search this blog

Wednesday, April 11, 2018

Bronze Age Central Asia: terra incognita no longer


I've updated my Global25 datasheets with the samples from the Narasimhan et al. 2018 preprint (look for these labels). Feel free to use this output for anything you like, and please show us the results in the comments below.

Global 25 datasheet

Global 25 datasheet (scaled)

Global 25 pop averages

Global 25 pop averages (scaled)

Also, here's my Principal Component Analysis (PCA) of ancient West Eurasia featuring most of the new samples. Note the cline made up of ancient and present-day South Asians running from the likely Indus Valley diaspora individuals (from the Gonur Tepe and Shahr-i Sokhta archaeological sites, in present-day Turkmenistan and Iran, respectively) towards the Bronze Age steppe. The relevant datasheet is available here.


I have little doubt that these are indeed migrants from the Indus Valley Civilization (IVC). Their relatively unusual genetic structure - which includes ancestry from an West Eurasian ghost population that is inferred to have been exceedingly poor in Anatolian-related ancestry, as well as significant indigenous South Asian ancestry - leaves little scope for plausible alternatives. If you're wondering what they may have been doing so far north of the IVC, Frenez 2018 has a detailed discussion on the topic. From the paper:

An alternative and intriguing hypothesis is instead supported by significant archaeological and textual data from comparable socio-economic or geographical contexts, which suggest that the likely high commercial and ideological value of ivory and of the expertise required to carve it made also possible and economically profitable the presence in Central Asia of independent itinerant ivory carvers native to or trained in the Indus Valley. These itinerant artisans might have provided at the same time both the raw material and the unique skills to transform it into finished objects.

...

Moreover, the existence of itinerant ivory workers in ancient South Asia is also described in a few literary sources. The Guttila Jātaka mentions a group of ivory carvers who traveled from Benares to Ujjain to offer their products and skills to the local elites (Pal, 1978: 46), while a Buddhist Sanskrit Vinaya tells the story of an Indian master ivory carver who traveled “up to the land of the Yavanas”, most likely the Hellenistic Bactria, to put his superior expertise at the service of a renown local artist (Dwivedi, 1976: 19).

Citation: Frenez, D., Manufacturing and trade of Asian elephant ivory in Bronze Age Middle Asia. Evidence from Gonur Depe (Margiana, Turkmenistan), Archaeological Research in Asia (2017), http://dx.doi.org/10.1016/j.ara.2017.08.002

See also...

On the doorstep of India

255 comments:

1 – 200 of 255   Newer›   Newest»
Ariel said...

I made a quick graph with the new samples https://imgur.com/a/Hxyao

capra internetensis said...

Awesome Davidski, thank you.

OK, using Corded Ware Baltic, Yamnaya Samara, West Siberia N, Balkans Chl and N, Sappali Tepe BMAC, Tepe Anau, Tepe Hissar, Geoksyur, Sarazm, and Shahr-i-SokhtaBA3 (the most Indian of the probable Harappans) as sources:


Archaeologically, IIRC, Sarazm is kind of a Geoksyur colony with Kelteminar people around:

Sarazm Eneolithic - 90.8% Geoksyur, 6.0% WSN, 3.2% S-i-S3 - distance 3.44%

So Kelteminar maybe is pretty much WSN-like, the Indian shift could be there or Geoksyur structure? Or Kelteminar could already be mixed with Geoksyur-like ancestry as well.

Afanasievo - 98.6% Yamnaya, 1.4% Sarazm - distance 1.66%

Srubnaya Outlier - 51.2% Yamnaya, 40.4% WSN, 8.4% Sarazm - distance 3.29%

Sintashta mainstream - 45% Baltic CWC, 39.8% Yamnaya, 11.4% BalkansChl, 2.6% BalkansN, 1.2% WSN - distance 2.01%
Sintashta Outlier 1 - 55.2% WSN, 25.2% Yamnaya, 19.6% Sarazm - distance 2.69%
Sintashta Outlier 2 - 52.8% Yamnaya, 20.4% Baltic CWC, 17.4% WSN, 9.2% BalkansN, 0.2% S-i-S3 - distance 2.51%
Sintashta Outlier 3 - 73.6% WSN, 26.4% Baltic CWC - distance 3.52%

lots of WSN, seems like Sarazm is the best representative of Anau-Namazga-BMAC type ancestry further north?

Alberto said...

Great! Thank you!

This is going to take quite a while to process and find the best setups for different populations, examine individuals, etc... So I'm sure that results will vary as we test more. But here is a quick run with the same setup that I used previously with Matt's simulations, except that I replaced the Indus_periphery_simulation for the 3 real samples in the sources.

At the bottom I added the new SC Asian samples as targets. Not sure if I missed any:

https://docs.google.com/spreadsheets/d/12_O8xooAUn3-QYvEhDpa6YIW6Ad45b0cUKSfV1uxH1s/edit?usp=sharing

Mr. Kulkarni said...

https://twitter.com/anupampom/status/982284507293331456?s=19
A good geographical and caste wise jati wise breakup thread of modern indian population using ancestry data from narsimhan paper.

Matt said...

Awesome. I'm a bit stymied by waaay too many populations to really visualise what's going on, and the huge diversity within some of these populations (to a far greater degree than the simpler MN and EMBA) makes it really hard to visualise as well, without averages which are a bit misleading, and a bit wary of walking into simulations without really being able to "see" what's going on.

Will say, Gonur2_BA and Shahr_I_Sokhta_BA2 are pretty much where the Indus_Diaspora_related sim I previously made are... but a little closer to present day South Asia and all real populations, which should reduce the affinity to Steppe_MLBA somewhat. The Iron Age samples look displaced slightly between Pashtun/Kalash and Gonur2_BA / Shahr_I_Sokhta_BA2, suggests that they harbour less Steppe_MLBA ancestry than present day Pashtun/Kalash?

@Alberto, nice to see that try - I'm not sure actually now about using the AASI zombie there with the real Indus_Periphery... seems like there might be some issue with that?

Rob said...

So HF Outlier is probably a Majkop like group that crashed onto the steppe
Aweosme !

Ariel said...

Matt

There are samples that are incredibly steppe shifted like one Gonur and Hajji Firuz BA, others in the same period and location have very little to no steppe.

Davidski said...

There was clearly a movement of steppe people, maybe Yamnaya itself, into what is now Northwestern Iran during the Bronze Age.

Awesome!

Lenny Dykstra said...

EARLY Bronze Age, is the most important thing. I think most expected IE admixture in North Iran after maybe 1,000 BC, with the Indo-Iranians. But this EBA steppe admix in the north Zagros may indeed be some proto-Anatolian branch speakers...

This is the date & context for the heavily steppe-shifted BA individual in your PCA (could the R-Z2103 male, with less steppe-shifted admixture, have a similar C14 date??)

• F11, 3 (I4243): Date of 2465-2286 cal BCE (3875±25 BP, PSUAMS-2113). Genetically
1716 female. This individual is a genetic outlier and is also an intrusive burial from the Bronze Age based on its radiocarbon dating, so we remove them from the main analysis dataset (thus, the main analysis dataset only contains 6 individuals).

capra internetensis said...

@Rob

Yamnaya Samara - 69% Khvalynsk, 19% CHG, 8% Armenia Chl, 2.8% Hajji Firuz Chl outlier, 0.6% Sweden MN, 0.6% Tepe Anau - distance 3.8908%
Yamnaya Samara - 69% Khvalynsk, 19% CHG, 10% Armenia Chl, 1.6% Tepe Anau, 0.6% Sweden MN - distance 3.8918%
Yamnaya Samara - 69% Khvalynsk, 20% CHG, 9% Hajji Firuz outlier, 2.2% Sweden MN - distance 3.8994%
Yamnaya Samara - 70% Khvalynsk, 20% CHG, 4.6% Balkans Chl, 4.6% Tepe Anau, 0.8% Sweden MN - distance 3.9282%
Geoksyur, Sarazm, and Iran Chl are marginally worse than Tepe Anau.

It actually does work but doesn't really seem much different from Armenia or Iran Chl. Does look like something from the direction of northern Iran is wanted. I think less likely to have come via Central Asia because there is no WSN or extra ANE?

Khvalynsk itself can take a few percent of HF outlier and such also, though again the improvement is ~0.01%.

Davidski said...

@Rob

When the BAM files are released this sample's R1b lineage will look even more downstream of Z2103 than it already does, making it impossible that it's from the Chalcolithic, C14 dating or not.

Samuel Andrews said...

Tajik are truly a unique people group. Their speak Persian but their ancestors were eastern Iranian speakers. They alone in a region full of Turkic & Indo Aryan speakers.

Both mtDNA & genome-wide data confirms they are probably of about 50% Andronovo origin. They're basically a Iran farmer+Andronovo East mix like how northern Europeans are a European farmer+Yamnaya-like mix.

They have much more Andronovo and less ASI than Pathan or Kalash and their Turkic neighbors. This could be because they are the last east iranian grouo left? Turks definitly reduced the level of Andronovo ancestry. In the Iron age, I'd imagine Central Asia with east Iranians & west iranians was mostly of Andronovo origin.

Samuel Andrews said...

A key question for India is, how much of their ASI+Iran Neo is from the Indus Valley Civilization and how much is from unidentified peoples who lived south & east of the Indus Valley.

mzp1 said...

@samuel

Pashtuns still speak East Iranian. They are super-macho tribal and very much dominate the tajiks in most of Afghanistan. Their culture is very IE.

Rob said...

@ Dave
As I said, we all agree that the call is suspect
It’s hard to believe such a downstream Clade existed in 5000 BC

Seinundzeit said...

David,

Thanks!

This is going to be so much fun (I'll start on it, once I get home). I already have plans for various analyses...

mzp,

Yeah, I think it is interesting how, despite having less Indo-Iranian steppe admixture, Pashtun culture has turned out to be much more similar to early Indo-Iranian culture than the culture of Pamiri peoples.

I mean, Pamiri peoples aren't tribally organized, none of them are nomadic, and they are very peaceful people. By contrast, Pashtuns are organized into patrilineally structured tribes, have a history of nomadism, and have a culture in which violence is an essential component.

I guess the terrain suits it better?

Anthro Survey said...

@Alberto

Nice runs. So, those Brahmins do have considerably less steppe than 40%(lol) like I've been saying for months now. Awe-some. I'm a bit curious about those Swat samples, though. Have you managed to obtain successful fits without using any steppe_MLBA?

@All: That Hajji Firuz BA sample is interesting, to be sure, but hard to make any firm conclusions. The steppe-Caucasus interaction zone predates formation of Yamnaya groups. Bit of a chicken-n-egg as things stand.

mzp1 said...

I was referring to the Tajiks of Agfhanistan (and Tajikistan) not the mountain Pamiris (who are not Tajiks are far as I know).

Pamiri peoples are mountain tribes (badakshan region) so they are more similar to other mountain tribes in their surrounds and dont have much need for warfare as outsiders rarely come their way. No one would go to India through there.

Tajiks are people like Ahmed Shah Masood (Soviet-era Mujahidin leader) who come from Northern Afghanistan and Tajikistan (though most of these are mongolian admixed). They have been dominated and/or settled for a long time.

Pashtuns have probably had a martial culture from very early on, having neither an isolated mountain terrain, nor much urbanisation throughout history. They are physically very strong and look rather different to Pakistani Punjabies. Though Iranian by language they are close in many ways to Pakistani Punjabies.

I imagine both Punjabies and Pashtuns had martial cultures in the past, and perhaps there was little difference in the two. Punjab though became 'civilised' due to urbanisation and trade, but the Pashtun territory doesnt allow for this so it seems they maintained the martial society.

Punjabies are also very aggressive in many ways but 1000s of years of sedentary 'civilization' probably made them somewhat more cooperative. I was talking to a Punjabi about this just the other too.

Anthro Survey said...

@Seinundzeit/Dave/Alberto/anyone with knowledge about this:

Where are those new Tajik_S samples from, by the way? They lack virtually any Turkic-related ENA ancestry, it looks like. Must be either from Kabul or Herat area, rather than Oxus or Bactrian Tajiks.

Davidski said...

@Anthro Survey

That Hajji Firuz BA sample is interesting, to be sure, but hard to make any firm conclusions. The steppe-Caucasus interaction zone predates formation of Yamnaya groups. Bit of a chicken-n-egg as things stand.

Nope, because those Hajji Firuz samples with Yamnaya-related input both probably date to around 2465-2286 cal BCE. So maybe they're not even as old as Yamnaya, but closer to Catacomb.

Where are those new Tajik_S samples from, by the way? They lack virtually any Turkic-related ENA ancestry, it looks like. Must be either from Kabul or Herat area, rather than Oxus or Bactrian Tajiks.

I don't know, but they're from the SGDP.

https://www.simonsfoundation.org/simons-genome-diversity-project/

Anthro Survey said...

@mzp and Sein

Yeah, the violence in Pashtun culture may not be easily attributable to steppe admixture. Could indeed have been an organic development prompted by the environment or rooted in the culture of the local people genetically resembling BMAC.

Take the Albanians. Very violent culture, a lot of near-genocidal sentiments towards outgroups like Greeks and Balkans Slavs, uber-patriarchal(honor killings are still big issue, and, no, Islam's not the cause because Catholic Ghegs have more of this). YET, they aren't exactly high on steppe admixture, much of which arrived with the Slavs who were linguistically assimilated.

Anthro Survey said...

@Dave

That's true and I'm not discounting this possibility(in fact I've suggested a separate steppe stream in Western Iran before), but we can't exclude intra-Caucasus dynamics in play. Look at how EHG-rich those Armenia_Chl were.

I see, thanks. I'll have to email those people, I guess. May not be bona-fide Bactrian Tajiks but East Iranic tribals.

AWood said...

@All: That Hajji Firuz BA sample is interesting, to be sure, but hard to make any firm conclusions. The steppe-Caucasus interaction zone predates formation of Yamnaya groups. Bit of a chicken-n-egg as things stand.

"Yamnaya" is just a label, it doesn't mean that EHG wasn't already creeping south east along the steppes of the Caspian/Azerbaijan a little earlier. In fact, depending on the radio carbon dating of the Z2103 'Hajji Firuz' sample, that might be exactly what we see.

The evidence still suggests Uruk and related Trans-Caucasus cultures moving north spreading J1, J2 and so forth, and R1 related groups moving southwards.

Arza said...

@ Davidski

I think that I know why they didn't detect steppe admixture in I2327.
He can be modelled as Iran + a tiny fraction of Iron_Gates/Latvia_HG:

Mbuti.DG Iron_Gates_HG Hajji_Firuz_C Hajji_Firuz_C_I2327 0.0060 1.696 467101
Mbuti.DG Iron_Gates_HG Iran_ChL Hajji_Firuz_C_I2327 0.0143 4.136 459370
Mbuti.DG Iron_Gates_HG Iran_N Hajji_Firuz_C_I2327 0.0427 12.323 474404
Mbuti.DG Latvia_HG Hajji_Firuz_C Hajji_Firuz_C_I2327 0.0100 2.772 465830
Mbuti.DG Latvia_HG Iran_ChL Hajji_Firuz_C_I2327 0.0177 4.944 457940
Mbuti.DG Latvia_HG Iran_N Hajji_Firuz_C_I2327 0.0449 12.470 472060

numsnps used: 161195
chisq: 6.144 . tail: 0.863597653
Iran_N . Hajji_Firuz_C . Latvia_HG
best coefficients: 0.174 0.762 0.064
std. errors: 0.104 0.108 0.029

Here is something interesting if you'll try Iran_ChL:

w/o CHG in right pops
numsnps used: 175904
chisq: 5.583 . tail: 0.89969675
Iran_ChL . Ukraine_Mesolithic . Yamnaya_Samara
best coefficients: 0.998 0.358 -0.356
std. errors: 0.130 0.176 0.298

w CHG
numsnps used: 175815
chisq: 8.395 . tail: 0.753529974
Iran_ChL . Ukraine_Mesolithic . Yamnaya_Samara
best coefficients: 0.880 0.178 -0.058
std. errors: 0.094 0.119 0.203

W/o CHG he's strongly repelled from Yamnaya, at best -5% if you'll take std. err into account.
With CHG at best 15%, but it shows Yamnaya at negative score anyway.

