search this blog

Saturday, December 17, 2016

Mobile and then some

I've now read the new Valtuena et al. prerprint a couple of times, and even though the plague angle is interesting, what really sticks out for me is how incredibly mobile our steppe ancestors were.

It's been obvious for a while now that during the Early Bronze Age there was a massive expansion from the Pontic-Caspian Steppe in all directions, as far as the Atlantic shores in the west and the Altai region in the east. However, considering the phylogeny of Yersinia pestis genomes in Valtuena et al., it looks like the Yamnaya and/or closely related peoples also regularly roved across the entire length of the Eurasian Steppe.

The oldest and phylogenetically most basal sample of Yersinia pestis to date comes from an Afanasievo skeleton from the Altai region dated to 4836-4625 YBP. Hot on its heels, the next oldest is a sample derived from the Afanasievo strain, but from a Corded Ware skeleton from the East Baltic dated to 4571-4422 BP. The two burial sites are separated by something like 3,000 kilometers.

It's unlikely that Yersinia pestis spread from the Pontic-Caspian Steppe, say, with the Yamnaya, both to the Altai region and the East Baltic, because it hasn't yet been found in any of the Yamnaya remains. It's also generally accepted that Yersinia pestis is indeed from Asia. So it must have been picked up by migrants from the Pontic-Caspian Steppe somewhere in Asia, probably the Altai region, whose descendants then returned to Europe with the bug, and may have taken part in the Corded Ware expansion.

But of course it doesn't end there, because it appears that the descendants or close relatives of the Baltic Corded Ware people then moved back to Asia, because a sample of Yersinia pestis derived from the Corded Ware strain is found in an Andronovo skeleton from the Altai region dated to 3694-3575 BP. Note also that the genome-wide genetic structure of most of the Andronovo individuals sampled to date is Corded Ware-like, in fact more so than Yamnaya-like. In other words, ancient human DNA also shows a very close relationship between Andronovo and the older Corded Ware.

Why were these people roving around so much? I know they were pastoralists and that the Eurasian Steppe became progressively drier during the Bronze Age, so it may have been necessary to move wherever suitable pastures were available. But still, it seems rather ridiculous that they spent so much energy on long range intercontinental travel when, by all accounts, all they had were wagons pulled by oxen.


Valtuena et al., The Stone Age Plague: 1000 years of Persistence in Eurasia, bioRxiv, Posted December 15, 2016. doi:

See also...

So how many of you really read the preprints at bioRxiv?

Bronze Age dope dealers


Nirjhar007 said...

Its not easy to imagine a logical reason or even to establish that such migrations took place at all . The case of Afana. is yet to be fully investigated with y-dna , We just hope that it don't turn out to be R1a , as that will really complicate the things . I like to think that metals perhaps were the main thing that were behind the migrations of nomads, but of course not every site of them had metals! .

I hope the upcoming massive aDNA results from Asia and Europe will shed light and we will be able to describe in more clinical manner .

If we were too take the Mouse evidence , then Its India from where the migrations happened . But of course we can't rely on that, can we ? ....

Davidski said...

Don't be ridiculous, of course these migrations happened.

And of course Afanasievo is from Eastern Europe. It's identical to Yamnaya and very similar to Corded Ware. So there's no reason why Afanasievo skeletons can't belong to R1a and/or R1b.

I can't see any complications at all.

Nirjhar007 said...

You can read for example the works of Frachetti , who has criticized the connection of ''Repin Culture'' with that of Afana. with alternate explanations.

Of course they show similar components as of Yamnaya , but there are valid reasons to think that Afana had a different source of influence, from rather south .

If Afana. belongs to R1a , that will mean that it didn't derive from Yamnaya area . Its plain and very simple , there is no reason to think yet, that Yamnaya area had R1a-M417s in considerable degree , there is no mystery in this shortcoming. And if do belong to R1b then it will mean that it didn't create any considerable effect, since R1b is insignificant for Indo-Iranians and Tocharians etc...

And already by the hint of that Ulug Tepe aDNA, it looks Yamanya like instead of Andronovo even during late bronze age (and we do know EBA samples are very good fit for Indic and SC Asian IE groups) , there is a parallelism with Frachettis suggestions IMO.

Davidski said...

Afanasievo is largely a mixture of Eastern European and Caucasian foragers and less southern than Yamnaya_Kalmykia.

So it comes from the Pontic-Caspian Steppe, and is closely related to the older Khvalynsk. Don't fool yourself otherwise.

Davidski said...

You know, you're just fooling yourself, despite all of the data available now.

You'll be devastated when the ancient DNA from India comes out.

Nirjhar007 said...

I am not getting fooled and neither are you David . You can suggest what you are suggesting , its also a possibility but not the possibility . To reach from the level of 'a possibility' to the level of 'the possibility' , we will need to see further .

