In recent years you may have read academic papers, books and press articles claiming that the Early Bronze Age Yamnaya culture of the Pontic-Caspian steppe was founded by migrants from the Caucasus, Mesopotamia or even Central Asia.
Of course, none of this is true.
The Yamnaya herders and closely related groups, such as the people associated with the Corded Ware culture, expanded from the steppe between the Black and Caspian seas, and, thanks to ancient DNA, it's now certain that they were overwhelmingly derived from a population that had existed in this region since at least the mid-5th millennium BCE (see here).
So rather than being culturally advanced colonists from some Near Eastern civilization, the ancestors of the Yamnaya herders were a relatively primitive local people who still largely relied on hunting and fishing for their subsistence. They also sometimes buried their dead with flint blades and adzes, but hardly ever with metal objects, despite living in the Eneolithic epoch or the Copper Age.
As far as I know, this group doesn't have a specific name. But in recent scientific literature it's referred to as Eneolithic steppe, so let's use that.
It's not yet clear how the Yamnaya people became pastoralists. Some scholars believe that they were basically an offshoot of the cattle herding Maykop culture of the North Caucasus. However, the obvious problem with this idea is that the Yamnaya and Maykop populations probably didn't share any recent ancestry. In fact, ancient DNA shows that the former wasn't derived from the latter in any important or even discernible way (see here).
On the other hand, Yamnaya samples do harbor a subtle signal of recent gene flow from the west that appears to be most closely associated with Middle to Late Neolithic European agropastoralists (see here). Therefore, it's possible that herding was adopted by the ancestors of the Yamnaya people as a result of their sporadic contacts with populations living on the western edge of the Pontic-Caspian steppe.
Eneolithic steppe is currently represented by just three samples in the ancient DNA record, and all of these individuals are from sites on the North Caucasus Piedmont steppe (two from Progress 2 and one from Vonyuchka 1).
As a result, it might be tempting to argue that cultural, if not genetic, impulses from the Caucasus did play an important role in the formation of the Yamnaya and related peoples. However, it's important to note that the North Caucasus Piedmont steppe was the southern periphery of Eneolithic steppe territory.
Below is a map of Eneolithic steppe burial sites featured in recent scientific literature. It's based on data from Gresky et al. 2016, a paper that focused on a specific and complex type of cranial surgery or trepanation often practiced by groups associated with this archeological culture (see here).
Incredibly, one of the skeletons from Vertoletnoe pole has been radiocarbon dated to the mid-6th millennium BCE. My suspicion, however, is that this result was blown out by the so called reservoir effect (see here). In any case, the academic consensus seems to be that the roots of Eneolithic steppe should be sought in the Lower Don region, rather than in the Caucasus foothills (see page 36 here).
Considering that nine Eneolithic steppe skulls from the Lower Don were analyzed by Gresky et al., I'd say it's only a matter of time before we see the publication of genome-wide data for at least of couple of these samples. Indeed, the paper's lead author is from the Deutsches Archäologisches Institut, which is currently involved in a major archaeogenetic project on the ancient Caucasus and surrounds. Unfortunately, the study is scheduled to be completed in about four years (see here).
But whatever happens, the story of Eneolithic steppe deserves to be investigated in as much detail as possible, because it obviously had a profound impact on Europe and its people.
In my estimation, at least a third of the ancestry of present-day Northern Europeans, all the way from Ireland to the Ural Mountains in Russia, is ultimately derived from Eneolithic steppe groups. It's also possible that R1a-M417 and R1b-L51, the two most frequent Y-chromosome haplogroups in European males today, derive from a couple of Eneolithic steppe founders. If so, that's a very impressive effort for such an obscure archeological culture from what is generally regarded as a peripheral part of Europe.
See also...
1,260 comments:
«Oldest ‹Older 1201 – 1260 of 1260re: Italy_MBA:S18130.E1.L1_scaled
Interesting, this individual is still closest to Sardinians, with only 5.2% Steppe.
@Ezio Auditore
Where exactly is this from?
@Rob,
"In Central Europe Europe there was gradual admixture too (2800-2200 BC); with a final result of ~ 50/50 % mix of “ early CWC” with GAC/ Bernberg / late TRB types"
No, Corded Ware's was not a gradual thing. They spread from Netherlands to Russia in only about 3-4 generations.
Quick migrations like what Corded Ware did, are why IE languages became so widespread. Yamnaya did a rapid migration across the European Steppe. Bell beaker did a rapid migration in Western Europe. Andronvoo did the same kind of thing in Central Asia.
This makes more sense as an explantion for why IE languages are widespread than gradual, councidental, accidental spread of a language. Basically, 3th-2nd millennium BC, IE speaking people were moving all over the place, mystery solved for origin of IE langauge family.
