peloponnese_N 55 %
Anatolia_ChL 19.3 % + Armenia_ChL 5.5 %
Ukraine_Eneolithic:I4110 14.4 % + Samara_Eneolithic:I0122 4.4 %


Rob said...

@ Alberto

"...avoiding the difficult to interpret Balkans_BA and Hungary_BA. "

Yeah, just said that too, a multitude of groups.

For Myceneans & Minoans, I think it's can be communicated thus:

https://imgur.com/a/BX3J8

It overall shows an Aegean origin of Minoans and circum-Pontic origin of Myceneans.

Alberto said...

@Davidski

Mycenaeans clearly have ancestry from the steppe, while Minoans don't. This was the nightmare scenario for you guys looking for an alternative, any alternative, to the steppe theory.

It's funny how everyone makes the necessary assumptions to convince themselves that the data proves them right. I guess that's why Gioiello keeps thinking with every new paper that he's winning all his battles against his enemies. With every new paper, everyone was right!

I can imagine the linguists who have dedicated half of their lives to study the Minoan inscriptions and argued for it to be an Indo-European language how happy they must have been when the ancient DNA showed Minoans to be 80% similar to Mycenaean Greeks. Not only they now have the strong cultural links, but also a very strong genetic support (apart from their own research in linguistics). No one will convince them that Mycenaeans showing 5-10% Yamanaya admixture that is lacking in Minoans is a strong argument against both populations speaking closely related languages.

I personally have nothing to change about my goals. When those samples proving that Mycenaean elites came from the steppe come, I will still be able to stand by my assertion that the 4 current samples show very weak genetic support for a cultural/linguistic link to the steppe. That's not going to change. It will be those new samples that will seal the deal, not these ones.

Not that I think it's likely that those samples will ever show up, but I don't care much either way.

zulla said...

Sorry, unrelated topic again. I'm trying to understand something.

Basu et al (2016) model North Indian Khatris as 97.9% ANI, 1.5% ASI and the rest Austroasiatic and Tibeto Burman. As I understand, this is significantly less ASI than some people here are willing to accept. So what is wrong with Basu's model?

Anonymous said...

Alberto said...
"I can imagine the linguists who have dedicated half of their lives to study the Minoan inscriptions and argued for it to be an Indo-European language"

There is no such who would have claimed it. Therefore, there is no one to shake his hands. On the contrary, it has been clearly proven that the Minoan language is not Indo-European.

Tested samples fundamentally re not the Achaeans, they specifically chose that they were not the Achaeans. It is their slaves, except one, which belongs before the Mycenaean Royal dynasty linkage to Minoans, but still in a culture where nothing is Achaean.

Al Bundy said...

@Rob I might have asked you this before, but when is Greek actually supposed to be in Greece?Is there some kind of consensus, or is it all over the place?PIE would be 5000, IndoIranian 4500, GraecoArmenian 4000 or around there are some sensible split times I've read.I know some who argue those dates are way too late.Again, Myceneans were expected to have some steppe, but is it enough for them to have been IndoEuropeanized?Greek is much closer to IndoIranian than anything in Europe.

Al Bundy said...

@Alberto I'm just assuming Minoan is not IE I haven't kept up with that.I agree that the Mycenean paper's results were really expected and we'd have to wait for more to clarify things.

Anonymous said...

@Alberto

If it can be proven Linear A was non-IE we have a documented case where the arrival of an IE language and a steppe component is clearly linked, which would pretty much be solid proof for me.

Arch Hades said...

Balkans_BA I2163 is too late to be Proto Greek.

Anyway, like Mike The Jedi, I agree with the first premise. That Yamnaya outlier from Bulgaria could have been a Proto Greco-Armenian speaker. Or maybe even Proto Paleo-Balkan [which would include Thracian and Illyrian as well].

But if you look at Matheison et al 2017 [the genetic history of Southeastern Europe] there was significant steppe ancestry in the Balkans even earlier dated to the Chalcolithic. So steppe ancestry was making it's way to Southeastern Europe and the Southern Balkans to a very early date. So going by the steppe hypothesis it's possible even IE came to Southeastern Europe even earlier than the time of the Yamnaya culture.

Al Bundy said...

I understand the interest because if Greek didn't come from the PC steppe then neither did PIE.If Greek did come from the steppe, which of course it might have, IndoIranian must have too.

Simon_W said...

@ supernord

Are you suggesting that the Mycenaeans were not Greek speaking? That they spoke Luwian instead? That would be crazy, because the Mycenaeans had a script and they wrote in an archaic form of Greek. Not Luwian.

Arch Hades said...

Supernord.

Uhh there have been plenty of scholars to propose the hypothesis that Linear A represents an Indo-European language.

https://en.m.wikipedia.org/wiki/Linear_A

Please stop asserting as fact what nobody actually truly knows at the moment. To assert as fact what you actually don't know is a form of dishonesty.

Rob said...

Well dave mentioned more Mycenean shaft grave DNA will be out shortly, and with Folker & Epoch Setting out the final proofs of their Minoan inscriptions thesis, lingering questions will be answered imminently

Al Bundy said...

@Rob Great

Simon_W said...

I'm going to lean out of a window now, but at the moment I don't quite believe that Etruscan came from the Aegean. At least it seems rather likely to me that the Villanovan culture of the early Iron Age (starting around 900 BC) was already Etruscan.

