search this blog

Thursday, September 1, 2022

Dear Iosif #3


Back in 2016 I made this prediction about the origins of the Yamnaya people (Steppe_EMBA):

But here's my prediction: Steppe_EMBA only has 10-15% admixture from the post-Mesolithic Near East not including the North Caucasus, and basically all of this comes via female mediated gene flow from farming communities in the Caucasus and perhaps present-day Ukraine.

The relevant blog post is still here. Looking back, my analysis is a bit sloppy and I didn't articulate my ideas too well. But that was a pretty good prediction for its time, and I believe it still has a chance of being confirmed, more or less.

On the other hand, the widely publicized hypothesis that the Yamnaya population is a ~50/50 mixture between indigenous Eastern European hunter-gatherers and Near Eastern or West Asian migrants never looked right to me. So I'm glad that it's now dead and buried.

Those of you not up to date with this topic, all you need to know is that the Yamnaya genotype existed in Eastern Europe at least a thousand years before Yamnaya, and, moreover, the Yamnaya people are largely derived from Eastern European foragers already rich in Near Eastern-related ancestry. The relevant ancient genomes are on the way (for instance, see here).

Nevertheless, the narrative that waves of Near Eastern migrants moved into prehistoric Eastern Europe, leading to the emergence of the Yamnaya culture and even the Proto-Indo-European language, is still being pushed by some notable scientists working with ancient DNA.

My hope is that, considering the latest revelations about the genetic origins of the Yamnaya people, these scientists can embrace a more nuanced view. How about something like this?

- people moved around, and they were especially mobile on the Eastern European steppe from the Eneolithic onwards

- when they made contact they sometimes mixed, so there was admixture between far flung steppe groups

- since population densities on the steppe were low until the Yamnaya period, minor admixture that entered the steppe during the Neolithic and Eneolithic wasn't dilluted easily.

See also...

Dear Iosif #2

But Iosif, what about the Phrygians?

Dear Iosif, about that ~2%

Dear Iosif...Yamnaya

58 comments:

Andrzejewski said...

I still think (like Rob) that something in the interaction with the farming EEF communities to its West -GAC, CTC, etc - had some formative impact. Sredny Stog was about 80%WSH :20% EEF. IMHO of 2 cents, PIE was formed as something new, unlike any Anatolian farmer, WHG, CHG or EHG spoken before.

Modas Califa said...

I am very skeptical about this analysis of Lazaridis in which there is an excess of EHG ancestry on the X chromosome. The two 7000+ year old Middle Don individuals from the "Population Genomics of Stone Age Eurasia" study (that already show DNA introgression of CHG) also have male lineages typical of EHG. It still seems to me that CHG DNA actually arrived on the steppe mainly mediated by women.

Samuel Andrews said...

They never even tried to model Yamanya as Piedmont_Eneolitic, plus Neolithic Europe, plus extra HG.

Technically they can successful model Yamnaya as Piedmont_Eneolithic, plus EHG, plus Eneolithic West Asia.

But they never considered the alternative. The alternative is more realistic and therefore should be favored.

Andrzejewski said...

@Modas Califa

CHG was already diluted when Progress (fem) immigrants who were EHG/CHG rich went up and met with Lower Don males.

But the fact that CHG was >33% itself only complicates things.

epoch said...

@Modas Califa

One of the more interesting things is that Eneolithic Piedmont steppe has a far larger discrepancy, and Khvalynsk has even no CHG admixture on X.

Indeed, the Don foragers have non-CHG Y-dna, R1a and even I2, as have all CWC samples. So there are three groups where a non-CHG lineage took over. (I didn't include Piedmont here, bear with me)

So what if R1b V1636 is a minority CHG lineage, typical for a CHG subgroup that lived at the steppe side of their realm? It picked up a ANE lineage, just like Villabruna did, which was a prevalent in little numbers only among the CHG subgroup on the north side of the Caucasus, maybe even up the coast of the Caspian or Black Sea. The Piedmont Progress samples then are an only recently admixted population.

If CHG was male mediated east of the Don and females from the resulting population introduced it west of the Don we might have explained a lot of things. R1b V1636 came to being 17k years ago but has a most recent common ancestor 6500 years ago. That would more or less fit in this.

Big problem with this is that while Eneolithic Piedmont steppe has an 50/10 autosmal/X ratio of CHG all the samples have non-steppe mtDNA.

Copper Axe said...

I noticed something sneaky about Lazaridis' arguments about "male CHG bias". In it he modeled Yamnaya as 50% CHG and 50% EHG which we know can't be right and so does he. The x chromosome amount was 34% with a possible deviation of 8.5%. At least on G25 that is pretty much amount of CHG Yamnaya gets when also using EHG, thus the sex bias should basically not even be there by his methods.

Yamnaya:
51.9+/- 1.3% (autosomes)
34.2+/- 8.5% (chrX)

And he then also posted these:
Results on RUS_Khvalynsk_Eneol (autosomal/chrX)
21.5 78.5 (s.e.=1.7)
-7.6 107.6 (s.e.=15.4)

Results on RUS_Eneol_Piedmont
55.5 44.5 (s.e.=1.7)
10.8 89.2 (s.e.=7.1)

Davidski said...

@epoch

Big problem with this is that while Eneolithic Piedmont steppe has an 50/10 autosmal/X ratio of CHG all the samples have non-steppe mtDNA.

Haha.

Trust the uniparental markers.

Matt said...

@Copper_Axe, if I remember G25 tends to slightly to substantially more Euro HG than qpAdm or the f4admix. Why, I don't know, it could be that some element of projection bias still compresses the Euro HG, even though Davidski does far better on this than and has far surpassed any published PCA projection.

Euro HG are still very significantly differentiated from present day, so getting projection right might be hard for smartPCA.

Or it could be that g25 is better. But either way we can't compare proportions computed with two different methods so easily.

Which does mean Laz has a slight advantage here is being able to compare x to a with the same method. While if we compare x computer with qpAdm to a computed with g25 + vahaduo there's a clear confound.

Matt said...

(i.e. I haven't done base CHG+EHG models in qpAdm for a while, due to the ill fit, but as far as I remember they were 50:50 or close to it.)

Copper Axe said...

@Matt

Base CHG+EHG models are not accurate for the simple reason that there is ANF present as well. At best it adds up to 40%.