@vAsiSTha
Again
"1. Minimal steppe ancestry in BMAC/Turan till 1500bce."
That's right. BMAC is not Aryan.
"2. Upto 20% Steppe ancestry seen in Swat iron age, but no population turnover."
Because it's not the Vedic Aryans, it's Dasyu after the Aryans conquered them. The Aryans were cremated, and the SPGT was buried in the ground.
"3. Steppe Migration was female mediated, overturning earlier hypothesis of a male mediated migration/invasion."
Gradual migration was male, conquest, but they exchanged women with the local population. The Aryans were not buried in the SPGT LBA because the Vedic Aryans were cremated, this is known with absolute certainty and accuracy. But they appear when the IA Aryans move on to Buddhism.
"1. No BMAC ancestry in SPGT: This is blatantly false, E1b ydna also found in SPGT."
E1b doesn't mean it's BMAC.
"2. Steppe LBA has no role to play in migration."
There is no such statement.
"3. western/central steppe people were those who migrated"
Absolutely the right statement, CWC > Sintashta > Petrovka > Seven rivers > HinduKush.
"4. Sakas/scythians had no role to play"
For the time in question, it's a perfectly correct statement.
Target: Italy_MBA:S18130.E1.L1
Distance: 1.0769% / 0.01076873
29.6 Bell_Beaker_ITA
24.2 UKR_N_o
19.2 Bell_Beaker_HUN_EBA_o
16.0 ITA_Remedello_BA
7.2 Anatolia_Barcin_N
3.6 Bell_Beaker_ITA_o
0.2 CZE_N
Other BBC, CWC, Unetice, Sintashta, WHG, EHG, CHG, Yamnaya = 0
Without UKR_N_o
Target: Italy_MBA:S18130.E1.L1
Distance: 1.1320% / 0.01132034
31.6 Bell_Beaker_ITA
21.2 Anatolia_Barcin_N
16.0 CZE_N
13.6 ITA_Remedello_BA
7.8 Bell_Beaker_Iberia
6.2 Bell_Beaker_ITA_o
3.6 Bell_Beaker_HUN_EBA_o
BBC + Remedello + farmers mainly
@ Archie the run you put is illogical, CHG in Central Asia as well ANF , that Bustan individuel has Steppe Maykopish ancestry, which was conveniently left out in your run as well Central Asian farmer, as they do exist in Central Asia lol . His Mtdna is like Central Asian farmers, which clearly indicates his mother was a local lady or mixed and his father from Indie. E1a/b exist in Central Asian farmers who else is going to bring it to SPGT which is on border with Central Asia. Central Asian and Steppe markers/input are absent in the Indus people but show up in the SGPT. Also Dasyu were also Aryans its just Iranic Aryans.
@Vashishta agree with your points but I have to also agree with David for these modern Indien demographic , it has to be male mediated R1a, also its common in the patrician class.
"The Aryans were not buried in the SPGT LBA because the Vedic Aryans were cremated, this is known with absolute certainty and accuracy. But they appear when the IA Aryans move on to Buddhism."
Archi, stop saying things like this. You don't know with any certainty, and you're making assumptions, and where the hell do you come up with the idea that Buddhism led to burials? If that was so, South Asia would be full of graveyards and human remains.
We don't know who the Dasyuv were, or what their funerary customs were.
@ Sam
It’s true, the first arrival of cwc reaches all the way to Atlantic rapidly; but this was groups of men Venturing far and wide ; more or less isolated
This was then supplemented by further migrations of a longer period
It took a while for the immigrants to intermix with natives
A more definitive shift occurred with BB ; and finally; with the Bronze Age when the very last of the old megalithic groups lost their cultural cohesion
To understand the final impact; one would need to analyse the later generations in very detailed way for each region
Nobody has done that; and only a handful of people in the world can do that
@Archi
"BBC + Remedello + farmers mainly"
True, I deleted my premature comment about Etruscans. But your model is a nonsensical overfit. More sensible would be:
Target: Italy_MBA:S18130.E1.L1_scaled
Distance: 2.5768% / 0.02576799
50.0 Bell_Beaker_ITA_o
29.6 Bell_Beaker_ITA
16.2 ITA_Remedello_BA
4.2 Anatolia_Barcin_N
Still a bit overfitted with that Barcin_N, but it shows that little else is needed besides Bell_Beaker_ITA/_ITA_o + Remedello. And quite remarkable that in Northern Italy at 1300 BC there were still people running around like this one.
@kouros
" agree with your points but I have to also agree with David for these modern Indien demographic , it has to be male mediated R1a, also its common in the patrician class."