And finally:
numsnps used: 163467
chisq: 4.049 . tail: 0.982527695
Hajji_Firuz_C . Latvia_HG . Yamnaya_Samara
best coefficients: 0.808 -0.011 0.203
std. errors: 0.061 0.043 0.088
Output without Latvia: 4.428 0.985723 0.815 0.000 0.185

So exactly as your G25 model with 2.8% of West_Siberia_N (Yamnaya pushed away from Latvia_HG by a tiny fraction).

With Ukraine_Meso (note the higher std. err.):
numsnps used: 177406
chisq: 3.686 . tail: 0.978210001
Hajji_Firuz_C . Ukraine_Mesolithic . Yamnaya_Samara
best coefficients: 0.808 -0.015 0.206
std. errors: 0.090 0.118 0.196

So the reason of the whole confusion is here:
3.686 0.97821 0.808 -0.015 0.206 infeasible
5.006 0.957778 0.899 0.101 0.000
3.851 0.985985 0.816 -0.000 0.184

Orange component maximized in IIRC 3 Srubnaya samples covered up this slight WHG shift and probably due to the lack of CHG and proper WHG and Anatolian components algorithm has forced a model in which Yamnaya scores are negative (local minimum?).

BTW D-stats aren't helpful here too:
Mbuti.DG Ukraine_Meso Hajji_Firuz_C Hajji_Firuz_C_I2327 0.0104 2.638 457435
Mbuti.DG Ukraine_Meso Iran_ChL Hajji_Firuz_C_I2327 0.0186 4.929 449651
Mbuti.DG Ukraine_Neo Hajji_Firuz_C Hajji_Firuz_C_I2327 0.0069 1.963 462305
Mbuti.DG Ukraine_Neo Iran_ChL Hajji_Firuz_C_I2327 0.0159 4.528 454440
Mbuti.DG Yamnaya_Samara Hajji_Firuz_C Hajji_Firuz_C_I2327 0.0087 2.458 465917
Mbuti.DG Yamnaya_Samara Iran_ChL Hajji_Firuz_C_I2327 0.0104 3.037 458375

PS Thanks for that low_res.

Davidski said...

Well, the genome-wide and Y-chromosome evidence strongly suggest that Hajji Firuz I2327 has Yamnaya-related ancestry.

It'll be interesting to see what his Z2103 lineage ends up being after the BAM files are released. If it's even more derived than currently, then we could be looking at Catacomb incursions into West Asia here.

Davidski said...

@Alberto

Why are you using the old Andronovo and Srubnaya_outlier samples in your models? They're low quality and no longer relevant.

Move onto the Steppe_MLBA_East groups. Try looking for the best fitting group from Kazakhstan for the steppe ancestry in upper caste Indians.

Rob said...

@ Capra

Thanks for the preview, I hadn;t yet looked at the data, but was merely entertaining metaphors
I agree with your analysis - Haji Garenz look just like other Iran Chalc, even the alleged R1b guy.

For Yamnaya, they remain the usual EHG/ CHG model, with the caviet of finding the source of CHG.

Interesting is the Dali EBA:
W Siberian N 81%/ CHG 8% / Dai 2/ Haji Firuz 2

Davidski said...

@Rob

I agree with your analysis - Haji Garenz look just like other Iran Chalc, even the alleged R1b guy.

Very funny.

Lenny Dykstra said...

David,

I just downloaded the scaled datasheet and pop averages, and for some reason Iran_N and Iran_Chl appear to have been deleted from the sheet? Strange...

Also, I know I asked this in other threads so apologies if you answered and I missed it, but would you be able to add the two negrito Hoabhinian hunter gatherers from the McColl paper to the Global 25?

I think Hoabhinian could combine with Onge to form an "AASI" cluster that would be totally free of West Eurasian admixture...

Davidski said...

@Lenny

Look for Abdul_Hosein_N, Ganj_Dareh_N, Wezmah_Cave_N and Seh_Gabi_ChL.

And I can't download the Southeast Asian ancient data yet, because it hasn't been published online.

Samuel Andrews said...

Tajik are more or less a two way mixture between BMAC-types and Andronovo-east types with little or no South Asian input. Kashkarchi=Steppe_MLBA_East.

1.4402"

Tajik_Yagnobi

Gonur1_BA,46.6
Kashkarchi_BA,38.2
Barcin_N,7.4
Mongola,6
CHG,1
Bustan_BA,0.5
Natufian,0.3

namedguest said...

Sintashta keeps appearing for the R1b Hajji Firuz guy, no matter how much I tweak things.
So, it's pointing out to be what David says, there might be a radioncarbon dating problem here.

About what's written in the paper:
"Hajji Firuz, can be modeled parsimoniously as a mixture of just 2 sources: a population related to that from Neolithic Iran and a population related to the agriculturalists from Anatolia."

Yeah, that's what's appearing, but they forgotten that ~15~20% Steppe signal appearing as well (not ANE, not West_Siberia).

Seinundzeit said...

Anthro Survey,

That makes sense; the high levels of violence (blood feuds, tribal conflict, territorial disputes, extraction of tolls, raids, etc) could certainly be a local development.

mzp,

"They are physically very strong and look rather different to Pakistani Punjabies."

True; I've been to the tribal areas, and there is no doubt that tribal Pashtuns are very impressive looking folk.

On average, they're burly, broad-shouldered, hirsute, and have very angular/chiseled facial features, giving them an interesting appearance.

All,

Okay, finally have to time to do some exploration.

First off, I wanted to have a feel for what we're working with.

(Sidenote: "AASI" is a simulation, based on the "Bhumji1" sample. As per David's qpAdm exploration, Bhumji1 is 75% AASI, 25% West Eurasian, and thus the most AASI sample from contemporary South Asia. This simulation was based on a utilization of the assumption of 75% AASI for that sample, and from there I went. If you people want, I can post the coordinates)

West_Siberian_N:

70.95% AfontovaGora3
18.30% EHG
6.95% Ulchi
3.80% Mongola

"distance%=0.3474 / distance=0.003474"

Yeah, West_Siberian_N are simply ANE, and with some very minor East Asian admixture.

Eneolithic Central Asians:

Sarazm_Eneolithic

57.45% Iran_N
23.85% CHG
12.55% AfontovaGora3
6.15% West_Siberia_N

"distance%=0.4086 / distance=0.004086"

Yeah, as I expected, mostly Iran_N, CHG, with some ANE/West_Siberian_N.

BMAC:

Bustan_BA

39.40% Iran_Chl
38.20% Iran_N
7.65% AfontovaGora3
6.15% CHG
5.20% Steppe_MLBA
3.40% AASI

"distance%=0.2397 / distance=0.002397"

Shahr_I_Sokhta_BA1

66.45% Iran_N
9.95% Iran_Chl
8.95% CHG
5.55% Steppe_MLBA
5.20% AASI
3.90% AfontovaGora

"distance%=0.3268 / distance=0.003268"

Dzharkutan1_BA

43.55% Iran_N
30.80% Iran_Chl
9.55% CHG
6.85% AfontovaGora3
5.00% Steppe_MLBA
2.55% West_Siberia_N
1.70% AASI

"distance%=0.2287 / distance=0.002287"

Sappeli_Tepe_BA

40.30% Iran_N
37.95% Iran_Chl
8.85% CHG
8.15% Steppe_MLBA
2.70% AfontovaGora3
2.05% AASI

"distance%=0.1891 / distance=0.001891"

In contrast to Eneolithic Central Asians, we start to see Iran_Chl influence, and some minor Steppe_MLBA. Not to mention persistent percentages of very minor AASI.

The essentially West Eurasian IVC migrant:

Shahr_I_Sokhta_BA2

64.25% Iran_N
17.45% AASI
11.55% AfontovaGora3
6.75% CHG

"distance%=0.569 / distance=0.00569"

This sample is essentially West Eurasian (only 15%-20% AASI), and is quite distinct from BMAC, but not shockingly different from Eneolithic Central Asians. Take away the 15%-20% AASI, and it'll be quite similar.

Also, we can say with confidence that the paper's "West_Siberia_N" percentages do not reflect an influx from Siberia. Rather, considering the preference towards AG3, I think we are on solid ground if we construe the West_Siberia_N percentages as reflecting local hunter gatherer ancestry (folks like the Kelteminar people).

Finally, Steppe_MLBA_East:

79.95% Steppe_MLBA
13.45% West_Siberia_N
4.75% CHG
1.85% Ulchi

"distance%=0.1363 / distance=0.001363"

As one would expect.

Now, we can start to explore contemporary South Central Asia and South Asia, using these ancient populations. I'll start posting results, later tonight...

For the king said...

@Sein nice work. You should try modeling Sappali teppe outlier, Parkhai LBA outlier and the Afghan sample as well.

Lenny Dykstra said...

The fit for Burusho is significantly improved by adding Scythian_ZevakinoChilikta. Burusho strongly prefers Zevakino over any of the other Steppe_MLBA_East type groups. Still, pretty remarkable continuity between IA Pakistan and Burusho.

Without Scythian_Zevakino:

[1] "distance%=2.7642"
Burusho

Udegram_IA,77.6
Onge,13.2
Zevakinskiy_LBA,3.4
Zevakinskiy_BA,3
Krasnoyarsk_MLBA,1.6
Sintashta_MLBA,1.2


With Scythian_Zevakino:

[1] "distance%=1.802"
Burusho

Udegram_IA,78.6
Scythian_ZevakinoChilikta,10
Onge,8
Sintashta_MLBA,3.4

Note that West_Siberia_N cannot improve the fit; only Scythian_Zev can get Burusho to fit this tight.

Playing around with Kalash, Pashtun, and Pamiri Tajiks, the former have basically zero affinity to Scythian_Zev; Pamiris pick up 4% Scyth_Zev but that's nothing compared to the ~54% combined Sintashta and Zevakinski steppe they pick up.

I think what's going on here is that Burusho have affinity not just with the "West Siberian" signal (which is accounted for by Udegram and Sintashta), but also a deeper relation to Baikal HGs than other South Asians have. And the Baikal_HG (or "Central Siberian) element is probably present in Scythian_Zevakino.

Jijnasu said...

@seinundzeit
Do the bhumij lack the east asian ancestry that other austro-asiatic speakers in India like the juang have?

Seinundzeit said...

Jijnasu,

East Asian admixture in this population is rather minimal; their ENA is almost completely AASI, unlike the Bonda.

For the king,

Thanks; and I'll work on those samples pretty soon. Although, I'm not seeing the Afghan sample in the Global_25 data-sheet?

Right now, I'm running contemporary South Central Asians, South Asians, and the ancient samples from the Swat valley. So far, the output is incredibly interesting. Many patterns of great interest.

I've always dreamed about this day.

Also, the Steppe_MLBA_East population seen above is Dali_MLBA. I'm using it in my contemporary models, and it works great.

Once I'm done running all these populations, I'll post the output.

Chetan said...

So we are back to square one. At least, if the sample turned out to be in the range 4500-4000 BCE, we could have speculated about a Proto-Anatolian migration through the Caucasus. But even that looks unlikely now. So the search has to be continued in the Balkans I guess

Rob said...

@ Dave said "very funny"

My bad. Didnt realise he was labelled separately.


Hajji_Firuz_ChL_o:I2327
Iran_N 37.7 %
Barcin_N 24.7 %
CHG 17.2 %
Levant_N 12.6 %
EHG 3.9 %
West_Siberia_N 3.9 %


Yes, he's got some EHG.

EastPole said...

After Narasimhan et al. publication there is a lot of talk about Aryans conquering India and Europe, for example in “The Economist”:

“The wider study not only confirms that “Aryans” (geneticists avoid the term) probably migrated from the steppes around the Volga and Don rivers to both India and Europe at around the same time.”

https://www.economist.com/sites/default/files/imagecache/640-width/images/print-edition/20180407_ASM994.png

https://www.economist.com/news/asia/21740048-aryans-did-not-come-india-they-conquered-it-new-study-squelches-treasured-theory-about

Some people interpret this as a confirmation of Hitler’s theory.
I think it is a total BS. Hitler’s Aryans never existed. The only true Aryans, Indo-Aryans or wider Indo-Iranians, seem to be related genetically and linguistically to Balto-Slavic people. They have nothing to do with Germanics, Western European ‘Aryans’ and they didn’t come from Yamnaya.

Mr. Kulkarni said...

@eastpole
Except for the fact that the word arya is only found in historical iran and india. And it did not mean 'race' at all.

Davidski said...

@Rob

Yes, he's got some EHG.

No Rob, he's got Yamnaya ancestry. You'll be able to see it if you use more proximate and sensible reference samples.

But don't worry Rob, the BAM will be out soon, and it'll show that his Y-chromosome lineage is a subset of the diversity found in Yamnaya.

epoch2013 said...

The EHG admixture in West-Siberian hunter-gatherers is interesting. We now have a smoothclade from WHG in the West, via a tad EHG in Iron Gates, more in SHG, even more in Ukranian Mesolithics, 100% EHG in Samara to 20% EHG admixture with 70% ANE in WSHG.

We also know that at the onset of the Gravettian there was no ANE in the area of Moscow.

I know Chad thinks ANE is Sunghir + ENA. But whatever it is, it's an admixture that was remarkably absent pre-LGM in large parts of Eurasia for an admixture found from Han to WHG post-LGM.

Alberto said...

Nice to see many people testing the samples. It's going to take us a little while to get to understand them all. Still many details to sort out.

@Matt

Re: the use of the AASI ghost, I thought not to use it first and stick to real populations. But there's still a caveat to that. The Shahr-i-Sokhta_BA3 sample has some 42% AASI (from the paper), which is not quite enough for many/most Indian populations. However, the AASI it contains is a much better match than Onge, which forces populations to get as much as possible from that former sample so they get as little as possible from Onge (because every 5% Onge increases the distance significantly and gives a poor model). This does not allow for them to make any other choice of West Eurasian populations and remain a 2 way mix of Shahr-i-Sokhta_BA3 and Onge. In other words, the differences between different West Eurasian streams are smaller than the difference between AASI and Onge. So unfortunately we might still need to use an AASI ghost to allow freedom in the West Eurasian side (which is what we're more interested in right now). Better a model with a degree of uncertainty because of using a ghost than one with certainty of being wrong because of using Onge. But I'll keep testing to see what works best.

@Davidski

Yes, that was meant as a quick run to get a rough idea with familiar populations and models. I'm sure we'll find better setups, but it requires a few days and hundreds of runs to figure things out.

@Anthro Survey

I still didn't have the time to test much more than what's shown in that run. I'll keep posting like all the others what I find along the way.

@all

Things interesting to figure out:
- What's the best proxy for the ANE-rich part of SC Asians? Srubnaya_outlier is probably steppe admixed, so I agree with Davidski that it might be better avoided, unless we find out that non-steppe admixed samples prefer it over the other choices: West_Siberia_N, Dali_EBA, a combination of high ANE pre-steppe samples from outliers,...
- Iran_N vs. CHG in non-steppe admixed populations? Sein's models suggest that CHG works there, so it would be interesting to explore that relationship. In the end we might not need either of those anymore by using proximate sources, but still given the rather unique position of CHG it's intriguing to check that.
- Finding setups that can more or less reliably distinguish steppe from other streams (for example, Sein's model above of Shahr-i-Sokhta_BA1 shows 5.5% Steppe_MLBA, which seems not possible given the date and location). That's why I think that we'll see different setups giving different results. We just have to try to find the most realistic ones regarding the dates, geography, archaeology,... Once we get reliable Y-DNA results, that should help too.