Andrzejewski said...

@Copper Axe @Matt “ Base CHG+EHG models are not accurate for the simple reason that there is ANF present as well. At best it adds up to 40%.”

20% at best, but still…

I am wondering what part of the WSH vocabulary derives from some extinct Tripliyan language, and whether certain parts of Sredny or Yamnaya’s lifestyle, language, phenotype and so on comes from either culture and/or from Globular Amphora, Bug Dniester and Narva HG.

epoch said...

@David

"Haha.

Trust the uniparental markers."

You need a lot of favourable uniparental take-overs. Far too many. Which points to a highly complicated situation.

The Khvalynsk 0% CHG on X is maybe interesting. It may point to arranged bride swaps, which makes stuff even more complicated. Maybe even deals where sons were sent back. Alan on AG once explained that Celts sent second sons to other tribes to grow up.

Rob said...

The ‘OG’ CHG were J1. Then in proto-Yamaga times CHG was linked to J1, Q1a & R1b-V3616. Then R1b-M269, R1a-M17, I2a.. , picked it up. Don’t need schmanzy tests, just follow simple analogue data

Richard Rocca said...

It doesn't take much for a EHG/CHG population that is pretty much even on the X side to push a little further into territories with less CHG and take on local brides to have EHG on the X-side increase. Pushing the X narrative and dismissing the Y-DNA evidence is silly IMO.

bellbeakerblogger said...

I'd by the very reasonable notion that CHG was already present the mountainous region of the northern Black Sea shelf. The best opportunity to prove that would be to look at skeletons from cave sites in the Crimean mountains in the years before the entire sea shelf was submerged (10k-7500bc).
So I'd buy that whatever admixture had already occurred between EHG and CHG in the Paleolithic, that it rapidly began increasing in these border communities around 7500bc for about a thousand years.

Not to sound like a zealot here, but it's very likely that the Northern Pontic and Southern Caspien was settled by Mediterrean Impresso peoples. Perhaps a lot of them. The salinification of the Black Sea would have been catestrophic for their communities. So if your looking for an additional Levantinesque admixture 70/30 ENF/CHG about 6kto 5k bc, well , there ya go.

Synome said...

In the paper the authors emphasize how the source of the additional West Asian (ANF related) admixture can't be from the Balkans because Yamnaya tests negative for Iron Gates. But in g25 ROU_N works perfectly well as a substitute for Barcin_N. The admixture percentage is nearly identical.

Yamnaya_RUS_Samara(GEO_CHG, ROU_N, RUS_Karelia_HG): distance 0.05841304 36.8 6.8 56.4

Yamnaya_RUS_Samara(GEO_CHG, TUR_Barcin_N, RUS_Karelia_HG): distance 0.5879835 36.4 7.0 56.6


Wise dragon said...

Prof. Alwin Kloekhorst an Indo-Europeanist and Hittitologist and an expert on the oldest Indo-European language has evaluated the main conclusion of the Southern Arc paper on Twitter and gave his verdict:

@AlwinKloekhorst
My reaction in a Dutch newspaper (@nrcwetenschap) on the new @ScienceMagazine paper by @iosif_lazaridis et al. on the location of the Proto-Indo-Anatolian homeland. Despite the wealth of new genetic data, the conclusion of this paper doesn’t convince.


25. Aug.

Antwort an

@AlwinKloekhorst

It argues for a West-Asian homeland, South of the Caucasus, for Proto-Indo-Anatolian, i.e., the ancestor of both Core Proto-Indo-European and the Anatolian language family. This is based on the influx of Caucasian Hunter Gatherer (CHG) genes both into the Pontic-Caspian steppes



Alwin Kloekhorst

@AlwinKloekhorst

and into Anatolia that has been reported in this paper. It is therefore argued that population groups spreading these CHG-genes probably spoke Proto-Indo-Anatolian. However, the earliest spread of CHG-genes from the South Caucasus into Anatolia took place already



@AlwinKloekhorst

25. Aug.

before 6000 BCE, which is much earlier than the time period to which — on the basis of linguistic arguments — the ‘Anatolian split’ can be dated. This split probably took place around 4200 BCE, which is nowhere near 6000 BCE implied by Lazaridis et al. This clear (and painful)





@AlwinKloekhorst
·25. Aug.
mismatch with evidence from linguistic data could have been avoided if this paper (with 100+ authors) would have had input from historical linguists as well. Clearly a missed opportunity, and in this way genetics, archaeology, and linguistics will not get any closer.

VerdianoBR said...

Honestly, I'm not convinced by the paper, either. Didn't geneticists learn anything from the well known Hungarian case (Early Magyar immigrants totally different from present Hungarians) or from the less well known Estonian case (a very eccentric one, with high Y-DNA impact, but extremely low autosomal one), or even that of parts of Turkey that today speak mainly Turkish (where Northeast Asian signal is exceedingly small or even undetectable)? Linguistic shift by elite dominance is rarer than previously thought, indeed, but it does happen occasionally.

I mean, if you have a language family that can be securely linked to the Pontic-Caspian steppe in each and every branch of it, except in one that happened to be spoken in the very hotbed of early development of farming (so, presumably, very large populations, implying a much lower likelihood of large-scale genetic impact via male-biased conquest), is it really sensible to immediately shift from one hypothesis to another, instead of considering the possibility that that specific branch spread and developed in a unique way? I don't think so.

You don't even have to go for a very implausible scenario, especially if you're dealing with a conquering migration from a low-density region to a high-density one. Something like this would be enough:

50% A - western Ukrainian steppe + 50% B - adjacent EEF people = C [50% steppe]
40% C + 60% eastern Balkanic people after migration southward = D [20% steppe]
30% D + 70% northwestern Anatolian people = E [6% steppe]
25% E + 75% western and central Anatolian people = F [1.5% steppe]

In each particular migration, there was large-scale movement with considerable genetic impact (50%, 40%, 30%, 25%), but steppe admixture was increasingly diluted, nonetheless.

VerdianoBR said...

Honestly, I'm not convinced by the paper, either. Didn't geneticists learn anything from the well known Hungarian case (Early Magyar immigrants totally different from present Hungarians) or from the less well known Estonian case (a very eccentric one, with high Y-DNA impact, but extremely low autosomal one), or even that of parts of Turkey that today speak mainly Turkish (where Northeast Asian signal is exceedingly small or even undetectable)? Linguistic shift by elite dominance is rarer than previously thought, indeed, but it does happen occasionally.

