search this blog

Saturday, August 17, 2019

A surprising twist to the Shirenzigou nomads story

Remember those potentially Afanasievo-derived and Tocharian-related Shirenzigou nomads from the Ning et al. paper? Well, in my opinion, they're probably neither. The genotypes and other data for these Iron Age individuals from the eastern Tian Shan are available here.

Below are a few successful and not so successful qpAdm mixture models for them. Note that I tried to use a wide range of relevant "right pops", but also retain a lot of markers, specifically to be able to discriminate between different types of steppe and steppe-derived sources of gene flow (refer to the full output). Admittedly, the Shirenzigou nomads can be modeled with Afanasievo-related ancestry, but...

KAZ_Botai 0.161±0.023
KAZ_Wusun 0.490±0.023
NPL_Mebrak_2125BP 0.349±0.019

chisq 5.793
tail prob 0.926172
Full output

KAZ_Botai 0.143±0.022
NPL_Mebrak_2125BP 0.295±0.019
Saka_Tian_Shan 0.562±0.024

chisq 6.796
tail prob 0.870794
Full output

KAZ_Botai 0.185±0.023
NPL_Mebrak_2125BP 0.428±0.021
RUS_Sintashta_MLBA 0.270±0.026
TJK_Sarazm_En 0.117±0.027

chisq 11.351
tail prob 0.414345
Full output

KAZ_Botai 0.032±0.027
KAZ_Zevakinskiy_LBA 0.567±0.025
NPL_Mebrak_2125BP 0.401±0.019

chisq 15.157
tail prob 0.232961
Full output

NPL_Mebrak_2125BP 0.452±0.031
RUS_Afanasievo 0.435±0.025
RUS_Okunevo_BA 0.114±0.049

chisq 19.808
tail prob 0.0708003
Full output

NPL_Mebrak_2125BP 0.409±0.031
RUS_Okunevo_BA 0.173±0.050
Yamnaya_RUS_Caucasus 0.418±0.026

chisq 20.453
tail prob 0.0589872
Full output

NPL_Mebrak_2125BP 0.464±0.033
RUS_Okunevo_BA 0.104±0.053
Yamnaya_RUS_Samara 0.432±0.027

chisq 27.189
tail prob 0.0072566
Full output

Both the Wusun and Saka are generally accepted to have been the speakers of Indo-Iranian languages. So it's possible that the Shirenzigou nomads were Indo-Iranian speakers too, or at least derived from such peoples.

Surprisingly, NPL_Mebrak_2125BP was the key to obtaining the best statistical fits. This is a trio of samples, roughly contemporaneous with the Shirenzigou nomads, from a burial site high up in the Himalayas in what is now Nepal (see here).

To be honest, I'm not quite sure why the Himalayan ancients work so well in my models. Perhaps they're just a really good proxy for an Iron Age population from the northern edge of the Tibetan Plateau?

By the way, most of the Shirenzigou nomads made it into the latest Global25 datasheets (see here). They can be analyzed in a variety of ways described in this blog post: Getting the most out of the Global25. Below is a screen cap of me comparing the effectiveness of Afanasievo, Sintashta and Wusun samples as proxies for the steppe ancestry in the Shirenzigou nomads with an online tool freely available here. As expected, the algorithm picks Sintashta ahead of Afanasievo, and the Wusun ahead of both.

See also...

They mixed up Huns with Tocharians

Some myths die hard

The mystery of the Sintashta people


Davidski said...

I haven't really checked this out properly yet, but it seems like the ancient Iberian pops do a good job at discriminating between Steppe_EMBA and Steppe_MLBA.

Samuel Andrews said...

Is NPL_Mebrak_2125BP more related to Chinese than to Lokomotiv? Does it have Indian admixture? There is Chinese-related East Asian ancestry in central Asia today.

Samuel Andrews said...

So overall the Shirenzigou nomads were only about 25% Andronovo which means maybe 15-20% Yamnaya (because Tian Shan Iranians were about 50% Andronovo). David, you were right the published paper did a poor job modelling them. But, modelling ancient ancestry has become so complicated it isn't big deal.

Francesco Brighenti said...

FYI, I reproduce here an exchange on the Facebook Group “Harappan Archaeology” I had with Indian geneticist Niraj Rai two days ago (see at ):

NIRAJ RAI: Our DNA based research on R1a1 suggest its origin in South Asia. For this finding we have analyzed 10 thousand DNA samples across India and findings will be published soon.

ME: When will you disclose to all of us the genomic data you have collected from Rakhigarhi skeletons? The paper on ancient DNA from Rakhigarhi had been announced years ago, yet it has not been published to date. Due to political reasons?

NIRAJ RAI: I am extremely sorry for this delay. I would like to inform you that most awaited article on Rakhigarhi will be out by next month.

ME: This is good news: next month, for the publication of a paper I have been longing for for years! So will you challenge Reich over the issue of R1a being native to the Steppes (as he and his colleagues have been claiming in their recent books and papers)? Does your evidence suggest R1a originated in South Asia?

NIRAJ RAI: We are not discussing R1a issues with Rakhigarhi. R1a will be a separate paper on modern DNA. Furthermore, we are not challenging Prof. Reich since he has never mentioned that R1a is exclusively Central Asian in origin.

Matt said...

As these guys are pretty heterogenous, looking at them through PCA may be complement to qpAdm of them as a group:
(Above is reprocessed PCA).

Not a position to look at them in a too sophisticated way atm; the general cline looks to fit with one end fitting close to roughly contemporary Kazakh steppe and Tian Shan samples (Hun and Saka), though the most "Western" sample may be a tad more West_Siberian influenced.

Does seem like in some high reprocessed dimensions where you do get a Sherpa+Han+Naxi vs Nganassan+Austronesian dimension (with Sherpa and ancient Nepal/Tibet most extreme), basically distiguishing Chinese and related populations from a combination of North+SE Asian, that these samples outbend towards being intermediate the on a cline specifically towards the specifically Chinese/Tibetan related though. So that is consistent with qpAdm. I'm not sure that's as true for other Iron Age samples, though hard to be immediately sure.

qpAdm group fit looks sensible to me. Basically similar to contemporary Iron Age nomads and what common sense would have predicted from the paper (and if I may say, pretty much what I can remember saying they'd probably be like as soon as the abstract first got mentioned before the paper even got released!).

FrankN said...

Answering to a question raised in the previous thread:

Drago: „Which leaf-shaped points do you refer to exactly ? Anything to do with bullet core technology ?

No – bullet core technology spread from Central Siberia during the Epi-Paleolithic/ early Mesolithic, and is IMO connected to the ANE ancestry found in Sidelkino, to a lesser extent also in Iran_Hotu.

Leaf-shaped points, alternatively also labelled „laurel-shaped“, „willow-shaped“, „almond-shaped“, „fish-shaped“ or „lanceotic“, constitute a subgroup within bifacially-retouched pressure-flaked points/arrowheads. Their shape sets them aside from other points such as triangular, tanged, or barbed arrowheads. Some authors have questioned whether „leaf-shaped“ points represent arrowheads at all. E.g., it has been noted that the Ice-Man's dagger blade (of Remedello origin) would, for its length of only 5 cm, have been classified as leaf-shaped arrow-head, hadn't the ice also preserved the wooden dagger handle. As such, while the literature often speaks about leaf-shaped arrowheads, I prefer the more open term „point“ instead.

Intuitively, the shape doesn't look very special – it is pretty close to Clovis fluted points, otherwise also known from the European Paleolithic Solutrean (which has given rise to all sorts of speculations), some Epi-Gravettian sites in N. Italy, and earliest attested from Middle Paleolithic S. Africa. Leaf-shaped points dominated PPPN assemblages, but became rare afterwards. Apel 2012 (Link 1) explains this with the increased demand for sickles consuming most of the large blades previously used for arrowhead/ spearpoint production, so Neolithic farmers shifted to transversal arrowheads ( „Querschneider“ in German terminology) produced from small blades and flakes. Transversal arrowheads are well attested from Caucasia (Shulaveri-Shomu/ Sioni) and the Pontic-Caspian Steppe (Kremennaya II) to Central Europe (EEF) and beyond. These comparatively light „Querschneider“ could also have had ballistic advantages, while still being pretty effective, as demonstrated in Eulau/Saale, where Corded Ware inhabitants were murdered by Schönfeld Culture attackers by a/o such transversal arrowheads.

Some of the Apel 2012 chronology is questionable. E.g., he still dates Khvalynsk to 5,000 BC instead of the now prevalent reservoir-effect corrected dating around 4,700 BC (Lower Volga) to 4,500 BC (Middle Volga). Similarly, he suggests presence of bifacially-retouched points in CT already from 5,700 BC on, while most researchers relate them to Trypillya B1 (ca. 4,500-4,000 BC, see link 2). Conversely, his 2,400 BC date for English leaf-shaped points is a millenium too late (link 3). While there is quite a lot to amend as concerns the Danubian spread of leaf-shaped points (a/o Surovo, see Anthony 2007) to ultimately Remedello daggers, Apel does a quite good job in tracing their spread from Khvalynsk up the Volga to Combed Ceramic Finland (ca. 3,900 BC) and ultimately Younger Pitted Ware (ca. 2,500 BC), and another, more westerly trail marked by the Skelya [SS II]-Yamnaya-Corded Ware sequence.
He furthermore notes that contemporary to Khvalynsk/Skelya, similar points also appear in the Atbasar Culture in N. Kazakhstan. The Atbasar Culture is considered the first Neolithic, i.e. sheep/goat pastoralist culture in the area. By ca. 3,700 BC, it was replaced by Botai.


FrankN said...

The earliest European (e-)Neolithic evidence of leaf-shaped points appears to come from the (pre-)Caspian Culture – see Fig. 4 upper row in Vybornov 2016. Comparison with Fig. 2 ibd. illustrates the massive break with earlier microlithic traditions. Since similar microlithic assemblages are also known from the Lower Don, local (re-)invention of leaf-shaped points seems unlikely.

An idea of S. Caucasian lithics is provided by Chataigner e.a. 2014 and makes obvious that they aren't related (pre-)Caspian points and heavy duty tools.
Even more intriguing in this respect are lithic implements from the Armenian EBA (Kura-Araxes) and MBA (early Trialetian). They contain just a single bifacially-retouched point, apparently an import, and in addition not leaf-shaped but tanged and barbed. Obviously, leaf-shaped points were alien to Chalcolithic E. Transcaucasia and remained so during the EMBA.