Even I agree that R1a in modern indians is mediated through males. That is the only way it can propagate after all.
before Swat samples, the hypothesis was that a large no of R1a males from steppe came into india in the bronze age. The new data contradicts this hypothesis. Davidski cannot digest this new data and claims that SPGT is not a representative sample, which is just his way to cope.
With all the data we have so far, it is clear that R1a was scant in NW india in the bronze age and expanded much later. When new aDNA comes out, we can change this theory.
Also consider this, from Iain Mathieson's blog http://mathii.github.io/2019/10/12/the-spread-of-the-european-lactase-persistence-allele
"Looking at the data from Narasimhan et al. 2019 it seems that the allele appeared in South Asia much later than in Europe. Sampling is a bit limited, but its earliest appearance in data is around 2000 BP (ie 0ce, emphasis mine) in Butkara and Swat. The present-day frequency of ~25% in some South Asian populations (e.g. 1000 Genomes PJL) suggests strong, recent selection perhaps similar to what we see in Iberia. In any case seems like selection on the alelle was in parallel in Europe and South Asia. It is probably not the case that the Steppe ancestry in South Asia was from a population that already had a high frequency of the allele."
Its he same sort of selection that caused R1a to expand in terms of population %
@vAsiSTha
Steppe admixture in South Asia was male mediated. This has been obvious for a long time and ancient DNA from India will show it clearly.
@davidski
"will show it...." keep coping
@vAsiSTha
Keep praying, if you're religious, because many more samples from South Asia are on the way.
Haha.
@archi
SPGT does not have ancestry from western/central steppe, but eastern steppe/IAMC. SPGT has more ancestry from BMAC than the steppe.
qpAdm
Target - SPGT (78 samples)
InPe (10 samples, leaving I8726) - 0.497 +- 2%
Kazakhstan_MLBA_Zevakinskiy - 0.229 +- 1.3%
Uzbekistan_BA_Dzharkutan1 - 0.274 +- 1.8%
chisq 7.6
tailprob: 0.3689
output file https://pastebin.com/3bdhbRkx
@Aniasi
You're wrong, Dacia was never an Aryan, it was Daha who lived on the site of BMAC became one of the Iranian tribes. Dasa/Daha means sedentary/local. It's a local BMAC population assimilated by the Iranians. Dasa and Dasyu are different words. Dasya were black people with whom the Aryans fought and won. Dasa was just a slave, a servant.
So your text is wrong and completely illogical, you just don't know anything.
Stop writing that you know nothing, you just shame yourself by showing that you know nothing. For a long time, science has firmly established that the Vedic Aryans were cremated, without exception, as written in the Vedas and all sources. Anyone who knows even a little bit about Buddhism knows that it does not generally recognize cremation, although it does not prohibit it, there are many rites in it. It was with the transition to Buddhism that the Aryans began to bury the dead rather than burn them, which has long been known to science. But you don't know how to scream.
Learn!
Simon_W said...
" @Archi
"BBC + Remedello + farmers mainly"
True, I deleted my premature comment about Etruscans. But your model is a nonsensical overfit. More sensible would be:
Target: Italy_MBA:S18130.E1.L1_scaled
Distance: 2.5768% / 0.02576799
50.0 Bell_Beaker_ITA_o
29.6 Bell_Beaker_ITA
16.2 ITA_Remedello_BA
4.2 Anatolia_Barcin_N
"
I show that apart from these populations there was some eastern population of farmers connected with the Czech Republic or Ukraine, UKR_N_o is a farmer with a small admixture of EHG/WHG. But the basis in it is the BBC, and there is no Yamnaya.
One thing for sure, is Steppe migrations into India weren't as extreme as Steppe migrations in Corded Ware & Bell Beaker. Those expansions created almost complete Y DNA replacement of Neolithic farmers and big imprint of steppe ancestry.
This obviously didn't happen in India because we see lots of Iran farmer Y DNA in India and much less Steppe ancestry than in Europe.
It's difficult to estimate the entire sequence sex-bias from present day y-chromosomes, which only really preserve the latest "layer" of interaction.
Like, if you have a scenario where initially you have female biased migration, then a male biased migration, you'll only really see the latest migration in the y, though the autosome would be balanced. And also an excess of y and mtdna relative to autosomal dna.
So it is hard to know for sure what exact population history is being concealed. I still say that for the best possible estimate of what is the case in modern genomes they should simply take high coverage modern genomes and estimate X:autosome ratios using modern proxies.
Regarding the Narasimhan 2019 paper's assertions that Indus_Periphery both lacks Anatolian related ancestry, and further that SPGT are then a complex mix of the most and least West Eurasian Indus_Periphery and steppe samples in very complex and specific ways, with no need for any contact with BMAC, these seem very questionable to me.