I mean, if you have a language family that can be securely linked to the Pontic-Caspian steppe in each and every branch of it, except in one that happened to be spoken in the very hotbed of early development of farming (so, presumably, very large populations, implying a much lower likelihood of large-scale genetic impact via male-biased conquest), is it really sensible to immediately shift from one hypothesis to another, instead of considering the possibility that that specific branch spread and developed in a unique way? I don't think so.

You don't even have to go for a very implausible scenario, especially if you're dealing with a conquering migration from a low-density region to a high-density one. Something like this would be enough:

50% A - western Ukrainian steppe + 50% B - adjacent EEF people = C [50% steppe]
40% C + 60% eastern Balkanic people after migration southward = D [20% steppe]
30% D + 70% northwestern Anatolian people = E [6% steppe]
25% E + 75% western and central Anatolian people = F [1.5% steppe]

In each particular migration, there was large-scale movement with considerable genetic impact (50%, 40%, 30%, 25%), but steppe admixture was increasingly diluted, nonetheless.

There is also another problem: using a set of diverse G25 source samples to avoid misattribution due to the lack of a more proximate source or overfitting, I can definitely get some steppe-related or at the very least EHG-shifted signal in some of the MBA, EBA and Chalcholithic from Western Anatolia, but NOT in those from Eastern Anatolia nor in those from the Neolithic period (except for 2 sample from Mentese, in the Marmara region virtually in Europe, who get ~0.8% and ~4% EHG admixture each, but no Yamnaya).

If it's a statistic artifact, then why don't any of the other Neolithic Anatolians get it, too, and Yamnaya-related admixture only appears from the Chalcolithic onwards and, particularly, from the LBA onwards?

Doesn't look like mere coincidence, to me especially when it fits so neatly with geography (western Anatolia) and timing (mostly from the Copper Age onwards) right when and where Anatolian IE emerged. Very small proportions? Yes, but not much lower than what you find, say, in some Indians, Iranians, Iron Age Anatolians and Mycenaean Greeks, so, you know... ;-)

VerdianoBR said...

Sources used (G25): Natufian, EHG (Karelia, Samara and Sidelkino), Iran-N (Ganj Dareh), Iran Mesolithic (Belt Cave), Anatolia Neolithic (Barcin), CHG, Levant Neolithic (PPNB), WHG (Cheddar Man), Yamnaya (combining Kalmykia, Karagash, Mereke, Samara and Ukraine).

TUR_Aegean_Izmir_Yassitepe_MBA:I5737 = 7% Yamnaya
TUR_Aegean_Izmir_Yassıtepe_EBA:I5733 = 6.4% Yamnaya
TUR_Aegean_Izmir_Yassıtepe_EBA:I5735 = 5.6% Yamnaya
TUR_Marmara_Barcın_ChL:I1584_v43,5_all = 5.6% Yamnaya
TUR_Marmara_Ilıpınar_ChL:I10545 = 1.4% Yamnaya
TUR_Med_Isparta_BA:I2495_all = 1.2% Yamnaya
TUR_Med_Isparta_BA:I2499_all = 1.2% Yamnaya
TUR_Med_Isparta_BA:I2495_all = 0.8% Yamnaya
TUR_Marmara_Ilıpınar_ChL:I10542 = 0.4% Yamnaya
TUR_Marmara_Menteşe_N:I0723 = 0.8% EHG_Sidelkino
TUR_Marmara_Menteşe_N:I0724 = 4.2% EHG_Karelia

-- 0% Yamnaya and 0% EHG in *all* the other samples from the Mesolithic to the MBA.

And there is still the fact that, as already expected, the Ezero Culture samples had substantial Yamnaya-like ancestry, but mostly EEF ancestry. That means IE languages must have already been present in the doorstep to Anatolia for those who come from the Balkans, and steppe ancestry would've arrived in Anatolia already very diluted (so, a ~3-5% steppe-related signal was likely to actually mean a ~9-20% SE European_Eneo/EBA input).

Rob said...

If we put aside for a moment uniparentals & archaeology which unequivocally demonstrate links between Soyutheast Europe and Anatolia, the paper suggests there can’t be movement via Balkans because

1. They can’t detect iron gates ancestry in Anatolia
2. The overall % of Anatolian farmer ancestry drops whilst CHG rises

These are faulty/ inadequate platforms to built an well rounded argument because it doesn’t consider dual/ multiple flows

A more useful approach would be like that taken in Papac et al, for example, which sought to understand the constantly shifting base on which new groups form

Davidski said...

@Copper Axe & Rob

What can you say about the Phrygians based on the data in the Lazaridis paper?

Rob said...

Havent looked in detail, however it is obvious that the 2 iron age individuals are shifted toward Balkans c.f. rest of group, which are Hellenistic, as well as c f Bronze Age Anatolian body

https://ibb.co/y0v9D45

Rob said...

Another interesting aspect is the “Phrygians” pulled differently than the Yassitepe MBA chap, although both obviously towards Balkans.
Does imply different waves into Anatolia

Davidski said...

@All

New updated and corrected G25 datasheets are up.

https://eurogenes.blogspot.com/2019/07/getting-most-out-of-global25_12.html

Please report errors and missing samples here, and I'll try to fix things asap.

If there are samples that I already ran but they're missing, please post a link to the raw coords and I'll update.

Morris said...

Here, samples from medieval Hungary
https://drive.google.com/file/d/19N5S2T6FjLyJB_8jz71KXt9wxV3Vwmu3/view?usp=sharing

John Smith said...

It seems that the letters "CHL" were replaced with "C", ending up misnaming some samples.

Chile's initials have only "C" instead of "CHL".

CZE_Unetice_C:CHL001
CZE_Unetice_C:CHL006.merged
CZE_Unetice_preC:CHL002
CZE_Unetice_preC:CHL003.merged
CZE_Unetice_preC:CHL005.merged
CZE_Unetice_preC:CHL007.merged
CZE_Unetice_preC:CHL008
POL_Chlopice_Vesele_Culture:I6531
POL_Chlopice_Vesele_Culture:I6537

These samples should be Medieval "MA", not "IA".