Check out the huge overlap between the Villanovan culture (dark shaded grey, on the right) and the Etruscan language area in northern Lazio (on the left, in yellow):

https://justpaste.it/1hz7m

And in fact the Villanovan culture also spread to parts of Campania. And guess what? Etruscan was also spoken in these Campanian places, see N2 in the right map:

https://justpaste.it/1hz7o

So if the Villanovans were not Etruscans, then the latter had an uncanny propensity to only settle areas where the Villanovans had been.

Moreover what else should the Villanovans have been, if not Etruscans? Umbrians is often heard as an alternative. But the distribution area of actual Umbrian inscriptions has very little overlap with the Villanovan area. In fact, the main concentration of Umbrian inscriptions (purple dots) lies clearly to the east of the Villanovans:

https://justpaste.it/1hz7p

So there seems to be something wrong with the idea that the Etruscans arrived with the Orientalizing influence that started around 720 BC. The latter wasn't restricted to later Etruscan areas anyway, Oriental influence is also found in Novilara for instance.

Moreover there is no evidence for war, violence and conquest at the onset of the Orientalizing phase, and no change in the settlement structure. If these Orientals were the real Etruscans, then they assimilated the Villanovans amazingly peacefully. That seems strange.

So the Villanovans seem to have been Etruscans. And the most obvious origin of the Villanovan culture is the Protovillanovan culture of the Final Bronze Age, with links to central Europe, where the related Raetic language is attested. Granted, the Protovillanovan culture was much more widespread than the Villanovan culture, it spread all over Italy. But its influence was nowhere deeper and more lasting than in the Villanovan culture - hence the similarity in the names. If the Protovillanovans should have been Italic rather than Etruscan - wouldn't it then be strange that their impact was stronger in the later Etruscan area than in the Italic area?

It's also striking how pottery like this bowl from early Iron Age Umbria

https://justpaste.it/1hz8q

is reminiscent of similar pottery from the LBA Subapennine culture:

https://justpaste.it/1hz8s

The latter predates the advent of the Protovillanovans.

Lee said...

Is it at all troubling to anyone that R1a1b1a2 is found in Iberia well before the Steppe migrations were supposed to happen?

I0410 Iberia_EN at around 7 kya before present or ~ 5000 BCE.

So....

Back to the early expansion of CHG-like heritage across Southern Europe? But the Dstats for Iberia_EN and ChL seem to indicate that no CHG was present in these populations?

IF not Steppe or CHG-like is Southern Europe/Iberia/Italy origin of for that form of R1b??

Was this discussed when the preprint came out and I just don't remeber?

Sincerely,

Lee


Lee

Davidski said...

@ Lee Albee

Is it at all troubling to anyone that R1a1b1a2 is found in Iberia well before the Steppe migrations were supposed to happen?

I'm pretty sure there's no R1a in any ancient Iberian remains tested to date. You must be thinking of the R1b(xM269) lineages that have popped up in one Neolithic and a few Beaker Iberians who don't show any steppe ancestry.

But these lineages are either confirmed or very likely to be R1b-V88, and there's plenty of R1b-V88 in hunter-gatherer samples from Eastern Europe without steppe ancestry.

So R1b-V88 at some point spread across Southern Europe along with WHG or EEF groups, as far as Iberia, and then into Africa. On the other hand, steppe ancestry moved out of the western steppe much later, mostly along with R1a-M417 and R1b-M269.

Rob said...

@ SimonW

And the the proto-Villanovan was essentially an offshoot of Urnfield ?

Anonymous said...

@Simon_W


Before Mycenaean. But even upon Herodotus in Greece spoke a non-Greek languages that Herodotus described. There remained Luviс place names.

@Arch Hades

After all Georgiev assumption that this Indo-European language was rejected, serious linguists have not carried on any linguistic imagination unknown visionaries. It's ridiculous when such nonsense fall into English Wikipedia, because such nonsense in any language of the world are printed mass.

Anonymous said...

For example, despite the archaic nature of the language and the monstrous distortion of the language completely unsuitable for the transmission of the Indo-European language, the language of Linear B was recognized immediately, it turned out to be an extremely distorted unknown dialect of the Greek language. For Linear A nothing like this happened, although in the mid-20th century all linguists were actively looking for these matches. The structure of writing and phonemic composition of linear A and B has nothing to do with any even hypothetical Indo-European language, it is impossible that it came up with at least one of the Indo-Europeans. And linguist freaks cause only laughter.

Archaelog said...

I think it is becoming more and more clear with each published paper that R1a, R1b and even I2 could have existed across a wide region from the Pontic-Caspian steppes to the northern forest cultures. Just because the elite graves associated with a particular time-period show only one of these Y haplogroups it doesn't mean the others were absent at the time. They all had a similar genetic structure and very possibly related linguistic-cultural elements. The forest steppes immediately to the west of Khvalynsk and Samara could have spoken a form of PIE for all we know which then expanded with the CWC. IMO there is no more reason to equate haplogroups to languages. Geographically and culturally closely related communities often become part of a linguistic culture. In the case of the steppe cultures, they were genetically related too. (occurrence of R1a and R1b in Khvalynsk with pretty much the same admixture)

Anonymous said...

@Rob

I was neither promoting a thesis nor a prediction or anything of the likes. I merely answered to Alberto's idea that such a small steppe admixture in Mycenaeans didn't proof a lot, pointing to the fact that this could be settled by other finds than just more samples.

Mind you, I actually liked his remarks that every side of this debate seems to think new papers strengthens their position.

Lee said...