Certainly that model where Indus_Periphery excludes ancestry from Turan_EN seems pretty inconsistent with what I see in G25 as the best fitting admixture cline for Indus_Periphery, which points much more at the likes of Shahr_I_Sokhta_BA1 than it does Ganj_Dareh+WSHG. I don't think G25's got it that wrong.
@Vashishta , excellente effort, this consistent with Bustan outlier. All French who have wrote on Gandhara Grave have said it melange of Indo Aryan, BMAC and IVC elements, the results depict this.
@ Matt, using Shahr I Sukteh samples in SPGT make no sense, what was Narsimhan thinking its Eneolithic culture of Jiroft. Quele Dommage! SPGT located near Pamir, those demographics have almost half Central Asian Farmer ancestry.
@vAsiSTha
You have already proved many times that you don't understand how to model at all, you don't know the principles of modeling, and you don't know the principles of proof, but you know the principles of deception that it is enough to print a single shot excluding inconvenient sources from the analysis. Taking all SPGTs is like measuring the average temperature in a hospital, excluding inconvenient sources is just a trick, not understanding that the eastern steppe is happening from the west is ridiculous.
Target Distance RUS_Sintashta_MLBA IRN_Shahr_I_Sokhta_BA2 UZB_Sappali_Tepe_BA RUS_Afanasievo UZB_Dzharkutan1_BA KAZ_Zevakinskiy_BA KAZ_Zevakinskiy_MLBA
PAK_Saidu_Sharif_H:I6893 0.01363483 37.8 51.0 11.2 0.0 0.0 0.0 0.0
PAK_Udegram_MA_Ghaznavid:I2959 0.01279876 36.2 29.2 23.2 0.0 7.2 0.2 4.0
PAK_Singoor_MA:I1805 0.00941797 34.4 30.6 27.8 6.4 0.8 0.0 0.0
PAK_Katelai_IA:I12475 0.02007767 26.0 42.2 0.0 0.0 30.2 0.0 1.6
PAK_Loebanr_IA_o:I12138 0.01291183 25.0 18.8 19.4 12.4 16.8 0.0 7.6
PAK_Barikot_H:I6890 0.00707281 24.0 46.2 14.4 0.0 14.0 1.4 0.0
PAK_Udegram_MA_Ghaznavid:I7716 0.00842732 23.2 26.4 35.4 7.6 7.4 0.0 0.0
PAK_Barikot_IA:I6548 0.01865257 23.0 46.0 11.0 8.0 9.2 0.0 2.8
PAK_Katelai_IA:I12477 0.01505069 23.0 39.8 19.6 5.8 11.8 0.0 0.0
PAK_Barikot_H:I7714 0.01692979 21.2 55.8 9.8 9.4 3.8 0.0 0.0
PAK_Saidu_Sharif_H:I7721 0.01301815 21.0 60.2 17.4 1.4 0.0 0.0 0.0
PAK_Barikot_H:I6889 0.01309472 20.0 40.6 0.0 0.0 39.4 0.0 0.0
PAK_Katelai_IA:I12472 0.02253709 20.0 67.0 0.0 0.0 13.0 0.0 0.0
PAK_Saidu_Sharif_H:I7719 0.00843783 20.0 49.0 18.0 0.0 7.4 0.0 5.6
PAK_Katelai_IA:I12141 0.01128986 19.4 43.8 3.0 5.4 28.4 0.0 0.0
PAK_Udegram_IA:I1985 0.01332081 19.4 52.4 6.2 0.0 22.0 0.0 0.0
PAK_Loebanr_IA:I8998 0.00652768 19.2 48.6 22.8 0.8 8.6 0.0 0.0
PAK_Udegram_IA:I8190 0.01087046 18.4 48.0 20.0 1.6 4.2 0.0 7.8
PAK_Loebanr_IA:I12458 0.00986782 18.0 55.0 17.4 2.0 7.6 0.0 0.0
PAK_Udegram_IA:I6897 0.00859619 17.6 40.2 17.2 11.2 13.2 0.0 0.6
PAK_Saidu_Sharif_H:I6896 0.01096035 17.4 53.0 17.2 2.8 0.0 0.0 9.6
PAK_Loebanr_IA:I10001 0.00873284 17.2 42.8 36.2 3.8 0.0 0.0 0.0
PAK_Loebanr_IA:I12982 0.00903310 17.2 40.0 24.6 4.2 14.0 0.0 0.0
PAK_Loebanr_IA:I10974 0.01129264 16.6 49.2 18.8 0.0 15.4 0.0 0.0
PAK_Aligrama_H:I8218 0.01257509 16.2 53.4 19.8 10.6 0.0 0.0 0.0
PAK_Saidu_Sharif_H:I7720 0.00779430 16.2 54.6 24.0 5.2 0.0 0.0 0.0
PAK_Loebanr_IA:I10000 0.00813485 15.6 61.8 0.0 5.2 17.4 0.0 0.0
PAK_Loebanr_IA:I6553 0.01427142 15.4 49.6 19.8 11.4 3.4 0.0 0.4
PAK_Aligrama_H:I8245 0.01377000 15.2 65.0 8.2 5.4 6.2 0.0 0.0