Baltic_EST_IA:s19_IIa_1
Baltic_EST_IA:s19_IIf_1
Baltic_EST_IA:s19_IVLS09KT_1

I would also like, if possible, for these samples to be named as follows:

JOR_PPNC:I1708
Levant_Natufian_EpiP:I1072
Levant_Natufian_EpiP:I0861

vAsiSTha said...

@wisedragon

Prof. Alwin Kloekhorst is misinformed and was immediately corrected by Lazaridis and others.

"However, the earliest spread of CHG-genes from the South Caucasus into Anatolia took place already before 6000 BCE, which is much earlier than the time period to which — on the basis of linguistic arguments — the ‘Anatolian split’ can be dated. This split probably took place around 4200 BCE, which is nowhere near 6000 BCE implied by Lazaridis et al. This clear (and painful)"

This is painful to read because there are clearly two separate waves of CHG-like ancestry in the steppe, one older about 5500bce, and the other much later. This is mentioned in the Southern arc paper two times. Ctrl + F "two streams of migration".

This is also mentioned in Allentoft et al preprint, and Chintalapati et al date Yamnaya formation to ~4100bce +- SE.


Rob said...

@ BBB

''Not to sound like a zealot here, but it's very likely that the Northern Pontic and Southern Caspien was settled by Mediterrean Impresso peoples. Perhaps a lot of them'''


how is it very likely when the evidence is very weak at best ? And why would these 'Mediterraneans' be CHG instead of Anatolian or Mesopotamian like ?

Davidski said...

@vAsiSTha

Both streams of CHG into the ancestors of Yamnaya came much earlier than 4,000 BCE.

The ~4100bce date for the formation of the Yamnaya genotype is the absolute minimum and probably wrong anyway.

Apart from that, the ancestors of the Anatolians came from the steppe via the Balkans.

Andrzejewski said...

‘@Davidski “ Apart from that, the ancestors of the Anatolians came from the steppe via the Balkans.”

Yep, the Novosovorvska or something like that.

Andrzejewski said...

A strong corroboration to my theory that Sumerian was an Anatolian Barcin LBK-related farmer language comes from the word for cow: “gud” in Sumerian. Some researchers believe that the Western Steppe Herders switched their lifestyle from foragers to agropastoralism following intense contact with their Cucuteni Triplyan neighbors to the west. It’s believed that the agricultural and animal husbandry terms as well as practices were learned from the EEF. So if it turns out, “gud” in Tripolye language —> “guous” in PIE —> “cow” in English.

The alleged “IE substrate” in some so-called pre-Sumerian Euphratic language may be non other but shared borrowing from Tripolyan —> PIE, thus Sumerian as an Anatolian farmer language can fit the mold of a similar vocabulary proxy to that extinct EEF language which provided us with words for “cow” or even “pork(as)”.

Rob said...

It doesn't matter when CHG arrived , or whether we call it CHG or Iran_N

It is now fairly clear to anyone with a more-than-basic understanding that CHG was amplified and spread within this steppe, and does not represent a migratory phenomenon which introduced a 'cultural package' incl. language into Europe.

Don’t see the need for the mental gymnastics, especially when the partner paper about Mesopotamia/ PPN was comparatively solid

Davidski said...

@All

Does anyone know how many authors wrote the Southern Arc paper on the PIE expansion?

vAsiSTha said...

"Apart from that, the ancestors of the Anatolians came from the steppe via the Balkans."

I mean if you want to rely on 1 I yhg and some 3-4% yamnaya ancestry in 2 samples as evidence, go ahead it's a free world. I, like a sane person, believe that a 40-50% ancestry from the east is a much more likely vector of language change.

Davidski said...

Yamnaya doesn't have 40-50% ancestry from the east. It's obviously in large part of local hunter-gatherer origin.

The rest of its ancestry is still a mystery, and there's no evidence that this minor ancestry had anything to do with any Proto-Indo-Europeans.

Details matter.

vAsiSTha said...

Huh? I said 40-50% ancestry in Anatolia from the east.

vAsiSTha said...

"The rest of its ancestry is still a mystery, and there's no evidence that this minor ancestry had anything to do with any Proto-Indo-Europeans."

It's not really too much of a mystery. Broad trends are known, only the last mile is required.

Target: RUS_Progress_En
Distance: 3.2616% / 0.03261607
59.6 RUS_Khvalynsk_En
23.0 TJK_Sarazm_En
17.4 GEO_CHG

0.0 ARM_Aknashen_N
0.0 ARM_Masis_Blur_N
0.0 AZE_Caucasus_lowlands_LC
0.0 AZE_Caucasus_lowlands_LN
0.0 DEU_LBK_HBS
0.0 IRN_DinkhaTepe_BIA_A
0.0 IRN_DinkhaTepe_BIA_B
0.0 IRN_Seh_Gabi_C
0.0 IRN_Seh_Gabi_LN
0.0 IRN_Shahr_I_Sokhta_BA1
0.0 IRN_Shahr_I_Sokhta_BA2
0.0 IRQ_Nemrik9_LBA
0.0 IRQ_Nemrik9_PPN
0.0 POL_Globular_Amphora
0.0 RUS_Karelia_HG
0.0 RUS_Samara_HG
0.0 RUS_Tyumen_HG
0.0 SRB_Iron_Gates_HG
0.0 TKM_Geoksyur_En
0.0 TKM_Geoksyur_N
0.0 TUR_Arslantepe_EBA
0.0 TUR_Arslantepe_LC
0.0 UKR_N

Target: RUS_Vonyuchka_En
Distance: 3.3511% / 0.03351106
49.8 RUS_Khvalynsk_En
28.6 TJK_Sarazm_En
20.6 GEO_CHG

1.0 SRB_Iron_Gates_HG
0.0 ARM_Aknashen_N
0.0 ARM_Masis_Blur_N
0.0 AZE_Caucasus_lowlands_LC
0.0 AZE_Caucasus_lowlands_LN
0.0 DEU_LBK_HBS
0.0 IRN_DinkhaTepe_BIA_A
0.0 IRN_DinkhaTepe_BIA_B
0.0 IRN_Seh_Gabi_C
0.0 IRN_Seh_Gabi_LN
0.0 IRN_Shahr_I_Sokhta_BA1
0.0 IRN_Shahr_I_Sokhta_BA2
0.0 IRQ_Nemrik9_LBA
0.0 IRQ_Nemrik9_PPN
0.0 POL_Globular_Amphora
0.0 RUS_Karelia_HG
0.0 RUS_Samara_HG
0.0 RUS_Tyumen_HG
0.0 TKM_Geoksyur_En
0.0 TKM_Geoksyur_N
0.0 TUR_Arslantepe_EBA
0.0 TUR_Arslantepe_LC
0.0 UKR_N