And we must always accept at the end, that Indo-European is a linguistic issue . A linguistic issue that requires also the application of Archaeology to describe the cultural aspects and evolution of it and genes don't speak language or describe cultural aspects . Genetics comes as a secondary tool , very good in case of proposing admixture, migrations etc etc, but its still in its infant stages . Good thing and the important thing is of course that, genetics is maturing fast and soon it will gain the maturity that Linguistics and Archaeology has , but we have to be patient and have to refrain from making bold and immature conclusions based on fragmentary genetic data .

Davidski said...

According to most archaeologists during the past 40 years there was no expansion from the steppe, even into Europe, which sounds pretty funny nowadays.

Nirjhar007 said...

We will all get fooled there and now , its the integral part of the package and that makes it fascinating .

Well thanks at least for suggesting again, that the Indian aDNA has 'devastating potential';).

Yes absolutely . As you know I am writing here what you told me to write and to keep Marnie happy! .


Samuel Andrews said...

"And if do belong to R1b then it will mean that it didn't create any considerable effect, since R1b is insignificant for Indo-Iranians and Tocharians etc..."

We don't have Tocherian DNA samples and there may be no modern populations with significant Tocherian ancestry. Most modern Turkic-speaking Central Asians who's predecessors conquered Indo Iranian speaking Central Asians have hardly any EHG and EEF. They don't appear to have a lot of Andronovo-like or Yamnaya-like ancestry. Instead their West Eurasian ancestry is mostly CHG/Iran_Neo. Our single Sycthian genome from Samara Russia does have an extra dose of CHG/Iran_Neo but he's basically the same as Andronovo. My point is we don't fully know the genetic history of Central Asia and so don't know what Tocherians were like genetically.

rozenfag said...

I wish we had decent aDNA from Central Asia. I know that there are samples from Central Asia analyzed in labs, but nothing is known when they will be published.

Nirjhar007 said...

Well applying Archaeology and Linguistics , Mallory I think rightly suggested xiaohe to be Proto-Tocharian and Xiaohe is predominantly R1a.

postneo said...

first thing to consider: in a contiguous land mass its far simpler for diseases without humans moving much.
animals can transmit, and neighbours can be infected.

mickeydodds1 said...

'All the best laid plans of mice and men come to naught'.

Davidski said...


But you're ignoring the human genetic evidence.

Afanasievo is identical to Yamnaya, and Corded Ware is mostly like Afanasievo and Yamnaya. Andronovo is basically like Corded Ware, rather than Yamnaya.

You must really be in pain seeing all of this.

postneo said...

also a virus will need to have low virulence in a host to spread over low density population areas.
linguistics is not a mature science. Its horribly under developed. A lot more distance to go... A lot of subjectivity and sometimes academic hubris.

phonemes are a lot more intriguing as compared to chance resemblance of isolated words. I heard a strange consonant(for europe atleast).. a retroflex flap , in some some kind of dubbed section in some techno music. For me it was like being hit by a brick. I suspect the language was Portuguese but it was too fleeting. Do you know anything more about it?

Apparently there is weak survival of retroflexion in swedish. I have heard there are other pockets. I find a lot of europeans cannot and even linguistic literature does not distinguish between these sounds.

For me and a lot of Indians, Its like some invisible ink that demarcates populations.

postneo said...

"But you're ignoring the human genetic evidence.

Afanasievo is identical to Yamnaya, and Corded Ware is mostly like Afanasievo and Yamnaya. Andronovo is basically like Corded Ware, rather than Yamnaya."

The connections seem temporal and not necessarily driven by people. Has it occurred to you that a lot of these mutations would have taken place in fleas.

I don't like to jump to conclusions. But if the spread was indeed rapid(??) then perhaps millets and associated rodent populations could be one of the vectors.

postneo said...

But frankly David I am guilty of not reading though?!! let me read first
I am just being cautionary about attributing all the spread to humans.

Davidski said...

No one's jumping to any conclusions. The phylogeny of this plague strain matches the movements of people as inferred from ancient human DNA.

And I don't think this strain of the plague has anything to do with fleas.

capra internetensis said...


It would be pretty strange if Uyghurs didn't have a ton of Tocharian blood. The trick is distinguishing it from all the Turkic, Soghdian, and other stuff they have mixed in.

Samuel Andrews said...

"It would be pretty strange if Uyghurs didn't have a ton of Tocharian blood. The trick is distinguishing it from all the Turkic, Soghdian, and other stuff they have mixed in."

It'd be strange but it isn't impossible. They don't seem to have a lot of Steppe ancestry. It would be strange if Tocherians didn't have a lot of Steppe ancestry. The same is true for Kazachs. They don't seem o have a lot of Steppe ancestry even though the Sycthians did. So maybe they don't have a lot of Sycthian ancestry. Maybe David can enlighten us on this subject.

andrew said...