In Central Asia, we a/o find the Kelteminar culture. Kelteminar points indeed were bifacially-retouched, but otherwise provide little analogy to Pontic-Caspian assemblages, since they were assymetrical. Consult, for a better idea about Kelteminar points, Plate 3 in the link below:
As such, Kelteminar doesn't qualify as origin of (pre-)Caspian arrowheads. However, the Neolithic layer (pre- 5,000 BC) at the Kelteminar site of Ajakagetma (Central Usbekistan, some 100 km N of Bukhara) is characterised by heavy duty tools not unsimilar to (pre-)Caspian ones (see Fig. 4 in the lik below). Ajakagetma has also supplied one example of a leaf-shaped point – apparently intrusive, but nevertheless indicating an origin from a culture that was in contact with Kelteminar.

Further south, leaf-shaped points are quite common. They have been found from Tepe Hissar (NE Iran) in the West to Mundigak (55 km N of Kandahar) in the East. Other notable findspots include Anau, Geoksyur, and Saraszm. However, all these finds date at best to the 4th, more often to the 3rd mBC, i.e. they postdate (pre-)Caspian and also Atbasar Culture assemblages. Acc. to Shakun 2003 (link below): „Due to the lack of arrowheads in the Neolithic sites of Turkmenia, it is difficult to argue for a local development of these artifacts. The Neolithic cultures of hunters and pastoralists in adjacent regions could be considered the most plausible source for this tradition. Near the Caspian Sea the multicomponent site of Djebel produced leaf-shaped arrowheads.

So, Djebel once again. It shows up in the Apel 2012 map with a date of 4,900 BC, which would make it roughly contemporary to the (pre-)Caspian culture. However, as I have said before, Djebel hasn't been AMS-dated, and some of the dates reported by Apel are problematic. As such, Djebel may represent the origin of (pre-)Caspian lithics, but may equally just have been another place that the expansion of (pre-)Caspian people reached during the early 5th mBC. In the second case, we may hypothesise about an origin on the Ustjurt Plateau, the Mangyshlak peninsula, maybe even in areas now covered by the Caspian Sea.

FrankN said...

Here the link to Shakun 2003 that was missing from my previous post:

Simon_W said...

@ F. Brighenti

Thanks for sharing. So 10'000 samples of modern Indian yDNA are supposed to prove an origin of R1a1 in South Asia. An interesting Approach. :D

Simon_W said...

That logic reminds me of Gioiello pointing out rare basal branches of R1a1 being found in Britain, supposedly proving an origin of the whole clade in NW Europe.

JuanRivera said...

R1a1 is ANE or EHG in origin (as R1a as a whole). So is R1b (there's basal R1b1 in Villabruna, but most diversity in ancient and modern R1b is located in Eastern Europe and Fennoscandia, in 10%+ EHG-admixed populations [and EHG itself, of course]). Given the extremely old age of R1a and R1b, an origin in an AG3-like ANE population followed by diversification in EHG is conceivable. Going further back in time, MA1 is the basalmost kind of R, while AG2, Anzick and Kennewick are the oldest examples of Q1a (as well as Q as a whole). Yana is P1.

Davidski said...

@Francesco Brighenti

Only ancient DNA can resolve where R1a originated.

The oldest R1a samples are from Eastern Europe, which basically means that R1a, including Indian and Central Asian R1a, is from Eastern Europe.

Map of pre-Corded Ware culture (>2900 BCE) instances of Y-haplogroup R1a

You can't challenge this with modern Indian R1a samples, because you don't know where the ancestral lineages to those modern lineages were a couple of thousand years ago.

You actually sound crazy. You need to find a good psychiatrist instead of worrying about the origin of R1a.

Francesco Brighenti said...


You misinterpret the sense of my communication! I am totally with you, and I have been following your blog with great interest since the publication of Narasimhan et al.’s 2018 preprint in order to get more clues about the scientific support recent aDNA studies can provide to the solution of the AMT vs. OIT controversy (which I have studied for the last 20 years or so).

I look forward to reading your comments soon after Niraj Rai and his colleagues have published the long-awaited paper on Rakhigarhi DNA (next month, as Rai publicly declared on Facebook).

Did you know that Rai also declared on Twitter (8 Jan 2019) they have genetic evidence for horses in Mature Harappan period, and that the findings will be published soon?

The OIT must have deeply pervaded the world Indian geneticists live in!

Davidski said...

@Francesco Brighenti

Ah, apologies for that last remark.

Davidski said...


By the way, I'm optimistic that at least a good portion of western scholars won't bend over and let their Indian colleagues roll over them in regards to the topic of R1a.

Some of them already do know the facts and are taking the honest position on this and related issues.

Here's an interesting comment article that came out recently. I think it's the first formal publication that actually acknowledges the fact that ancient DNA is at least pointing to Eastern Europe as the place of origin for R1a.

On Methodological issues in the Indo-European debate

AWood said...


Based on the evidence, R1b were foragers on the ANE -> EHG cline, just like R1a were. The reason why R1b Villabruna looks WHG is because the eastern hunter gatherers had already entered the Balkans where they encountered an indigenous group that presumably migrated from Turkey -> Europe. (YDNA I, and maybe C) It's that mixture of hunter gatherers that created WHG after the ice receded. Much later, M269 forms somewhere in Ukraine or southern Russia and we have a EHG genome. It is purely happenstance that the Villabruna reference sample is R1b, there were others in the vicinity who were I2, C1, and maybe other rare branches.

Drago said...

@ AWood

'' It is purely happenstance that the Villabruna reference sample is R1b''

R1b-L754 appears in Epigravettian Italy precisley because it is supposed to

Davidski said...

@Samuel Andrews

But, modelling ancient ancestry has become so complicated it isn't big deal.

This is a very big deal.

The authors of the paper, who included Johannes Krause by the way, failed to even test whether potentially key, and also geographically and chronologically obvious, samples such as the Wusun were relevant to the theory that they were testing.

You really have to wonder what's going on in these sorts of cases.

Open Genomes said...

The Shirenzigou nomads on the Global25 Ward's distance-squared clustering tree

This is a diverse group:
• SRZ_M820 clusters with Tien Shan Huns
• SRZ_X3 clusters with a Mongolian clade (DA27 appears to be a Mongol as well)
• SRZ_M012 clusters with Turkic Karakalpaks and Kazakhs and a Nogai
• SRZ_M010, SRZ_F004, and SRZ_M819 cluster with an Iron Age Scythian, a Medieval Karluk, Kimak, and Karakhanids, and Xiong Nu from Mongolia

So we have affinities with Sakas / Tien Shan Huns, Mongolic peoples, Turkic peoples, and Xiong Nu.

JuanRivera said...

I think WHG originated in Europe, given that a Villabruna-like component is seen as early as Vestonice and Paglicci. However, Villabruna could have got its R1b from a proto-EHG/proto-Iron_Gates_HG (Iron_Gates_HG is EHG-admixed) population traveling up the Danube and then south of the Alps.

Open Genomes said...

Has anyone considered that at least some of these Shirenzigou nomads were Xianbei, Di, or Qiang? The Xianbei were possibly a Mongolic people, and the Qiang were Tibeto-Burman speakers. The Di were either Proto-Tibetan or Proto-Turkic.

The Donghu Confederation, attested from the 7th century BCE consisted of "Five Barbarian" tribes of diverse origins. This could very well account for the genetic diversity we see among the Shirenzigou nomads, in spite of their common culture, and also their seeming Tibetan affinities.

The Donghu confederation of the "Five Barbarians"

which consisted of the:
Jie (likely Yeniseian speakers)

Chad said...

They should be Yuezhi, as I said before. They were pushed West by the Wusun, who would later be pushed West by the southern Xiong-nu. All should be Han and closely related admixed from alliances. Xianbei will be almost pure ENA, like the main Xiong-nu. Xiongnu are significantly more related to the Han than Shamanka. They came from the Chinese interior and may have minor Shamanka admixture.

Drago said...

Who were the Yuezhi ? A few vague early references ; and a whole lot of imaginative historiography
If these guys are really PH155 & Q1a; it’ll be consistent with where most of them sit genetically - native Inner Asians

Ric Hern said...

@ AWood

Yes it will be interesting to see how Villabruna picked up some CHG-relatedness after the Western and Eastern R1b split...Were the CHG-related population that R1b migrated through small or did Villabruna Ancestors breed the CHG out by means of many years of interaction with WHG, Pre-R1b related populations on their way to Villabruna...?

Did the Ancestors of Eneolithic Caucasus Piedmont area stay behind somewhere between the Urals and Caucasus while their Villabruna relatives migrated further Westwards ? Or did their Ancestors backmigrate towards the East at some time before the Early Mesolithic ?

If the CHG-related population were small at the time when Villabruna R1b Ancestors migrated through them then the bulk of CHG-related people only arrived later in the Pontic Caspian, or Villabruna Ancestors didn't migrate through the Core area of the CHG-related population...

I kinda doubt that the CHG-relatedness of Villabruna is something that survived the formation of Haplogroup K, P1 and R all the way to R1b....especially looking at where Ust Ishim, Yana and Mal'ta were found...thousands of kilometres from I and J rich populations and at least separated by 20 000 years...

Matt said...

OK, a few more visualizations of G25 data for the Shirenzigou by dropping them onto the scaffold of pops I used a few weeks ago to look at Saka and Hun:

It looks as if M820 is a fairly close match to Huns or Saka for Tian Shan, but not the most ANE shifted samples, and then the rest are more or less in a cline from between the bulk of Hun Tian Shan samples and SRZ_X3, the most East Asia shifted sample. Compared to DA127, the Tian Shan Hun "East Asian" outlier, SRZ XE is a bit more south shifted and tentatively shifted toward ancient Nepal/Mongolia (albeit SRZ_X3 is still closer to DA127 than he is to Han/ancient Nepal).

For y ref according to anthrogenica poster Kolgeh: SRZ_X3: Q1a2a2b1 (most East Asia shifted sample), SRZ_M012: R1b2. (The other designations may be in paper, or online elsewhere).

Slumbery said...

@Ric Hern

I won't argue for it, because it is just a casual speculation about possibilities, but it is possible that R1b went on a completely different way. We know that "CHG-Iran" had an ancient ANE ancestry, it can be imagined as a two way mixture of Dzudzuana + ANE (give or take). It is possible that R1b was a minor lineage of the ANE component in CHG-Iran that actually arrived in the South (as Iran had more ANE than CHG) and went westward via Anatolia, then went through a founder effect in WHG.
Of course this would have happened in UP time, so it won't make R1b Middle Eastern in Mesolithic context.
We can see on the example of R1a-M417 that sometimes small lineages can go boom.

Andrzejewski said...

@Slumbery @Ric so, Indo-Europeans mixing with farmers (EEF) is adding to the Dzudzuana-like component in Europe because ANF were like 50% Dzudzuana 50% UHG. But if UHG is similar to WHG it explains why Ötzi’s maternal side is mistaken as a Serbian forager, as @Samuel Andrews claimed in the previous blog entry. So Iran_N - CHG has a Farmer-like admixture and that’s perhaps why Haak 2015 wrote that the two populations split before LGM 25,000 years ago.