Target: Yamnaya_RUS_Samara
Distance: 3.0501% / 0.03050126
71.0 RUS_Progress_En
18.6 RUS_Khvalynsk_En
8.0 POL_Globular_Amphora
2.4 SRB_Iron_Gates_HG

0.0 ARM_Aknashen_N
0.0 ARM_Masis_Blur_N
0.0 AZE_Caucasus_lowlands_LC
0.0 AZE_Caucasus_lowlands_LN
0.0 DEU_LBK_HBS
0.0 GEO_CHG
0.0 IRN_DinkhaTepe_BIA_A
0.0 IRN_DinkhaTepe_BIA_B
0.0 IRN_Hajji_Firuz_C
0.0 IRN_HajjiFiruz_C
0.0 IRN_Seh_Gabi_C
0.0 IRN_Seh_Gabi_LN
0.0 IRN_Shahr_I_Sokhta_BA1
0.0 IRN_Shahr_I_Sokhta_BA2
0.0 IRQ_Nemrik9_LBA
0.0 IRQ_Nemrik9_PPN
0.0 RUS_Karelia_HG
0.0 RUS_Samara_HG
0.0 RUS_Tyumen_HG
0.0 RUS_Vonyuchka_En
0.0 TJK_Sarazm_En
0.0 TKM_Geoksyur_En
0.0 TKM_Geoksyur_N
0.0 TUR_Arslantepe_EBA
0.0 TUR_Arslantepe_LC
0.0 UKR_N

Davidski said...

You don't know shit.

Out of Armenia and Out of Sarazm are equally hilarious.

vAsiSTha said...

Don't get worked up now. These are the same tools you yourself used in Iosif #1.
But 4% yamnaya in 2 anatolian samples proves Balkan anatolian hypothesis? lmao.

Rob said...

% of “CHG” doesn’t matter either. It’s just statistical homeopathy.

Farmers from northern and western Europe have even more “west asian” ancestry, but they like steppe eneolithic represent native developments
But nobody has claimed they were inert or closed off to admixtures and influences.
But this was selective & they kept their own identity passed on dynastically
Again - evident to anyone who isn’t a con artist


bellbeakerblogger said...

@Rob,

Firstly, I don't believe that anyone could have been pure enough CHG to supply the volumn needed to make these herders. At least not in the form of any kind of capable or fighting people. They really would have to have parachuted from outer space. So I believe that a lot of the CHG we see appear quite suddenly is native.

Geologically the Taman Peninsula (south of azov) and Crimea are the northern end of the Caucasus range (submerged around 7500bc). So the Paleolithic forward, if we are to assume CHGs were native to this ecoscape, then I think it is reasonable that a high percentage of EHG/CHG admixture had already occurred long before the Black Sea flood in this specific area.

We can look to the hypothesized tidal rise that encouraged Anglo and Saxon to Britain. When your land goes to shit, you fight for someone elses. Maybe that is what we see in growing mixture on the Steppe, people fighting.


Rob said...

The east shift in Anatolia comes from pastoralists in northern Anatolia who have less AHG and more CHG as a base. Then follows another wave from Arslantepe
These have nothing to do with the CHG fishers who settled the Volga steppe

Western Anatolia then experiences a ~30% arrival of Balkan Bronze Age people who are steppe admixed. There is even a second LBA-IA wave of Phrygians too.

bellbeakerblogger said...

@Rob,

But specific to Impresso folks, they are emerging a much bigger phenomenon than they once were. More of a coherent culture than just a name for miscellaneous people with dark hair that poke pottery with sticks or shells. They seem to be a "people" who very quickly colonize large distances, probably heterogenous of sorts, but they went everywhere. Perhaps the salinifcation of the Black Sea brought them in like flies. Probably the reverse, that they were already settling the estuaries and maybe already mixing. But they are mostly ENF depending on where they come from. The Levant and the Adriadic would be the two probable sources

Kouros said...

Sarzm accumulate Steppe Siberien and Eneolithic Steppe from Kumsay , as well IVCp, it is barely present in contemporary samples in Turkmenistan which lay west. There is no possibility these population are ancestral. David is right here, it is absurd. Sarazm even has IVCp , a population which forms only apres 6000 BP, that eliminate any possibility .

sample: Sarazm En:Average
distance: 4.2494
Ganj_Dareh_N: 50.5
GEO_CHG: 20
Boncuklu_N: 0
Kumsay_EBA: 21.5
CG_IVCp: 8


sample: Parkhai En:I4635
distance: 3.2358
Ganj_Dareh_N: 75.5
GEO_CHG: 18.5
Boncuklu_N: 5
Kumsay_EBA: 1

sample: Tepe Anau En:I4085
distance: 2.7355
Ganj_Dareh_N: 76.5
GEO_CHG: 16
Kumsay_EBA: 3.5
Boncuklu_N: 3
CG_IVCp: 1

Sarazm like Agyrizal but less Siberien, look at the distances


Bustan BA o1 Average
5.40

Aigyrzhal BA Average
9.54

Shahr I Sokhta BA2 I8726
10.42

Progress En Average
17.02



epoch said...

@Vasishta

This is painful to read because there are clearly two separate waves of CHG-like ancestry in the steppe, one older about 5500bce, and the other much later. This is mentioned in the Southern arc paper two times. Ctrl + F "two streams of migration".

Kloekhorst talks about Anatolian languages which, according to the paper you mention, was spread into Anatolia 6500 BC from Iran. The point he makes is that split of Anatolian languages is not old enough to have originated that long ago. Unless off course you want to propose that Anatolian languages originated on the steppe.

I mean if you want to rely on 1 I yhg and some 3-4% yamnaya ancestry in 2 samples as evidence, go ahead it's a free world.

How many of these samples were from actual Anatolian speakers, you think?

Rob said...