"it seems rather ridiculous that they spent so much energy on long range intercontinental travel when, by all accounts, all they had were wagons pulled by oxen.

IIRC, they had domesticated horses by then. And, you don't need the entire tribe to relocate itself to spread disease, just a handful of people riding horses on missions of trade, diplomacy, exploration, selling services as mercenaries, wife seeking, or whatever.

Davidski said...

Khvalynsk and Yamnaya did have domesticated horses, but is there any evidence that these horses from the Western Steppe made it to the Altai with the ancestors of the Afanasievo people? I don't remember anything about horses in Afanasievo.

Also, it's hard to believe that there were regular contacts between the steppe tribes in the Altai and near the Baltic. But I could be wrong here.

Grey said...

"Why were these people roving around so much?"

seems to me there's three broad options:

1) tribal groups moving en masse to a new home - in which case why would they go back and forth?

2) transhumance - i imagine that was a thing in many cases but Ukraine to the Baltic and back?

3) long distance traders - that would explain the back and forth but what would they be trading that could bring pestis?


running with the trader option

iirc the plague flea is carried around by rodents (marmots in central Asia) and the explanation for the historical plagues is the fleas were carried by rats on ships so a relatively fast journey with food so the rats could survive the journey

so if this was an early version of the same thing you'd think these traders would need to be carrying food in large enough quantities for rats to feed on and hide in?

however ships needed to carry their own food - cos there's no malls on the sea - whereas on land much easier to visualize trading posts along the route selling food to traders than them carrying it with them (and if for some reason that was necessary boats down the Volga would take the weight better)

so... traders to the Baltic carrying bulk grain in barges to eat along the way?

i can just about see it as a neolithic fantasy horror movie: tribes along the Volga too hostile to trade with for food so caravan of beaker barges with copper and grain sail down the Volga with a plague rat on board... dun dun dun

but otherwise seems implausible

so what else?


other mammals e.g. cats, can catch the plague apparently - and can pass it to humans by sneezing etc

so what if these traders were trading horses - bringing herds from the steppe?

would they need to bring fodder as well as letting them eat grass? in which case you get the grain barges again

otherwise (unless i've missed a branch) if the horses carried the fleas wouldn't they need to be unaffected?


Anyone ever tested Przewalski's_horse to see if it was immune to plague?

Seinundzeit said...


It seems Uzbeks have some very serious steppe-related admixture. So, I'm fairly certain that most Central Asians (including Uyghurs and Kazakhs) have ancestry from streams of West Eurasian ancestry which were highly steppe-admixed.

Here is a model:

43.50% Mongola
27.85% Srubnaya_outlier
22.35% Iran_Chalcolithic
5.55% Barcin_Neolithic

In this context, Lithuanians and other Northern Europeans score 30% Srubnaya_outlier, as do most South Central Asians from Afghanistan and northwestern Pakistan. Pamiri Tajiks score around 40%.

So, despite being 40%-45% East Asian, Uzbeks are the same level of EHG/ANE-rich steppe ancestry as Northern Europeans and South Central Asians.

Anyway, I'm pretty sure there is some Tocharian ancestry among the Uyghurs.

Samuel Andrews said...

"So, despite being 40%-45% East Asian, Uzbeks are the same level of EHG/ANE-rich steppe ancestry as Northern Europeans and South Central Asians."

I'd like to know what type of test produced those results. That model isn't good for Northern Europeans because they don't have Iran_Chalcolithic or Mongolian-like ancestry(except for some) and because it doesn't include WHG. We know Uzbeckistan doesn't have as much Steppe ancestry as Northern Europe anyways.

All results I've seen for Central Asians say they're heavily admixed with CHG/Iran_Neo in a way which ancient DNA can't explain. They definitly have some Steppe ancestry but they aren't Turkic Siberian+Steppe like we'd think based on ancient DNA. They have huge amounts of CHG/Iran Neo. Our Iron Age Sycthian had some extra amounts of IranNeo as well, maybe people from Western/Southern Asia were moving north.

Result for Kyrguz using David's Global10 PCA.

"Ket" 64.75
"Barcin_Neolithic:Avg" 17.15
"Iran_Neolithic:Avg" 10.45
"Levant_Neolithic:Avg" 7.65
"Srubnaya:Avg" 0
"Kotias:KK1" 0
"Baalberge_MN:I0560" 0
"Esperstedt_MN:I0172" 0
"Iberia_MN:Avg" 0

Results for Uzbeck and Uygur using one of David's old D-stat spreadsheets. Iran Neolithic isn't aviable in this spreadsheet.