Ariel said...

So I made this PCA graph and I added some notes/highlights trying to interpret these results, I'm looking for feedbacks... https://imgur.com/a/h4Tlm

Aram said...

EBA in NW Iran is the KA. This BA Hajji Firuz is from MBA. And it is not related to Hittites.

---

Material was found from the Dalma period (5000–4500 BC), and then following to the Pisdeli period, Chaff-Faced Ware horizons, and Kura-Araxes I and II periods. This is the Early Trans-Caucasian or Kura-Araxes culture, which spread through the Caucasus and the Urmia Basin around 3500 BC.[2]

Later, the Middle and Late Bronze Age (Urmia Ware), and Iron and Urartian/Achaemenid periods are also attested.

Seinundzeit said...

Okay, here it goes.

My reference populations:

AfontovaGora3
West_Siberia_N
Wezmeh_Cave_N (Iran_N)
Abdul_Hosein_N (Iran_N)
Ganj_Dareh_N (Iran_N)
Seh_Gabi_ChL
Geoksiur_Eneolithic
Sarazm_Eneolithic
Dzharkutan1_BA
Sappali_Tepe_BA
Shahr_I_Sokhta_BA1
Bustan_BA
Shahr_I_Sokhta_BA2 (IVC_Diaspora_West_Eurasian)
Dali_MLBA (Steppe_MLBA_East)
Ulchi
Mongola
Cambodian
AASI

So, we have coverage for ANE/West_Siberian_N, Iran_N, Iran_Chl, Central Asian_Eneolithic, BMAC, predominately West Eurasian IVC, and Steppe_MLBA_East.

With Shahr_I_Sokhta_BA2, our logic is that it’s the closest sample we have to the West Eurasian element that introduced/spread agriculture into South Asia.

I mean, the AASI admixture is pretty minor (somewhere between 10% and 20%, probably 15%), so it should be a decent representative for the West Eurasian ancestry at the root of IVC/Dravidians. Although, the percentages of Shahr_I_Soktha_BA2 will obviously absorb some AASI, due to the approximately 15% it has, but this is really minor issue.

To be continued….

Seinundzeit said...

Continuing from where we left off…

First off, let’s try South Asia.

Northern India/Pakistan:

Brahmin

53.7% Shahr_I_Sokhta_BA2
23.6% Steppe_MLBA_East
22.6% AASI

"distance%=0.3445 / distance=0.003445"

Punjabi

54.05% Shahr_I_Sokhta
31.00% AASI
13.20% Steppe_MLBA_East
1.75% West_Siberia_N

"distance%=0.3536 / distance=0.003536"

Chamar

51.4% Shahr_I_Sokhta_BA1
42.4% AASI
4.9% AfontovaGora3
1.3% Steppe_MLBA_East

"distance%=0.5325 / distance=0.005325"

South India:

Brahmin_TN

59.5% Shahr_I_Sokhta_BA2
26.1% AASI
14.5% Steppe_MLBA_East

"distance%=0.3911 / distance=0.003911"

Velamas

64.05% Shahr_I_Sokhta_BA2
33.20% AASI
1.90% Steppe_MLBA_East
0.85% AfontovaGora3

"distance%=0.4122 / distance=0.004122"

Observations:

The preprint has South Asia pinned correctly; there is a West Eurasian stream of ancestry related to Iran_N (but much more "eastern", due to an excess of ANE, and even less Natufian/Levant_N/Anatolia_N-related ancestry), Steppe_MLBA_East, and AASI.

Although, for some reason, the preprint uses Steppe_MLBA, which is odd, considering that the steppe ancestry would have been spread by Steppe_MLBA_East (they’re in the right geographical position, and have the right genetic structure).

Anyway, in South Asia, north Indian Brahmins are the only population with enough Steppe_MLBA_East to warrant attention (around 25%). Lower castes, and non-Brahmin South Indians, are quite close to 0%.

To be continued…

Seinundzeit said...

Continuing from where we left off…

Now, we can turn briefly turn our attention to ancient samples from the Swat valley.

In our current context, the Swat valley is a transitional region; it can be construed as a heavily "Indian" influenced area of southeastern Central Asia, or a heavily Central Asian influenced region of northwestern South Asia.

Looking at the ancient samples, this seems to have also been the case during the Iron Age (and it is here that the preprint is somewhat off the mark).

Ancient Swat Valley:

Barikot_IA

46.00% Shahr_I_Sokhta_BA2
13.75% Sappali_Tepe_BA + 8.80% Shahr_I_Sokhta_BA1
18.85% Steppe_MLBA_East
8.90% AASI
1.95% Mongola
0.95% West_Siberia_N

"distance%=0.1831 / distance=0.001831"

Udegram_IA

44.75% Shahr_I_Soktha_BA2
11.20% Shahr_I_Soktha_BA1 + 8.75% Sappali_Tepe_BA
15.75% Steppe_MLBA_East
9.55% AASI
4.45% West_Siberia_N
3.85% Seh_Gabi_ChL
1.70% Mongola

"distance%=0.155 / distance=0.00155"

Unlike actual South Asians, these samples do display 20% BMAC-related admixture, which ties well into the geographical position of the valley.

Not to mention the minor East Asian percentages, which don't occur in South Asia proper. These are problematic facts, since these samples are modelled like Indians/eastern Pakistanis in the preprint (only Indus_periphery, no BMAC, and no Iran_Chl)

Although, it seems that the preprint is right about the Steppe_MLBA admixture in these samples (to the tune of 15%-20%, although we are using Steppe_MLBA_East, rather than Steppe_MLBA).

Out of interest, I also tried the Brahui.

They certainly aren’t South Asians, but they also aren’t southern Central Asians. They are probably best construed as an isolated population on the southeastern fringe of West Asia. I’ll just include them here.

Brahui:

27.10% Wezmeh_Cave_N
24.80% Shahr_I_Sokhta_BA2
23.70% Steppe_MLBA_East
18.40% Seh_Gabi_Chl
6.00% AASI

"distance%=0.2244 / distance=0.002244"

Exactly as one would expect; they don’t simply take the mostly West Eurasian IVC sample. They need the more western Iran_N as well (not to mention Iran_Chl). Interestingly, no BMAC-related admixture.

To be continued…

Seinundzeit said...

Continuing from where we left off….

Finally, a look at southern Central Asia.

Tajikistanis:

Yaghnobi

46.85% Steppe_MLBA_East
24.55% Seh_Gabi_Chl
24.05% Sappali_Tepe_BA
4.55% Mongola

"distance%=0.1541 / distance=0.001541"

Rushan

51.20% Steppe_MLBA_East
26.65% Sappali_Tepe_BA
7.70% Seh_Gabi_Chl
7.70% Shahr_I_Sokhta_BA2
6.15% Mongola
0.60% AASI

"distance%=0.1406 / distance=0.001406"

Shugnan

43.95% Steppe_MLBA_East
17.55% Sappali_Tepe_BA
15.30% Seh_Gabi_ChL
8.50% Shahr_I_Sokhta_BA2
5.60% Mongola
4.20% West_Siberia_N
2.75% Shahr_I_Sokhta_BA1
2.15% AASI

"distance%=0.1217 / distance=0.001217"

Ishkashim

39.05% Steppe_MLBA_East
30.65% Sappali_Tepe_BA
7.35% Shahr_I_Sokhta_BA2
5.65% AASI
4.75% Mongola
3.95% Bustan_BA
3.70% West_Siberia_N
2.95% Seh_Gabi_ChL
1.95% Shahr_I_Sokhta_BA1

"distance%=0.1608 / distance=0.001608"

Pashtuns:

Karlani Pashtun, Central Highlands

37.35% Steppe_MLBA_East
26.05% Shahr_I_Sokhta_BA2
25.30% Seh_Gabi_ChL
4.35% AASI
3.60% Bustan_BA
3.35% Mongola

"distance%=0.1728 / distance=0.001728"

Batanri Pashtun, Nomadic

30.05% Steppe_MLBA_East
24.45% Shahr_I_Sokhta_BA2
16.70% Sappali_Tepe_BA
13.55% Seh_Gabi_ChL
7.35% Shahr_I_Sokhta_BA1
4.20% AASI
3.70% Mongola

"distance%=0.2085 / distance=0.002085"

Sarbani Pashtun, Southwestern plateau

55.40% Sappali_Tepe_BA
26.10% Steppe_MLBA_East
8.65% Shahr_I_Sokhta_BA2
5.60% AASI
4.25% Mongola

"distance%=0.2217 / distance=0.002217"

Central Asians of Indo-Aryan extraction:

Kalash

38.25% Shahr_I_Sokhta_BA2
31.05% Steppe_MLBA_East
10.80% Sarazm_Eneolithic
8.90% Shahr_I_Sokhta_BA1
4.90% AASI
4.25% Seh_Gabi_ChL
1.85% Mongola

"distance%=0.1311 / distance=0.001311"

Kho_Singnali

26.90% Steppe_MLBA_East
23.25% Shahr_I_Sokhta_BA2
21.20% Bustan_BA
7.60% AASI
6.70% Shahr_I_Sokhta_BA1
5.30% West_Siberia_N
4.45% Seh_Gabi_ChL
3.20% Mongola
1.40% Ulchi

"distance%=0.2029 / distance=0.002029"

Chetan said...

@Seinundzeit Interesting that the difference in steppe_MLBA ancestry between North Indian and South Indian Brahmins is lesser than predicted by the Harappa DNA project.

Seinundzeit said...

General Observations:

It seems that Southern Central Asians, whether they are Dardic, Pashtun, or Pamiri/Yaghnobi/Tajik, are complex mixtures between BMAC, Steppe_MLBA_East, and IVC, with a consistent layer of extra Iran_Chl.

Looking at subpopulations, the Tajikistanis are essentially 50% Steppe_MLBA_East, while we see a very strong layer of extra West Asian affinity for the Karlani Pashtun (25% Iran_Chl, on top of whatever BMAC has).

The Kalash have a very interesting Central Asian Eneolithic dynamic.

Anyway, this is a very, very preliminary analysis. There is much more juice to be squeezed from this data.

Over the coming days, I have a ton of ideas to test. I'll also explore the other samples.

Not the final word...

(Although, I'm off to bed)

Rob said...

Srubna Outlier
Srubnaya_MLBA_o
Yamnaya_Samara:I0370 45.4 %
West_Siberia_N 37.3 %
Gonur1_BA_o 16.15 %
d 0.036%

Rob said...

@ Aram

"EBA in NW Iran is the KA. This BA Hajji Firuz is from MBA. And it is not related to Hittites."

Seems you;re right about KA

Hajji_Firuz_ChL_o:I2327
Armenia_EBA 55.9 %
Ganj_Dareh_N:I1947 20.4 %
Levant_N 15 %
Yamnaya_Samara 7.1 %

BTW Dave, there is nothing wrong with using more distant sources (eg EHG/ CHG in this case).,as long as one is aware of what it;s showing
In fact, its a good idea to analyse from various angles.


Rob said...

So many sources to consider, so I started with Meso-Neolithics as a 'zoomed out' look

https://imgur.com/a/dJdeS

Is it genuine archaic ANE in South Asia , perhaps diluted with time by migrations from West Asia ? Indeed, Siberian HGs were said to be ANE + EHG + ENA, another 'dilution' but up north due to westward (EHG) and Eastward (ENA) admixture.

Seems to be similar to the paper's qpAdm admixture in Fig 2.3

Rob said...

As per the paper, Steppe ancestry very much depends on which IVC periphery proxy one uses.
Gonur BA 2 vs Shahr_I_Sokhta_BA3 are going to yieled different results
The first, for non-Iranians, yields very minimal if any steppe. The latter, as demonstrated in Sein's runs, gives steppe admixture (Whether Yamnaya or MBA)
hhmm

Rob said...

For example

Punjabi
Gonur2_BA 68.7 %
Onge 26.8 %
West_Siberia_N 4.5 %
Yamnaya_Samara:I0370 0 %


Punjabi
Shahr_I_Sokhta_BA3 77 %
Yamnaya_Samara:I0370 6.5 %
Anatolia_BA 6.2 %
Onge 4.2 %
West_Siberia_N 2.7 %

Brahmin
Gonur2_BA 64.7 %
Onge 17.8 %
West_Siberia_N 7.8 %
Anatolia_BA 6 %
Yamnaya_Samara:I0370 3.7 %


Brahmin
Shahr_I_Sokhta_BA3 69.8 %
Yamnaya_Samara:I0370 13.2 %
Anatolia_BA 10.3 %
West_Siberia_N 4.1 %
Levant_BA 2.3 %

Davidski said...

Gonur2_BA:I2123 doesn't have any steppe ancestry. It's one of the Indus_Periphery samples; in other words a migrant from the Indus Valley.

The problem with Rob's models is that he's ignoring the real source of steppe ancestry in South Asians, which of course is Andronovo east, not Yamnaya.

Ariel said...

Sorry I confused it with the gonur outlier. Also about gonor 2, I was looking at the PCA that I made earlier and gonur 2 is actually significantly shifted towards Neolithic Iran compared to Shahr_Sokhta_BA 3, that makes it quite an outlier, so I don't think is a good sample to model modern South Asian who are mostly on that ASI ANI cline between Paniya and Kalash/Pashtun.

Rob said...

@ Dave

“Gonur2_BA:I2123 doesn't have any steppe ancestry.”

Did not claim it did Dave; in fact it’s the complete opposite

I intentionally used 3000-2500 BC for simplicity, because there are so many sources and options after 2000
But if Andronovo SE is a source then so should Yamnaya.
You obviously misunderstood because of your usual defensive paranoia

Davidski said...

@Rob

But if Andronovo SE is a source then so should Yamnaya.

It won't be, because you're making the algorithm skew the outcome due to the lack of European Middle Neolithic input in your model, so it avoids Yamnaya and picks Anatolia_BA, which has nothing to do with South Asia.

The algorithm will always attempt to make the best with what it's offered, but it won't choose the real sources of ancestry if it's forced to make an extreme correction to produce coherent results overall.

Ariel said...

To explain myself I little bit better, I'm seeing a cline between something broadly related to populations with more steppe (like modern Tajiki, Gonur outlier, Hajji Firuz bronze age, Sappali Tepe outlier) and the hindus periphery samples that are closer to Paniya. In simple terms a cline between the late BMAC/SWAT with steppe and Dravidians. That's quite important because you don't need a lot extra Iran Neolithic (outside the admixture that was already in the hindus periphery) to model modern South Asian populations.

[1] "distance%=3.3783 / distance=0.033783"
Brahmin
"Paniya" 47.95
"Gonur1_BA_o" 22.65
"Hajji_Firuz_BA" 12.4
"Sappali_Tepe_BA_o:I7493" 9.6
"Ganj_Dareh_N" 7.4

[1] "distance%=2.4473 / distance=0.024473"
Kalash
"Gonur1_BA_o" 33.95
"Hajji_Firuz_BA" 21.6
"Paniya" 19.5
"Ganj_Dareh_N" 16.55
"Sappali_Tepe_BA_o:I7493" 8.4

[1] "distance%=2.7612 / distance=0.027612"
Tajik
"Hajji_Firuz_BA" 39.75
"Gonur1_BA_o" 22.75
"Sappali_Tepe_BA_o:I7493" 19.6
"Scythian_ZevakinoChilikta" 9.85
"Paniya" 4.85
"Ganj_Dareh_N" 3.2

ROB-Player. said...