@Davidiski

Sorry that was a typo on my part Iberia_EN I0410. it was listed as R1b1a(xR1b1a1a2)

so that is R1b-V88?

Are you sure? I get autosomal data, and mtDNA data. The y chromosome haplotypes always trip me up for some reason.


Who retested it? Do you have the source?

Thanks

Lee

Arch Hades said...

So when are these new Mycenaean samples supposed to be released? I've also heard there's going to be some classical era Greek genomes from Ambracia to be released soon too.

Anthro Survey said...

@Simon_W

It's not a far-fetched idea that Etruscans were largely descendants of the Villanovan culture. If that was the case, it still doesn't contradict the notion of padanic Villanovans speaking IE and giving rise to IE-speaking Italic cultures, though.

Perhaps Etruscans of Tuscany were analogous to Basques: Local retention of a non-IE language despite a good chunk of Central European influence.

As for Anatolia_BA-like influence: indeed, that was probably widespread all across Italy at the time and dates back much earlier than Etruscan times. I don't suspect that Etruscans shad any more Anatolia_BA-like ancestry than Picenes, for example.

Anthro Survey said...

@Arch Hades

"So steppe ancestry was making it's way to Southeastern Europe and the Southern Balkans to a very early date. So going by the steppe hypothesis it's possible even IE came to Southeastern Europe even earlier than the time of the Yamnaya culture."

Yeah, this is also what I suspect sometimes. Though, they may have been speakers of a more basal language, ancestral to Hittite and such, as opp to Greek. The terms Suvorovo, Novodanilovka, Cernavoda are relevant here.

Ever seen this diagram?
https://www.researchgate.net/figure/The-first-migration-4200-4000-bc-of-the-Suvorovo-type-immigrants-into-the-Danube-v_fig2_270961451

Davidski said...

@Lee Albee

You should read the discussion about how different types of R1b relate to the spread of steppe admixture in the Beaker paper (Olalde et al. 2018).

Davidski said...

@Chetan

Corded Ware spread into the forest steppe, but the ancestors of Corded Ware came from the steppe. Hence the high similarity between Yamnaya and early Baltic Corded Ware, including in their levels of Caucasus ancestry.

And of course the oldest R1a-M417 sample to date does come from the steppe. See map here.

https://eurogenes.blogspot.com/2017/12/corded-ware-as-offshoot-of-hungarian.html

Davidski said...

@Alberto

Not that I think it's likely that those samples will ever show up, but I don't care much either way.

Those samples will show up because this area of science and technology is improving rapidly, so the data will come thick and fast and there won't be anywhere to hide from the facts.

And you do care, because if you didn't, you wouldn't have posited various desperate theories on the topic, like the one about the Georgian-like people migrating to the Baltic and mixing with SHG to form Corded Ware.

I told you Corded Ware and R1a-M417 came from the steppe. But you wouldn't have any of it, because you do care, and you obviously want the steppe hypothesis to fail.

I'm totally confused by this position of yours, because it makes no sense to me to want something to fail so badly, despite the fact that it's on such solid scientific grounding and might well turn out to be true, but that's your business.

Michalis Moriopoulos said...

Gimbutas herself was clearly no fan of steppe people, but she had the maturity to report the facts as she saw them. If anything she had a bias toward the "Old Europeans," whom she regarded as more egalitarian or even matriarchal, in contrast to the patriarchal kurgan warriors who ostensibly destroyed the Eden that was pre-IE Europe. She's been criticized heavily for this rather rosy view of Old Europe, of course, but I think her example is instructive: You don't have to like that something happened to accept that it happened.

Hopefully in a couple of years the fine details of IE's origin and spread will be settled to every reasonable person's satisfaction. I don't think it would be too optimistic to see a consensus on that form by 2020, if the relevant samples keep rolling in.

Once this happens, I'm curious how your focus here might shift, Dave. Will you try your hand at Afro-Asiatic, the other major West Eurasian language family? Its origin is also very controversial, so you'd probably have a blast arguing with people about it.

Davidski said...

@Mike

I don't know much about Afro-Asiatic, there are few people out there who do, even in academia, so it's not an easy topic to base a blog like this on.

But two years is a long time, especially considering how fast things are moving in the field of ancient DNA. I'm sure there will be several hot topics to cover once the PIE question is solved to most of our satisfaction.

Ric Hern said...

Yes I wonder how much R1b(V88) contributes towards the formation of Afro-Asiatic especially the Chadic Branch...Could there be a detectable Mesolithic European language substrate in both Indo-European and Afro-Asiatic ?

Folker said...

@all
I am not a linguist, and I have only limited interested for linguistic. It happens I studied perhaps 6/7 languages in my life, in majority IE languages, but also Japanese (and since Japanese is including some Chinese, some Chinese as well). For most people who have done the same, it's clear how languages in the same family can be related.

Now, about Minoan and Linear A. If Minoan was written in Linear A (and nobody is seriously contesting it), it means that by a very large consensus, Minoan was not an IE language. Why is the consensus so large?

Because we have enough writings in Linear A to find some clues about its morphology, and of its phonology by similarities with Linear B (who was writing an archaic form of Greek).