"Iranian" 45.75
"Nganasan" 41.95
"Srubnaya" 11.05
"Kotias" 1.25
"Yamnaya_Kalmykia" 0
"Anatolia_Neolithic" 0


"Nganasan" 55.35
"Iranian" 30.5
"Srubnaya" 10.6
"Kotias" 3.55
"Yamnaya_Kalmykia" 0
"Anatolia_Neolithic" 0

Davidski said...

Holy shit, I just realized that there's a huge mistake in this paper.

They confused Afanasievo with Andronovo. Check this out...

"The Baltic Y. pestis genomes are genetically derived from the strain that was found in the ‘Andronovo Complex’ from the Altai region."

Nope, RISE509 with the oldest plague strain is from an Afanasievo burial.

Seinundzeit said...


Here is how Lithuanians fare, using the same reference populations:

42.05% Villabruna
30.00% Srubnaya_outlier
22.75% Barcin_Neolithic
5.2% Iran_Chalcolithic


(the Iran_Chalcolithic represents CHG in Lithuanians, since there is a close relationship between CHG and Iran_Chalcolithic. I don't include CHG, because they do strange things in the South Central Asian modelling)

Again, the same reference populations as the ones used on the Uzbeks.

Whenever the steppe reference is heavily EHG/ANE-shifted, northern Europeans and South Central Asians tend to be equal, while Pamiri Tajiks tend to exceed Northern Europeans.

It is a common pattern, and it is reason for why Lazaridis et al. claims that the Kalash have around the same amount of Yamanya-related admixture as Lithuanians.

In reality, it actually boils down to how ANE-shifted the the steppe reference happens to be.

When the steppe reference has much less EHG/ANE, it skews much more towards Northern Europeans, rather than towards South Central Asians.

As the less EHG/ANE-shifted steppe populations are usually shifted in an Anatolian Neolithic/WHG direction, this makes them more similar to contemporary Northern Europeans.

Regardless, since Uzbeks are basically around 60%-70% genetically derived from a population very similar to the most "northern"-shifted Pamiri Tajiks, these results aren't surprising.

Harry Parihar said...
This comment has been removed by the author.
Seinundzeit said...

Hey Harry,

Definitely, I'll try to post some stuff tomorrow.

Rob said...

@ Dave

Yes. Andronovo is after CWC. Unless the plague had time machines. ;)

Rob said...

@ Sein

I think this is a good model :

"Karasuk_outlier:RISE497" 36.35
"Barcin_Neolithic:I1097" 16.6
"Kotias:KK1" 14.1
"Altai_IA:RISE600" 14.05
"Paniya" 9.4
"Armenia_MLBA:RISE408" 5.85

"Karasuk_outlier:RISE497" 67.9
"Barcin_Neolithic:I1097" 17.6
"Satsurblia:SATP" 8.8
"Greece_LN:Klei10" 3.8

(source Pops included Iran Neolithic)

Rob said...

....and Karasuk Outlier is:

"Ulchi" 82.35
"AfontovaGora3:I9050.damage" 17.65
"Loschbour:Loschbour" 0

Seinundzeit said...


It's a good model, but I always try to have a setup which can accommodate any population I throw at it.

Basically, I tend to avoid the creation of region-specific reference population setups, because I like to see how different populations behave when they are all tested under the same conditions.

This is a good way to gauge broad affinities, and it lets us avoid some confounding factors.

The only problem, one has to spend an annoyingly long time playing around with different reference populations, to see which set allows one to accommodate every tested population in a sensible manner.

I guess the best strategy might be to use reference populations that aren't recently admixed between each other, and/or whose general affinities are well examined within the academic literature.

In my case, this strategy has worked well.

Rob said...

@ Sein

As do I. I group them by Eurasia-wide, with varying chronologic subsets, and the sets can be used for Europe & south Asia. So, based on those results, there seems to be little direct steppe input in those groups. Lithuanians, look wholly different to Uzbeks; as do Tajiks, for example.

"Armenia_EBA:I1633" 36.6
"Srubnaya_outlier:I0354" 32.55
"Paniya" 16.4
"Iran_Neolithic:I1945" 4.95
"Srubnaya:I0431" 3.75
"Karasuk_outlier:RISE497" 2.25

Seinundzeit said...

@ Rob

Regardless, the use of the term "direct" is of great interest.

Uzbeks, Kyrgyz, etc, are Turkic peoples, so no one would expect them to have "direct" ancestry from the general stream of peoples ranging from Samara_Eneolithic/Srubnaya_outlier all the way to Sintashta/Andronvo.

In addition, the whole use of the term "direct" is exceedingly vague/undefined in the spatio-temporal context that is being examined. This sort of talk is very philosophically underdetermined.

But, what isn't vague is the fact that these Turkic Central Asians do have substantial steppe-related admixture, and not unusually high/or unexpected levels of CHG/Iran_Neolithic.