As for Middle Eastern populations, Iran_N has Dzudzuana and Natufians have Dzudzuana, plus there was a strong Anatolian_N incursion between Natufians to Levant_N. Maybe that explains the relatively high (30%) of ANF in Roman times ME populations?

Leron said...

So if the Uygurs are not partial descendants of the Tocharians. And these Shirenzigou nomads were not Tocharian ancestors either. Where did this IE population came from?

Maybe the mystery lies among the Xiongnu. Depending on how far west they originally extended, could they have picked up a far off IE speaking group that became heavily mixed, making their genetic ancestry very opaque to us. While the later coming Iranians erased whatever trail we could detect.

Davidski said...


Yep, in regards to the origin of the Tocharians, ancient DNA has further confused the issue, rather than clearing it up.

Considering the ancient DNA already available from the Tarim Basin and surrounds, it seems unlikely that an Afanasievo-derived population migrated to the region and became the Tocharians.

The Huns and Sakas from the Tian Shan, the Wusun, and all other steppe-related groups in the area are derived from the Andronovo population, local foragers and farmers, and eastern Asians.

So Tocharians are probably derived from one of these groups, but I doubt that they were part of the Xiongnu expansions. They were either one of the groups taking part in the early Andronovo migrations, or they moved into the Tarim Basin from somewhere in the west later, perhaps during the early formation of the Silk Road?

rozenfag said...

@Davidski - by " ancient DNA already available from the Tarim Basin " you mean Xiaohe mummies or there other studies? It would be nice if Xiaohe mummies were tested using modern techniques.

Davidski said...

Yes, I mean the Y-DNA and mtDNA from the different levels of the Bronze Age Tarim Basin cemeteries.

It's not much to go on, but that's all we have. And it seems to fit with the newer ancient DNA from Bronze and Iron Age sites from around the Tarim Basin.

Andrzejewski said...

It’s interesting that the Jie are described as having Europoid features in Chinese sources

Andrzejewski said...

I can understand the origins of Yenisseyan speakers: some sources claim that Dene-Yenisseyans are basically Beringians who migrated in both directions, notwithstanding others who think that they are direct descendants of MA1 populations. Note that the formation of the ancestors of American Indians can’t support the Beringian theory.

Davidski said...


You're way off topic.

There are papers from the last year or so that you can read on the peopling of the Americas. You can find them here...

Davidski said...


You're focusing too much on archeology and not enough on the genetics, especially the most obvious aspects of it, such as...

- there is no evidence that Piedmont_En was a new population and that any of its recent ancestors came to the Caspian steppe from somewhere in the east

- eastern cultural influences on the Caspian steppe are easy to explain without the need to posit that the Piedmont_En population came from the east, considering that already during the Maykop period parts of it appear to be swarming with highly mobile people with up to 50% Western Siberian-related ancestry

- the difference in the ratios of eastern ancestry between Piedmont_En and Steppe Maykop is huge, and Steppe Maykop is probably mixed, which means that a large proportion of its recent ancestors were even more eastern; so both Piedmont_En and Steppe Maykop can't be from east of the Caspian, unless their ancestors came from very different regions, which is difficult to accept as anything but special pleading.

Ric Hern said...


At this stage it doesn't look to me as if ANE passed through Central Asia into Iran during the Upper Palaeolithic because Central Asia was even more Arid than today. So if there was a North to South migration during this time it could most likely only have been via the Caucasus which is also doubtful but plausible...

I personally can not see how a Big Game Hunting society (High Protein Diet) simply preferred to migrate into a desert and become Gerbil Hunters. I think the more familiar environment would have been along plus, minus the same latitude which stretched all the way, deep into Europe...

JuanRivera said...

Iran has more ANE than CHG, with Gonur and other SC Asian cultures having even more ANE (of the same MA1-like kind as the former; along with West_Siberia_N). One can also keep in mind that people could have travelled either via the Caspian Sea coast or through the Altai, Tian Shan and Pamir mountains. Plus, other people have radically changed their substenance modes, as seen in Indigenous Australians (who changed from the tropical rainforests in the north to the Outback, which is a desert, in a very little time after entering Australia).

Ric Hern said...

@ JuanRivera

Desert is a broad term...Ranging from Kalahari like desert which is mostly Dry Woodland, like large parts of the Outback compared to the Namib Desert and Sahara with large areas covered in Sanddunes devoid of vegetation. Dryer Mountain areas in Central Asia, mostly devoid of vegetation could have been the state of Central Asia during the Upper Paleolithic because it was dryer than what we see today...

However Australia seems to have been more Moist in some areas looking at Lake Mungo which is desert today where Ancient Human remains were found.

So directly comparing South Siberian Hunter Gatherers to what happened in Australia doesn't make sense.

Why did the Ancestors of Amerindians follow the herds into America ? Why wouldn't their relatives not have followed the herd migrations towards the West ?

zardos said...

@Ric: The most likely explanation is that they were blocked by stronger, more competitive people.
For the same reasons there was little back migration and the Proto-Americans largely disappeared from Eurasia.
The ANE ancestry in West Eurasians split earlier and took a different path.

I think we just don’t know enough about all the populations involved in the formation of the PC steppe populations.

@David: Did you consider the possibility of a majority local, but small more recent South Eastern influences?
More like the Western Neolithic influence, largely hidden by the predominantly local EHG-CHG ancestry?
Can you exclude that too?

Davidski said...


You mean minor admixture into Piedmont_En from the southeast? If so, no I can't exclude that possibility, but based on what I've seen in my qpGraph and qpAdm runs, the HotuIIIb sample isn't exactly a prime candidate as a proxy for such gene flow.

I can design qpGraph trees in which Piedmont_En takes minor admix from a Sarazm_En node, but I haven't really been able to reproduce such results with qpAdm, probably because of the excess Siberian-like ancestry in Sarazm_En.

In fact, it seems to me that to model the ancestry of Piedmont_En you need EHG, CHG plus minor AfontovaGora3-related gene flow, and I doubt that this is a smoking gun for recent gene flow from the east. Rather, it suggests to me that Piedmont_En might be some local remnant of a once much more widespread ANE-rich population that contributed gene flow to Sarazm_En.

Davidski said...

By the way, the interesting news is that the Himalayan-related (Tibetan?) ancestry in the Shirenzigou nomads can also be seen in their mtDNA.

Their mtDNA haplogroups such as A17, D4j1b and G3b might easily be from Tibet or even Nepal.

music lover said...

Run qpAdm with Afanasievo as a source and Botai on the right. Now repeat the process with Botai as a source and Afanasievo on the right. In theory if the Xinjiang samples share additional drift with Afanasievo this should be revealing.

Davidski said...

@music lover

Adding Afanasievo and then Botai to the right pops doesn't really change anything. But yes, worth a try as an experiment...

KAZ_Botai 0.177±0.022
KAZ_Wusun 0.492±0.023
NPL_Mebrak_2125BP 0.331±0.017
chisq 10.466
tail prob 0.655467
Full output

NPL_Mebrak_2125BP 0.442±0.026
RUS_Afanasievo 0.437±0.022
RUS_Okunevo_BA 0.121±0.040
chisq 22.379
tail prob 0.0497605
Full output

Also, you should keep in mind that the Shirenzigou R1b isn't Z2103+ or even M269+, so there's no direct paternal link to Afanasievo, but rather to some other R1b-rich population, probably more similar to Botai.

Gabriel said...

If not Indo-Iranian, what other language or ethnicity could have they spoken or been?

Davidski said...


If not Indo-Iranian, what other language or ethnicity could have they spoken or been?

How about Tibetan?


One thing that I didn't do in my post, that I should've done, was to demonstrate clearly where the western steppe ancestry in the Shirenzigou nomads was ultimately from. So I added this model to my post.

KAZ_Botai 0.185±0.023
NPL_Mebrak_2125BP 0.428±0.021
RUS_Sintashta_MLBA 0.270±0.026
TJK_Sarazm_En 0.117±0.027
chisq 11.351
tail prob 0.414345
Full output

For comparison, here's the same model with Afanasievo instead of Sintashta.

KAZ_Botai 0.066±0.033
NPL_Mebrak_2125BP 0.474±0.024
RUS_Afanasievo 0.448±0.048
TJK_Sarazm_En 0.012±0.039
chisq 19.617
tail prob 0.0508796
Full output

JuanRivera said...

So, it shows that Afanasievo didn't contribute to populations of the Tarim basin. It only contributed to LN/BA Baikal, BA Mongolian and maybe Hezhen populations.

Bob Floy said...

So Tocharians probably were Sintashta/Andronovo derived after all.

Davidski said...

It's impossible to say for now where the ancestors of the Tocharians came from and when they moved into the Tarim Basin. Tocharian languages are attested there very late in the game.

But considering that all of the direct western steppe ancestry in the Bronze Age Tarim Basin and Iron Age/Medieval Tian Shan looks decidedly Sintashta-related, then I think there are two options which can probably be corroborated or contradicted with ancient DNA from sites where Tocharian languages are attested...

- Tocharian was first spoken somewhere on the steppe or the forest steppe by Sintashta-related populations, and spread east alongside early Indo-Iranian languages within the Andronovo horizon

- Tocharian first moved from the steppe into the Near East, say, via the Caucasus along with Catacomb-related populations, and then eventually into the Tarim Basin along the Silk Road.

Bob Floy said...

"It's impossible to say for now where the ancestors of the Tocharians came from and when they moved into the Tarim Basin. Tocharian languages are attested there very late in the game."

Of course.

"Tocharian was first spoken somewhere on the steppe or the forest steppe by Sintashta-related populations, and spread east alongside early Indo-Iranian languages within the Andronovo horizon"

That's basically what I'm thinking.

Matt said...

It's imaginable that Tocharians could come from just about any source.

Even if they did ultimately come from an Afanasievo source, extensive admixture with Sintashta related populations would most likely have happened by the Iron Age or at least Medieval period, and since those populations seem numerous, it would likely dominate their genome.

Not that these samples are particularly likely to represent Tocharian speakers, either.

The attestation is probably too late too pin them down to a homeland by trying to find a genetic correlation, due to this factor. Attestation is over 3000 years later in a region which is thinly populated, highly waypoint between East and West. Though in any case trying to evidence that a particular language expanded from a particular place at a particular time purely through genetic arguments (patrilineal markers / autosome) is in any case not sound, linguistically speaking.

Drago said...

Genomics alone indeed leads to problematic models - even if (& because of) it was thought to fit a “correct” model.