What Rob is saying makes sense. In short he is showing that there need not be Steppe MLBA East only to
explain steppe in South Asians.


@davidski


The algorithm will always attempt to make the best with what it's offered, but it won't choose the real sources of ancestry if it's forced to make an extreme correction to produce coherent results overall.


How is what you said above different from what the paper is doing? they have adhoc samples from space and time,
it is not like they have levelled the playing field with samples fed into the alogirithm from all slices of space and time.

Forget about time depth they do not even have all the samples in the same radius of SA to MLBA_East in the same time slice. If they could do that anyone would be sold on the stats, but not the way it stands now.

Davidski said...

@ROB-Player

You're an idiot.

ROB-Player. said...


@ROB-Player

You're an idiot.



I am ok being an idiot calling spade a spade.

Dont want to be a genius who calls spade something else ;).

Rafs said...

Thanks for the graphs, Ariel!

EastPole said...

I modified your PCA.

https://s18.postimg.cc/wqdbw0709/Steppe_MLBA2.png

Red triangles are modern Slavic speakers.
Open circles are Steppe_MLBA.
Red are Steppe_MLBA_West: Sintashta.
The rest are Steppe_MLBA_East:
aqua are Krasnoyarsk_MLBA,
cornflowerblue are Kazakh_Mys_MLBA,
gold are Dali_MLBA,
chartreuse are Oy_Dzhaylau_MLBA.

It is incredible how many Eastern Andronovo samples are similar to Sintashta and Corded Ware/modern Slavs.
It means that probably the forest steppe and steppe was almost empty when they started to migrate east from Vistula-Dnieper Corded Ware homeland. Small admixture with Yamnaya and that’s it. The rest is from HG.

epoch2013 said...

@FTR1A

This is the first possible find of R1b in pre-BA Iran. There are more samples from Iran and they are J and E. There is none of that in Yamnaya. While technically this is possible there is another thing. There is no high diversity of R1b in pre-BA Iran. There is a large diversity in the Pontic steppe and Eastern Europe.

What you want me to believe is that that one single sample originated in Iran, which doesn't have any different clades and not in a place where there far more kinds of R1b. Is that parsimonious?

That is, if the sample legitimately Chalcolithic. If it is, it would be parsimonious that it came from the steppe.

postneo said...

The paper fits a 2600 bc darra-i-kur Afghan sample as 12 % iron gates which is higher and older than steppe mlba east

EastPole said...

This model of expansion:

https://s18.postimg.cc/hunzg4fkp/screenshot_360.png

would fit very well linguistic wave theory:

https://s18.postimg.cc/re6v0ymyh/image.png

Samuel Andrews said...

@Rob,

Tiny amounts of ANE-West Siberia HG often signifcantly reduce Steppe percentages for South Asians. In Global25, Yamnaya is about 35-40% ANE. That's a lot. So, West Siberia HG is in fact able to swallow up real Steppe ancestry. Combined West Siberia HG and IranNeo can make a bad but good enough approximate for Steppe ancestry.

Several of the Eneolithic-bronze age genomes from SouthCentral Asia, like GonurBA, have West Siberia HG ancestry. When they are used to explain IranNeo ancestry in Southcentral Asia they also mask some of the Steppe ancestry.

Realistically, no one in Southcentral Asia has much more than 5% West Siberia HG stuff. I'd say Tajik are about 40% Andronovo, Kalash & Pathan 30%.

Samuel Andrews said...

@EastPole,
"Hitler’s Aryans never existed. The only true Aryans, Indo-Aryans or wider Indo-Iranians, seem to be related genetically and linguistically to Balto-Slavic people. They have nothing to do with Germanics, Western European ‘Aryans’ and they didn’t come from Yamnaya."

The association between Aryan and white people in general needs to stop. It makes it impossible to talk about them without the cloud of racism. You're replacing Aryan association from Nordics to Slavs. That doesn't help.

Think about it, why are you so interested in the distant link between Slavs and Aryans? Obviously the reason is all the racist hopala about an Aryan master race. There's nothing special about the Aryans. They were one of many Bronze age IE groups, probably of mixed ancestry. They weren't the creators of Greek & Indian civilization. All that, master race stuff is false.

Samuel Andrews said...

@EastPole,
"It is incredible how many Eastern Andronovo samples are similar to Sintashta and Corded Ware/modern Slavs."

Yes, very similar but northern Slavs also have an extra dose of farmer & WHG admixture. From the mtDNA perspective, you guys have lots of farmer mtDNA that Andronovo did not.

Polish

Krasnoyarsk_MLBA,68.1
Globular_Amphora,24.2
Narva_Lithuania,7.7

capra internetensis said...

Okay, in order to test whether Steppe could easily be replicated by a mix of other sources in scaled Global25 PCA nMonte (pen=0 for consistency):

Brahmin - 73% Shahr-i-SokhtaBA3, 27% SintashtaMLBA - distance 2.01%

A surprisingly decent fit, of course it can be improved.

Now, with no Steppe, the following sources:
Indus Periphery - Shahr-i-SokhtaBA2 and 3, Gonur2BA
Turan - Shahr-i-SokhtaBA1, Gonur1BA, Geoksyur, Tepe Anau
North/East Asian - West Siberia N, AfontovaGora3, Mongola
Anatolian - Boncuklu, TepeCiftlik, BarcinN, AnatoliaChl
Iran-Caucasus - CHG, GanjDarehN, SehGabiChl, ArmeniaEBA, ArmeniaChl, Tepe Hissar
Levantine - LevantBA, LevantN, BantuSE

Brahmin - 69% Shahr-i-SokhtaBA3, 16% ArmeniaChl, 10% West Siberia N, 6% Boncuklu - distance 2.08%

I gotta say I wasn't actually expecting the Sintashta model to come out ahead :D

adding Mala, Paniya, and Juang because SIS3 may not be good enough ASI ref:

Brahmin - 33% Mala, 21% SiS2, 18% SiS3, 10% Boncuklu, 10% WSN, 5% ArmeniaChl, 2% CHG - distance 1.87%

adding Mala and SiS2 to Brahmin + Sintashta:

Brahmin - 37% Mala, 24% Sintashta, 23% SiS2, 16% SiS3 - distance 1.47%

Sintashta seems hard to beat

capra internetensis said...

^ errata - "SiS3 + Sintashta", also Onge was an East Asian source

epoch2013 said...

Anyone that still is busy contemplating Hitlers theories seems to have missed that his regime lost the war almost 3/4 of a century ago.

You could have been born after that war and be great-grandfather today.

Hitler and his theories are utterly, utterly, utterly irrelevant to either today's politics or the DNA papers on archaeology.

Matt said...

@Alberto, hmmm... I see your logic re:simulated AASI.

In truth, I'd personally probably be happiest just to use a minimally admixed modern on the cline (and in general I would prefer the minimal admixed India cline probably Mala or Malayan to Bhumij or Birhor which are obviouslyon the Austroasiatic cline). But it is a thorny problem if you want true independence of component, even with the very high levels of AASI survival in South Asia (relative to what we see in Europe, or for the Hoabinhians in SE Asia, Jomon in Japan, Iberomaurasians in North Africa, East African HG, etc.!).

This said, I probably won't try modeling the South Asian samples myself, and will focus on the West-Central Asian samples, which seem like a simpler story maybe. So will leave this decision to others ;).

@Sein: Although, for some reason, the preprint uses Steppe_MLBA, which is odd, considering that the steppe ancestry would have been spread by Steppe_MLBA_East (they’re in the right geographical position, and have the right genetic structure).

Well, if the Steppe_MLBA west samples work in their models, and they do, often better than the Steppe_MLBA_East(!), then it's hard to think of a definition of "right genetic structure" that doesn't fit.

And from what I know, it's not like authorities like Mallory have ever really been strongly decisive that it *had* to be from the Inner Asian Mountain Corridor, rather than via more westerly Andronovo streams (e.g. Mallory had some models where the stream was through BMAC, which could be true, though if so, not with much genetic admixture). So "right geographical position" is arguable as well.

So as I understand it, there's no real warranted reason to focus on Steppe_MLBA_East. So the paper doesn't.

Rob said...

@ Dave
Yes i agree that if MNE is missing the the system compensated by choosing something else with ANF but I don’t think that diminishes or eats away from steppe itself
Anyhow, as I said, I’m still working downwards in time
But some of those steppe MLBA east samples are rather late in time (1200 BC). Is that a historically viable option, and wrt to Seins models some samples are LBA some Eneolithic
Is unappealing to me (which is why I haven’t delved too much into European moderns )

Shaikorth said...

@Capra
Yeah, Sintashta stays for Indo-Aryans. I threw a lot of alternative ANE sources on the scaled sheet:

Kshatriya

Onge 40.1
Abdul_Hosein 38.4
Sintashta_MLBA 14.8
West_Siberia_N 6.7
Dai 0.0
Dali_EBA 0.0
Sintashta_MLBA_o3 0.0
Barcin_N 0.0
Ganj_Dareh_N 0.0

North_Kannadi takes no Sintastha, which wasn't an unexpected result. The Onge may be elevated by something though.

Onge 64.90
Abdul_Hosein 29.85
West_Siberia_N 5.25
Dai 0.00
Dali_EBA 0.00
Sintashta_MLBA_o3 0.00
Sintashta_MLBA 0.00
Barcin_N 0.00
Ganj_Dareh_N 0.00

Arza said...

@ epoch2013
Hitler and his theories are utterly, utterly, utterly irrelevant to either today's politics or the DNA papers on archaeology.

How they are irrelevant if they are present in the mainstream so called science (history, archaeology and linguistics) till this day?

We should accept lies as truth just because someone lied enough time ago?

Rob said...

@ Matt
“And from what I know, it's not like authorities like Mallory have ever really been strongly decisive that it *had* to be from the Inner Asian Mountain Corridor, rather than via more westerly Andronovo streams (e.g. Mallory had some models where the stream was through BMAC, which could be true, though if so, not with much genetic admixture). So "right geographical position" is arguable as well.”

Mallory never used the idea of an IAMC
It’s a relatively new concept
The Paper pushes the classic andronovo model yet on the other hand the image within it demonstrates the IAMC position, & on the third hand the Data doesn’t support the Sintashta elite conquest model (& is showing something potentially different still). That’s why I think the paper was probably written by Anthony or someone with a strong Kurganist push (or frankly, I thought some of the comenetators here wrote the paper lol).

Alberto said...

Yes, finding the best setup to try to tell apart the different streams of ancestry is quite tricky with samples from after 2500-2000 BCE (depending on the place). Here I focused on finding a setup that would not require Srubnaya_outlier (it still worked better than any other high ANE sample available) so that it wouldn't take any steppe admixture (given that it's half Yamnaya) and at the same time avoiding false Steppe_MLBA signals.

A work in progress, but after several hundred runs this is my best choice for now:

https://docs.google.com/spreadsheets/d/1XqI31rghEb4GlGr-yGPxWXiv5fObNGleAuZu8Sx7Q3w/edit?usp=sharing

The first tab has the ancient samples, the second tab the modern ones. Both with the same setup. Looks reliable enough to detect steppe_MLBA admixture without false positives, though the still scarce ancient samples from South Asia could leave some doubts.

I found one Gonur1_BA sample (main cluster, I3374) that seems to have 10% Steppe_MLBA admixture, even if it's not marked as outlier (the sample is from around 2050 BCE). The Hajji_Firuz BA and outlisers also show Steppe_MLBA. And the Swat Valley samples have a quite variable level of Steppe_MLBA, between 0% and 20% (average 9.3%).

The AASI is still Matt's simulation that can be found here: https://pastebin.com/feVZKjmP

All using the scaled datasheets.

Alberto said...

Forgot to say: I'm also not very convinced that Steppe_MLBA _East is any better than West. In the end I used 2 sources that are both labelled as East, but Krasnoyarsk_MLBA is quite similar to West, while Dali_MLBA has more ANE/ENA.

Seinundzeit said...

Matt,

"West-Central Asian samples, which seem like a simpler story maybe."

On the contrary South Asians are very simple mixtures. Just IVC-related West Eurasian admixture, Steppe_MLBA_East, and AASI. These are the only streams from Punjab to Tamil Nadu.

Tajikistan, Afghanistan, and northern/northwestern Pakistan are far, far more complex. There is obviously BMAC-related ancestry, IVC-related ancestry, some Eneolithic Central Asian-related heritage for isolated people like the Kalash, excess West Siberian for Khalils, perhaps some Scythian admixture, a recent layer of West Asian influence, and the older Steppe_MLBA_East influence. It's a lot to take in.

With regard to Steppe_MLBA_East, I would note that some of these samples are from the Ferghana valley, in South Central Asia! So, we know that people with this sort of genetic structure were in South Central Asia.

Furthermore, Steppe_MLBA_East is the primary reason for why Alberto was seeing Srubnaya__Outlier in his models (even though excess ANE/West_Siberian_N was accounted for via the IVC samples).

But, I guess it is possible to use Steppe_MLBA in the modelling.

Matt said...

@Sein, sorry, I'm talking more about modeling the earliest *ancient* West-Central Asian groups, basically testing how well these groups mix as earliest Iran_N, CHG, Levant_N, Anatolian, West_Siberian, EHG, Iron_Gates_HG, and whether the models or Iran_N+Anatolian+WestSiberian are best case models.

You're correct that modern West-Central Asian groups are no less (or more) complex to deal with than modern South Asia.

Matt said...

@Sein: "Furthermore, Steppe_MLBA_East is the primary reason for why Alberto was seeing Srubnaya__Outlier in his models (even though excess ANE/West_Siberian_N was accounted for via the IVC samples)."

And yet, the Steppe_MLBA west groups work just as well in the paper! And may work just as well in G25 with further testing.

Davidski said...

This is where the Kashkarchi_BA samples from Ferghana, on the Uzbek/Tajik border cluster. Basically in Eastern Europe.

https://drive.google.com/file/d/1_X41wez12kXX58JiSRuOq89bZhPeJjmU/view?usp=sharing

The Ferghana Valley is in South Central Asia, and not far from India.

Simon_W said...

@ Arza, east Pole, Samuel Andrews and Epoch2013

Hitler wasn't much of a theorist, so that talk about "Hitler's theories" is nonsense. There was among the National Socialists the belief that the PIE were members of the Nordic race from Northern Europe (or from Atlantis, the allegdly lost continent) and essentially identical with the Germanics. But this belief wasn't a creation of Hitler. And the fact that the word Indo-European is still being translated as Indogermanisch in German doesn't mean that Germans believe in a special relatedness between Germanics and PIE. The German word Indogermanen means exactly the same as the English word Indo-European, although its etymology is obviously different. So that word doesn't imply a certain theory on PIE origins at all. It's completely neutral, and if German speakers prefer to speak of Indogermanen instead of the synonym Indoeuropäer, they're akin to people using the German acronym DNS for DNA. I personally use Indogermanen when talking in German about Indoeuropeans, yet I subscribe to the steppe theory.

Seinundzeit said...

Matt,

"And yet, the Steppe_MLBA west groups work just as well in the paper! And may work just as well in G25 with further testing."

Sure; I haven't even started trying out the various setups I've had in mind for years. With further exploration, we'll see if Steppe_MLBA does as excellent of a job as Steppe_MLBA_East. Although, again, would note that Steppe_MLBA rather than Steppe_MLBA_East doesn't make much geographical or historical sense (a priori).

Matt said...