Their West Eurasian ancestry is basically identical to Pamiri Tajiks (referring to Uzbeks), and their Turkic ancestry comes from people who were on the ENA-ANE/EHG cline (I can model the Karasuk outlier you used as 70% Mongola and 30% ANE), so this is far from surprising.

Rob said...

@ Seindundzeit

Yes, i know who Uzbeks & Krygyz are bud.

"In addition, the whole use of the term "direct" is exceedingly vague/undefined in the spatio-temporal context that is being examined. This sort of talk is very philosophically underdetermined."

Id beg to differ, if one understands the historical - cultural context of samples we are dealing with. I do, so you can allay your theoretical concerns.

"But, what isn't vague is the fact that these Turkic Central Asians do have substantial steppe-related admixture, and not unusually high/or unexpected levels of CHG/Iran_Neolithic."

Yes, they have low levels of Iran Chalc. But their 'steppe' admixture looks like Mongolian/ Altain admixture. Calling it steppe is sort of wishful thinking.

But let's look at them another way - using EpiPalaeolithic source Pops.

"Ulchi" 34.2
"Kotias:KK1" 23.15
"MA1:MA1" 18.85
"Israel_Natufian:I0861" 13.8
"Dai" 5.7
"Villabruna:I9030" 4.3

"Kotias:KK1" 38.8
"MA1:MA1" 35.6
"Ust_Ishim" 15.2
"Israel_Natufian:I0861" 8.9
"Dai" 1.5
"Villabruna:I9030" 0

"Villabruna:I9030" 58.05
"Kotias:KK1" 35.6
"Israel_Natufian:I0861" 5.1
"Ulchi" 1.25
"MA1:MA1" 0

Uzbeks look very different to their Tajik neighbours. The problem is we don;t have a diverse enough sampling of Altai-Siberia to model these Ulchi who are required for Turkic CAs. But this is a fair representation:

"Dai" 44.25
"AfontovaGora3:I9050.damage" 32.65
"Israel_Natufian:I0861" 23.1
"Villabruna:I9030" 0

Very basically, Mesolithic Altains plus new 'East Asian" admixture, I;d bet arriving in Neolithic.

So again, my conclusions is proven; the "steppe' in CA Turkics is actually Mesolithic Afontova-stuff, ie not "steppe at all.
On the other hand the steppe in Indo-Aryans looks like Yamnaya stuff, but could be from much older Mesolithic populations in south-central Asia.

Seinundzeit said...

Here is my standard setup, but with that Karasuk outlier added:


23.25% Mongola
23.05% Karasuk_outlier:RISE497
14.7% Srubnaya_outlier
14.05% Barcin_Neolithic
9.9% Iran_Chalcolithic
8.85% Iran_Neolithic
2.65% Sintashta
2.55% MA1
1% ASI


The fit is extremely good, so I think this may involve some overfitting, but whatever.

Anyway, the Srubnaya_outlier percentages have not disappeared, and now we see some Sintashta and MA1.

Also, my standard setup doesn't really have the Pamiri Tajiks as resembling Lithuanians.

With it, Pamiri Tajiks are at 40% Srubnaya_outlier, while Lithuanians are at 30%. Also, Pamiri Tajiks are construed as 0% Villabruna versus 40% in Lithuanians, they are construed as having very substantial Iran_Chalcolithic/Iran_Neolithic versus very little to none in Lithuanians, and as having 5%-10% ASI + 5%-10% Mongola versus 0% among Lithuanians. It all makes sense.

rozenfag said...

Comment: The people of Pamir(Ishkashim, Khorog, etc) are very different from other Tajiks. Technically they are not even Tajiks - they speak languages, which are closer to Pashto, rather than to Tajik.

Seinundzeit said...


"Yes, i know who Uzbeks & Krygyz are bud."

That's wonderful.

"Id beg to differ, if one understands the historical - cultural context of samples we are dealing with. I do, so you can allay your theoretical concerns. "

Unfortunately, I think you don't quite understand my concerns.

It boils down to the basic epistemological framework in which these questions are framed, but I think it'll turn out to be too complex of a discussion.

"Yes, they have low levels of Iran Chalc. But their 'steppe' admixture looks like Mongolian/ Altain admixture. Calling it steppe is sort of wishful thinking."

So, you agree with me, which makes the whole discussion rather circular.

"Dai" 44.25
"AfontovaGora3:I9050.damage" 32.65
"Israel_Natufian:I0861" 23.1
"Villabruna:I9030" 0"

To have the Ulchi as 25% Natufian-related is highly fanciful, which is why I advise experimenting with different dimensions, and why I advise a more detailed examination of your reference population setup.

"So again, my conclusions is proven; the "steppe' in CA Turkics is actually Mesolithic Afontova-stuff, ie not "steppe at all.
On the other hand the steppe in Indo-Aryans looks like Yamnaya stuff, but could be from much older Mesolithic populations in south-central Asia."