But can Andronovo explain IE expansion from Europe to Asia ? Within a set model; it can. And if that model is correct; can it link to other IE languages; which spread through elsewhere in Europe (even if not directly) - yes it can . And can this be demonstrated via collateral lines of evidence ? Probably.

Drago said...

Also; when 2 different groups encounter , without time machines; we can never definitely know how language shift occurred; but the most probable direction of language language shift can be discerned. But that requires a comprehensive cognisance of how it occurred from a micro regional to broad scale perspective .

Matt said...

I suppose you could propose a model to explain Asian IE languages via Andronovo only, depending on what is being defined as Asian, and how seriously or loosely methods for phylogenetic structure and dating within IE are taken.

I don't think there's anything particularly arguing for it though, and it would be an argument that breaks down the equation of Andronovo=proto-Indo-Iranian for the argument that the horizon contained multiple unrelated IE branches without much sign of mutual influence (and probably you lose ability to make those arguments that Indo-Iranian specifically of IE shares features which correlate with archaeology in Andronovo+Sintashta).

Slumbery said...

@Ric Herm
"At this stage it doesn't look to me as if ANE passed through Central Asia into Iran during the Upper Palaeolithic because Central Asia was even more Arid than today. So if there was a North to South migration during this time it could most likely only have been via the Caucasus which is also doubtful but plausible..."

This article says the last time the precipitation in Central Asia was clearly lower than today for an elongated time is the LGM itself and right after it it became at least as rainy as today:
Combined with the much higher glacier coverage in the mountains bordering CA from the South (means more melt-water than today after the LGM) this suggest that South-Central Asia had at least as good a vegetation let's say 18kya ago as today or better.

Also in my mind a migration of ANE to Northern Iran is not really a North -> South migration, more like an East -> West one or at least a NE -> SW one. Not via the central plains of CA, but via the foothills of the mountain range.

Nevertheless, I am really not sure this happened this way. The idea that CHG-Iran (and their similarity) is a result of similar mixing between a local basal rich population and incoming ANE is actually from Davidski, but the statistical fits in nMonte are super bad. Possibly because every non-ANE reference we have in the G25 database is totally wrong (waiting for Dzudzuana...), but (among others) a possibility that this did not really happen this way or the incoming population was very different from MA1 and AG3.

epoch said...

I read somewhere in responses that Iranian ancestry is more related to MA1 and EHG more to AG3. As MA1 is pre-LGM and AG3 is post-LGM, maybe this indicates that the ANE in Iranian is also pre-LGM, and the admixture event resulting in EHG is post-LGM.

Ric Hern said...

@ Slumbery

That article seems to have concentrated mostly just on the two caves. Several things they say is rather confusing and does not point towards the wider area going through the same changes as the mentioned caves...

Other studies I read, investigating vegetation etc. did not come to the same conclusion as the study you mentioned...So I think until there is clear consensus by all, the Dry Paleolithic Central Asia thing works for me...

Ric Hern said...

@ Slumbery

Wild desert Asses, Camels, Fat Tailed Sheep, Gerbils etc. and all other desert like animals should not be forgotten...they certainly took time to evolve...

Slumbery said...

@Ric Hern

The article actually gives a good explanation why they can use the results of the two caves to infer the climate changes of the region. I agree though that is not very clearly written. It took me reading to extract the information I was looking for. Still, it is not "just two caves".

Interesting you mention those other studies, because in other articles archaeologist kept complaining how sparse and insufficient the vegetation data in the region (from the Paleolithic).

As for camels et al., those would a good counter against an argument I have not made. I said that based on the results of the cited article and the probable effects of the mountain glaciation the region was at least as wet and habitable after the LGM as it today. Look how wet it is today. I never claimed that Central Asia was a lush rain-forest in the last five million years.

Ric Hern said...

@ Slumbery

I have no problem with what you prefer to extract from what you have read. I'm just explaining my point of view.

My personal view is that domesticated animals made it much easier for people to survive and trive in those areas than Hunter Gatherers. If I remember correctly there is a huge gap regarding Upper Paleolithic Campsites or Settlements between plus/minus Tadjikstan and the East Coast of the Caspian, even as far South as Central parts of Iran...? So yes there were some small Paleolithic Hunter Gatherer groups mostly from the Tadjikstan towards Mongolia but that huge gap towards the Caspian Sea makes me doubt a migration via a Southern Route towards Northwestern Iran. If you can throw some light on any, I'll be grateful.

JuanRivera said...

Has anyone noticed that Yana RHS prefers K14 over Sunghir?

vahaduo said...

@ all

Vahaduo: Global 25 Views

List of currently available PCAs:

- North Europe,
- Europe 1 - intra-European PC 1 vs. PC 2,
- Europe 2 - intra-European PC 1 vs. PC 3,
- South Asia.

I will add West Eurasia probably later this week.

- use only Global 25 scaled coordinates,
- you can project more than one sample, but you need to add them one by one,
- some errors in the pasted data are corrected without notice,
- some ancient samples have old coordinates and they won't match with projected ones when up to date Global 25 spreadsheet will be the source,
- ideas how to further divide South Asia PCA or West Eurasia PCA into clusters are welcomed,
- if you find any bugs - let me know.

Drago said...

@ Matt

''I don't think there's anything particularly arguing for it though, and it would be an argument that breaks down the equation of Andronovo=proto-Indo-Iranian for the argument that the horizon contained multiple unrelated IE branches without much sign of mutual influence''

Actually; it’s the most solid link evident at present
According to Blazek; Tocharian is most similar to Balto-Slavic

Bob Floy said...

Yeah, but is it really? Every opinion about what language Tocharian is closest to seems to be related to the researcher's own ethnicity.

Drago said...

@ Bob
Haha true
I suspect Balto-Slavic might have initially been overlooked because it’s Satem features & RUKI which it also shares with Indo-Aryan. Hence Tocharian was seen as a out of place; (but it really might not be an issue given that languages are dynamic; not static trees; and there was continuous contact and back migrations between Scythian and Eastern Europe; to the exclusion of Tarim basin). Thus; along with the “tartan” motifs; Adams & mallory suggested they had to be Celts ;)

At the end of the day; we have to account for IE languages expanding east & southeast somehow.
Afensievo doesnt' seem to be the link; as noted by myself & also Outram
So what links these disparate groups from Central basin to South-central Asia to eastern Europe ?
Well; there is the Andronovo horizon; which despite claims to the contrary- may explain Tocharian; I-A & Balto-Slavic
; and if there is a later migration to Tarim basin; that’s okay too

What I’m saying is that very early balto-Slavic night have been similar to pre-east Asian admixed Tocharian

Leron said...

If the connection is to Andronovo, then can we find some Tocharian ancestral link to the Tagar culture?

Bob Floy said...


I'm currently leaning toward "Andronovo" being a larger and more complex phenomenon(linguistically) than most of us have thought up to this point, and I continue to think that those Tarim basin mummies are related to the Tocharians, although this dosen't seem like a popular idea anymore.
As for the centum/satem issue, I'm not a professional linguist, but it's hard to imagine that the universal classification of Tocharian as centum has no meaning. You've expressed in the past that you think it's more of a fluid thing than a hard-and-fast, either/or issue. Is that classification based on only a handful of words?

Davidski said...


Awesome work!

Davidski said...


I don't think there's anything particularly arguing for it though, and it would be an argument that breaks down the equation of Andronovo=proto-Indo-Iranian for the argument that the horizon contained multiple unrelated IE branches without much sign of mutual influence (and probably you lose ability to make those arguments that Indo-Iranian specifically of IE shares features which correlate with archaeology in Andronovo+Sintashta).

I don't understand what you're trying to say here, considering that we know nothing about the languages in the Andronovo horizon.

All we know is that it's the most plausible and about the only realistic way to link the Indo-European speakers of Europe and South Asia, and that it was vast and populated by many closely related but potentially linguistically very distant populations.

Bob Floy said...

It could just be that the Andronovo phenomenon represents early, barely differentiated IE speakers, whose languages evolved into indo-Aryan, Tocharian, etc., where they ended up settling. Iranian languages proper, closely related to IA, would have developed back on the steppe with Srubnaya, an Andronovo related group who stayed behind.

Or something like that.

JuanRivera said...

Sintashta is a mixture of Corded Ware and Yamnaya/Catacomb. So, maybe Indo-Iranian is from the Corded Ware side and Tocharian from the Yamnaya/Catacomb side. One can keep in mind that Sintashta_MLBA_o2 has more Yamnaya/Catacomb (in addition to Botai-related ancestry) than the main Sintashta cluster.

JuanRivera said...

However, Sintashta_MLBA_o2 hasn't been proven to contribute to any population.

JuanRivera said...

I'm seeing that Sintashta prefers Afanasievo over Yamnaya and Catacomb. Can anyone verify on qpAdm?

Bob Floy said...

"Sintashta prefers Afanasievo over Yamnaya and Catacomb"

Interesting if true.

Matt said...

@Davidski: I don't understand what you're trying to say here, considering that we know nothing about the languages in the Andronovo horizon.

It's not incomprehensibly complicated; it is argued that there is an linguistic affiliation of Indo-Iranian with the Andronovo Horizon ( place names, parallels between reconstructed rituals and archaeology, location.

So I am simply saying that if you argue that a whole bunch of IE languages - or even for some reason just two phylogenetically unrelated varieties - were spoken in the horizon, you can't at the same time make the argument that this horizon must have been Indo-Iranian speaking. You can't have both, they're mutually contradictory.

Davidski said...


The main Sintashta cluster is basically like Sredny Stog and Czech Corded Ware, while the outliers are more or less like Yamnaya and/or Botai.

The mystery of the Sintashta people

It's impossible estimate with qpAdm whether the Early Middle Bronze Age ancestry in Sintashta is more like Yamnaya or Afanasievo.

Davidski said...


I also have this one...


Not sure how to effectively break up the South Asian PCA. I'll have to think about that.

Drago said...

“So I am simply saying that if you argue that a whole bunch of IE languages - or even for some reason just two phylogenetically unrelated varieties “”

The final branches appear unrelated ; for reasons explained above
But in fact they probably have the common origin
The other options (extinct/ non-IE Afanasievo & the ““noiveau nihilistic theory) are not worth any attention.

Davidski said...


So I am simply saying that if you argue that a whole bunch of IE languages - or even for some reason just two phylogenetically unrelated varieties - were spoken in the horizon, you can't at the same time make the argument that this horizon must have been Indo-Iranian speaking. You can't have both, they're mutually contradictory.

It's basically certain that multiple languages and even language families were spoken within the Andronovo horizon and perhaps even within Sintashta settlements.

There's no way that all of the vast areas of Central Asia and West Siberia that were dominated by these archeological cultures were uniformly proto-Indo-Aryan and proto-Iranian speaking for something like a thousand years.