One of the major trait of the language written in Linear A is the fact it was an agglutinative language.
Agglutinative, means that the morpheme are easy to identify, and each morpheme has an unique fonction. Examples of agglutinative languages are Japanese (and Korean), and Turkish languages
IE languages (with the exception of Armenian) are not agglutinative, and never were. They are fusionnal languages (cf wiki: "Fusional languages or inflected languages are a type of synthetic languages, distinguished from agglutinative languages by their tendency to use a single inflectional morpheme to denote multiple grammatical, syntactic, or semantic features")

Please note that Ancient Armenian was not an agglutinative language, but became so since then, probably by influence of a substrate and neighbouring languages, some of them being agglutinative.

So, it is completely excluded that PIE was agglutinative, and any agglutinative language existing around 2000 BC could simply not be derived from PIE.

Why are we so sure Linear A wrote an agglutinative language? Because with statistics it is possible to identify some morphemes used. Not their meaning, but it is enough to identify general traits of the language.

Archaelog said...

@David Yes we have the oldest R1a-M417 from Alexandria but there is a contesting hypothesis that Northern Pontic/ Ukraine Eneolithic cultures were not IE. Like Sredny Stog and Alexandria. Do you think the Alexandria R1a guy could have been non IE? Just because he was not part of the Yamna culture. Would love to hear your opinion on this.

Also we have both R1a and R1b samples from Khvalynsk which is believed to be precursor to Yamna.Also from Ukraine_Mesolithic. That shows that from an early date, these two haplogroups could have been distributes across the Pontic-Caspian steppe and maybe even the forest steppe. The fact that the Yamna samples were all R1b could mean that an R1b lineage formed the elite in that time period.

I don't trust any claims of a purely R1b Indo-European and an R1a Uralic community as some people just keep on making.

Anonymous said...

@David

"I'm totally confused by this position of yours, because it makes no sense to me to want something to fail so badly, despite the fact that it's on such solid scientific grounding and might well turn out to be true, but that's your business."

What surprises me is that a certain group of people claim that the Steppe Hypothesis is dead, or nearly dead. Or some of the posters here that will literally go for anything but the Steppe Hypothesis. The first is clearly not true, the second is a clear sign of bias.

Davidski said...

@Chetan

There is no competing hypothesis. There's just one idiot online desperate to make Corded Ware and R1a-M417 Uralic. Why do you take him seriously?

The academic consensus is that the Corded Ware people came from the steppe, that they were rich in R1a, and also the first Indo-Europeans in Northern Europe.

Several genetic, archaeological and linguistic papers have come out in recent years and months arguing this, and the only thing that is up for debate now are the details. For instance...

https://www.academia.edu/35980445/Comments_to_Olalde_et_al._2018_on_the_Bell_Beaker_phenomenon

Aram said...

Simon W

What You say about Etruscans makes sense. Thete is even an Y dna for that theory. G2a L497. See the map here.
https://www.eupedia.com/europe/Haplogroup_G2a_Y-DNA.shtml

Notice it's presence in Raetian lands and Etruria.

Archaelog said...

@Davidski The "Indo-European Demic Diffusion Model" published by the source we are talking about here is overall a decent read and includes some neat maps. But it suffers from the problem we discussed - an over reliance on linguistic theories, that too hardly mainstream. I do believe that linguistic theories should now be cross-examined with the empirical evidence that is coming in. The time for any "pure" humanities disciplines is over.

This paper gives a lot of importance to the fact that the Yamna samples obtained so far are uniquely R1b but it ignores the presence of R1a in the region from the Mesolithic onward (Khvalynsk, Ukraine).

The so called North-West Indo-European group he proposes (including Celtc, Italic, Germanic and Balto-Slavic), derived from a single R1b community in the Carpathian basin according to this paper, is anything but the consensus. I am yet to see any mainstream work in Indo-European linguistics which openly supports this grouping. The truth is, Balto-Slavic shares more isoglosses with Indo-Iranian and Germanic than with Italic-Celtic.

It also gives too much credit to far-fetched hypotheses like Nostratic and Eurasiatic proposed by people like Allan Bomhard which are generally not accepted as mainstream.

Davidski said...

@Chetan

Despite the recent major aDNA papers, Carlos is still going on about Corded Ware and R1a being Uralic, and R1a-Z645 not being from the steppe but native to East Central Europe.

I've never seen such contrived and contradictory theories on this topic as those on his blog.

I mean, hundreds of samples from the Baltic region and East Central Europe have just been released, including Globular Amphora from Poland and Ukraine with no R1a, and no mention of anything about Corded Ware being Uralic rather than Indo-European.

There were also those Czech Corded Ware samples, with the two R1a individuals clustering with steppe samples, and the one I2a2 with Neolithic farmers. It's as if they were especially sampled to stick it up Carlos and his theories.

http://eurogenes.blogspot.com/2018/03/awesome-substructure-within-czech.html

So obviously nothing will convince him that he's on the wrong path. I expect he'll just keep going and making ever more crazy assertions against the data for years, until all you'll see on his blog are the ramblings of a mad man.

Anonymous said...

"That shows that from an early date, these two haplogroups could have been distributes across the Pontic-Caspian steppe and maybe even the forest steppe."

Actually, Khvalynsk and Alexandria are forest steppe.

Davidski said...

Actually, Khvalynsk and Alexandria are forest steppe.

That's a negative on both counts.

Alexandria is below the forest steppe boundary, and Khvalynsk is on the boundary.

https://drive.google.com/file/d/1hgsjlOijUdgGXP6_KazWnL9TxHW_8fO2/view?usp=sharing

Archaelog said...

Steven said "Why doesn't the Greek language show any resemblance to other Indo-European languages that were spoken by carriers of I2a2 and R1b-L-51 if theses are the haplogroups that brought Yamnaya DNA to Greece?"