PAK_Butkara_H:I6549 0.01423760 15.2 64.8 7.8 1.6 10.6 0.0 0.0
PAK_Loebanr_IA:I12983 0.00895534 15.2 44.4 12.0 12.6 15.8 0.0 0.0
PAK_Loebanr_IA:I12456 0.01296824 15.0 39.2 26.0 5.4 14.4 0.0 0.0
PAK_Aligrama_H:I8220 0.01418860 14.4 46.2 15.0 0.0 17.6 0.0 6.8
PAK_Udegram_IA:I6901 0.01794641 14.4 39.8 31.8 14.0 0.0 0.0 0.0
PAK_Katelai_IA:I12462 0.00623070 14.2 35.2 13.0 7.0 30.0 0.0 0.6
PAK_Loebanr_IA:I13228 0.01663058 14.2 52.8 15.4 0.0 17.6 0.0 0.0
PAK_Loebanr_IA:I13227 0.01382944 14.0 53.4 2.2 9.2 21.2 0.0 0.0
PAK_Loebanr_IA:I12984 0.01573615 13.8 36.4 13.8 7.6 28.4 0.0 0.0
PAK_Loebanr_IA:I12987 0.01122349 13.8 50.8 7.6 1.8 26.0 0.0 0.0
PAK_Loebanr_IA:I6292 0.01397785 13.8 36.6 12.8 14.8 22.0 0.0 0.0
PAK_Loebanr_IA:I12137 0.01373400 13.6 48.0 0.8 9.6 28.0 0.0 0.0
PAK_Katelai_IA:I10523 0.01296151 13.4 46.2 31.0 5.6 3.6 0.0 0.2
PAK_Loebanr_IA:I13222 0.01465388 13.4 42.8 1.6 2.8 39.4 0.0 0.0
PAK_Loebanr_IA:I13223 0.01029478 13.4 47.6 26.6 4.8 7.6 0.0 0.0
PAK_Loebanr_IA:I13226 0.01206551 13.4 53.4 19.0 0.0 14.2 0.0 0.0
...
PAK_Saidu_Sharif_H_o:I7722 0.02501166 6.6 93.4 0.0 0.0 0.0 0.0 0.0
...
Average 0.01251844 11.8 48.2 16.9 7.8 12.8 0.3 2.2
@Matt,
"It's difficult to estimate the entire sequence sex-bias from present day y-chromosomes, which only really preserve the latest "layer" of interaction."
Good point. modern France & Spain have around 50-605 R1b, but in the Bronze age they had basically 100%.
Y DNA R1a Z93 may have been more popular in India 2,000 years ago or whatever. There may have been many groups in India with more Steppe admix than today. But, I do think modern Indian DNA does leave out the possibility of truly 'massive' migrations and Y DNA turnovers like what Bell Beaker and Corded Ware did in Europe. Admixture may have been sex bias but there weren't enough Andronovo invaders to make a huge Y DNA turnover.
@ Avista
“ SPGT does not have ancestry from western/central steppe, but eastern steppe/IAMC”
Zevakinsky = Andronovo + residual Dali & afansievo
@Samuel Andrews
India has always had a tribal structure that has been replaced by a caste. There was never a single people there, as there was no single language. All the units practiced different traditions, and funeral rites were different. So, we can't talk about replacing the Y-haplogroups there.
Good point. That's always important to keep in mind with India.
New paper on mesolithic to neolithic Sicily: https://www.biorxiv.org/content/10.1101/2020.03.11.986158v1
Nothing too unusual it seems, early mesolithic Sicilians seem to be, just like the recent central Italian HGs, to be relatively drifted and pure while later ones show ancestry related to EHG and Pinarbasi HG.
I am generally convinced that every old (before trade time) epochal change has somehow been brought about by some kind of migration.
R1b/WHG brought the Final Paleolithic to Europe.
R1a/EHG brought the Mesolithic to Europe.
Q brought the Pottery Neolithic to Europe.
Here in Sicily one can see the migration from south-East Europe bringing the Late Mesolithic, from the Balkan Cardial Ceramics bringing the Early Neolithic.
@matt and @kouros
Shahr-i-sokhtaBA1 cannot be ruled out as source instead of BMAc in SPGT. It has a uniques ancestry compared to BMAC - higher wshg, less anatolian. Which might indeed be required for certain SPGT models.