Drago said...

@ Bob

Yep something like that.
You can refer to H. Andersen's chapter in ''Language Contacts in Prehistory: Studies in Stratigraphy.'& Z Golabs wotk on the ''strata'' in Balto-Slavic. They highlight Baltic & Slavic languages were progressivly satemized over time; & not completely so. There's also the finding that some of the 'early IE'' loans in Uralic have some proposed Tocharian -like features, with indo-Aryan being from significantly later period, in all likelihood- Scythian migrations.. The implication is that balto-Slavic, Indo-Aryan & Tocharian could unproblematically descend from the broadly similar set of dialects spoken by Andronovo communities, if some variant of the steppe model is correct. As I said, there are few other options which even come close.

vahaduo said...

@ Davidski

Caucasus - Near East PCA is online in two versions (1: PC 1 vs. PC 2, 2: PC 1 vs. PC 3), although I'm not sure if the second one is really needed.

Davidski said...


Yep, the first Caucasus - Near East PCA does the job well enough. The second one is basically just one dimension of variation.

Matt said...

Sure, that argument can be made, I'm just noting that it contrasts with the view which a lot of scholars take, and that they can't take both. For reference that various scholars do take that view, for instance, Mallory - The Problem of Tocharian Origins:

"The problem here ( for Tocharian origins in the Bronze Age north of the Tianshan,) is that the Bronze Age north of the Tianshan can be seen as an extension of the Eurasian steppe Andronovo horizon (Debaine-Francfort 2001, 57), which one generally presumes to involve an (Indo-)Iranian identity. So the north-south flow of steppe metallurgical types and other artifacts may be seen as evidence for the potential spread of Iranian languages. While these correlate with the later Chinese historical tradition of locating presumably Iranian-speaking tribes such as the Wusun in the north, it does set up a problematic pincher movement of Iranians coming from both the west (the Pamirs and onwards to the historical Saka sites of western Xinjiang) and northern Xinjiang. If we also label all those buried in the cemeteries south of the Tianshan as Iranians we effectively eliminate every known Iron Age occupant of the region as a potential Proto-Tocharian. This essentially proceeds from the probably dubious paradigm in which everyone is presumed Iranian unless they can prove otherwise. The issue here is that anything known from the Eurasian steppe from the beginning of the Andronovo horizon through the Iron Age is generally interpreted as Indo-Iranian, or specifically Iranian."

Though as parts of the Andronovo horizon are excluded from Indo-Iranian, that means that they are no longer "available" for divisions of Indo-Iranian into East Iranic, etc.

That doesn't seem necessarily a problem, as weakening the equation of Indo-Iranian with Andronovo, in favour of Andronovo as a diverse horizon that may encompass other languages such as proto-Tocharian and Uralic languages, may allow for other hypotheses for Indo-Iranian origins to come to the fore.

For example, Indo-Iranian can branches with Armenian and Greek in phylogenies (whether of lexical traits, or phonological and grammatical innovations), or else splits early, and the first attestation of Indo-Aryan is under Mitanni. If Rakhigarhi proves to be fairly ASI, and the Indus Periphery individuals unrepresentative of IVC, a degree of BMAC and further Iranian plateau/Turan related admixture may also be necessary to explain South Asian population structure, which would suggest some movement relating to Indo-Aryan along along this vector from NW to South Asia.

Davidski said...


I honestly can't see how the idea that Andronovo was linguistically diverse weakens the argument that it was dominated by Indo-Iranian speakers and opens up more of a possibility that Indo-Iranian languages spread from West Asia.

I actually always considered that a linguistically diverse Andronovo horizon was the most sensible assumption, and thought that the focus on it being Indo-Iranian was just that, without the implication that other language families couldn't have been present within it.

But I seriously doubt that Uralic was widespread within Andronovo. It was probably only present along its northwestern fringe, otherwise we'd be seeing Y-hg N pop up regularly in Steppe_MLBA populations, but it doesn't even make an appearance in any Scythians tested to date.

Davidski said...


Here's another one. It might be useful to do two views for it, because PC1 & PC3 are very informative together.


Davidski said...

By the way, is there any way to add a feature to the mixture modeling tool which limits the number of reference populations to the best 5 no matter how many populations and/or individuals are pasted on the source page?

Janko Raven Johnson said...

Have been wading through Kuz'mina's Origin of the Indo-Iranians, which appears to make the case for identification of Andronovo with that origin. Topical quote fwiw:

"It is probable from the linguistic data that in the middle of the 2nd millennium BC the Indo-Iranian language family was an aggregate of independent related language dialects that constituted a language continuum. It is possible to suppose that separate local variants and types of sites of the Andronovo culture corresponded to separate tribes who spoke certain dialects. Such an understanding of the Indo-Iranian continuum allows us to attribute the Timber-grave culture, which was a possible ancestor of the Iranian-speaking Scythians, to Proto-Iranian." (p. 167)

Matt said...

@Davidski, it kind of seems like it does in the sense that if we're going "Oh, well this part of the horizon possibly did not speak Indo-Iranian languages", and that's a part is vital source of evidence for the claimed correspondence between Andronovo and II - rituals and place names which are presumed to be older than the Iron Age - then that weakens the argument more generally.

Or on the other hand if we were to do the opposite and cherry pick that those parts of the horizon that are sources for that evidence just happened to be Indo-Iranian, then that leads open to charges that you would be making a "Just So" argument.

And there's the whole thing of "If part X of horizon could speak a totally other language without it being obvious from ritual or material culture (e.g. the rituals are move across linguistic borders), what confidence do we have in any of the reconstructions of correspondences between Indo-Iranian and Andronovo?"

Anyhow, those kind of seem like that's how it is to me, but it would be more useful to bounce off one of the architects who argue for equation of Indo-Iranian=Andronovo.

zardos said...

@Matr: Whats the problem? We see similar things elsewhere. For example Hallstatt is highly likely to contain Proto-Celtic, but its highly unlikely all of Hallstatt, all its provinces were early Celts.
Andronovo is stretched over a much larger territory with distinct provinces and over a long period of time. Actually some even considered related groups to form Tumulus culture on top of Unetice people. So there might be even a direct connection to Celts and in turn Tocharians. :)

zardos said...

It is interesting to follow the spread of chariots in this respect and for Hallstatt and early/British Celts the chariots were of central importance.

Matt said...

I have to say, talking about Central Asia again has made me go back to the Narasimhan preprint and look at the outliers and chronology. Champing at the bit for the full print.

Elements I must not have noticed or thought about too much originally but seem noteworthy, are

- The outliers at Gonur which are Steppe admixed don't really separate cleanly much in time from those which are West Siberia N admixed.

Gonur1_BA_o2, steppe admixed, is two samples at 4104 BP and 4010 BP (average 4057).

Gonur1_BA_o1, West Siberia admixed, is at 4280 BP and 4100 BP (average 4190 BP)

When you hear Reich discuss their work, there's a bit of an impression of a big temporal gap between West Siberia and steppe phases, but that doesn't seem to be the case.

I suppose perhaps that's the case elsewhere in the Turan, considering Eneolithic+all BA, but it isn't very apparent at Gonur.

(Indeed the supplement tends to emphasise that these four and the Indus Diaspora related sample are together in the early phase of Gonur:

"Of particular interest in the evaluation of a steppe influence in the BMAC is that the recently dated outliers (genetically steppe related [meaning both O1+O2]) fall within the early period of the BMAC, c. 2300-2000 BCE during the main floruit (flourishing) of the BMAC. The social context of these burials includes a variety of tomb types, i.e., pit (3454), shaft grave (N4329), cist (3201), and even a burial (3240) within the ‘Royal Cemetery’, which may indicate social integration within the BMAC community. The burials were completely compatible with the rites of the BMAC and the only grave to yield an array of grave goods, the shaft grave, produced typically BMAC artifacts. Unfortunately, the plundering of the ‘Royal Cemetery’ does not permit an assessment of the status of the individual buried there and it might be noted that the warrior burial (2380) was genetically local."

There's an absence of outliers from the later phases at Gonur, where all individuals tend to be "genetically local". If we were expecting a tendency of acceleration of merging between genepools over time, it's more the opposite, that unalike individuals came in mostly during the early 'flourit' of the civilization and tapered off over time, without affecting the general population structure too much).

Sadly only 2 of these Gonur outliers made it through to Global 25 (low quality on the others).

Matt said...


- The Gonur1_BA_o2 outliers (steppe admixed) are really pretty early compared to the Steppe_MLBA samples. The date is older than any of them (though not by a huge amount) and contemporary with the youngest Potapovka samples. Looking at the Haji Firuz BA sample, she is also earlier than any of the Steppe_MLBA, 4326 BP (2326BCE).

Also, on Haji Firuz BA, how weird is it from an IE model like David Anthony's for there to be very early substantial steppe ancestry almost contemporary with Europe in Northwest Iran, near the same regions that would later show evidence of Armenian, Indo-Aryan Mittani, and Western Iranic languages? My impression is the relatively "consensual" steppe model tends to stress that these regions would see much later "incursions" of IE speakers only by the late Bronze Age to Iron Age.

Such that the Mitanni are seen to represent a "first wave" of IE speakers moving through Iran. (IE studies are never very clear on what is actually happening linguistically in Iran before the Iron Age though - everyone seems to agree that Elam existed all through the Bronze... but it only represents a small slice of Iranian territory, so the rest is 'Here Be Dragons').

- One of the Gonur1_BA_o1 outliers was a female buried aged 50-60 with "a large rectangular pit near tomb 3235 in the “Royal Cemetery” where 17 individuals were found together with a four-wheeled wagon, two camels, two dogs, one calf and two lambs." Wonder what happened there...

- I don't think I appreciated how late you would get BMAC related clusters with no trace of steppe ancestry. Bustan_BA is 1400 BCE, and there are 2x Parkhai and Sumbar samples with no steppe either at 1250 BCE (and one outlier with a low level). Quite close to the earliest Iron Age samples. Almost overlapping with the earliest Pakistan samples. Sappali Tepe BA is between Gonur and these late Bronze Age set. The Dzharkutan samples, including Dzharkutan2_BA Caucasus related outliers (wrongly identifed as Steppe_MLBA admixed in the paper?) are earlier at around 1900BCE.

(On on this note it looks like the earliest Steppe related outlier from BMAC that made it through to G25, Sappali_Tepe_o, doesn't quite fit as admixed between BMAC samples and the main set of Andronovo+Sintashta:

Note PC3 in the reprocessed PCA and how I7493 isn't really intermediate the Steppe_MLBA samples and BMAC.

The Sappali Tepe outlier seems to fit better between Hajji Firuz BA and Hajji Firuz CHL in some ways a little better.)