But this is becoming precisely the wrong question to ask. Every IE language other than Anatolian seems to have split from the core within a narrow time frame now 3300 - 2500 BCE. Which means all of the dialects at the time of split would have been speaking pretty much late PIE. All of the remaining features of the branches seem to be later developments which came about due to interactions with other dialects in their places of attestation. My views in this regard have changed a lot. No I think it no longer makes muchmore sense to speak of IE subgroupings such as Graeco-Aryan. Sure there are broad similarities but they could just as well be archaic retentions as common innovations.

In other words there is nothing contradictory about the idea of an R1b dominated pre Greek and an R1a dominated Indo-Iranian being linguistically close to each other. Just shows these two haplogroups came to dominate the two respective branches. Even a small amount of R1b went to India with Indo-Aryans who were nonetheless a clear z94+ group. Doesn't mean any "R1b related languages" came to India with the Indo-Aryans does it?

Ryukendo K said...

@ Davidski

Been away from this blog for a while, first off, congrats on the high-res analyses you've been able to do. The splitting of steppe ancestry in South Asians into Andronovo and Srubnaya outlier (H/T Sein) is incredible, and finally brings unlinked autosomal variation in line with haplotype results and the uniparental connections between Steppes and S Asians.

Pretty sure now that Srubnaya/Andronovo/Sintashta-like source for Steppe ancestry in S Asia is supported by all lines of evidence.

Few questions, general really, anyone can answer them:
1) I'm surprised no one has done fine analysis on N Eurasian populations from Finland to the Baikal yet using Global25. Capitalising on recent drift should separate all the confusing ENA waves in Uralics and Turkics. E.g. the Baltic paper plus your analyses with Baltic HGs plus M Myllyla's work suggest that Finnics and Saami, vs all other Uralics + Chuvash/Mordvin/Erzya, vs all other Turkics East of Bashkir have different ENA compositions. Pretty busy right now, not up for it, but if anyone else wants to do it I'm just throwing it out there that there should be very interesting patterns.

Whoever does it, you might want to test the Tajiks as well, their ENA seems East Uralic-like, i.e. like that found in Scythians & Sarmatians+the Chuvash/Erzya/Mordvin group. Karasuk_Outlier is a great reference to use here. Otherwise I'll get back in 2 weeks.

2) Likewise there are extremely interesting patterns in the modern Middle East starting from Levantines and moving to Mesop. and north to N Cauc. even in Global10 with the different Near Eastern aDNA references, e.g. Armenia MLBA dominates in S Cauc and Assyrians, Anatolia Chalc+CHG+Scythian/other steppe in N Cauc, Iran Chalc in Mesop. A panel analysis of all E Mediterraneans and Caucasians should fine-tune these patterns and allow us to make predictions.

3) Now we have Greek Neolithic, Peloponnese N, Anatolia Chalc, Minoan+Mycenean and BB Sicily we should be able to pinpoint the sequence and sources of movements into the E Mediterranean, which should suggest patterns that match (or not) with Tyrsenian languages. Also should check out exactly how far the Anatolia-->Balkan spillover reached, though for some reason qpAdm and nMonte diverge on that case like in Yamnaya Bulgaria, so David might want to corroborate.

4) Surprised no one has looked at ancient Balkan samples for modern Balkan modelling yet, it seems Steppic ancestry was pretty low in that area for a long time.

Ryukendo K said...

@ Folker

Do you have archaeological references for what happened in Anatolia? If not could you send some links to Freu's work in Archaeo? I've only found his popular works.

Anonymous said...

@Davidski

The boundaries were moving from south to north.

Ryukendo K said...

About linguistic evolution, secondary demographic expansions causing linguistic leveling effects from already-kurganized "in-between regions" could create a branching tree with high levels of vertical structure, i.e. long temporal distance between the earliest branching and latest branching languages and not just a "starburst", even if Steppe ancestry burst into the scene into most areas very quickly.

E.g. the weird discrepancy between the "90% replacement in Britain by Netherlands BB with no Iberian Neolithic ancestry" and the modern elevated haplotype sharing between Britons and Iberian HGs/Neolithics could be explained if, say, some more Western continental post-BB population subsequently contributed to Britain much later. If we then add that this population was closer in space to the proto-Italics and Ligurians, then all West European IE would be unusually close and branch unusually late compared to what we would expect if we dated all the languages to the earliest intrusion of Yamnaya. There's also the added benefit of explaining why Britons are not super elevated in Steppic ancestry today compared to C Europeans, which is what the BB data would have us expect if we postulate population stasis after the Bell Beaker intrusion.

Of course this will require much more time-sensitive analyses like haplotypes and rare alleles on more data. We also cannot discount later massive shifts because the populations doing the shifting are closely related, so to push the needle of EEF:Steppe ratio would need quite a bit of migration.

Hmm Matt if you don't mind what happens if you compare the England Iron Age genome to BB Britain? What continental sources does the England Iron Age demand on top of BB Britain?

Davidski said...

@ryukendo kendow

Thanks.

It might take me a few months to blog about all of the interesting things in the new datasets that have been released. Just steadily working my way through it all.

Rob said...
This comment has been removed by the author.
Michalis Moriopoulos said...

@Rob

I was just being fair. She was a human being; she had both silly biases and interesting insights. My point was that, even though she didn't care for the steppe people personally, she thought their contribution to European culture was undeniably important. A less scrupulous individual with a similar distaste for patriachy and martiality might have just outright denied they mattered at all.