I will have to disagree, but due to a matter of differences in degree (with respect to emphasis), rather than differences in kind.

The "steppe" in CA Turkics is actually Yamnaya-related via their local West Eurasian ancestry, and Mesolithic ANE-related via their Proto-Turkic ancestry.

That's the correct conclusion.


Yes, Pamiri Tajiks aren't actually Tajiks, as in sedentary Persian people. Their languages are much closer to Pashto.

rozenfag said...

Regarding Bronze Age Steppe admixture in CA Turkics: however small it is, it should exist, because it is the best way to explain prevalence of R1a among some Central Asian groups(Altayans, Kyrgyz, etc.), related to East European and South Asian R1a's.

But without aDNA samples from Central Asia itself, all of this is speculation. Waiting for them.

Seinundzeit said...


That's an important angle which I totally forgot to mention, all of these populations have noticeable R1a (with some showing extremely substantial percentages), not to mention the presence of some R1b.

Samuel Andrews said...


How could Lithuanians be 42% WHG if they have less than 10% U5b and >5% I2? WHG is being confused as Steppe. No way Lithuanians are only 30% Corded Ware/Srubnaya.

Seinundzeit said...


This is typical of direct PCA-based analyses, they always give Europeans much more WHG ancestry than what the formal methods show. Who knows which is more correct.

Although, much of that WHG disappears with Iberia_Chalcolithic in the mix, although Srubnaya_outlier stays at 30%.

Also, that 30% isn't in the context of them being tested with Srubnaya or Corded Ware. It's using the "Srubnaya_outlier", which is very different from Srubnaya/Corded Ware. It has far more ANE, and much less WHG and Anatolia Neolithic affinity.

It's like testing them with an ANE-admixed spin on Yamnaya, so 30% makes sense.

Rob said...

@ Sein

Yes, that Natufian figure seems off., it was generated using PCA data. It would be good to confirm using a DStat data, when Dave can generate a DStat column for Uzbeks later. But back to our original issues, yes the 'steppe' in Turkic has two probable sources - an eastern/ Mongolian one and a western. For the latter, my PCA-based run preferred something Caucasus -like (Kotias) for some reason instead of Samara. I'm not too confident that is the correct scenario.

Seinundzeit said...

@ Rob


All I've been saying is that people like Uzbeks have ANE ancestry from deep inside Asia (via their actual Turkic ancestry) and Steppe EMBA/MLBA ancestry via their local West Eurasian ancestry, with a skew towards the latter in the case of Uzbeks.

In addition, they don't display any excess of Iran_Neolithic/Iran_Chalcolithic admixture, which was the original point of contention, in my conversation with Sam.

So, we actually agree on this.

Grey said...

"I just realized that there's a huge mistake in this paper. They confused Afanasievo with Andronovo"

That's shame - i was enjoying my

"if pestis is native to the steppe and horses are native to the steppe maybe horses have plague resistance"


Janko Raven Johnson said...

Kristian Kristian send at Vilnius in September

Olympus Mons said...

But there are things (a lot) that have a strange “ring” to it… here are a few:
3.35 – if they brought Y. pestis , wouldn’t they be the least successful guys in the world as per all following arguments? – Women they “stole” would die, their children would die… unless they brought their women with them that were also immune to bear children that were also immune. Right?
5.20 - He says that they were milk drinkers. Do we have dna evidence that they could digest milk? And all of them?
7.20 – Does he says women came from Urnfield culture? 1300bC did he meant Unetice?
7.50 – If it were male groups marrying locals how come so many make the argument for them bringing U5a1? If they were marrying local Neolithic…so…?
8.10 – I didn’t like that to make the argument that it were mainly males … he has to go to ….Denmark?? pretty fringe. Or even Eulau.
10.35 – If at mature phases of CWC they were marrying women that had a “Neolithic” diet at childhood… then lets forget the U5a1 from the steppe, right?
13.25 - His argument is strange. So… Yamnaya did not have words for agriculture? So “Neolithic” women teach the language…. But all language ended up being PIE from the steppe taught by those women that were not original PIE speakers?!

I suppose all this conundrums are all normal if you are an Yamnayalist! 

Ric Hern said...

I wonder if the Okunev Culture pushed the Afanasevo back into Europe ? Could the Botai Culture be relatives of the Okunevo Culture ?

Ric Hern said...

Or maybe the Plague was due to the 5.9 Kiloyear event ? I read somewhere that the oldest plague victim died 3783 bC. ?

Ric Hern said...

Sorry I was thinking about the early diversion date and branching of Y. Pestis.But this date correlates with the 5.9 Kiloyear Event....

Seinundzeit said...


Truthfully, I'm deeply sorry.

When it comes to posting those results, I've taken far too much time, much longer than I promised.

So, here are some experimental attempts.