@ Bell beaker Blooger

Sure anything is possible, but I dont believe there was a coherent, primordialistic 'Impresso" people . I've described the Impresso phenomenon before here - it is an artistic reportoire within a common Neolithic pool of ideas. In the Ionian and Adriatic regions, it became emblemic before spreading to southern France where it was taken up by Neolithiczed hunter-gatherers. Impresso decoration was also present inland in western Asia, as far as the Tigris, but within a genetically different subset of people within the larger Neolithic koine.

Impressed decoration was present in the North Pontic region, but as Reinguber has outlined, this was part of the local HG -pottery phenomenon. This is backed up by aDNA which shows local Neolithics to be hunter-gatherers which moved down as far as the Danube. I linked this article for you before but you might have missed it.

As for CHG, its alleged mass -migration is a statistical mirage. Its rise between Sredni Stog & Yamnaya was because of dynamics within the steppe itself - people from the Dnieper-Don area mating with those from the lower Volga - where it had arrived ~ 6000 BC due to Caucasian fisher-hunters moving episodically toward the Volga.

Rob said...

@ Epoch

Yeah I'd say I5737 was an Luwian-Hittite

I wouldn't take any heed of uninformative distal models esp. when done by hacks

Simple but accurate 2-way model based on people who were actually around western Anatolia at relevant time



Target Distance Beli_Breyag_EBA Camlibel_Tarlasi_LC
TUR_Aegean_Izmir_Yassitepe_EBA 0.02471016 21.2 78.8
TUR_Aegean_Izmir_Yassitepe_MBA 0.03588514 30.4 69.6
TUR_Marmara_Ilipinar_C 0.01635408 15.2 84.8




The movements & contacts began as early as 4000 BC, but peaked during EMBA. There was a large movement of people from SEE into Anatolia, in fact much of Thrace was left depopulated by the MBA. On the other hand, there is a large settlement gap in west Anatolia between 4700 and 4000 BC. The neo-Anatolians who arrived came from the east





H₂ŕ̥ḱtos said...

@vAsiSTha

The thing is, your models of steppe_en and Yamnaya there show exactly the opposite of what Lazaridis et al. are saying in this paper. Your models show:

1. That steppe_en is EHG + something CHG-related, without Anatolian- or Levantine-related admixture,

and

2. That Yamnaya is steppe_en + more European HG admixture + local European farmer admixture.

This is exactly what proponents of the kurgan theory of the origins of PIE are already saying: That CHG-related admixture was already present on the Pontic-Caspian steppe from a very early date (Allentoft's paper's pre-print talks about Middle Don foragers with 20-30% CHG-related admixture, from ~5300 BCE), and that Yamnnaya has ancestry from Eastern European farmers but does not have trans-Caucasian ancestry from late neolithic or eneolithic Anatolians or Iranians.

Maybe the pre-PIE language was brought into the Pontic-Caspian steppe by a a pre-~5300 BCE migration of CHG-related individuals, who introduced it to the EHG-related people living there, rather than being a continuation of the "EHG language" already spoken there? This suggestion can't be falsified, certainly. But even if this is true, this date is just too early to fit Lazaridis suggestion that the language ancestral to Anatolian and nuclear IE was spoken by this CHG-related group in Armenia-ish. If true, this would push the date of what he's calling proto-Indo-Anatolian back to, well, surely at least ~5500 BCE or so, if I'm being generous. As @epoch has pointed out, as Kloekhurst explains, this is just too early, and this is why they should have had a historical linguist on board.

If proto-Indo-Anatolian was spoken in West Asia at the time that linguists agree Anatolian branched off, and then the remaining branch - i.e., pre-nuclear IE - spread trans-Caucasus to the Pontic-Caspian steppe ca. 4200 BCE, then we have to accept that it doesn't seem to have spread with any genetic influence. Possible? Sure, but then, you're arguing that it can't have spread steppe->Balkans->Anatolia on the basis of "no gene flow", so we're now looking at a double standard.

Even worse than a double standard, we're looking at what doesn't even appear to be true, as there are more than "one or two" CA and EBA Anatolian samples who pick up steppe ancestry when modelled thoroughly, and they pick up more than 2-3% of it. Again we need to bear in mind that nobody's saying that there was a direct migration of a Yamnaya-like population into Anatolia, and nobody ought to be expecting 50% Yamnaya-like samples.

H₂ŕ̥ḱtos said...

@vAsiSTha

Cont.:

Consider this guy - https://amtdb.org/sample/I2181 - from about 4500 BCE, in Bulgaria, with about 30% Yamnaya-like ancestry and the rest being local European farmer ancestry:

Target: BGR_C_outlier:I2181
Distance: 5.1636% / 0.05163649
55.2 UKR_Globular_Amphora
31.6 Yamnaya_RUS_Samara
13.2 TUR_Marmara_Barcin_N
0.0 ARM_Masis_Blur_N
0.0 Baltic_LTU_Narva
0.0 GEO_CHG
0.0 IRN_Ganj_Dareh_N
0.0 Levant_PPNB
0.0 POL_Globular_Amphora
0.0 ROU_Iron_Gates_HG
0.0 RUS_Karelia_HG
0.0 RUS_Samara_HG
0.0 UKR_Meso
0.0 UKR_N
0.0 UKR_Trypillia
0.0 WHG


Or, more accurately, since he pre-dates Yamnaya:

Target: BGR_C_outlier:I2181
Distance: 5.6150% / 0.05614985
66.4 UKR_Globular_Amphora
27.2 RUS_Progress_En
6.4 TUR_Marmara_Barcin_N
0.0 ARM_Masis_Blur_N
0.0 Baltic_LTU_Narva
0.0 GEO_CHG
0.0 IRN_Ganj_Dareh_N
0.0 Levant_PPNB
0.0 POL_Globular_Amphora
0.0 ROU_Iron_Gates_HG
0.0 RUS_Karelia_HG
0.0 RUS_Samara_HG
0.0 RUS_Vonyuchka_En
0.0 UKR_Meso
0.0 UKR_N
0.0 UKR_Trypillia
0.0 WHG

Or, in terms of only pre-eneolithic components:

Target: BGR_C_outlier:I2181
Distance: 6.5089% / 0.06508901
58.2 UKR_Globular_Amphora
15.6 RUS_Karelia_HG
15.0 TUR_Marmara_Barcin_N
9.6 GEO_CHG
1.6 RUS_Samara_HG
0.0 ARM_Masis_Blur_N
0.0 Baltic_LTU_Narva
0.0 IRN_Ganj_Dareh_N
0.0 Levant_PPNB
0.0 POL_Globular_Amphora
0.0 ROU_Iron_Gates_HG
0.0 UKR_Meso
0.0 UKR_N
0.0 UKR_Trypillia
0.0 WHG

In reality, I expect this amount of Globular Amphora ancestry may be inflated, and there's "really" more Ukrainian forager ancestry and more local Balkan farmer ancestry, but the important thing is that we're looking at a ~4500 BCE (i.e., not-that-long-before-proto-Anatolian) sample in the East Balkans, who is steppe_en (i.e., EHG+CHG) and Eastern European farmer, and nothing else.