Drago said...

In summary; these early steppe individuals left no lasting impact? Similar to Afanasievo

Davidski said...


My impression is the relatively "consensual" steppe model tends to stress that these regions would see much later "incursions" of IE speakers only by the late Bronze Age to Iron Age.

The earlier steppe ancestry in that region probably represents the incursions of Catacomb related groups into the Near East. But no one knows what languages they spoke, and, for instance, sample Hasanlu_IA F38 from near Hajji Firuz is likely to be from a Hurrian-speaking site.

The spread of Iranian languages across Iran is generally seen as an Iron Age phenomenon that started in the vicinity of the present-day Iranian/Turkmenistan border.

Matt said...

Yes, generally that's the consensual view, I'm remarking that the consensual view would not predict early spread of steppe ancestry in NW Iran.

I'm not sure if there is a basis to believe that populations that produced individuals like I4243 would have been non-IE speaking, and what would happen if arguments that were made consistently for the association of the spread of IE languages with steppe ancestry were made for this population.

Drago said...

Again; context is everything. Merely “fishing for steppe ancestry” isn’t useful.

The high point of BMAC featured some contact between early steppe communities & BMAC. But what does it actually show? Limited contact & intermarriage; with some steppe chiefs offering daughters to the landed elites. The steppe communities; however; stayed north. The parallels to Varna are similar; although D Anthony claimed the Varna baby represented proof of a steppe conquest of SEE.

When BMAC collapsed; this was all “reset”. New ; and different steppe communities arrived- Andronovo. They had to negotiate new relations with post-BMAC groups. And instead of negotiating via a more centralised chiefs in “forts”; it was more scattered agricultural communities; and took a one-on-one form. Hence were going to see some relatively unadmixed groups (eg LBA Bustan) & Andronovo groups infiltrating around them.

Matt said...

@Drago, at Gonur and BMAC no particularly lasting impact seems evident. Sporadic.

In NW Iran, it seems not enough samples yet to tell. It may have been more substantial.

I'd also comment that on the Parkhai_LBA outlier, I6667, another one of that sample does not seem clearly Steppe_MLBA admixed rather than West_Siberia admixed. Using cline extension on Parkhai_LBA against the BMAC main cluster generally, doesn't work too well, but to the degree it works, it suggests West Siberia ancestry relative to the other members of the cluster. (Seems contrary to the claim in Reich's presentation on the subject that "Before 2000 BCE" West Siberia admixed, "After 2000 BCE" Steppe_MLBA admixed - ?).

Drago said...

The high point of BMAC featured some contact between early steppe communities & BMAC”

And also “Dali EBA” - like groups

Davidski said...


Yes, generally that's the consensual view, I'm remarking that the consensual view would not predict early spread of steppe ancestry in NW Iran.

Nope. The early spread of steppe ancestry via the Caucasus was a mainstream topic well before ancient DNA.

I mentioned it in a few blog posts, like here...

A potentially violent end to the Kura-Araxes Culture (Alizadeh et al. 2018)

But, as far as I know, no one ever claimed that these were early Iranian speakers.

Drago said...

Yes there was movement from steppe to Caucasus to West Asia. And hence Kristiansen accounts for proto-Anatolian, for ex, coming via this route. The problem is - Kura-Araxes & post-KA movements map out the non-IE groups in Anatolia; whislt the IE groups can very easily be delineated as coming from the Balkans.

In the same vein- what IE groups moved into Haji Firuz ? There seems to be no connecting thread to Iranians-speakers

Davidski said...

Is that Kristiansen's map? I thought it was done by a German paper based on one of the Damgaard papers.

Anyway, it does look strange, because it seems to derive basically all European Indo-European language families, including Baltic and Slavic, from Hungarian Yamnaya. That won't work out.

Davidski said...

@Janko Raven Johnson

Kuz'mina didn't have ancient DNA to work with, so she had to rely on archeology to corroborate linguistics and make the strongest case possible.

But we now know that the Andronovo groups that moved into South Central Asia were very closely related to the Andronovo and Srubnaya groups that moved into the area that is now the Iranian/Turkmenistan border.

So no matter what else was happening within the Andronovo horizon, these groups could have spoken Indo-Aryan and Iranian languages that spread into Anatolia and eventually across India and Iran, along with early Indo-Iranian religion as well as steppe horses and spoked-wheel chariots of course.

On the doorstep of India

An early Iranian, obviously

Matt said...

David, seems you may be confused? I am not arguing that in a general sense literally no one argued for the movement of steppe population ancestry across the Caucasus, but that Anthony and other theories of IE from the steppe that see no language dispersals into Iran until relatively late with Indo-Iranian would not predict early movements of ancestry as described by Hajji Firuz.

Matt said...

Davidski: But, as far as I know, no one ever claimed that these were early Iranian speakers.

Yes, I'm wondering aloud what all this is based on. The claimed features connecting II to Andronovo-Sintashta seem to be:

- Correspondences between proto-Indo-Iranian rituals (reconstructed from Rig Veda, Gathas), and Andronovo/Sintashta. But are those correspondences that secure or exact compared to archaeological evidence from more southern sources? Can they be compared firmly to what is going on in Iran outside Elam in the Bronze Age, since no one seems to know? ('Here Be Dragons'). Those were made by Russian archaeologists early relative to when BMAC began to be excavated... And in any case how tightly bound to language are the religious rituals?

- Iranic toponyms are said to correlate better with Andronovo as indexed by sites than with later horizons in the Iron Age which are more firmly Eastern Iranian speaking. But that was made in the 1980s, so how firmly clear does this idea remain in 2019, and how convincing and striking is the pattern?

- Groups like Hurro-Urartians do exist in the Northern ME, but they hardly saturate it in attested distribution. (Hurrian -, Urartu - There is plenty of linguistic terra incognita sufficient space for IE groups to have existed since the Bronze Age. It doesn't seem like a secure argument that there is no available territory for early IE branches.

- There are claims that Indo-Iranian shares vocabulary with Finno-Ugric (and other IE branches not) and phonological features with Balto-Slavic.

But it seems like this is all from Indo-Iranian->Finno-Ugric at a pan-II level, and while this is explained by various stretches ("II languages had 'high status'!" etc), it would make sense if Indo-Iranian derives from point where there was no opportunity to take loans from Finno-Ugric, then only a subset of II expanded to be in contact with them. Similarly shared phonological innovations possibly can be explained as later areal innovations from contact with Eastern Iranic languages (see -

More generally, overall, it seems that it could be more parsimonious to have a centre of ancient diversity of Indo-Iranian within Iran, as you then have less far in total to travel to get Indo-Aryan to both Mitanni and to India, and you do away with the rather inelegant proposal of Burrow for expansions of Indo-Iranian to the Near East in multiple waves whereby the later "defeated" the earlier ( In favour of only a single dispersal of Eastern Iranian languages and none of the confusion about the general absence of clear proto-II or Indo-Aryan evidence elsewhere to the north.

You also get a better fit with the lexical models that see an early branch off and deep subdivisions with II, or analysis of innovations that sees a "Southern" clade vs "Northern" clade of IE languages (as in Ned's final model above, where it is moderated by contact between II of some sort and BS to explain remaining phonological isoglosses).

(On a side note, when reading Mallory's IE Encyclopedia entry on the Indo-Iranians (, that he argued that the Swat Valley IA should be "most easily" identified with early Eastern Iranian Dardic and Nuristani speakers, not Indo-Aryan languages, presumably crossing over from the Inner Asian Mountain Corridor. May be worth bearing in mind when considering the samples).

Davidski said...


The only realistic link between Indo-European speakers in Europe and South Asia is the Middle to Late Bronze Age steppe. Everything else is just special pleading.

Typically Indo-European human and horse DNA, religious beliefs and rituals, and military technology moved deep into both Europe and Asia at this time, and made massive impacts. So if Indo-European languages didn't as well, then that'd be some kind of miracle.

Vara said...

"Typically Indo-European human and horse DNA, religious beliefs and rituals, and military technology moved deep into both Europe and Asia at this time"

That's nonesense. There's little of that in the Iranian Plateau and Jiroft had horses since 2500BCE.

Matt has the right idea. Proto-Indo-Iranian had to develop in Ariana. There is little actual I-I rituals that can actually be attributed to Andronovo, that's why you have the Kulturkugel theory in the first place. Not only that, but whenever an Andronovo ritual is looked at it just completely contradicts the actual literature we have. Take the Srubnaya dog sacrifice of Kuz'mina and Anthony and go search the whole Rigveda for it and let me know if you find it as they claim. The funny thing is sometimes Srubnaya is considered the homeland of Zoroastrianism I guess that totally makes sense why the Magi never sacrificed dogs?

A late entrance simply can't explain the I-I contacts with the Near East groups in the 3rd millenium.

More and more work has been done on Iran now and they're finding that a lot of these I-I rituals are related to those of the Iranian plateau like here:

Where the actual Indo-Iranian culture derives from is here:

Joey said...

How do you postulate that proto-Uralic got loans from PII if it developed in Iran? As far as I know, Jiroft is not even considered a distinct culture by others than Yusef himself.

Vara said...

"How do you postulate that proto-Uralic got loans from PII if it developed in Iran?"

Matt already went over that but I have a few more scenarios.

1. BMAC- Seima Turbino interactions
2. BMAC-Abashevo interactions (Some Andronovo groups were carrying material from Abashevo)
3. Issedones and other Scythian groups

The thing with Proto-Uralic is that there are so many theories on the homeland that there are infinite possibilities. However, it's very clear that they were not close neighbors looking at the very few II loans.

"Jiroft is not even considered a distinct culture by others than Yusef himself."

What does that mean?

Joey said...

Seima-Turbino being proto-Uralic is not likely at all. What is the most likely scenario is that proto-Uralics interacted with proto-indo-iranian Sintasha and Andronovo in the steppe-forest periphery.

"What does that mean?"

That its a culture hypothized by Yusef Majidzadeh but has little support from academics in general.

Vara said...

You're just preaching now. Would you like to tell us about the PU homeland.

Seima-Turbino is one of the candidates for the PU homeland. Abashevo is considered the PU culture by some, right? Abashevo pottery has been found in the vicinity of BMAC. There you go.

Now how did Andronovo interact with Kassites, Gutians and Lullubeans?

"That its a culture hypothized by Yusef Majidzadeh but has little support from academics in general."

That's nonesense. When Shahr Sukhteh was first excavated Sarianidi proposed that it is basically a part of a larger complex culture. He was opposed at that time cause no other site was excavated. However, now all of that is considered as part of the Helmand Culture.

If you have actual evidence then go ahead enlighten us but please stop repeating the mantra cause I'm tired of it by now.

Davidski said...