Re: tangent on comparing the relatedness of Eastern Andronovo (Steppe_MLBA_East) to other populations, using simple plot of euclidean distance across all G25 dimensions to compare the relatedness of east and west steppe_MLBA groups to other populations: https://imgur.com/a/QRwNT

Regarding modern populations in the north of Europe, we can see that on average present day a) Finno-Ugric and North Russian populations are closer to east steppe_MLBA than other populations are, while b) Celtic+Germanic populations generally seem to edge relatedness to west steppe_MLBA and finally c) other Slavic populations intermediate (generally about as close to west steppe MLBA as a) and as close to east steppe MLBA as b) ).

This seems in line with how Finno-Urgic+North Russian have some more ANE+ENA ancestry (bringing them closer to steppe_MLBA_East), while the Celtic+Germanic populations are closer to simple fits of MN farmer+Steppe_EMBA (bringing them close to steppe_MLBA_East), and then the West Slavic+Baltic populations are slightly disrupted off the simple MN farmer+Steppe_EMBA cline by have higher levels of likely Baltic HG which very slightly further.

There's no particular sign anyway it seems to me in Global25 of a particularly close population genetic relationship between Steppe_MLBA and Balto-Slavic people as a whole. It seems pretty likely that some rare allele sharing or something will come up given the linguistic arguments. (That is, it's more likely that Steppe_MLBA groups share more recent actual ancestors with most Baltic-Slavic speaking populations). But at the moment, in general genome wide terms, at least as represented by G25, the particular Balto-Slavic signature linking them to European HG if anything pushes them if anything further away from Steppe_MLBA than present day ancestrally Celtic and Germanic speaking groups. This was the case before this paper with more limited Steppe_MLBA samples and roughly seems still similar now.

Similar graphs compared distance from present day populations: https://imgur.com/a/wL5Mm

(Incidentally, this doesn't seem like it's bad for RK's idea that early Uralic speakers were a population autosomally like Steppe_MLBA_East?)

@Rob, apologies if I've upset you by not reading this paper which you've previously referred me to and apparently very badly want to discuss with someone. Perhaps explain briefly why you feel it's critical to the questions of population genetic structure over time, and to the linguistic spread of Indo-European, specifically, rather than early Bronze Age cultural change in general? I'm not just going to read it because you call me a wimp (if anything I'm less likely to read it).

Seinundzeit said...

Matt,

Ah, looking at the Dali_MLBA comparison, seems that South Central Asians are certainly closer to Dali_MLBA than they are to Sintashta (in the models I posted yesterday, the Steppe_MLBA_East pop was Dali_MLBA)!

For what it's worth, in a couple of mins, I'll be adding my old Steppe_MLBA average, with the addition of Scythians, just to see what happens...

Rob said...

@ Alberto
Yep sensible estimates
10-20% MLBA admixture
Overal, demographically negligible ; and no signs of elite conquest scenario
So probably just assimilated andronovo traders and females
Very unlikely steppe spread PIE

Seinundzeit said...

Okay, so I added Kashkarchi_BA, and my old Steppe_MLBA average, with some Scythians (alongside Dali_MLBA).

Basically, it seems Kashkarchi_BA is the best bet. We should have expected that, since as David noted, these are samples found in South Central Asia.

Most populations gravitate towards it, or have a mix of it and Steppe_MLBA, or a mix of it and Dali_MLBA. So overall, for the sake of consistency, probably the best population to use.

Also, as I expected, no Scythian ancestry in South Asia.

But, in South Central Asia/West Central Asia (they mean the same thing), various Pashtuns and Pamiris show Scythian percentages.

I'm going to try something different (a few extra additions, and the removal of something), and then I'll post the output.

Open Genomes said...

Here's the nMonte2 from the Global_25 (not scaled) for I8728 Shahr_I_Sokhta_BA3:S8728.E1.L1, 4600 ybp / 2550-2450 BCE, Y-DNA J2a and mtDNA R7.
This is the most "extreme" "Indus Valley Diaspora" sample.

[1] "distance%=3.568 / distance=0.03568"

Shahr_I_Sokhta_BA3:S8728.E1.L1

Gonur2_BA:I2123 71.4
Onge:ONG-1 28.6

I8728 Shahr-I-Sokhta BA3 is a straight up mix between I2123 from the BMAC site of Gonur2 from 4246 ybp / 2452-2140 calBCE, mtDNA M30a and the Onge.

The question is whether I2123 has admixture from the the Indus Valley, or the Indus Valley has admixture from BMAC. Since Gonur2_BA:I2123 has an additional "green V5" component which is seen at high percentages in some Sintashta samples, and this is lacking in Shar-I-Sokhta BA3. Other Gonur sampled don't fit. It would seem that the IVC people admixed with just some individuals in BMAC, not the other way around.

Interactive table of autosomal K=6 results for all samples

Interestingly, his closest match today is NA20861 (GIH Gujarat) and this one is Y-DNA L-Z42497 under L-Z5933. Other Gujaratis are also relatively close, too.

Davidski said...

@Open Genomes

Both Shahr_I_Sokhta_BA3 and Gonur2_BA are likely migrants from the IVC.

The only difference between them is that the former has a lot more indigenous South Asian ancestry.

Neither can be native to Turan, or even have admixture from BMAC, because their West Eurasian ancestry is unique, and unlike that found among the BMAC population, it has much less, or even no, Anatolian farmer-related input.

Seinundzeit said...

David,

Absolutely; BMAC clearly represents a movement from further west. There's tons of Iran_Chl influence, which the preprint represents via "Anatolian" admixture.

With that being said, there is something going on between Eneolithic Central Asians (especially Sarazm) and IVC.

If you strip away the 15% to 40% AASI, you'd end up with something quite similar to Sarazm EN. Of course, it would still be slightly different (Sarazm_EN has more West_Siberian_N, while the IVC samples still prefer AG3/MA1).

Which makes me wonder about the ethnogenesis of the deeper streams which went into IVC.

Essentially, does this mean that the first West Eurasian agriculturalists of South Asia came via South Central Asia, rather than West Asia?

Or, perhaps the eastern Iranian plateau was home to populations quite similar to Sarazm EN, and IVC had a relationship with those eastern West Asians?

Interesting stuff...

Gill said...

Looks like

Neolithic_Sarazm + AASI = Indus_P

Neolithic_Sarazm + CHG = Eneolithic_Sarazm

Gill said...

I always thought there would be a really high ANE and ENF combo somewhere in South Central Asia (like 40/60, maybe even 50/50). I was using populations like that in my "simulated" admixture runs for a while.

ANE probably extended in a gradient across Asia with a high-ANE frontier that interacted with East Eurasian (Amerindian), WHG (EHG), or both (W Siberia N) and ENF (precursor to Eneolithic Sarazm). I mean it had to since it can't teleport. The question was how that mixing took place, which stages corresponded to actual populations. There's too much geography between Siberia and the Indus Valley for there not to have been more ENF-rich populations with even higher affinities to ANE than those of the Indus Valley.

Gill said...

Also, there is remarkable continuity from the Bronze Age to now in the Indus Valley.

Saidu_Sharif_IA S7720.E1.L1's results in Harappa admixture calculator look like a modern day Kashmiri/West Punjabi. Some of the others look like modern South Indians.

It looks like after the Bronze Age, the only changes in South Asia were internal. No more new major changes to the makeup from the outside (well, since after whenever the Austroasiatic migration also ended).

mzp1 said...

Just a bit of context for davidski regarding the haji-firuz outlier...

You do know that very early Kurgans are found in Leila Tepe and Maykop.

Davidski said...

@mzp1

Just a bit of context for davidski regarding the haji-firuz outlier...

You do know that very early Kurgans are found in Leila Tepe and Maykop.


Totally irrelevant even if true.

There was obviously a Yamnaya or Catacomb incursion, or incursions, into what is now northwestern Iran sometime during the Early to Middle Bronze Age.

The Hajji Firuz BA sample that was successfully radiocarbon dated is at least 50% Yamnaya.

Jaydeepsinh Rathod said...

Sein,

Or consider this - Sarazm_EN like groups were migrants from IVC. Consider the fact that it has so little ANF - an anamoly among Copper Age Turan samples. Also consider how close it is to non-AASI ancestry in South Asians. Check the figure S3.10 right panel, in the supplement of the paper - pg 122. Have a look at the allele sharing between Sarazm and the Birhor Austro-asiatic group from South Asia. Sarazm has significantly higher sharing with the Birhor than the main BMAC cluster - Gonur1_BA, Sappali_Tepe_BA, Dzharkutan1_BA & Bustan_BA. Remember that the main BMAC cluster has as much as 5 % of AASI ancestry. There is a South Asian yellow component even in the admixture graph for Sarazm. We should be ready for the possibility that IVC people in a pre-3000 BC era could have had very low or non-existent AASI percentages. Merely because AASI is indigenous to South Asia does not mean it was spread all across South Asia early on. This is a bit of common sense that I m afraid alludes most people here.

Seinundzeit said...

Gill,

"Looks like

Neolithic_Sarazm + AASI = Indus_P

Neolithic_Sarazm + CHG = Eneolithic_Sarazm"

Seems like a sensible idea.

Speaking of which, I've found a setup that works even better than the previous one I used for the models I posted up-thread.

I think I'll post the output tomorrow, although I may tweak even further.

It's such a treat to have these samples; they clarify so much.

Jaydeepsinh Rathod,

"Or consider this - Sarazm_EN like groups were migrants from IVC. Consider the fact that it has so little ANF - an anamoly among Copper Age Turan samples. Also consider how close it is to non-AASI ancestry in South Asians. Check the figure S3.10 right panel, in the supplement of the paper - pg 122. Have a look at the allele sharing between Sarazm and the Birhor Austro-asiatic group from South Asia. Sarazm has significantly higher sharing with the Birhor than the main BMAC cluster - Gonur1_BA, Sappali_Tepe_BA, Dzharkutan1_BA & Bustan_BA. Remember that the main BMAC cluster has as much as 5 % of AASI ancestry. There is a South Asian yellow component even in the admixture graph for Sarazm. We should be ready for the possibility that IVC people in a pre-3000 BC era could have had very low or non-existent AASI percentages. Merely because AASI is indigenous to South Asia does not mean it was spread all across South Asia early on. This is a bit of common sense that I m afraid alludes most people here."

I don't know Jaydeepsinh; this seems somewhat less parsimonious than assuming that Sarazm_Eneolithic-related populations constituted the West Eurasian portion of IVC genetic ancestry, and that such ancestry was quite widespread across the far eastern Iranian plateau and southern Central Asia.

It's a case of whether the "chicken or egg" came first.

epoch2013 said...

@Arza

"How they are irrelevant if they are present in the mainstream so called science (history, archaeology and linguistics) till this day?"

They aren't. Or else show me - with the relevant articles, papers or books - where and what Hitler contributed to history, archaeology and linguistics.

Huck Finn said...

@Matt: many thanks for the comparisons. Re: Incidentally, this doesn't seem like it's bad for RK's idea that early Uralic speakers were a population autosomally like Steppe_MLBA_East?)It is easy to agree and Ryus's reasoning was and still is very convincing. It indeed seems that there are good reasons to connect at least some part of Uralic expansion to archeological events such as Seyma Turbino phenomenom and Krotovo culture on the Asian side of the Ural moountains, Garino Bor (Turbino) culture on the European side of those same mountains.

Samuel Andrews said...

Good job, guys for noticing the ANE signal in Steppe MLBA outliers as well as in modern Southcentral Asian.

One thing many of us were wrong about is that Kalash & Tajik have as much true Steppe ancestry (Yamnaya-like) as northern European. Adjusting for West Siberia HG (and Central Asian HG) ancestry in them their Steppe_MLBA_East numbers drop to about 33% for Kalash and 42% for Tajik.

Considering Steppe_MLBA_East is 69% Yamnaya-like, that would make Kalash 23% Yamnaya and Tajik 29% Yamnaya.

It's hard to find the correct West SIberia HG ancestry proportion so maybe those Yamnaya numbers are way off. But if it is correct, it is a big suprise.

Davidski said...

@All

I've updated the post. It now includes a little explanation of who the Indus periphery (or diaspora) individuals may have been.

By the way, I've got a new post coming later today with a somewhat different PCA, and a discussion about the origin of steppe ancestry in South Asia.

Samuel Andrews said...

@Sein,

Nice stuff. Where is your AASI from? I don't see it in the G25 spreadsheets. I've been using Mala instead even though I think they might be 25% iranNeo.

Seinundzeit said...

Thanks Dave, I appreciate it.

Sam,

"Nice stuff. Where is your AASI from? I don't see it in the G25 spreadsheets. I've been using Mala instead even though I think they might be 25% iranNeo."

Thanks. It's an AASI simulation which I always use.

If you want the coordinates, I can post them.

All,

Sam is referring to these models, below. I had these posted earlier, but due to complications (lol), it was better to delete the post in question.

Essentially, I tried to produce models that were based on more "distinct" ancestral streams, so as to avoid "overfitting".

I actually have a new setup with "proximal" sources; very solid stuff. I might post that tomorrow.

But before that, these are still pretty cool.

Reference populations:

AfontovaGora3
West_Siberia_N
Ganj_Dareh_N (Iran_N)
Seh_Gabi_ChL (Iran_Chl)
Sarazm_Eneolithic (Central Asia_EN)
Kashkarchi_BA (Steppe_MLBA_East)
Ulchi
Mongola
Cambodian
AASI

Good for the "big/broad picture".

Shahr_I_Sokhta_BA1:

39.15% Sarazm_Eneolithic (Central Asia_Eneolithic)
36.95% Ganj_Dareh_N (Iran_N)
18.55% Seh_Gabi_ChL (Iran_Chl)
5.35% AASI

"distance%=0.2211 / distance=0.002211"

Sappali_Tepe_BA:

46.6% Seh_Gabi_ChL (Iran_Chl)
30.6% Sarazm_Eneolithic (Central Asia_Eneolithic)
17.4% Ganj_Dareh_N (Iran_N)
3.5% Kashkarchi_BA (Steppe_MLBA_East)
2.0% AASI

"distance%=0.1335 / distance=0.001335"

Shahr_I_Sokhta_BA2:

59.4% Sarazm_Eneolithic (Central Asia_Eneolithic)
22.6% Ganj_Dareh_N (Iran_N)
17.9% AASI

"distance%=0.4254 / distance=0.004254"

Gonur2_BA:

57.6% Sarazm_Eneolithic (Central Asia_Eneolithic)
24.4% AASI
18.1% Ganj_Dareh_N (Iran_N)

"distance%=0.4816 / distance=0.004816"

Shahr_I_Sokhta_BA3:

46.4% AASI
43.0% Sarazm_Eneolithic (Central Asia_Eneolithic)
10.0% Ganj_Dareh_N (Iran_N)
0.6% AfontovaGora3

"distance%=0.7122 / distance=0.007122"

Is this cool, or is this cool? (lol)

Lots of Iran_Chl in BMAC, but none in IVC.

IVC West Eurasian ancestry can be tentatively modeled as a mix of Sarazm_Eneolithic and Iran_N.

Not meant to be taken literally, but way closer to reality than Iran_N + West_Siberian_N + Onge.

Chetan said...

@David I wonder why David Reich would go against consensus and propose something radical like this. if this supposedly Z2103 sample is not even radiocarbon dated? Maybe he has other data he considers to be evidence for an Iranian/Caucasian incursion into the steppe?

Jaydeepsinh Rathod said...

Sein,

Can the South Asian samples be modelled as a two way mixture between Sarazm_EN + AASI ?

Seinundzeit said...

Jaydeepsinh,

I'm sure they could, but only if one were to exclude Iran_N.