One thing is for sure. SPGT is not a simplistic indus_periphery + Sintashta mixture.
Another thing is also for sure - the native population of Swat in 1500bce was closer to I8728 than I8726 in terms of AHG ancestry. This can be easily seen from the PCA plots and the Clines that are made on those plots.
i8726 is one of the oldest inpe samples and in the next 1500 years the region saw increasing AHG ancestry from the south.
@archi
The tribal nature of India in bronze age was exactly same as tribal nature elsewhere. There were tribes at war with each other.
For SPGT, g25 vahaduo consistently selects Molaly_lba over any andronovo source, take your pick and molaly will be picked over that.
@kouros
Zevakinskiy MLBA is dated 1608-1443bce. so it is perfect. This ancestry existed in the eastern steppe till 1100bce as zevakinskiy LBA samples are all similar to mlba.
The youngest Zev_lba sample is dated 950bce and is much more east asian shifted than others, that one is obviously not a source.
As far as tibetan input is concerned, its a stretch and a forced fit. There is 0 arch evidence of such a movement. A steppe source for that east asian component is the most parsimonious explanation. There is 0 presence of modern tibetan y dna or mtdna in SPGT.
@vAsiSTha, yeah, I would estimate agreement with most/all of your comment that was @me.
When I extrapolate along the Indus_Periphery cline using my standard methods*, the ghost seems pretty specifically like Shahr-i-sokhtaBA1 or Gonur1, and not really like a point on a cline between Ganj_Dareh and WSHG.
Visually: https://imgur.com/a/WgAbEh8
1) Two fits with Vahaduo for Shahr_I_Sohkta_BA1 with Anatolia, GD and WSHG and the most ASI sample on the Indus_Periphery cline (I1828, as you note). The fit with I1828 is better.
2) PCA showing my extrapolation of Indus_Periphery cline (green dot) and the above models (red and blue dot). The cline extension very specifically overlaps with Gonur1 and Shahr_I_Sohkta_BA1, and not with a point on the Ganj_Dareh->WSHG cline.
So I don't know what Narasimhan's paper must have done to get fits with SS1 and Gonur1 being less preferred than GD+WSHG. That seems inconsistent with what G25 finds.
(ScaledG25 Indus_Periphery_Extrapolation,0.067846,0.068325,-0.14305,0.031975,-0.12358,0.031041,0.015521,-0.0038512,-0.060691,-0.03916,0.012186,0.0042098,0.0023807,-0.014312,0.016129,0.024705,-0.013088,0.0064607,0.0031161,-0.029674,0.0089136,-0.028938,0.0037957,-0.029548,0.013302)
*Identify largest PC of variation within set of samples, use variation on that PC to fit multivariate equation and estimate beyond limits of samples.
@Vashishta, pointed noted, I was confusing dates, it strongly seems they mixed with WSHG related populace whether Botai or Maykop, Zevakinsky MLBA has both and apres /after with Central Asian farmers THEN with late IVC populace.
Can you try Loebanr outlier? He has low ENA so would need more Androvon source , Zevakinsky BA or Dali MLBA. Though your QpAdm has great resulte, it looks like the R1a band arrive arround time of SPGT or bit later, not earlier (1000 BCE).
@vAsiSTha, on the point about the endpoint of Swat_IA cline, I almost agree, but not totally. I agree that clines seem to suggest that the ASI related endpoint of the Swat_IA samples cline is closer to I8728 than I8726.
That said, Narasimhan's paper Fig4B seems to place it about 60% along the distance from I8726->I8728... (like I11459). The method far as the above method for clinal G25 modelling suggests it would have to be at least 70% along that distance, though could be 75%. So I think Narasimhan's paper is perhaps lowballing it bit.
The weirder bit about the model is to propose that Loebanr_IA_o is a very specific mix of the most I8726 and Steppe_MLBA, when even their lowballing cline suggests people in Swat pre_IA/MLBA must have been more like I8728.
(G25 Scaled: Steppe_related_Swat_IA_Ancestor,0.10289586,0.074141432,-0.036794652,0.042597728,-0.05825432,0.023013168,0.0049298308,-0.009324096,-0.058039152,-0.048991332,-0.0001123977,-0.0048768332,-0.001900358,-0.0210507696,0.0258824056,0.0291413108,-0.0013911984,0.0036479804,-0.0004487004,-0.0181001488,-0.001018866,-0.0091920516,0.0039271752,-0.0046973564,0.003298228
ASI_related_Swat_IA_Ancestor,0.04030067,-0.057706896,-0.168507694,0.088410416,-0.09849604,0.058519096,0.0017587426,0.009375088,0.013079056,0.011636346,-0.009351368,0.0057997346,-0.002574051,0.0015186888,0.0004453532,-0.0015835474,-0.0013053348,-0.0027654362,0.0002971712,-0.0037789136,-0.001124827,-0.0031880902,0.0003773644,-0.0063995358,0.004106816).