Update yourself.

Abashevo couldn't have been the PU homeland because there's no sign of Y-haplogroup N on the steppe or forest steppe even as late as the Iron Age. And there's no N in any of the ancient DNA from Central Asia.

It was always going to be this way

Conan the Barbarian probably belonged to Y-haplogroup R1a

Also, I wasn't talking about just any type of horse, but a specific breed of horse that first shows up in Sintashta and eventually moves into Iran. See HERE.

Vara said...


As I said, it doesn't matter where the PU homeland is. There many possibilities as to how Uralic and II interacted.

We do not know how the horse moved into the Iranian Plateau or when because we do not have horse DNA from there anyways. It could've been just traded you know.

What we do know is that during the late 3rd Mil Indo-Iranians were interacting with Zagrosian tribes and by 18th century BCE they were already a military power in the Near East. What we also know is that the actual presumed Indo-Aryans of Swat and Iranians of post-BMAC had little steppe ancestry and almost no cultural relation to Andronovo. Connect the dots.


Drago said...

Yeah In-Ir might have diversified in Iran or Turan through bidirectional contacts; then some moved back with Scythians; imparting their influence on Slavic etc

Davidski said...


I can connect the dots but you obviously can't. Horse DNA is just one piece of the puzzle. Human DNA is another.

They both show that typically Indo-European humans and horses moved from the steppe into Central and West Asia during the Bronze Age. And then there is the archeology, Indo-European religion etc.

Denial is not a good long term strategy.


Scythians and Slavs don't explain anything here, because obviously Baltic languages are similar to Sanskrit not because of recent Iranian influence but because they're conservative.

Vara said...

Lovely. Ignoring my arguments while repeating the same PR BS.

"They both show that typically Indo-European humans and horses moved from the steppe into Central and West Asia during the Bronze Age."

Show me one archaeologist that found an Andronovo artifact in Iran.

"And then there is the archeology, Indo-European religion etc."

The archaeology that pushed the Kulturkugel theory because there is almost no influence from Andronovo on the actual Indo-Iranians?

Since you're suddenly an expert on IE religions care to explain this:

I thought you said you weren't going to stick to a homeland theory. Hell, don't listen to me I just predicted this years ago. If you remember when your boys were saying Swat and early Iranians were going to be 70% Andronovo I was the only one saying that the actual Indo-Iranians will have less steppe ancestry than their modern day descendants.

You're in denial.

Davidski said...


Show me one archaeologist that found an Andronovo artifact in Iran.

Why the straw man argument? You don't have any real arguments? Read this paper carefully. It's by people who know what they're talking about.

A palimpsest grave at the Iron Age cemetery in Estark-Joshaqan, Iran

And it fits very nicely with this. Make sure you look closely at the map.

An early Iranian, obviously

And I've got something better for you than 70% Andronovo in South Central Asia. I've 80% and more Adronovo in South Central Asia. I remember clowns like you denying not long ago that something like this was even possible.

On the doorstep of India

Btw, Indo-Iranians interacting with Zagrosian tribes in the 3rd Mil BC? Hahaha. You're obviously a comedian.

Drago said...

@ Davidski

“Scythians and Slavs don't explain anything here, because obviously Baltic languages are similar to Sanskrit not because of recent Iranian influence but because they're conservative”

Yes but there was also secondary convergence;
Scythians made it quite far into EE.
So first there there was the movement from EE to south Central Asia; then a countercurrent back

Drago said...

The possibility that some of the descendants of Majkop spoke Hattic & Minoan should be looked at; as per Schriver’s broad suggestion

Andrzejewski said...

@Drago I’ve been posting all along that Hattic and maybe Minoan may derive from Kura Araxes.

Davidski said...

Kura-Araxes is normally associated with Hurro-Urartian languages.

Hattic might be from the Caucasus, but it hasn't been specifically linked to Kura-Araxes by anyone yet. The chances that Minoan is derived from Kura-Araxes is basically zero.

Drago said...

Well Minoans lie on a cline from Greek N. to Barcin_CHL to Majkop& Kura-Arxes. So their language might be local; but I there’s something to Hattic-Minoan; then I wouldn’t discount it . There’s archaeological Links (indirect )
Davidski can perhaps look into which is better Stats for Minoans out of Majkop & K-A.

Andrzejewski said...

I thought that Armenians, Georgians and Chechens among others all descend mainly from KA (along with Kurds) but I may be wrong. I know that KA is mostly ANF + CHG. As for Maykop I read that it is a lot of Ubaidians and Halafians originally from places like Leyla Tepe (Uruk Expansion) and that NW Caucasus speakers are their direct descendants but everything Caucasus related is very confusing to me

Andrzejewski said...

And based on recent articles I read, isn’t it possible that we largely underestimate the ratio of WHG as Drago asserted earlier?

Matt said...

@David, I am not so sure. What is happening in most of Northern Iran is still archaeologically very obscure. Archaeological trails are lacking, the direction of ritual influence unclear, timing of arrival of horses, chariot technology also unclear, etc.

Early Bronze Age connections may be sufficient, supportive of early divergence in some linguistic models.

The mainstream Indo-Iranian argument (even if "the best we've got") still seems a weak argument in evidence with some backing by preference for consequences and narrative by strong adherents.

Many of the details seem vague, or suggestively wrong on the basis of Narasimhan. For instance, Parpola 2017 suggests the following sequence of events - "The horse-drawn chariot was developed around 2100 BCE in the Sintashta culture, and spread quickly in many directions. Through the Petrovka culture it came to the Bactria and Margiana Archaeological Complex (BMAC) (2500–1500 BCE) of southern Central Asia, where Petrovka pastoralists speaking Proto-Indo-Aryan took over in the 20th century BCE, and continued in the BMAC garb to the Indus Valley on the one hand (e.g. BMAC cemeteries at Quetta and Sibri 1900 BCE) and to northern Iran (Tepe Hissar 1900 BCE) and Syria on the other. (Fig. 6.)". (See -

But this series of events seems wrong looking at the adna so far, with no evidence of a sustained genetic contribution by Andronovo related peoples to late BMAC, only early outliers at Gonur, which mostly overlap in time with West Siberia admixed outleirs. No takeover.

Even of the later outliers in the BMAC zone:

- G25 suggests that Dzharkutan_2 BA are mostly CHG admixed, not really representative of movements from the Steppe_MLBA zone ... (Narasimhan 2018 does the classic Harvard paper thing, following Reich's book's tendency to ignore anything they personally did not publish, and feigns ignorance of the existence of CHG, of course! And yet boats can cross the Caspian Sea...).

- ... the latest outlier in time Parkhai_LBA_o fits even in Narasimhan at highest P only as West Siberia admixed*, with no or negligible evidence of steppe ancestry - ...

- ... leaving only Sappali_Tepe_BA outlier, but even that is a bit questionable as a two-way combination of BMAC populations and Steppe_MLBA -, fitting better with mostly Armenia_EBA related ancestry.

If we are to disregard the outliers, of course, we can stay that mostly Steppe_MLBA populations existed at Dashti-Kozy... but this is very far from connecting with both Mitanni and the Indo-Aryans of India; for which reason Parpola took the above stance, to try and manufacture a common thread in Indo-Aryans moving through BMAC into NW Iran and to India both...

More adna from Bronze Age Iran (if we can get it) will be vital.

*Which fact Narasimhan's supplement seems to have rather mischeviously glossed over in favour of contradicting their own model two pages earlier and claiming that Parkhai_LBA_o represents a shift towards Steppe_MLBA in Northeast Iran at exactly the time predicted by this model of Indo-Iranic expansions (see the text in above link from p139 - "If this is the case and the Indo-Iranian language family branched from a source ultimately in the Steppe, the likely geographic entry point of these people into Iran would be in the northeast of Iran. Within our dataset, we can trace a time transect from a single location on the border between Turkmenistan and Iran—at Parkhai—where we have a set of samples with dates of 3550 BCE, 2600 BCE, 2189 BCE, and 1455 BCE. We observe from PCA (Fig. S3.9) that there are no shifts in Steppe related ancestry until the late Bronze Age and the sample dated to 1455 BCE." . Well, yes, West Siberia is vaguely steppe related, but...).

Vara said...

David, please.

You posted that paper many times and I told you a gazillion times that dude is a Scythian at a time where Scythians were making it into Western Iran. The Medians weren't buried that way.

Yeah there were a few outliers in BMAC, so what? Look at the context. The Andronovo were nothing but traders they couldn't have changed the language or culture of anything and their genetic impact was very small.

Come on David, read some papers.

"It is probable that this move was preceded by successive spearheading forays of (non-IIr. speaking) mountain peoples into Mesopotamia, such as the Guti, Lullubi, and Kassites (c. 2250-1750 BCE), who were as yet only marginally influenced by IIr. languages and customs. Some of them are perhaps represented by the sudden expansion of BMAC materials into Susa, Shahdad, Tepe Yahya, Hissar, the Gulf,Baluchistan, the S. Indus area (Lamberg-Karlovsky 2002: 72, 74, 84) and Harappa (R.Meadow, pers. comm.)"

Matt said...

On Dashti-Kozy and feasibility for a direct movement into Swat from there, a couple more notes:

- Although this population is in a place (e.g. an extension of the IAMC that links to the Northern Central Eurasian Steppes) that makes a migration to the Swat Valley feasible, that population wouldn't really fit with the "first wave from the steppes" theory of Indo-Aryan, since they are already at the same time at roughly which Mitanni is attested (1400 BCE while DK samples mean date roughly 1550-1600 BCE), and there is not much real time for them to get there.

- The Indus_Diaspora samples so far are from locations quite remote from Swat Valley and India, Gonur in Turkmenistan, and the city of Shahr-i Shokhta along the Iran+Afghan border*.

It's no more (or less) likely that they represent un-admixed representatives of Indus Valley and Swat in the BA than the Steppe EMBA/MLBA and Dali_EBA and Armenia EBA admixed samples in BMAC represent the Steppe or Northern Central Asia or the Caucasus.

A model of Dashti-Kozy+Indus_Diaspora may work for Swat (or almost but not quite!), but if the Rakhigarhi leaks are correct and their samples are like the Irula, then Dashti-Kozy probably would not work at all using actually representative samples...

*The earlier term in preprint of Indus_Periphery being inaccurate and changed for obvious reasons!

Davidski said...

This paper looks interesting...

‘Children of the light’: On yoga, body schemes and altered states of consciousness in the Nordic Late Bronze Age – a link to India?

It sort of reminds me of a blog post I did years ago...

Children of the Divine Twins

Davidski said...


There won't be any revelations in the final Central South Asian paper as a result of the new samples, or the Harappan samples from Rakhigarhi.