It has to be Sarazm_EN + Iran_N + AASI. Although, I would note that Sarazm_EN isn't perfect for these samples. It has a specifically West_Siberian_N skew, while the IVC samples still prefer AG3/MA1.

But, they do prefer Sarazm_EN to Iran_N.

Jaydeepsinh Rathod said...

Sein,

Why does it have to include Iran_N ? There is no real evidence that Iran_N like groups migrated to South Asia from Zagros. They are related for sure but the Iran N like groups may have already existed since Mesolithic in South Asia with a bit more ANE/WSHG just like Sarazm_EN. That is what I am trying to figure out.

Seinundzeit said...

Jaydeepsinh,

Sarazm_EN isn't basal enough for the West Eurasian ancestry seen with IVC; the algorithm will always attempt to make the best with what it's offered, so naturally we see Iran_N.

Aram said...

Y dna assignements is not so precise science as some people could think.
Having 3 positive calls on Z2103 level means little. There is always the problem of false positives, unreliable SNPs.
You need to see if it has positive calls on L23 level also. Or even in completely different branch like V1636. (found in KA).
If it had a dozen SNPs on that level then the assignement can change. But the huge problem is that the ISOGG has much lower number of SNPs on V1636 level than on M269. That is why we shouldn't be surprised that their script transformed the Kura Araxian V1636 into M269.


Davidski said...

@Aram

I can assure you that this supposedly Chalcolithic (but in all likelihood Bronze Age) Hajji Firuz sample has steppe ancestry.

This is fully in line with the C14-dated Hajji Firuz Bronze Age sample, which is obviously mostly of steppe origin.

Kura-Araxes doesn't have any steppe ancestry. So the Z2103 call for the Hajji Firuz guy is probably legit.

Samuel Andrews said...

@Sein,
"If you want the coordinates, I can post them."

Please do.

Jaydeepsinh Rathod said...

Sein,

Alright thanks. I suppose we will have to patiently wait for Neolithic and Mesolithic samples from South Asia. I hear that there are ongoing attempts to get aDNA from such samples. God willing they will succeed and we may see the paper in a couple of years.

But does the use of Sarazm + Iran N + AASI obviate the need to use any steppe ?

And another thought experiment -

Can the Afanasievo be modelled as EHG/WSHG + Sarazm_EN + ANF ?

Aram said...

Davidski

Yes Kura Araxes has no Steppe ancestry. Which makes sense. But it has that V1636 branch (v1636 has 15.000 formation age). That is why I was wondering is it possible that in Eneolithic Iran we are seeing the arival of that V1636 branch. After all V1636 also could carry some EHG.

Anyway Your version is also very plausible so let's wait and see what happens.

Davidski said...

@Chetan

I wonder why David Reich would go against consensus and propose something radical like this. if this supposedly Z2103 sample is not even radiocarbon dated? Maybe he has other data he considers to be evidence for an Iranian/Caucasian incursion into the steppe?

I haven't discussed this with David Reich, so I don't know what he's basing his theory on exactly and how confident he is in his prediction.

You would have to e-mail him and ask what the story is behind that. Keep in mind, however, that guys like this never have pet theories. They will always move with the data.

If I had to guess, then I'd say that it boils down to a very simple and logical inference from the currently available data, which show that a new culture and way of life appeared on the steppe just as the genetic structure of the people there shifted in a big way and became more southern.

Hence, it's likely that a new language came along with the new culture and economic package, so if Yamnaya was Indo-European, which it probably was, then Indo-European languages came from south of the steppe.

I actually doubt that this has anything to do with the Hajji Firuz R1b.


Chetan said...

@Davidski Hmm that may be the reason after all. But we already knew southern migrations brought agriculture to the steppe.

But to be frank, there is no evidence that proto-languages belonging to PIE or Proto Indo-Uralic family were ever spoken south of the Caucasus. Most previous guesstimates by linguists have usually converged on the Pontic-Caspian steppe. That's why the theory holds precedence in all standard academic works dealing with PIE.

Matt said...

@Davidski, if you are currently able to do qpGraph on these data, I have a quick test model I'd like if you could run:

https://pastebin.com/xfJA6ErX

The idea is to model all of the varying set (EHG, Yamnaya, CHG, Abdul_Hosein_N, Gonur2_BA) with varying proportions of the same ANE, UHG and Basal Eurasian, and also allow an admixture edges into Gonur2_BA for AASI.

It simplifies the relationships between population slightly, so I'm not so much interested in the worst stat, but if you could give me the output for the full stat fits, if you are able and willing to run, that would be useful. Thanks. Only if you have time and are willing and able.

I'm really interested in Gonur2_BA and Abdul_Hosein_N can be modeled with the same or different proportions of the ANE, UHG and Basal Eurasian, net of the AASI being accounted for.

Samuel Andrews said...

I'm just starting to discover Central Asia is really interesting region. Look at the difference between Kyrgyz and Tajik two neighboring ethnic groups one Turkic and one Iranian.

Kyrgyz
Steppe_MLBA_East 15.3
Iran_Chl 10.8
West_Siberia_N 3
Mongola 36.3
Ulchi 21.6
Yakut 11.5
Mala (ASI) 0

Tajik
Steppe_MLBA_East 38.9
Iran_Chl 20.7
Iran Neo 21
West_Siberia_N 9.8
Mongola,3.7
Ulchi,0.1
Mala,6.1

When you compare another pair of neighbors, Turkmen (Turkic) and Pathan (Iranian), it's the same trend. Most of the Asian ancestry in Central Asia, maybe even in Kazahstan, looks like it is from the Turks.

I'd say the mixed Gepid genome from Serbia shows that Huns were Turkic and overwelmingly Asian. Probably some Scythian ancestry but mostly of Asian origin (excluding signifcant Asian admixture in Scythians).

2.2716"

Gepid_Serbia_ACD

Germany_Medieval,36.1
Balkans_IA,33.9
Mongola,16.7
West_Siberia_N,4.9
Nganassan,3.1
Mala,2.8
Yakut,2.3
Hajji_Firuz_ChL,0.2
Kashkarchi_BA,0
Ganj_Dareh_N,0
Abdul_Hosein,0

West Siberia N might be masking Andronovo ancestry. When you remove it....

2.3905"

Gepid_Serbia_ACD

Germany_Medieval,37.4
Balkans_IA,29.5
Mongola,16.8
Kashkarchi_BA,6.4
Nganassan,4.5
Mala,3.1
Yakut,1.8
Hajji_Firuz_ChL,0.5

So, still this Gepid woman's European ancestry looks like it has very little Andronovo influence. His Asian ancestry is mostly explained by "Mongola" like Turks today.

Hun, plus Goth, plus local Balkan makes the most sense. If so, that would probably mean Huns & early Turks in general did not have a lot of Scythian ancestry.

epoch2013 said...

@Samuel

Tajiks and Kyrgyz differ very much in looks, the former basically West-Eurasian and the latter Mongolid.

Elliv J said...

Kyrgyz is still very R1a though. How turks were formed is a big puzzle.

Rob said...

@ Dave

“I can assure you that this supposedly Chalcolithic (but in all likelihood Bronze Age)"

Supposedly because you don't want it to be ?
Isn't that Khvalynsk Q male that you cling to also not individually C14 dated? He could be from 2800 BC, like the supposedly Ukraine Eneolithic outlier
Pandoras box if you will

"....Hajji Firuz sample has steppe ancestry.”

Yes dave About 7%. So probably maybe had a great grandmother on the steppe

Perhaps like the Bronze Age female, who actually does have an amount of steppe worth talking about

a said...

Blogger Rob said...

“I can assure you that this supposedly Chalcolithic (but in all likelihood Bronze Age)"

Supposedly because you don't want it to be ?
Isn't that Khvalynsk Q male that you cling to also not individually C14 dated? He could be from 2800 BC, like the supposedly Ukraine Eneolithic outlier
Pandoras box if you will

"....Hajji Firuz sample has steppe ancestry.”

Yes dave About 7%. So probably maybe had a great grandmother on the steppe"

Have you ever had to replace a flat tire in the middle of the bush in the middle of inclement weather- with no garage/tow truck/friends.The logistics of placing the ancestors of Afansievo culture in together/in the same package with R1b-Z2103-from Hajji Firuz are going to be very interesting to say the least. The most logical path to Samara Volga is a straight path 1900km+/-[its closer to Sumerian culture-760km+/-] the next leg of the journey to where the Afanasievo culture sites were found from Samara-Volga to Afanasievo would be roughly 2600+/- km. Some of the early Afansievo samples are -I5270 T1a1 R1b1a1a2a2 Afanasievo 3322-2939 calBCE (4435±20 BP).The distance from Iran samples to Afansievo is roughly 4500km of different terrain and climate/food/weather from Iran.To break a wheel or axles would require native wood and special tools to fix-if you agree the trip was made with oxen and wagon and self-sufficient home steaders . This was done in the same time frame when the earliest texts come from the cities of Uruk and Jemdet Nasr and date back to 3300 BC, or the accession of first dynasty of Egypt-Hor-Aha—was placed at 3100 bc give or take a century (3218–3035, with 95% confidence), Stonehenge-- Archaeologists believe it was constructed from 3000 BC to 2000 BC.
That is just the logistics for wagons- were not even touching the issue of pottery.

Palacista said...

If Reich believes that IE came originally from south of the Caucusus he has to explain how IE is utterly different from all the attested languages from that region. He is surely can't relying on the clearly intrusive Anatolian branch of IE.

mzp1 said...

Proto-Indo-European is so 2017. Yamnaya spoke Iranian. Get with with the times people.

Rob said...

@ Dave

Help me out here. I added Sintashta - the founder of steppe MBA- as you recommended.
Along side I added Gonur (regular), Shahr_I_Sokhta_BA3 (IVC periphery), Onge (for extra AASI), Dali EBA (for siberian end of steppe), Dai, & Levant / Anat BA to cater for the middle eastern admixture in Persians.


The results are here:

https://i.imgur.com/Ak33pBF.png


5-15% in SBPT and modern Indians, the rest is Turan Eneolithic, IVC periphery, extra AASI; + near Eastern admixture in Iran (i.e. still some BA Anatol & /or Levant)

The overall results are really no different than before.

What matters now is the social relevance and context.

Rob said...

@ Palacista

"If Reich believes that IE came originally from south of the Caucusus he has to explain how IE is utterly different from all the attested languages from that region."

Not claiming that it is, but how do you know what was spoken in South Caucasus in 4000 BC ?
Which exact languages were spoekn there at that time which were just so different to PPIE ?
Is there not evidence that *PIE had contact with Kartvelian, and NW Caucasian (which might itself be from south of the Caucasus)

" He is surely can't relying on the clearly intrusive Anatolian branch of IE."

It certainly appears to be intrusive to the Hattic substratum, sure.
But when did Hattic arrive to E Anatolia, and from where ?

It would be rather naive, given what we have thus far found out, to see 1700 BC as a mirror to 4500 BC.

Rob said...

@ "a"

Not 100% following you there, are you suggesting there is some kind of exclusive link between the wheel and Z2013 ?
Whatever the case, yes, I think overall R1b is from the north, but that doesn't rule out bilateral connections which might predate Yamnaya (as we see between Khvalynsk and Kura-Arax).

Kristiina said...

PIE reconstructions are very rich in consonants, e.g. *ph₂-ter- 'father', *gʷḗn (oblique stem *gʷnéh₂-) 'woman', *kʷékʷléh₂ 'wagon', 'wheel'. From Wals we can see that consonant-rich languages abound in Caucasus and India
http://wals.info/feature/3A#2/19.3/152.9

Gender system is also a very important aspect of IE languages. Wals shows that apart from IE languages, sex-based gender systems are typical of Indian, Afro-Asiatic and many Caucasian languages http://wals.info/feature/31A#2/36.3/70.3

PIE also shows ergative/active traits and these systems are also frequent in Caucasus and India
http://wals.info/feature/98A#2/15.0/143.0

Many PIE constructions are monosyllabic and many Indo-Europeanists favour reconstructions with very few vowels and a widely accepted vowel system includes only *e, *o, *ē and *ō. These characteristics are the reason to see a connection between PIE and Northwest Caucasian languages. Most roots in Northwest Caucasian languages are monosyllabic and, moreover, ablaut is common as in IE languages.


Alberto said...

@Rob

What matters now is the social relevance and context.

Yes, I think that once we've digested a bit better all this new data it will come down to that. I'm awaiting for some confirmation of the Y-DNA results, because that will be important too.

If those results are correct (except for the strange R1b cases), and I think they should be, then I think that the picture from the sampled area is quite clear. We saw that recent paper about the context of the samples, and it was already very clear that these samples had nothing to do with the steppe. Matriarchal society (at least in the burial rite), no weapons, etc...

And here the sampling of that area is pretty good. 41 samples that date to 1300-900 BCE from 6 different sites, plus another 15 from 500 BCE to 1 CE from 2 sites. From those earlier 41 samples 21 yielded Y-DNA, and if results are correct none of them belong to R1a. Given what we know about the steppe people, I think that's a pretty strong evidence that these were not steppe guys. They do have some admixture. From the ones dating to that period, I'm getting a average of 8.7% Steppe_MLBA_East, with many individuals having 0% and there seems to be no difference in the statuses. So I don't think the admixture there is too relevant, but rather circumstantial.

From the 15 later samples, it's not too different. 10.2% Steppe_MLBA_East admixture, and 1 out of 6 R1a.

This is of course just one small area and rather peripheral. But at the same time, it was deemed as central to the steppe "invasion". More importantly, these look clearly like the ancestors of Dardic speakers. So while limited in scope, I do think it's relevant data to add to the few other bits we have already regarding IE languages.

All this said, it's still early times. And I'm absolutely not in a rush to make any claim about IE languages. Any further data can change the outlook of the few bits we have so far. I just think that the data has to be analysed for what it is, and not for what one wants it to be. Something I'm missing in the latest literature on the subject (this very paper being a good example).

Not everything is steppe. I'll finish by saying that in addition to the 8.7% steppe admixture I see in those SPGT samples, they also have an extra 12% Iran_ChL, or an extra 21.3% SC Asian admixture when compared to the Indus diaspora samples (though part of the latter could be local, given the location). We might have to start looking into those things too. And SC Asian/Iran_ChL admixture in the steppe, etc...

epoch2013 said...

@Rob

We don't see much in Sumerian, which is attested from 3000 BC. There is one Euphratic theory claiming to have found a IE substrate in it, but that seems to have been refuted.

For the king said...

@All
Can you guys model Kurds/Iranians and central Asians using these populations:
Hajji Feruz Chl (Main cluster)
Shahr_I_Sokhta_BA3
Sappali tepe BA outlier
Parkhai LBA outlier
Han or Mongola

AWood said...

@a

Yes dave About 7%. So probably maybe had a great grandmother on the steppe"

--More realistically, a great-grandfather. It isn't a coincidence that different geographic regions form admixture clusters due to prolonged isolation from one another. It shouldn't be a surprise when Y haplogroups turn up more frequently, in some cases exclusively in one region over another. Simply because these regions came in contact with one another more frequently when domestication and agriculture came into fruition, let's not start pretending that the odd Z2103 signal south of its most likely point of origin, suddenly spurred civilizations like Sumer. That's just mad hatter talk.

AWood said...

My comment should have been directed at @Rob/Gravetto-Danubian. He's still upset that his ass was handed to him by the R1a Corded Ware guys.

For the king said...

Rob/Gravetto-Danubian was alive during the Corded Ware era? good to know.

Slumbery said...

@Kristiina

Consonant density:
The same map shows that consonant/vowel ratio is variable even among closely related languages with recent branching/connection. This means it is a feature that can change easily and unreliable as a marker for ancient connections.