Folker said...

@Rob
I have no proofs to give to you. No need to.
By the way, I formulated a theory or an hypothesis based on different data and scholars work. Future research will tell if this hypothesis is confirmed or not.
Obviously, if already proved, it will not be an hypothesis.
It's basical scientifical process.
So, again, you are completely missing the point by faulty logics (but it's a common trait among many people).

Moreover, if I love to discuss openly many things, I usually do so with people able to take some distance with their own beliefs or situation.

You have clear difficulties to do so, not only because of some bias in interpretating date, but also because you are often overeacting.

And since you have limited knowledge outside the anglo-saxon academic world (and it happens that in many fields, academic publications are not written in English, especially concerning Anatolia), it would be wiser to be more humble.

Well, it doesn't matter anyway. It's your problem, not mine.

@Ryukendo kendow
Not much is available through the web.
Espcially in Europe, as past publications from the last 50 years are still unvailable.

Anyway, some French academics are available on (more will appear in time, as more and more publications are scanned and made available):
www.persee.fr

articles by Jacques Freu
http://www.persee.fr/authority/274869

He also wrote in "Res Antiquae" (paper only):
http://www.safran.be/resantiquae/

Several issues of "Anatolia Antiqua" are available:
http://www.persee.fr/collection/anata

You should read papers from Isabelle Klock-Fontanille, as she works on Hattic and Hittite:
https://hal-unilim.archives-ouvertes.fr/search/index/q/authIdHal_s%3A%22isabelle-klock-fontanille%22/

Regarding archeology:

About burials in Anatolia (EBA)
https://halshs.archives-ouvertes.fr/halshs-00808189/document

About Barcin Höyük
https://ais.ku.edu.tr/AR/ROZBAL200914__Anatolica.pdf

About Ilipinar
http://rjh.ub.rug.nl/Palaeohistoria/article/download/24880/22328
http://www.persee.fr/doc/anata_1018-1946_1997_num_5_1_864

I find this website very interesting:
http://www.tayproject.org/enghome.html

Folker said...

just an example of the difficulties pf archeology in Turkey, and why it's difficult to prove a Balkan route (click on photos):

http://www.tayproject.org/TAYmaster.fm$Retrieve?YerlesmeNo=1094&html=masterengdetail.html&layout=web

Obviously, no way to know what was this mound.

Simon_W said...

@Rob

„And the the proto-Villanovan was essentially an offshoot of Urnfield ?“

Yes, especially of eastern Urnfield. Though the Proto-Villanovan in the narrow sense is limited to Italy and Southern Switzerland. Certain affinities are found along the northern Adriatic up to Slovenia and Croatia.

But since the Urnfield area on the whole probably was IE and partly Celtic, so the logic goes, that the Proto-Villanovans must have been IE too, and maybe Italic. But it's totally possible that some non-IE people found Urnfield influences attractive and adopted them without adopting the language. After all Urnfield influence was also found in Catalonia and southwestern France, where non-IE Iberian and Aquitanian were spoken in antiquity.

Simon_W said...

@Aram
re: G2a. Well observed. ;-)

@Anthro Survey

We have to be careful here to distinguish Villanovans and Proto-Villanovans. The Villanovan proper, from the Iron Age, is much too late (starts around 900 BC!) and geographically too limited to be ancestral to Italic (it's just found in Etruria, and around Bologna, around Verucchio, around Fermo and in a limited area in Campania).

But sure, it's possible that the Proto-Villanovans were Italic. Though at the moment, for the said reasons, I rather don't suscribe this theory.

But definitely, if the Proto-Villanovans and Villanovans were Etruscans, then the latter must have had a good chunk of IE admixture just like the Basques, because Villanovan mtDNA is quite northern in the MDS plot.

About Anatolia_Chl/Anatolia_BA admixture in Italy: At the moment I rather think that in northern Italy it mostly goes back to the Roman era, and to colonists from Latium and further south. I doubt it was important in pre-Roman northern Italy. But we'll see when more ancient DNA comes out.

Matt said...

ryu:Hmm Matt if you don't mind what happens if you compare the England Iron Age genome to BB Britain? What continental sources does the England Iron Age demand on top of BB Britain?

Interesting post with some interesting possibilities and directions for future study which academia should take (though I think don't count out the "simple" formal stats and PCA yet). And the preceding post as well. Re: the above, are we talking about in D-stats or in the Global25 measures?

Rob said...

@ Folker

No, the problem appears to be with you.

You clearly stated "It is interesting by itself, as the theory formulated some years ago postulating an early arrival in Anatolia (perhaps around 4000 BC), and a wide era of IE languages in Anatolia, has been disproved."

https://www.blogger.com/comment.g?blogID=4123559132014627431&postID=2108955732756925868&isPopup=true

All I asked was for you to construct a semi coherent argument for this. I, we, would be happy to learn. Instead, after dancing around like a weasel for half a thread, you now say - "I have no proofs to give to you. No need to.
By the way, I formulated a theory or an hypothesis based on different data and scholars work. Future research will tell if this hypothesis is confirmed or not."

Let's put aside the fact that you're rather ignorant of the subject matter - not only at a synthesis level (you understand neither archaeology nor genetics), but you pontificate advice to others ("Not much is available through the web. "Espcially in Europe, as past publications from the last 50 years are still unvailable.") which is completely incorrect (there are tons of works, entire volumes !- which is why you rely on outdated works and have no idea about the matter); you make false assumptions about my (mono)linguiality - the works in any case are also available in English !
It doesn't matter if you/ I are wrong or right, but you obviously can't back your claims or even discuss like a real man. So the bottom line is - you're full of crap, and it's clear to see for all.