Anyway, I hope the most important thing is that this paper is tested to its limits; they are proposing fairly complex models between closely related populations.
@matt, thanks for the coordinates
I love the swat native ghost. Its perfect. The cline from swat native ghost to swat steppe ghost is also perfect.
But, The swat steppe ghost should be a bit closer to SPGT on that cline, in my opinion, requiring slightly more Okunevo type(wshg/botai+eshg) ancestry. Here's what i mean https://imgur.com/gallery/1L7VbwD
As far as SiSBA1&2 is concerned, GD+WSHG is too simplistic as minor Anatolia_N was present in the region by 2500bce. I reckon pops like Parkhai_en played a role in the formation of indus periphery samples, if not SiSBA1. for all we know these type of ancestries may be representative of NW india around 4000bce. Why is this important to you? do you feel the anatolian hypothesis is still in play?
@Kouros, I could have a quick go if you really want, but I would just want to caution first about why that would be harder and/or more unreliable.
Visually, check out: https://imgur.com/a/QJUclky
Both the "Indus_Periphery" and Swat_IA set span quite a lot of space in the PCA relative to the positions of the estimated extrapolations. Most of the Swat_IA cluster around a particular point, but there are quite a few samples particularly at Aligrama that represent a less plausibly admixed population (you could say, "stragglers", I guess).
This means I can be fairly confident that they when I extrapolate, I'm not just amplifying noise in the G25.
This is because they probably represent relatively recently admixed populations that haven't had much time to go panmictic and lose the last traces of their pre-admix structure.
(You might notice the distance of the estimated Steppe_related_Swat_IA ancestor to Loebanr_outlier. IMO this shows that Loebanr_outlier is probably not quite right for explaining the entire cline, and how using PCA produces more robust estimates of the ancestor than just extrapolating from outliers; which is close to what Narasimhan's approach is, treating the Loebanr_o as one end of a cline).
The Rors are a lot more of a tight, panmictic population without much variance. We can extrapolate from their internal cline for sure, but I'm not sure that we would actually get a realistic estimate of two ancestral populations, rather than things that are affected by noise and assortative reproduction within the population.
Btw, note that when I've constructed my clines, I've just used are just the IA Swat/PAK samples. If I'd included the historical period (H), the population means for those are quite obviously shifted relative to the IA to positions that would need some enrichment of both AASI and steppe related ancestry, and there are outliers that quite obviously have a different combination of AASI and steppe related ancestries than can fit with the IA cline (including the Ror-like I6893, which is an outlier despite Harvard not recognising it as one).
See: https://imgur.com/a/70JhAke.
This seems pretty plausibly to me due to the more "cosmopolitan" Buddhist-historical period pulling in outliers from further afield (huge scale, cross ethnic religious movement), and a feeding in of both ASI enriched ancestry from the south and more steppe related ancestry from the north, to sites of pilgrimage and high status burial that were sampled.
(This has some analogies to the "tilting cline" in Europe of NW Europe->MN_chl farmer tilting to a cline of NW Europe->East Med like. Or in the case of Eastern Europe, the more dramatic tilt of CW->Balkans_Chl to Baltic_BA->East_Med. Though obvs shift between IA->H in PAK seems less dramatic change and more subtle. )
Matt
Could you also use a subset of the indus periphery 11 samples and average out the coordinates to check which subset falls closest to your swat native ghost? does leaving out i8726 and averaging the other 10 work? I will use that subset for qpAdm.
@kouros ill post qpAdm for SPGT_o soon.
@vAsiSTha, cheers. Well, it's certainly always possible that these ghosts could be slightly off in some ways from over-representing some set of outliers. I can only really say it's my best estimate from just using the method and an eye of how far clines can plausibly be extended without postulating essentially impossible combinations of ancestry. Definitely you could plausibly move up or down these clines to some degree.
Re; importance to me, I think I just find it annoying that to me they seem to have pushed their data beyond where it will go, and perhaps I'm arrogant enough to think I can spot patterns that they've missed.
I do get the impression that Narasimhan paper appears to have intentionally tried hard to exorcise the Anatolian hypothesis by genetic means, rather than simply been neutral on it (which fits with Nick Patterson's and David Reich's preferred hypothesis I think, and so probably influences the entire lab). They have maybe pushed their data in the wrong directions to exclude any post-early neolithic connections between "Gedrosia"+"Turan" and South Asia to do so, when in reality it seems to me likely there probably was some bidirectional admixture at some levels in all periods. I suspect it won't really stand up that robustly over time and re-analysis. They've also seem to have somewhat obscured parts of their method in the paper and key stats etc, which seems irritating.