For comparison, I'll also post models which use only the Onge (but besides that, all of the other reference populations are identical to what I used in the experimental ASI models).

Austroasiatic Indian populations:


70.65% Onge
16.65% Iran_Neolithic
12.7% Ami



60.4% ASI
24.35% Ami
15.25% Iran_Neolithic



74.9% Onge
19.75% Iran_Neolithic
5.35% Ami



65.3% ASI
17.4% Iran_Neolithic
17.3% Ami



67.95% Onge
29.95% Iran_Neolithic
2.1% AG3



70.2% ASI
24.05% Iran_Neolithic
5.75% Ami



68.05% Onge
30.85% Iran_Neolithic
1.1% AG3



67.7% ASI
25.15% Iran_Neolithic
7.15% Ami


We see substantial improvement (with regard to the fits), after the inclusion of a South Indian-based ASI simulation. I think this is a good sign.

Also, I always included the Onge, for the ASI models. Yet, despite that, all South Asians always picked the ASI simulation (they always received 0% from the Onge). So, I think this is another point of interest.

In addition, the East Asian percentages make much more sense, with the ASI simulation. I mean, Austroasiatic languages are now recognized as being intrusive to South Asia; they came to this region from Southeast Asia. So, one would expect some of these people to have noticeable-to-moderate East Asian ancestry.

Concerning this very point, I think it might be important to remember that some Austroasiatic Indian tribes often include a minority of individuals who seem to show an evident East Asian-vibe (with regard to their facial features).

For example, I was watching a documentary on the Bonda, and a minority of them almost approximated Cambodians in terms of facial features, which was very surprising to me. I expected them to almost always look predominately Jarawa/Onge-like, but with larger builds than those folks. Not the case at all though, they seem very diverse, some almost look vaguely Andamanese, some almost look vaguely like West Eurasian-leaning low caste Indians, and some almost look vaguely like deeply pigmented Cambodians or Malay.

My point is, some of these people should have substantial East Asian admixture, which the ASI simulation demonstrates quite well.

Now, the same reference population setups, but with Chamar and Punjabi_Lahore:


54.15% Iran_Neolithic
36.35% Onge
9.5% AG3



52.25% ASI
42.15% Iran_Neolithic
5.6% AG3



54.6% Iran_Neolithic
31.5% Onge
13.9% AG3



45.35% Iran_Neolithic
38.9% ASI
15.75% Samara_Eneolithic


Again, substantial improvement over using only the Onge, and again the simulation is always preferred (despite the inclusion of the Onge).

Also, 0% East Asian admixture is found, another demonstration of the distinctiveness of the Austroasiatic peoples, with regard to that sort of ancestry. I should make mention of this too: South Indians also show 0% East Asian admixture, only score percentages for the ASI simulation. So, this is an airtight conclusion.

Finally, the Lahori Punjabis pick Samara_Eneolithic rather than AG3, and the inverse for the Chamar. This is exceedingly sensible, just as one would expect. This result doesn't happen in any other context, including when one only uses the Onge.

Later, I'll post some models of those South Asian populations which are predominately West Eurasian, and some models of South Central Asians. This time, I’ll only use my ASI ghost, as I hope I've sufficiently demonstrated the superiority of using the simulation rather than the Andamanese_Onge.

Again, I apologize for the wait.

Seinundzeit said...


Here are models for high caste Northern Indians and Sindhis. Next, I'll show models for South Central Asians and Balochistanis:

UP Brahmins

41.9% Iran_Neolithic
33% ASI
24.95% Samara_Eneolithic + 0.15% Sintashta



46.6% Iran_Neolithic
22.15% Samara_Eneolithic + 6.1% Sintashta
25.15% ASI



55.75% Iran_Neolithic
22.4% ASI
12.15% Sintashta + 9.7% Samara_Eneolithic


Seinundzeit said...

South Central Asian populations:


31.8% Iran_Neolithic
25.65% Samara_Eneolithic (0% Sintashta)
17.55% ASI
15.4% Iran_Chalcolithic
9.6% Mongola


Pakistani Pashtun, Khyber Pakhtunkhwa (a true Yusufzai)

43% Iran_Neolithic
15.65% Samara_Eneolithic + 12.75% Sintashta
16.95% ASI
11.65% Iran_Chalcolithic



44.8% Iran_Neolithic
23.6% Samara_Eneolithic + 11.9% Sintashta
14.4% ASI
5.3% Iran_Chalcolithic


Pakistani Pashtun, tribal areas (FATA)

31.7% Iran_Chalcolithic
26.55% Samara_Eneolithic + 3.9% Sintashta
14.65% ASI
1.2% Mongola


Afghan Pashtun, Ghazni (Ghilzai)

21.85% Samara_Eneolithic + 10% Sintashta
30.1% Iran_Chalcolithic
11.7% ASI
1.7% Mongola