Sex based grammatical gender:
Better, but we can see that it developed in multiple regions independently (Australia, America) and there are IE languages that lack it (Persian, Armenian).

Ergative/active:
There is some uneven sapling density problem on that map. Sure, there are some closely packed languages in the Caucasus that fit the bill, but other regions are not sampled very well, and actually IE languages seem to be very varying themselves.

Open Genomes said...

@David

I don't doubt at all that Shahr_I_Sokhta_BA3 is an immigrant from the IVC, and it sure seems that Gonur2_BA is also an IVC immigrant, but with some additional interesting BMAC ancestry which includes this "green V5" ancestry from Central Asia (or at least, some Sintashta samples) absent from the IVC Diaspora.

I don't recall that the study actually said that Gonur2_BA was basically from IVC ancestry, and this at least proves that some IVC immigrants were present in BMAC, something that has been shown archaeologically to be possible.

The question then is, does the IVC have any Anatolian Neolithic influence, or just Iranian Neolithic?

Kristiina said...

@ Slumbery

The purpose of my comment was to show that PIE is not utterly different from all the attested languages from the Caucasus region. There are several traits that PIE shares with the languages of the area, including Kartvelian family, Northwest Caucasian family, Sumerian and Semitic family.

Santosh said...

Here is an naive amateur question. Sarazm historically was a mining area. Could it be that the IVC folks went there to get their metal fix?

Santosh said...

More intriguing. The Fergana valley, where the indo aryans came from (later), is just a couple of mountains away from Sarazm?

Slumbery said...

@Kristiina

Sorry, I just skipped through the comments and did not catch any earlier post you reacted to.

And of course IE is not from the moon, but there are some features that are rather uninformative for the purpose of deep language relatedness. I think that the consonant/vowel ratio is one of these.

Samuel Andrews said...

@Alberto,

It looks like at least three homogenous communities are represented in the data from Pakistin. The main community has about 20% Steppe_MLBA_East.

"if results are correct none of them belong to R1a"

Yes, but most Scythians belonged to R1a Z93. Also, high percentages of people in essentially every Indo iranian ethnic group has lots of R1a Z93.

"We might have to start looking into those things too. And SC Asian/Iran_ChL admixture in the steppe, etc."

The closest thing to IranChl ancestry EUropeans have is CHG ancestry (it almost definitly is not from Iran). They recieved it from the Steppe. If you think that is the source of Indo European languages in Europe, it isn't crazy to say Andronovo who had 90%+ R1a Z93 and was of 70% Steppe decent spoke an Indo European language.

Santosh said...

This whole story looks like a story of valleys. (Valleys are where you can do farming). Fergana valley, Serazm valley, Swat valley, wanna ring more bells?

Shaikorth said...

@Sein
I'm getting very similar results with the scaled sheet by using just Onge as AASI stand-in. Have you tried your zombie on the scaled sheet?

Gonur2_BA

Sarazm_Eneolithic 59.5
Onge 25.0
Ganj_Dareh_N 15.4
West_Siberia_N 0.0
Abdul_Hosein 0.0
AfontovaGora3 0.0

Gill said...

Sein,

Even in simulated admixture runs (recombining existing components) using Iran_Chl, Sarazm, and Indus_P as components, Afghans and Tajiks, including the half Tajik half Pashtun individuals, have a strong affinity to Sarazm over anything else (aside from HRP0370 who keeps getting a lot of Iran_Chl, using these components makes HRP0370 and the two Pashtun/Tajik mixes look very different for the first time in Admixture) and in some near the Pakistan area, it can suddenly swap between Indus_P and Sarazm.

The non-ASI part of Indus P is like 85% Iran_N + 15% AG3/ANE (so approx 37-38% ANE proportioned), and the non-CHG part of Sarazm is like 75% Iran_N, 17% AG3/ANE, 8% West Siberia N (that's something like 40-43% ANE proportioned) from the stats you posted above.

I think Indus_P is really just Neolithic South Central Asia (Sarazm percursor) plus extra Iran_N of some sort.

It would also explain the Baloch. Why a population like the Baloch, who seem similar enough to everyone else, generate components that are really high in ANE (zombies of the component usually have 35% ANE or more). That component is always labeled as South Central Asia. I think that's a shadow of the real South Central Asia which was, obviously, in South Central Asia. For whatever reason, the algorithm couldn't reconstruct it from modern Tajiks/Afghans as well as it could from Baloch. It could be due to the influx of CHG and Chalcholithic-era Caucasian (i.e, Iran_Chl) type admixture into the region. Even Sarazm_EN is chock full of CHG.

Ric Hern said...

I wonder if Hajji Firuz had some relation to the Gutians ? The R1b in some Kurds had to come from somewhere ?

Ariel said...

An overlooked fact is that in the "Andronovo" region there are many places with Indo-Aryan names. With the recent ancient dna discoveries it's hard to argue that Andronovo is not an early offshot of the Indo-Aryan node, also the evidence of a genetic impact of people from the Turan region or S Central Asia in the eastern steppe are non existent, so I can see only one solution here. With Hajji_Firuz outlier and her genetic affinity to Yamnaya Bulgaria and with the Andronovo affinities in the BMAC SWAT region, a theory that will argue that PIE stemmed from some southern region would imply a re-invasion of people that spoke a language from the same family, a re-invasion that happened on a large scale in different regions. That bothers my occam's razor sensibility. As I said many times, it's like arguing that in England the English language was spoken before the Romans, then it was forgotten, and then it was brought back by a totally different group of people, at the same time English was also brought to America, but not by the English but by a mystery population with no name, no archaeological relevance, that came from a place that has no evidence or traces of that said language. At least the anatolian theory was falsifiable, theories that are not falsifiable shouldn't be taken that seriously. The steppe theory is has always been super falsifiable, you need one population, a set of data that doesn't fit and everybody will lose their mind, it hasn't happened yet, over hundreds of regions and samples. At some point the data should start speaking for itself.

epoch2013 said...

@Ariel

"Hajji_Firuz outlier and her genetic affinity to Yamnaya Bulgaria"

Whoa! Where does that come from? This is the BA late burial in the mound, isn't it? The laste thread has so many reactions I might have missed it. The BA sample is linked to Yamnaya Bulgaria?

Kristiina said...

@ Slumbery

I cannot find traits that would make PIE/IE languages so unique. Verbal person marking (= subject marking) is mainstream. A similar marking exists in Afro-Asiatic, Uralic and Turkic languages. SOV word order is mainstream. It is the main word order in the languages of India and Caucasus, as well as in Uralic languages and in all Siberian languages. Beekes reconstructes 8 cases to PIE (nominative, vocative, accusative, genitive, ablative, dative, locative, instrumental). The same system exists in Dravidian Tamil. Georgian system is almost identical but it lacks accusative. Only six cases are reconstructed to PU (nominative, accusative, genitive, locative, ablative, lative). Articles, which are an innovation, are shared with Afro-Asiatic languages, and one could easily think that European Neolithic languages had articles. One interesting point is grammatical reduplication which existed in Greek and Sanskrit. Reduplicative forms are frequent in Dravidian and Afro-Asiatic languages. In ancient Greek there was aorist (verbal form, narrative and gnomic aorist), and an aorist, which is usually past tense but can mark other aspects as well, exists in Kartvelian languages.

In his book, Beekes gives an excellent explanation to the existence thematic and athematic endings in PIE: the subject of the transitive verbs are in the ergative, while their object was found in the absolutive. The absolutive also served as the subject of intransitive verbs. This system was valid for the athematic inflection, for the stative inflection, and for the aorist of the thematic verbs. With the present of the thematic verbs it was otherwise. Georgian, for example, has a difference of this kind between the present and the aorist.

Rob said...

@ Epoch

“We don't see much in Sumerian, which is attested from 3000 BC. There is one Euphratic theory claiming to have found a IE substrate in it, but that seems to have been refuted.”

I don’t think it has

Mr. Kulkarni said...

Are the kashkarchi samples better fits for steppe ancestry in the south asian IA and later samples?

mzp1 said...

@Ariel,

Except that Andronovo incursions into South Central Asia are not "invasions". These people are migrants with economic ties to BMAC.

Haji Firuz also not an "invasion". Where is the cultural transformation?

You got the whole thing back-to-front and keep referring to immigrants as invaders and conquerors.

Cultural transformation goes from Caucusus to Western Steppe. We find the earliest Kurgans in LeilaTepe (4000BC) and Maykop. Maykop also seems to have some relationship to South Central Asia but I need to look into that more.

There are no Kurgans anywhere outside the Steppe If PIE were the Kurgan people why did leave behind such an important element of their culture? Kurgans burials are not attested in any branch of Indo European, nor do we have Kurgans archeologically from any IE branch. Not Greek, not Tokharian, Latin, Hittite, only a few in Germanic likely due to cultural diffusion.

The closest we have of Kurgan burials come from Rigveda and early Iranian Mythology

Rigveda: "Let us bury death beneath this mountain" Funeral Hymn book 10
Shahnama: Zahhak (If I remember correctly) is imprisoned under a mountain

These ideas are not entirely consistent with burying loved ones under a mountain though it is the closest we have from IE literature. It could be Kurgan burials are later developments of the same idea.

Kurgans/Steppe cultures are closely related to Iranian speakers, and to no other branch of IE

Skythians appear from 1000BC occupying the Steppe/Kurgan regions. They are very close genetically to Yamnaya peoples. Skythians continued the same building traditions earlier seen in Sintashta with no break anywhere. Skythians continue to use Kurgan burials in much the same way with no break. Sarmations continue the same traditions of burying women in Kurgans in the same region it was practiced by the Yamna 2,000 years earlier.

No other IE branch is attested to have utilised Kurgan burials

You don't like back-migrations? Then explain how steppe PIE speakers change to Iranian speakers with no archeological cultural change?

Borrowings into Uralic are Iranian, if somewhat Indo Aryan shifted

Not PIE, but Iranian. The reason they look rather more IA is because Iranian is simply a divergence from IA. Uralic borrowings were not as Iranian shifted compared to other branches because they are earlier borrowings.

Proto Indo European is complete bullshit. It is not attested anywhere. Neither is Proto-Indo-Iranian. IIr words are not attested anywere, not in Uralic, not in Mittani, nor as you say, as place names in the Andronovo sphere.

No Proto-Indo-Iranian Mythology can be reconstructed.

The Steppe theory cannot explain the divergence of IA and Iranian from IIr. I have yet to hear a single coherent attempt.

How do you go from PIE in Western Steppe to IIr in Sintasta. What caused such a big change? Dont say Uralic influence because those borrowings are all one-way. And how do you go from IIr to Vedic and Avestan?

Mr. Kulkarni said...

@kristiina
Whats your take on the indo iranian loan words in finno ugric. Is that for real? Is the other way round true as well?
Also, is it true that FU has more sanskrit loanwords than iranian.
Whats your take on how this happened.

a said...

AWood said...
@a

Yes dave About 7%. So probably maybe had a great grandmother on the steppe"

"--More realistically, a great-grandfather. It isn't a coincidence that different geographic regions form admixture clusters due to prolonged isolation from one another. It shouldn't be a surprise when Y haplogroups turn up more frequently, in some cases exclusively in one region over another. Simply because these regions came in contact with one another more frequently when domestication and agriculture came into fruition, let's not start pretending that the odd Z2103 signal south of its most likely point of origin, suddenly spurred civilizations like Sumer. That's just mad hatter talk."



F38 is right next door to this sample. Even though its iron age, its downstream from L584*

Iran_IA F38 (2-way)
Iran_ChL 0.815±0.066
Poltavka_outlier 0.185±0.066
P-value 0.72807065
chisq 10.457
Full output

Iran_IA F38 (3-way)
Anatolia_BA 0.122±0.107
Iran_ChL 0.717±0.098
Poltavka_outlier 0.161±0.070
P-value 0.773758066
chisq 8.989

"Identical-by-State (IBS) affinity test. This method is generally pretty good at picking up recent ethnic-specific genetic drift. These are F38's top 25 matches out of over 100 present-day populations:

Georgian 0.676468
Armenian 0.676024
Abkhasian 0.675791"
http://eurogenes.blogspot.ca/2017/08/the-iron-age-iranian.html

Then you have the I4243--Hajji_Firuz_BA-sample 2465-2286 calBCE----- that had to be put as an outlier, for good reason.
1 Natufian 38.47
2 Ancestral_North_Eurasian 31.29
3 West_European_Hunter_Gartherer 25.92
4 Ancestral_South_Eurasian 3.96
5 East_Asian 0.36




Rob said...

@ AWood

“My comment should have been directed at @Rob/Gravetto-Danubian. He's still upset that his ass was handed to him by the R1a Corded Ware guys.”

No idea what you’re taking about there, Britney
I2a1 wasn’t even really in Poland but lived there later. I2a2 has been found already in 4 CWC and amply in later lombards and Goths, so not really sure about ass-handing
but what’s that got to do with R1b in Chalcolithic Iran ? Your strawman comment "let's not start pretending that the odd Z2103 signal south of its most likely point of origin, suddenly spurred civilizations like Sumer. That's just mad hatter talk." is typical AWood nonsense because in my follow-up comment to "a" i stated "I think overall R1b is from the north, but that doesn't rule out bilateral connections which might predate Yamnaya (as we see between Khvalynsk and Kura-Arax)."

It seems you’re a little frazzled and your facts are way off as usual, so run along to your Beaker thread so Alan and RMS2 can whisper sweet fairytales in your ear and you can freely handle each other’s asses all day long.

epoch2013 said...

@Rob

This is the paper:
http://science.org.ge/old/moambe/2-3/Gordon%20Whitteker.pdf

In it the following link, one among may, is laid:

Sumerian: hu ‘(phonetic value)’
IE: *h2au-i-2 ‘bird’

Furtheron the following link is laid:

Sumerian: •u8, us5, OS u3-wi ‘ewe’ (Ebla)
IE: *h3ou-i-s ‘sheep’

How on earth would Sumerian loose a preceding 'h' on one occasion but keep and fortify it on another? In am an utter amateur, so there is a substantial chance I miss something, but that seems like randomly allowing for sound-chances where they fit.

Rob said...

@ Ariel

"At least the anatolian theory was falsifiable, theories that are not falsifiable shouldn't be taken that seriously. The steppe theory is has always been super falsifiable"

ha ! What gold from Ariel

Look at the comments here - esp Sams usual insights "there is 7.5% steppe in SBPT & 10% steppe in Myceneans so the steppe hypothesis has been confirmed !!!!"

Ariel said...

mzp1

In fact Andronovo has higher level of EEF, that means that it came from the west, probably from a pseudo CWC area, that explains the balto-slavic Indo-Iranian connection and R1a-Z93 situation. We know that there are late eastern steppe samples similar to CWC, that is quite convenient for the theory, wouldn't you say? Andronovo is problably not at the balto-slavic Indo-Iranian "node level" (maybe not even Sintashta was the "split"), those name places I was talking about are in fact Indo-Aryan related, and connect to later BMAC and SWAT, by archeology and now by genes.

The reason behind the lack of Kurgan burials in South Asia is very simple, those burials were a much earlier phenomena, those steppe people arrived in Europe much earlier than everywhere else, think about this, where are the Kurganic burials in Spain? That doesn't mean that steppe culture didn't arrive in Spain.

a said...

Rob said...
@ Ariel

"At least the anatolian theory was falsifiable, theories that are not falsifiable shouldn't be taken that seriously. The steppe theory is has always been super falsifiable"

ha ! What gold from Ariel

Look at the comments here - esp Sams usual insights "there is 7.5% steppe in SBPT & 10% steppe in Myceneans so the steppe hypothesis has been confirmed !!!!"