Rob said...

@ Simon W/ Aram

Thanks. I wonder if Etruscan - if inherited from G2a-rich Italian farmers- has any relationship to any of the Caucasus languages ?

Folker said...

@Rob
It's rather obvious for everyone that an wide dispersal of IE languages in Anatolia in the IVth millenium is not backed by data, and is obviously not a consensus among scholars. If the fact that Hurrian or Hattic were widely spoken in Anatolia is not enough for you, what will be?

Rob said...

@ Folker, or FFoucart
What consensus - that on French page of Anthrogenica ? Such esteem

Folker said...

@Rob
Pleased to find your decency is back.

Mind since I studied History, I may be able to understand works on History and/or Archeology, and make my own opinion, even if the subject is not part of my main interest.
By the way, the fact that Nubian archeology was part of my university cursus means that I'm not completely ignorant in Archeology proper.

If you have read the references given to Ryukendo, with a large part of fairly recent publications (less than 10 years), you will see there is a relatively large consensus among those scholars (mainly European) to the IIId millenium BC as the period of arrival for IE in Anatolia. Some argumenting from Caucasus, others from Balkans. Often based on linguistics arguments, to the level and the way Hattic words are present in the tablets, and how they were used. Given the fact that part of these scholars took part of archeological in Anatolia, by digging up sites, I will not rule them out easily.

Moreover, I tried to find some possible kurgans in North Turkey dating from EBA. And they are good candidates, undigged. The problem is, a large part of them are not known to authorities, and often destroyed to make place before being digged up properly. Hence the difficulties.

Anonymous said...

@David

Something utterly off topic. If you can spare the time, what does this do?

Mbuti GoyetQ116-1 Karela_HG AfontovaGora3
Mbuti GoyetQ116-1 Karela_HG MA1

Ryukendo K said...

@ Folker

Thanks for the references. Now watch me as I go tackle in my rusty French. probably need to fortify myself with some French Fries. Gulp.

Rob said...

@ Folker
Yes thanks some decent links
I double checked about the linguistic consensus about theories of Anatolian dating

Garrett late 5th/4th
Ringe/ Anthony - late 5th/ 4th
Finkeelberg - between 6-4th
Melchert 4th
Nichols: 6/5th Mill BC

The Hattic substrate has long been over emphasised (Melchert). In any case it signifies contact and is not a clear marker of chronology (what came first). In any case Hattic could be neolithic or it could be new arriving group , with subsequent relative shifts of linguistic range within Anatolia

But ultimately, as linguists profess- “all the dates proposed by linguists are, as is often admitted by the linguists themselves,
purely speculative.“
But as you mentioned, your not a linguist, so you don’t have that problem ..

Back to your papers. The Ilipinar suggests strong links between Anatolia and Balkans throughout neolithic and subsequent Chalcolithic 5th mill

The burial evidence paper shows widespread adaptation of extramural cemeteries in Anatolia , with interesting hocker position becoming prominent in late 5th /4th miellonium, the earliest of which are found - yep you guessed it - in lower Danube 5th millennium

Thanks for confirming the consensus ;)

@ Ryu
I trust those fries were tasty

Simon_W said...

BTW @ Rob and Anthro Survey

I was influenced a lot by this paper on the Protovillanovans:

http://www.landesmuseum.at/_eisenzeiten/eisenzeiten%20I%20pdfs/Amann.pdf

However, it's not very fresh is it? I first downloaded it 7 years ago, and its most recent reference is from 2004.

From what I've gathered it has become more fashionable in recent years among Italian archaeologists to further dissect the Protovillanovan facies, to reject its cultural uniformity and to seek for the origins of various historical peoples of ancient Italy in the different groups of this final stage of the Bronze Age. And Emma Blake also wrote that the Protovillanovan is clearly more varied than the Apennine and Subapennine culture of the MBA and LBA. According to her, and she's drawing upon many, there is for example the Chiusi-Cetona facies which encompasses Umbria, the Marche, and the inner part of Etruria. And quite distinct from this, especially in the pottery, there is the Tolfa-Allumiere facies, which encompasses southern and coastal Etruria. And in the Final Bronze Age II, the Latial culture starts to go its own ways and splits from the Tolfa-Allumiere facies. According to Emma Blake eastern Lombardy and Veneto show interesting relationships with the Tolfa-Allumiere facies. Now we can puzzle if this is related to the special relationship between Etruscan and Raetic or the one between Venetic and Latin...

In any case, to me it looks pretty obvious that the Chiusi-Cetona facies is Umbrian, the Tolfa-Allumiere facies Etruscan and the Latial culture Latin. So all these peoples were there in the Final Bronze Age and participated in the Protovillanovan culture.

So if the Etruscans should have Aegean roots there would have to be evidence for Aegean cultural influence in coastal Etruria before the Final Bronze Age. Indeed there are some sherds, but they are rather scarce. It's not looking like an invasion.

Simon_W said...