Me (I have ancestry from a bunch of distinct Pashtun tribes, in both Afghanistan and Pakistan, with the addition of 12.5% Afghan Uzbek ancestry)

37.05% Iran_Neolithic
29.6% Samara_Eneolithic (0% Sintashta)
16.7% Iran_Chalcolithic
11.7% ASI
4.95% Mongola


Afghan Pashtuns (samples collected in the provinces of northern Afghanistan, probably a hodgepodge of Ghilzai and Durrani)

33.55% Iran_Chalcolithic
21.8% Samara_Eneolithic + 11.85% Sintashta
18.8% Iran_Neolithic
9.15% ASI
4.85% Mongola


50% Panjsheri Tajik + 50% Afghan Pashtun

37% Iran_Chalcolithic
32.5% Samara_Eneolithic + 1.5% Sintashta
16.65% Iran_Neolithic
8.45% ASI
3.9% Mongola


Afghan Pashtun, Khost (Karlani)

22.35% Samara_Eneolithic + 12.45% Sintashta
25.3% Iran_Neolithic
24.1% Iran_Chalcolithic
8.2% ASI
3.8% Barcin_Neolithic
3.8% Mongola


Afghan Pashtun, Kandahar (Durrani)

36.15% Iran_Chalcolithic
21.25% Samara_Eneolithic + 6.95% Sintashta
24.1% Iran_Neolithic
5.95% ASI
5.6% Mongola



38.25% Samara_Eneolithic + 3.75% Sintashta
37.05% Iran_Chalcolithic
7.65% ASI
7.4% Iran_Neolithic
5.9% Mongola



34.6% Samara_Eneolithic + 14.3% Sintashta
34.7% Iran_Chalcolithic
6.3% Iran_Neolithic
5.1% Mongola
5% ASI



40.05% Samara_Eneolithic + 2.15% Sintashta
44.45% Iran_Chalcolithic
7.85% Mongola
0.9% ASI



49.7% Iran_Chalcolithic
31.65% Samara_Eneolithic + 9.05% Sintashta
6.5% Mongola
3.2% Barcin_Neolithic


Seinundzeit said...

Finally, populations of Balochistan/southeastern Iran:


43.5% Iran_Neolithic
27.3% Iran_Chalcolithic
12% Sintashta + 7.7% Samara_Eneolithic
9.5% ASI



36.95% Iran_Neolithic
33.95% Iran_Chalcolithic
9.95% Sintashta + 9.5% Samara_Eneolithic
7.95% ASI
1.7% Mongola



44.15% Iran_Neolithic
29.5% Iran_Chalcolithic
13.05% Sintashta + 6.05% Samara_Eneolithic
7.25% ASI


Seinundzeit said...

General Observations:

Using the ASI simulation, most South Central Asians now show a combination of both Sintashta and Samara_Eneolithic scores, even ones which previously seemed allergic to Sintashta, like the Kalash (I'm an exception. Probably boils down to the different tribes that make up my ancestry, and my Uzbek admixture could make things confusing/complicated).

I think this means we need to be more cautious, when we see only some steppe references working successfully for South Central Asians. The oft-repeated fact that Yamnaya is preferred to Sintashta/Andronovo could really just be an artifact, due to the lack of a true ASI reference (the Onge/Jarawa are very inadequate for the task).

Of course, not fully, but it might play a partial role.

Also, one difference between the predominately West Eurasian Indians/Pakistanis and South Central Asians lies in Iran_Chalcolithic admixture.

East of the Indus River, this sort of ancestry just isn’t found, while to the west of the river it steadily increases on a westward and northward cline. For some Pashtuns, and all Tajiks, it becomes the primary ancient Iranian plateau element.

When it comes to ASI ancestry, the full HGDP "Pathan" set is probably somewhere between 15% and 20% ASI, the Kalash are around 15% ASI, Pashtuns like myself and most Afghan Pashtuns/Tajiks are more around 10% ASI, and the East Iranian peoples of Tajikistan range from almost 10%-ish ASI to pretty much 0% ASI.

With concern towards the latter, the Yaghnobi Tajiks seem to be the only South Central Asians with no ASI admixture; their ENA is wholly of the kind seen in Uzbeks, Uyghurs, and other heavily East Asian-shifted Central Asian peoples.

When it comes to the Balochistanis, they (like me and most of the Afghans) are more around 10% to 5% ASI. But unlike us Pashtuns, they have 0% East Asian admixture, only ASI.

That's pretty much the size of it, and I think ancient DNA from Mesolithic South Asia will corroborate this picture.

Seinundzeit said...


Lol, some corrections, I rushed through.

Pakistani Pashtun, tribal areas (FATA)

22% Iran_Neolithic

Afghan Pashtun, Ghazni (Ghilzai)

24.65% Iran_Neolithic


4.6% Iran_Neolithic