The paper will be going firmly with the Sintashta angle for the Indo-Europeanization of South Asia.

There's no way that this consensus is going to be overturned with ancient DNA or anything.

Matt said...

Eh, if the Rakhigari samples are Irula like, two-way models like Indus_Disapora+Sintashta/Andronovo won't work, and you will need a heavy dose of BMAC/Iran C like ancestry along the way, as well as some Steppe_MLBA like ancestry.

The rough details that there is not population continuity will not change, but the results may not support the directional model of the formation of ANI from Narasimhan, if they are Irala like.

Whether or they are going to be Irula like, I can't say. Leak suggests they are. 1 Irula like IVC sample certainly means a lot more than 3 (or even 6+!) BMAC+AASI admixed samples from BMAC and Shahr-I-Shokta.

Davidski said...

I think you're getting way ahead of yourself based on an old leak about a low coverage singleton.

I guess you haven't heard that there will be many more than just three Indus_Periphery samples, and the assumption will be that they do represent some part of the Indus Valley population, mainly because they look like obvious outliers in Central Asia.

Matt said...

Yeah, I've seen the presentation with the 6 samples, hence why I mentioned 6 samples above (screencap -

I don't see any indication that they represent an unadmixed population. The only 3 that are similar on the plot are similar to Shahr-i-Shokta outlier 2 and Gonur 2 BA, which are the least plausible representative of IVCs. The other two have more ASI ancestry, in varying amounts.

There's no real sense in assuming any of them are unadmixed with BMAC or populations from Southwest Iran, and any adna from Rakhigarri will carry more weight (even 1 sample).

Davidski said...

Apparently there's more than one sample now.

Matt said...

Of course, the legend on the above screencap reverse Turan and Steppe_MLBA poles and gets it wrong, but the full presentation is here:

Note, it indicates that are least two of the further new samples are from Gonur.

Matt said...

One other thing about the D. Reich slide depicting the new outliers, when they model ANI and the Swat main cluster, it's not like he's even using the average of the outlier samples, but rather a point much closer to the least ASI admixed outliers.

E.g. with my annotation, the lines here would trace midpoints from my guess of close actual average of the outliers to the Swat Valley outliers, maain cluster (even though I suspect that these labels are reversed since otherwise it makes no sense with the data we have):

Even if we are to assume that the outliers represent a single population representative of IVC without admixture from local in BMAC/Jiroft (dubious IMO) it seems very strange not to use their average and instead a "fantasy point" closer to the average of the three least ASI admixed samples.

Especially if we have no idea when the Swat Valley outliers which presumably have enriched Steppe ancestry are actually from and the Sway Valley set stretches up until the early historical period.

(At the extreme, I would hope Reich wouldn't be drawing his "ANI Cline" stretching back to the steppe MLBA using the contrast between a mass of 108 Swat samples and 2 >700 BCE samples though. Because that would be bonkers.)

Matt said...

The newer stuff on Indo-Aryan from the steppe has probably aged better, but if anyone wants to read a blast from the past that tries to explain Indo-Aryan, Asko Parpola's "Coming of The Aryans" (āsas._Studia_Orientalia_64_195-302).

Now Parpola argues very strongly for "conclusive proof of a foreign colonization that took place around 2000-1900 BCE (in the Indus Valley at Mehrgargh) ..." from an "entire cultural complex practically identical to that of sites like Tepe Hissar III in northeastern Iran, Namazga V in southern Turkmenistan and Sappali Tepe and Dashly in Afghanistan". He terms these latter societies as "the Greater Iranian Bronze Age of the Namazga V phase" and then goes on to make a very emphathic argument for "an Aryan identity of the Namazga V culture ... controlled by a semi-nomadic military elite" which he equates with equestrian tendencies, which he appear to trace back on a migratory route from Azerbaijan ("This, then, would appear to be the most likely direction from which the Hissar II culture acquired both its chariot technology and its ruling elite").

Main supporting figures desribing geographic movements under this model:

(This is supported by such colourful points as how "Particularly important evidence for their Aryan identity (of the above military aristocracy) is an evidently locally made cylinder seal of alabaster from the Hissar III B levle, so far generally dated to about 2350 BC. This dating would make the seal the earliest known representation of a horse-drawn two-wheeled war chariot". Similar information in Blench's 2012 text -

This model, still a frequent cite today, does then propose a relatively early dispersal of proto-Indo-Aryan across the Caucasus, then its further dispersal through North Iran from the west.

Whether Parpola's argument in 1988 stands at all in any sense in the present day, it's certainly of interest to me that he made this argument, and I wonder if the evidence marshaled in favour of it is simply wrong, or is correct but may be interpreted differently in the light of adna, or what. (What do me make of the early cultural movements that he suggests which are certainly attested by genetics in NW Iran at the time suggested, and are so a plausible vector for language, but seem to have had no great genetic impact on the wider sphere).

(Certainly in the last figure, the position of proto-West Iranian with Srubnaya seems unnecessarily complex, since all the attested Iranian languages North of the Caucasus are from East Iranian clade. It seems unnecessary to propose the complete replacement of West Iranic languages, when it can be solved with a simple dispersal of East Iranic languages from a place near where West and East Iranic are both found).

Arza said...

@ Matt

IIRC besides Steppe and Swat all of these points are hypothetical. It was written in the description... somewhere.

Drago said...

Current state of art:
“ Mixing metaphors: sedentary-mobile interactions and local-global connections in prehistoric Turkmenistan”
- Rouse, Carasetti

Davidski said...


Unfortunately, I have to say that you're going on quite a bit there, so there's no point continuing this discussion and rehashing all of the arguments that I've already posted a thousand times at this blog.

But one point that I need to clear up is that the currently available Indus_Periphery samples obviously have very little recent Central Asian ancestry, and what they do have was probably acquired by their ancestors already in the Indus Valley, and not Central Asia.

The reason I say this is here...

Global25 South Asia ancients

As you can see, the Indus_Periphery samples from Gonur and Shahr I Sokhta don't exactly form a cline with their contemporaries from Central Asia, including from Gonur and Shahr I Sokhta.

So it looks like we're missing a population on this plot that basically lacks Anatolian, Central Asian and South Asian ancestries, and thus should be sitting just below the ancient Central Asian samples.

That is, this is the ghost that represents the population that contributed most of the West Eurasian ancestry to the Indus Valley Civilization (IVC) people, otherwise it's impossible to explain the genetic structure of the Indus_Periphery samples and also that of the least Central Asian admixed Dravidians and tribals.

This means that the IVC population, when sampled thoroughly, is indeed very likely to fall mostly on the cline formed by the Indus_Periphery trio, although it might show a fair amount of variation depending on the location of the sampling sites, with some individuals or groups perhaps being almost exactly like the ghost that I described, and others possibly with significant recent Central Asian admix.

In other words, your suggestion that the IVC population might turn out to be, by and large, like the Irula and the Rakhigarhi IVC singleton that the old leaks were about is unreasonable and rather strange actually.

Matt said...

David, the BMAC and Shahr-i-Shokta ASI admixed samples tell us that an ASI population existed and that there were samples in the BMAC that were admixed between a genepool rest of the BMAC /S-i-S and ASI.

It doesn't tell us that the IVC people, actually within the IVC, were variable and "existed on a cline" between S-i-S 3 and S-i-S 2, and in fact that assumption is based on nothing at all*. In fact, it's highly likely all the Indus Diaspora samples have some level of local admixture.

It doensn't tell us that the IVC and post IVC people had enough BMAC/Iran related ancestry to cover present day and Iron Age Swat populations without further migration from that direction.

What will tell us the answer to both those questions, is the adna from Rakhigarri. If they turn out to be very close to "ASI" (basically like S-i-S 3) then "ANI" will not model with a single pulse in of steppe ancestry, but require further migration from direction of BMAC/Iran.

You're a fairly clever chap and this stuff should be obvious to you. It will be peculiar if it's not understandable. I think you've gone into rote "defender mode" here and that may be constituting a block to comprehension.

I don't know what the samples at Rakhigarri will show, but it is obviously they will be what will be used to model the IVC population itself in further models, rather than 3-6 probably admixed samples from outside the IVC. To do otherwise would be ludricrous, akin to using admixed LNBA samples from Europe to model Steppe_EMBA, in preference to using actual samples from that location. If they happen to be like the Irula, then simple 2-way models with Steppe_MLBA and IVC will not work for Swat, or for any South Asian population.

No one knows what they will be like at the moment. My main point here is that the assumption of Narasimhan's preprint, that Indus_Diaspora is a reasonable representation of the IVC not locally admixed people, is a dubious one.

*I myself have actually indulged in this speculation in the past that the IVC had a structured population with more and less ASI heavy people within it.

But this is a fanciful backprojecting of present day Indian population structure, when the variable ASI in these populations can be more easily explained by variable admixed with locals *once* they migrated out of the IVC.

Davidski said...


I myself have actually indulged in this speculation in the past that the IVC had a structured population with more and less ASI heavy people within it.

Well that's the sensible assumption considering the geography and varied nature of the IVC.

You'll understand all of this much better when the South Asian papers and new data come out, hopefully soon.

Matt said...

I think it's not a warranted or considerd assumption that either these subjects were unadmixed migrants, or that the IVC cultural area was necessarily more genetically varied than other large horizons like Andronovo, the BMAC sphere, etc. of comparable area.

I think you will likely understand this better when the Rakhigarri data is published.

@Arza, sure there are unknowns and theoretical populations. I guess I use the term fantasy point, because I'm not sure the basis for the cline they present is.

For instance, in "The State of The Field" (in which they note further sample outliers and Shahr-i-Sohkta out of interest), Reich presents a graphic suggesting that SwatIA samples formed a cline between close to an absence of ASI ancestry to close to a point on their India Periphery cline - (see:

(Note in any case this is not a cline between the Steppe_MLBA samples at Dashti_Kozy and Indus_Diaspora average, which was presented in the previous preprint's main text, so provides no support for such a model).

However, whenever I run all the available samples in G25 through analysis, this is not really evident at all:

Instead the samples from Swat form a triangular cloud varying in steppe+BMAC related and ASI ancestry, and the best fitting single axis of variation in these samples varies along a point close to SIS_BA3 and close to the BMAC samples.

Even if you limit to pre-800BCE Swat samples (probably a good idea) and limit outgroups, the cline Reich describes above still is not visible -

Ric Hern said...

Something interesting.

vahaduo said...

@ Davidski

Yesterday I've added East Asian PCA in both versions.

By the way, is there any way to add a feature to the mixture modeling tool which limits the number of reference populations to the best 5 no matter how many populations and/or individuals are pasted on the source page?

I'll try to create such algorithm, but the task is non-trivial.

Davidski said...

Yes, I saw the East Asian PCA. Thanks.