As for the Anatolia_ChL-related admixture in Italy, I think it started to appear in the Chalcolithic. According to Harrison and Heyd it was in Apulia just like in Sicily (the Sicilian Beaker). And the Chalcolithic Gaudo culture of Campania also shows cultural relations to the Aegean and Anatolia. And then the Laterza culture from Apulia expanded up to southern Lazio, and influence of the Gaudo culture can be found as far north as southern Tuscany. So this was the first wave, and at the northern fringe of its expansion the Anatolia_Chl admixture must have been rather low. Then there followed another wave of Aegean and/or East Med influence in the Bronze Age, around 1600 BC. It peaked in the period from 1445 through the 13th century BC. And this was mostly a South Italian and Sicilian thing. Only few sherds are found further north. This wave faded out at the end of the 12th century BC. And the next Aegean wave followed in the 8th century BC, with the historical Magna Graecia.

So I expect that the ancient Latins, Umbrians and Etruscans had rather low to no Anatolia_Chl admixture. It's also telling that this was low to non-existent in the two MBA Dalmatian samples. If it was so low there, why and how should it have been stronger in northern and central Italy at the same time?

And then it's archaeologically obvious that the Emilia was depopulated with the collapse of the Terramare culture at the end of the LBA. Modern-day Emilians and Romagnols are darker haired than Tuscans, Umbrians and Marchese people - that's hardly a legacy of the Terramare culture, considering the population drop at its end. The area was then colonised by Etruscans and Umbrians, and these were subjugated by the Gauls. And I suspect that the higher percentage of dark hair goes back to neither of these peoples, but to Romans and Romanized people from central Italy, who, at that date, may still have had lower Anatolia_Chl than today.

Alogo said...

I didn't catch who brought up the Carpathian Basin link but it's definitely there, including the cheekpieces that you've brought up a few times yourself (though you made reference to the group with more direct steppe ties). And that's basically the problem, that in the EHII to MH period you have plenty of influences that could be traced to east Central Europe, the steppe and even the Caucasus. The basic model is there, sound and already tentatively shown with aDNA too - a steppe(-influenced Balkan) population interacting with a south Balkan-Aegean one around that time - but until we get more samples, we've barely gone past what archaeologists and historians have been arguing about all along regarding the specifics, unfortunately. Something Beaker-like or Steppe_MLBA-like seems to work well.

On the PCA, it seems that you can more or less get via that one Vucedol sample to Armenoi to one Mycenaean if you use a Peloponnese/Minoan-like substrate. The problem is that there aren't enough samples or Y-DNA to tell which part of the northern steppe-presteppe cline is exactly responsible as you basically point out in the OP.

@namedguest,
I'm not getting your point there? The Yamnaya_Bulgaria sample has a lot of extra Neolithic ancestry. The increase isn't so much due to the Peloponnese_N/Anatolia_BA combo, which basically approximates the Minoans that Lazaridis et al. did use, but because David used Yamnaya_Bulgaria instead of the Steppe_EBA and Steppe_MLBA averages Lazaridis et al. did. So it depends on the sort of "steppe ancestry" used in the first place. The estimates in the paper increased between Steppe_EBA and Steppe_MLBA too, obviously.

@supernord,
That's an aspect I've been interested in all along too but the way you put it is basically just sheer supposition until we get more samples, considering that plenty of scenaria have been proposed. It's just as "likely" that there actually existed no pre-Greek IE language in Greece before Greek but just a non-IE one (not that an Anatolian IE substratum would have to be accompanied by significant steppe ancestry in the first place). Others have considered even multiple pre-Greek IE substrata, in more substratomanic times. What you wrote is a bit too speculative, and less parsimonious at that, until we get more data for you to be that certain I think.

Alogo said...

@vacuouswastrel,
That linguistic argument has always been a strong one and most archaeological scenaria, with few exceptions, also have proto-Greek arriving only c. 2200 or later (in the late EBA - early MBA), obviously with a 17th century terminus ante quem. All sampled later (at least post-Poltavka) steppe cultures seem to have this EEF-WHG increase so one could potentially posit a link of Greek to e.g. Late Yamna or Catacomb and still have it be part of a 'Greco-Aryan' continuum and on the steppe. And, to make an obvious point, the whole slew of less well-attested languages between Greek and Indo-Iranian (Phrygian, Thracian, Moesian-Dacian etc.) shouldn't be ignored. It's quite possible that (what was to be) proto-Indo-Iranian and proto-Greek weren't in that close proximity within the steppe/forest-steppe in the first place and later language extinction, which affected the whole area from northeast Greece to the Southeast Ukraine, makes it look that way.

As for the chariot itself, it's easier to connect it to the arrival of proto-Greek (though obviously not absolutely certainly associated even there too) than most other well-attested European branches but it also appears in other areas like e.g. Scandinavia at around the same time. I wonder what happened there if we connect proto-Germanic to Corded Ware or Beaker. Simple cultural diffusion via the Carpathian Basin or a 'Greco-Aryan' adstratum? A minority opinion I've mentioned before here btw is of an actual Indo-Iranian MH intrusion in already (since end of EHII) Greek-speaking territory. But that's a hard one to test, unless we somehow get blessed with very specific sets of Y-DNA and even then...

@ryukendo,
In a previous version of David's dataset, I tried some models with just ancient Balkan genomes, Baltic_BA or Slav_Bohemia and various sources on the Levant to Armenia cline. You get some good distances and plausible models. But it's still very early with such limited sampling for specifics obviously.

Ancient Balkan-Aegean genomes and (the likely imperfect proxy for early/proto-Slavs) Baltic_BA, along with some extra steppe and Near Eastern type of stuff, seem to dominate things even when you throw everything together.