More like fools gold. Look how R1b-Z2103-steppe dynamics worked from Steppe to Hungary Bell Beakers

http://bellbeakerblogger.blogspot.ca/2017/07/szigetszentmiklos-cemetery-santas-six.html


Beaker_Hungary

Barcin_N,49
Yamnaya_Samara,31.8
Narva_Lithuania,11.4
Blatterhole_HG,6
Ukraine_Mesolithic,1.8

[1] distance%=4.9659

Beaker_Hungary_no_steppe--------{I2a1a1}

Barcin_N,76.2
Blatterhole_HG,23.8

[1] distance%=2.4992

Beaker_Hungary_outlier

Yamnaya_Samara,76
Barcin_N,19
Koros_HG,4.4
Blatterhole_HG,0.6

Arza said...

136 ancient genomes from the upcoming "A population genomic history of the Eurasian steppe" study are online:
https://www.ebi.ac.uk/ena/data/view/PRJEB20658

Palacista said...

Rob can you not use autistic as an insult.

Rob said...

@ blogger "a"

Trying to understand your on the spectrum comments - you seem to be proving what is already known - plenty of steppe in the known steppe-admixed BB individuals, whose language provenance is unknown, but sweet FA in regions where indo-european is actually attested.
Gotcha !

a said...

@ blogger Rob said...

"Again, your comments don't make sense
Can you try to be less autistic"

Try taking the cob out[without breaking it], you might understand better

Rob said...

@ Palacista
Sorry.

Rob said...

@ epoch
So when you meant 'disproven' you meant you don't think it is correct.

Ariel said...

Rob

In fact most languages are spread by people who have distant relationship with the original speakers. I'm very much opposed to idea that there were scandinavian like people spreading Greek in the Minoan world. The same goes for South Asia or Iran, those IE speakers were mixed with the locals, at the end of the chain the steppe component would not have been predominant, Greece was at the end of the chain. I assume that the Mittanni or Hittite elite was mixed with the locals, so much that probably there wasn't a big genetic difference, but when we will have the data we will see a difference between Hurrian pre and post Mitanni. How much roman DNA there is France? Or Spain? Not much, but probably you will find some roman-like sample in roman Spain, like that Hajji BA or those Bactria Margiana samples with more steppe, like the Vucedol outlier or the Yamna in Bulgaria.

mzp1 said...

I not denying that EEF type ancestry moved into Andronovo. They obviously didnt come from CWC which is a non-steppe non Kurgan culture. Likely those people just became part of the Yamnaya grouping and ended up East when the Kurgan culture expanded there (to take advantage of Sintashta <-> BMAC trading routes).

Central/South Asian trades routes to the Middle East and Caucuses causes Iranian tribes to move West into Caucuse regions.

Iranian tribes move North of Caucuse (Yamnaya)

Sintashta set up to sell metal to BMAC and join Central Asian/South Asian/Middle Eastern trade triangle.

Yamnaya peoples move West to take advantage of trade between Sintashta and BMAC. (Andronovo)

Andronovo move further South to get closed to BMAC wealth and trading opportunities.

IVC, Sintashta, BMAC all decline (2000BC - 1500 BC)

Southern Andronovo merge with South Central Asians.

BMAC culture moves West (YAZ), Medians, Persians attested.

Rob said...

@ Ariel

"n fact most languages are spread by people who have distant relationship with the original speakers. "

Yes I agree, but that doesn't negate my reply to your comment which is the non-falsifiablity rampant in the steppe zealots which prevail on this blog.
For them, a spit of "steppe MBA' proves the kurgan hypothesis, despite the fact that hardly any of them consider context, or indeed have any real clue about what they;re talking about.
hence your original statement was wrong, and your follow up statement irrelevant and off topic

Davidski said...

@Matt

fatalx:
> 2 edges at A

old europe said...

Rob
Always remember what you call kurgan burials are in fact as for position neolithic Farmers and as for the weapons ( axe halberds, bow, arrowheads) they have a precursor in remedello ( see jeunesse study)

Samuel Andrews said...

@Rob,

The authors of this paper, including David Reich, agree that Steppe admixture in Iron age Pakistan go in line with a Steppe origin for Indo Iranian languages. So, am I and others really autistic Steppe fanboys who never do any critical thinking? Learn to respect other's opinons. Everyone who doesn't agree with you isn't mentally handicap.

R1a Z93, mtDNA evidence, Andronovo, Scythian, Pathan, Kalash, Tajik DNA are really what confirm Indo Iranian languages are from the Steppe.

Hindi is an Indo Iranian language hence is also from the Steppe. Steppe admixture in Iron age Pakistan which was absent in all pre-2000 BC genomes from central Asia is more confirmation but not the most important evidence.

Rob said...

@ Ariel

“ much earlier than everywhere else, think about this, where are the Kurganic burials in Spain? That doesn't mean that steppe culture didn't arrive in Spain.”


There are “kurgan” burials in Iberia
These are individual inhumations facing north with arrowheads, copper dagger etc - just like in B.B. Germany and even late CWC
In fact one was tested in Olalde and was L51

@ Old Europe
Of course
It’s nicely summarised in that article that Matt and everyone else here wish never existed because it’s yet nail in the coffin for their steppe fable

Matt said...

@Davidski, trifurcations are a no-no then. Forgot that. Can you try this instead?: https://pastebin.com/yEphP1gn

old europe said...

Rob
Where are you from?

Rob said...

@ OE
Near you

old europe said...

More specific please...

Ariel said...

mzp1
So, the are two possibilities that I can see here for the Yamna is not PIE route.
1) Yamna got IE from the caucasus or Iran (even if Iran neolithic DNA doesn't work for that but whatever). In this case the Kurgan theory is the same but there is a slight unfalsifiable variation at the very beginning (occam's razor needed).
2) All those steppe people never spoke IE or acquired it later on from different sources, either from the farmers when it comes to Europe, or somehow from mysterious IE speakers hidden in Mesopotamian caves when it comes to Mitanni and western Iranians, or from those rich BMAC people that your are talking about when it comes to the Indo-Aryan branch (btw all those people were totally unrelated and genetically cats and dogs, and in those regions there were no attested IE languages before and there were many attested languages with not even a word or a name that resembles IE).(Also you will have to explain Tocharian somehow).
Too me that biggest problem is going to always be the same, R1a and the Balto-Slavic Indo-Iranian connection. There is no late movement from the middle east or South Asia to Europe or the steppe. So it's really simple logic, it's a fact that the balto-slavic R1a node was already in Europe in the CWC era (modern popualtions are identical to those ancients), and there has to be a relatively recent common source with Indo-Iranian branch, there is just one possibility there: CWC moving east, one that has amazing archaeological and genetic evidence

Mr. Kulkarni said...

@samuel
Reich is a geneticist.

"R1a Z93, mtDNA evidence, Andronovo, Scythian, Pathan, Kalash, Tajik DNA are really what confirm Indo Iranian languages are from the Steppe."
lol wut? dna speaks languages now? do share what you are smoking.

Steppe entry into Swat is too little, too late for Vedic aryans.
Sorry bro.




Ariel said...

ROB

We were talking about THE kurgan burial with the mound and so on. And don't think that mzp1 was referring to any burial with a distant connection to the steppe.

Rob said...

Ariel

The mound is just a monument - a marker above the ground
Whether it’s a pile of dirt or stone cairn or reused megalith - with similar positioning and provisioning of the inhumed- it’s all the same concept and all stems back to the area where the steppe was acculturated by east Balkan farmers

old europe said...

Not only accultured....EEF in sredni stog was 40% and EEF in yamna was around 15% if I remember correctly Dave

Rob said...

@ Mr Kulkarni
Reich is paying lip service
Anthony is everywhere in these Harvard papers
Which is why they are postHoc in their analyses
Hopefully the European labs come up with more intelligent theories

mzp1 said...

How much EEF do South Asians have?

Mr. Kulkarni said...

@Rob
Yes. Their conclusion that steppe probably entered IVC 2000-1500bc after bypassing BMAC was quite laughable.
So, no invasion - no chariots, hardly any r1a males, hardly any steppe admixture overall till 1st millenium BC.

@all
How is Y haplogroup J2a from Kotias (CHG) found in Karelia 6200BC (EHG)?


Open Genomes said...

The R1b1a1a2a2-Z2103 from south of Lake Urmia, 7650 ybp / 5900-5500 BCE (but no radiocarbon date).

This is a bit of a mix of ancestry (the Natufians) and descendants (the others.):

[1] "distance%=1.2129 / distance=0.012129"

Hajji_Firuz_ChL_o:I2327

Seh_Gabi_ChL:I1662 44.50
Geoksiur_Eneolithic:S8504.E1.L1 19.95
Mycenaean:I9006 15.15
Varna_o:ANI163 8.75
Afanasievo:I5279 5.85
Natufian:I1072 5.80

It appears that the Shulaveri-Shomu culture had descendants no only in Chalcolithic Western Iran and Turkmenistan, but also in the Late Neolithic / Bronze Age Balkans region, and also, interestingly, in the Afanasievo culture, but apparently not Yamnanya or anything further west.

Even if the date is wrong and this is an intrusive burial, the dispersal would still hold true.

Archaeologically, there was a Late Neolithic / Anatolian Chalcolithic movement into the Balkans. It's hard to say how the influence went as far east as Afanasievo. Perhaps this had something to do with a connection between the Anatolian Indo-European languages, and Tocharian?

Anthro Survey said...

@Mr. Kulkarni

I wouldn't be so sure. What if the bulk of the invaders landed and set camp a bit east in modern-day Uttar Pradesh/Haryana on the outskirts of the decaying IVC?
Consider this:
1.Highest Brahmin population per capita(around 10% of the population) in India if we don't count those Himalayan states; historical Brahmin migrations southward also stemmed from this area
2.Epicenter of post-IVC cultural phenomena in Northern India involving pastoral and relatively rural societies
3.Epicenter of Vedic-era and Mauryan(a bit east-ward shifted here) India.

We need to be looking more on the Ganges side of things.

Also, it won't be easy to find remains from proto-Brahmins: cremation.

Mr. Kulkarni said...

@anthro survey
so you have no evidence then? ok.
Sindhu and Saraswati were the major rivers in the Rig Veda, not Yamuna or Ganga.
Your 'Aryan invaders setting up camps east of Haryana' makes no sense, doesnt match literary record.




mzp1 said...

Rigveda is mostly concerned with the Indus, not the Gangetic plain.

mzp1 said...

The Rigvedic people are nothing like similar IEs attested around 1500BC, Hittites, Myceneans, Mittani. I mean in terms of geographical size, clan structure, internal warfare, state organization etc etc Not to mention language, literary material and mythology.

The other three are quite similar i.e highly centralized political units in contact with attested non-IE speakers.

Anthro Survey said...

@Matt and Sein

Yeah, so far my models like Steppe_MLBA. Seems MLBA_east was less of a bus stop than it was a detour. After all, those Kashkarchi and Dashty Kozy samples are basically steppe_MLBA/sintashta-like. Same with Petrovka.

I'll post models later where Pamiris and Pashtuns are successfully modeled as a combo of Steppe_MLBA and Scythian-related Zevakinski_LBA. This makes a good historical sense, imo. I would have modeled Jats this way, too, but Dave removed them from the datasheet. :-(

In fact, Dave's avatar is actually a 1st century bust of a Scythian from Khalchayan, in what is today southern Uzbekistan.

mzp1 said...

@ariel,

Did you not read my posts..Balto-Slavic has Indo-Iranian connections because Kurgan culture spoke Iranian.

There are not many attested languages prior to 2000BC, none for South/Central Asia.

Tocharian doesnt make sense for Steppe Theory either, but I like the theory above from Open Genomes. That would explain why even the Afaniesevo didnt use Kurgans i.e because they were not Iranian, or rather not descended from the Yamnaya/South Caucuse Iranians. Because genetically they are the same, and according to you that seems to be everything, so why no Kurgans?

Basically there are a few things our theory doesnt work for. Lets however look at the issues in your theory, just based on language change.

You need to show how PIE becomes IIr in Sintashta, even though it is not moving far, even though it barely changes at all getting to South Asia, through BMAC and IVC.

You need to show where IIr becomes Iranian and IA.

You need to show archeological basis for linguistic change in BMAC

You need to show archeological basis for linguistic change in IVC

And probably the best one, you need to show archeological basis for linguistic change in Western Steppe from PIE to Iranian.

Anthro Survey said...

@Kulkarni
Not yet. :-)

Ok, the early Vedic period supposedly had a more westerly center of gravity according to the scriptures(which wasn't exactly Swat). It still doesn't take away from UP's subsequent significance and the population increase taking place there. Painted Grey Ware. Kuru Kingdom.

So, my point is, it'd be better if they sampled these regions instead for the dates indicated on Swat samples.

Rob said...

@ Anthro

A lot of the kurgan narrative rested on Swat.
Here is what Kuzmina said, what most of the Kurgan proponents rest on

"the evidence of cultural belief is of crucial importance to assigning an ethnic identification and origin for the population of Swat. These aspects of culture are often preserved, especially in the language, despite distant migrations. The analysis shown in Table 17 shows that all the parameters and all the elements of the burial rites of Swat and Gomal have close analogues in no other culture but the Andronovo.'''

She was waaay off, and so were Anthony, Parpola, Wilzel, etc and so were internet 'experts' like predicting a major 'crashing into Asia' with tons of 'bad ass R1a ' and 70% steppe.

Some people are choosing to toe around the reality of 1 R1a and 7% steppe in Swat.

At least you're coming up with new theories - like the conquering Indo-Aryans were beggars in temporary camps.

rozenfag said...

@Arza Thanks for the link. The authors seems to be very confident: " The steppe nomads later further admixed with Turkic-speaking groups of East Asian ancestry that spread westward across the steppe in multiple waves: firstly, the Xiongnu confederations that emerged in Mongolia around the 3nd/2nd century BC;" AFAIK there is no scholar consensus about the language of Xiongnu.

Davidski said...

@Matt

qpGraph appears to have gotten stuck while trying to interpret the new graph file that you posted.

Rob said...

Sam Andrews

"The authors of this paper, including David Reich, agree that Steppe admixture in Iron age Pakistan go in line with a Steppe origin for Indo Iranian languages. So, am I and others really autistic Steppe fanboys who never do any critical thinking? Learn to respect other's opinons. Everyone who doesn't agree with you isn't mentally handicap."

First off, i never called anyone handicapped just because they disagreed with me, so you're blowing smoke there. (rather i pulled "a" on his typical lack of effort posts) In fact, I don;t really have any stance other than understanding 2 or 3 possible explanations for the expansion of PIE across Eurasia.
Secondly, I don't need a lesson in ethics from you, as its somewhat hypocritical given your history of attacks and misrepresentations with anything that goes against your favourite steppe theory.
Thirdly, I am right to castigate you guys. With a few exceptions you seem to be steppe ideologues light on knowledge and heavy on the talk, i mean deep down i like y'all but expect shtick when you make flawed bold assertions and dish it out in the first place.

"
R1a Z93, mtDNA evidence, Andronovo, Scythian, Pathan, Kalash, Tajik DNA are really what confirm Indo Iranian languages are from the Steppe."

Maybe maybe not. Sure, there is evidence of mild steppe admixture in SA, but it's far from the reality you masterminds predicted. Anyone genuinely interested in prehistory will now reconsider things and come up with newer or adapted models to explain the expansion of PIE because your whizz -bang scenario has just been falsified.


BTW Sam, you don;t understand how academia works. I wouldnt worry about work Reich says now. Look to what the data says, and to do that, you need to go study for 10 years before commenting.

«Oldest ‹Older   1 – 200 of 255   Newer› Newest»