Is this guy a good candidate to be representative of a putative eneolithic Balkan population which took proto-Anatolian into Anatolia? Certainly.

Do any CA and EBA Anatolian samples score any ancestry from him?


Target: TUR_Aegean_Izmir_Yassitepe_EBA:I5733
Distance: 2.4424% / 0.02442411
49.0 ARM_Masis_Blur_N
29.8 TUR_Marmara_Barcin_N
15.0 BGR_C_outlier
4.8 IRN_Ganj_Dareh_N
1.4 GEO_CHG
0.0 Levant_PPNB


Target: TUR_Ikiztepe_LC:IKI017
Distance: 4.6753% / 0.04675260
37.4 ARM_Masis_Blur_N
29.6 TUR_Marmara_Barcin_N
20.2 GEO_CHG
12.8 BGR_C_outlier
0.0 IRN_Ganj_Dareh_N
0.0 Levant_PPNB


Target: TUR_Marmara_Barcin_C:I1584
Distance: 2.6973% / 0.02697251
34.2 TUR_Marmara_Barcin_N
30.8 ARM_Masis_Blur_N
13.8 GEO_CHG
12.0 BGR_C_outlier
9.2 IRN_Ganj_Dareh_N
0.0 Levant_PPNB


Target: TUR_Arslantepe_LC:ART018
Distance: 2.4222% / 0.02422227
63.0 ARM_Masis_Blur_N
14.4 GEO_CHG
11.2 BGR_C_outlier
9.2 IRN_Ganj_Dareh_N
1.4 TUR_Marmara_Barcin_N
0.8 Levant_PPNB


Target: TUR_Aegean_Izmir_Yassitepe_EBA:I5735
Distance: 3.8114% / 0.03811383
47.8 TUR_Marmara_Barcin_N
12.4 IRN_Ganj_Dareh_N
11.0 BGR_C_outlier
10.6 ARM_Masis_Blur_N
9.8 GEO_CHG
8.4 Levant_PPNB

Target: TUR_Marmara_Ilipinar_C:I10542
Distance: 2.5937% / 0.02593727
50.0 ARM_Masis_Blur_N
36.2 TUR_Marmara_Barcin_N
7.0 IRN_Ganj_Dareh_N
6.6 BGR_C_outlier
0.2 Levant_PPNB
0.0 GEO_CHG

H₂ŕ̥ḱtos said...

@vAsiSTha

Cont.:

These aren't the only ones, though admittedly, the signal from this putative proto-Anatolian-speaking Balkan population decreases from here. If anyone's interested, they can change the model around a bit, but I've noticed that actually, you can use straight Yamnaya instead of this guy, and you get numbers which are significantly more than a third as large, which suggests that these numbers may even be under-stated. Davidski has hinted at there being more relevant samples on the way from both sides of the Aegean, so I'm keen to see how things progress, but for now, we can show:

1. The CHG-related ancestry on the Pontic-Caspian steppe is too old to be relevant to proto-Indo-Anatolian.

2. There is "Balkan steppe" ancestry in, not all, but some, CA and EBA Anatolian samples. Funnily enough, it seems to peak in the west, but again, more samples would allow for more resolution, which would be grand.

With this, and the information from historical linguistics (Kloekhurst and others also explain that PIA originating among a population already living a heavily agricultural lifestyle seems very unlikely, given that there seems to be a very, very scant shared PIE farming-related lexicon, with many IE farming-related terms seeming to be post-PIE loans) in mind, a Pontic-Caspian steppe origin for PIA remains likely.

Rob said...

.. neoAnatolians in the sense speakers of Hattic related languages, ultimately coming from northeast

epoch said...

Actually, what the paper says about timing of the Anatolian/IE split is this:

"The Anatolian split is placed by that study at ~3700 BCE (4314 to 3450 BCE, 95% highest posterior density interval), a period during which the Caucasus hunter-gatherer ancestry first appears as far west as the Chalcolithic individuals from Northwest Anatolia (at Ilıpınar) sampled in our study"

But that makes for a rather strange proposition, as it positions the root of Anatolian languages in NW Anatolia and it needs to somehow explain why there is no trace of IE languages left in the entire area in between.

vAsiSTha said...

@kouros

I have rebutted these arguments of kumsay_eba ancestry in Sarazm already in the first few comments of my post here

To sum it simply:
1. Sarazm samples are 500yrs older than Kumsay.
2. Sarazm qpAdm requires 0% EHG or Khvalynsk ancestry. Therefore there is no chance of it deriving ancestry from Kumsay_eba or Progress_en or ehg which are all EHG heavy.
3. The affinity that you see, is actually due to Sarazm like ancestry in Kumsay_eba (which are very similar to Steppe_Maykop).

Davidski has covered the presence of Sarazm/Geoksyur like affinity in Maykop samples here

vAsiSTha said...

@H₂ŕ̥ḱtos

"The thing is, your models of steppe_en and Yamnaya there show exactly the opposite of what Lazaridis et al. are saying in this paper. Your models show:"

These are just G25 models. Formal stats reveal more nuances. There is indeed truth in what Lazaridis is saying and I see it when modeling Afanasievo in qpAdm. There is an inexplicable failure in the Generated Dstats related to PPN.

" That CHG-related admixture was already present on the Pontic-Caspian steppe from a very early date (Allentoft's paper's pre-print talks about Middle Don foragers with 20-30% CHG-related admixture, from ~5300 BCE), and that Yamnaya has ancestry from Eastern European farmers but does not have trans-Caucasian ancestry from late neolithic or Eneolithic Anatolians or Iranians."

Yes 25% CHG ancestry is already present in Middle don foragers by 5500bce sure (khvalysnk as proxy). But how does progress_en then get 40% more (CHG+Sarazm-like) ancestry on top of khvalysnk?

Target: RUS_Progress_En
Distance: 3.2616% / 0.03261607
59.6 RUS_Khvalynsk_En
23.0 TJK_Sarazm_En
17.4 GEO_CHG

Whatever the true source is, it is surely not from the steppe. To prove that it is from the steppe and has been local for a long time, you have to prove the presence of samples in the steppe from 6000bce which already have 55% CHG(like) ancestry. Nothing like that has been seen yet.

So the base case is that there was a second wave from the south from an unknown source. This is correctly claimed by both Allentoft and Lazaridis and is consistent with the Chintalapati date of ehg/chg admixture (4100bce +- Std errors).

With respect to the BGR_C outlier, you are showing me models with a distance of 5-6%, and it's hard for me to take them seriously. Even if what you are saying is correct, the inflow into Anatolia from the east is just much larger (40-50%) and is much more consistent with lasting language change than a few outliers. If that is the standard we are going by, that a minor % ancestry among 2-3 samples decides the vector of language change in spite of the presence of another source that presents a much larger chunk of ancestry, then we can overturn most of the consensus results using this logic.

vAsiSTha said...

@H₂ŕ̥ḱtos

Not to mention that the BGR_C_outlier has only 37k SNPS, and is contamination is marked 'very high' in the latest anno file.

The steppe signal you are seeing is likely from moderns.





H₂ŕ̥ḱtos said...

@vAsiSTha

I appreciate the respectful and well-reasoned reply!

'These are just G25 models. Formal stats reveal more nuances. There is indeed truth in what Lazaridis is saying and I see it when modeling Afanasievo in qpAdm. There is an inexplicable failure in the Generated Dstats related to PPN.'

You're right, of course, but then, on the note of formal statistics, I've seen qpAdm models of Yamnaya which either fail with Levant_PPNB, or which include a drop of it, with an error rate of around three times the actual quantity of admixture, but which happily include European farmer admixture. I think this makes a lot of sense when we consider that we have archaeologically-documented interactions between Yamnaya and related steppe groups, and European farming societies such as Globular Amphora and Cucuteni-Trypillia. Ultimately, I remain unconvinced, given the archaeological evidence and the modelling with formal statistics that I've been shown, that the ANF ancestry in Yamnaya came from a Levant_PPNB-packing Caucasian migration, rather than from admixture with European farmers.


'Yes 25% CHG ancestry is already present in Middle don foragers by 5500bce sure (khvalysnk as proxy). But how does progress_en then get 40% more (CHG+Sarazm-like) ancestry on top of khvalysnk?'

'Whatever the true source is, it is surely not from the steppe. To prove that it is from the steppe and has been local for a long time, you have to prove the presence of samples in the steppe from 6000bce which already have 55% CHG(like) ancestry. Nothing like that has been seen yet.

So the base case is that there was a second wave from the south from an unknown source. This is correctly claimed by both Allentoft and Lazaridis and is consistent with the Chintalapati date of ehg/chg admixture (4100bce +- Std errors).'

Yeah you're quite right - Essentially that's what I'm suggesting, that the source of this CHG-related ancestry in steppe_en is something as-yet-unsampled. As you say, we haven't found anything with 55% CHG-like ancestry from 6000BCE, but we have found such samples from later on, in Progress and Vonyuchka. I don't find it unreasonable that there could have been strong continuity in the Caucasus piedmont region from 6000BCE to 4000BCE or so, with the ancestors of Progress_en and Vonyuchka_en providing ancestry to groups further to the north. With regards to the dating of the EHG-CHG admixture event you mention there, I'll need to go and have a look at that, but I've seen mention of calculated dates of this admixture event going back to the mesolithic, so I don't know what to make of this conflicting information right now. We have mesolithic Karelian HGs with Y-hg J1, so there must have been some relevant admixture flowing into Eastern Europe in the mesolithic (and indeed, this is the only time period we have strong evidence (i.e., Y-DNA evidence) of male-mediated admixture from a CHG-like group into the area!).

H₂ŕ̥ḱtos said...

@vAsiSTha

'With respect to the BGR_C outlier, you are showing me models with a distance of 5-6%, and it's hard for me to take them seriously. Even if what you are saying is correct, the inflow into Anatolia from the east is just much larger (40-50%) and is much more consistent with lasting language change than a few outliers. If that is the standard we are going by, that a minor % ancestry among 2-3 samples decides the vector of language change in spite of the presence of another source that presents a much larger chunk of ancestry, then we can overturn most of the consensus results using this logic.

Not to mention that the BGR_C_outlier has only 37k SNPS, and is contamination is marked 'very high' in the latest anno file.

The steppe signal you are seeing is likely from moderns.'

Yep, all fair comments, though:

With regards to the question of ancestry proportions and linguistic affiliation, consider the amount of steppe ancestry in Sardinians, the amount of Siberian ancestry in Estonians and Hungarians, and the difference (or, to the point, lack thereof) in ancestry proportions between iron age Italics and Etruscans. I tend to agree with what you're saying, but we can't act like linguistic changes must always follow the DNA 100%, so to speak. I would say that generally speaking, a sudden pulse of new Y-DNA lines into an area seems to be more likely to accompany a linguistic shift than any other type of genomic change, but then, even that isn't a bulletproof statement because look at Basques. Strange things happen!

With regards to the BGR_C_o sample himself, yeah he's a poor coverage boi, which is a shame. I chose him because he's very relevant in terms of timescale to the question at hand. We can use Yamnaya_Samara, or Yammaya_Bulgaria instead - ultimately my point is that we can produce good, sensible models of CA and EBA Anatolians with steppe ancestry, despite what Lazairidis is claiming off the back of using EHG and CHG, instead of Yamnaya (or something Yamnaya-derived) and Iran_N.

I think ultimately, it's clear there are issues with some of Lazaridis' modelling choices. More samples and more models are bound to be coming out over the next... whatever span of time. Davidski has hinted that he's aware of some of them. We'll see how the picture develops, but my predictions about what is likely remain the same.