That said, genetics all aside, the linguistic big problem with the Anatolian hypothesis has always been and probably will always be the wheel words problem.
The only way to explain shared wheel terms under the Anatolian hypothesis is that diverging IE varieties preserved root terms for "roll" etc that forms the basis for the wheel words, and then systematically derived the same wheel words from those by homoplasy (parallel process of innovation from the same roots).
But even that's apparently tricky to reconcile with the etymology of the words which suggests that the wheel words formed before linguistic innovations in IE formed. This also apparently excludes borrowing from one variety of IE into others, as an explanation (since the linguistic history of wheel words is that it seems that they formed before sound changes and grammatical changes in IE varieties).
(I think there have been some suggestions that the way around this is that there's a lot more re-naturalisation (not the right technical term) in borrowed words than we think, which is to say that once borrowed, speakers can tell that a word "isn't quite right" and then adjust it in ways that become indistinguishable from as if it had been in that language all along. This doesn't seem to fit that well with cross-linguistic evidence of borrowing as far as I know though.)
Most of the other paleo-lexical and distribution arguments against Anatolian hypothesis seem weaker to me* and not insurmountable, if challenging. But the wheel words are the problem. I don't think Anatolian can be considered properly in play unless this is ever resolved. Although later homelands that are not the steppe may be (depending on how much we want to allow for language to become separate from ancestry).
I'll have a look at your other comment later today.
*Relative paucity of early attested IE in ME and Southern European writing not necessarily an insurmountable problem, though it is difficult. Horse words maybe or maybe not a problem - it's dumb to suggest horses need to be domesticated or at high frequency before people will have a word for them, but it's a decent criticism that vocabulary will not be preserved if we thought that IE speakers were spreading into horse free regions millennia before they were brought there by domestication.
qpAdm models for SPGT_o
SPGT - 0.506 +- 0.083
Dali_MLBA - 0.30 +- 0.037
Dzharkutan1 - 0.194 +- 0.067
tailprob: 0.75
output file https://pastebin.com/zfx8E4gH
SPGT - 0.664 +- 0.07
Sintashta_MLBA - 0.255 +- 0.03
Dzharkutan1 - 0.08 +- 0.06
tailprob: 0.81
output file https://pastebin.com/QiwVwP37
SPGT - 0.635 +- 0.07
Kashkarchi_BA - 0.285 +- 0.035
Dzharkutan1 - 0.08 +- 0.06
tailprob: 0.865
output file: https://pastebin.com/Z4mcvCFs
I prefer Kashkarchi as it is more proximal in space and time. The excess steppe ancestry is female mediated due to T1a and H2a2 mtdna markers.
@matt im open to any model which allows vedic culture to be present in NW india before 2000bce when the now dried up ghagghar was mighty, glacial fed and perennial.
The IAMC route has been operational atleast since 3000bce when grains and turan ancestry travels north to the dzhungar plains, and that is also the route taken in the bronze age to come south from east steppe to south asia. We should have seen a tocharian related language in south asia, but that is not at all the case.
As far as Narsimhan's paper is concerned, I agree it is very shoddy work.
Matt said:
"That said, genetics all aside, the linguistic big problem with the Anatolian hypothesis has always been and probably will always be the wheel words problem."
Please see below:
https://www.academia.edu/9452122/_The_Origins_of_the_Indic_Languages_the_Indo-European_model_in_Angela_Marcantonio_and_Girish_Nath_Jha_eds._Perspectives_on_the_origin_of_Indian_civilization_New_Delhi_259-287"
The words for wheel are in favor of the IIr territory as homeland. An analysis of the Rig Veda and the Avesta assuredly indicates a movement of the Iranians *away* from the Vedic people and not the other way round. Ir also shares isolgosses with Armenian and Greek that it does *not* share with IA.
Mayuresh M. Kelkar
@Kouros, although I'm not sure about the idea, here's a pair of ghost samples which can fit Rors as a 2 way model in G25 (scaled), where one of them is presumed to be like PAK_IA's main cluster:
Ror_Ancestor_Swat_IA_like,0.0686139,0.00097098,-0.1097497,0.0662602,-0.0809551,0.0418856,0.00331195,0.001100066,-0.0202721,-0.01608631,-0.00516642,0.00111724,-0.00268553,-0.00794932,0.0121765,0.0113439,-0.002992709,-0.000081378,0.000334834,-0.01101687,-0.001089736,-0.0059691,0.00198699,-0.00582553,0.00418643
Ror_Ancestor_Other_Far,0.10914,0.050478,-0.014887,0.11488,-0.02278,0.04721,-0.0029921,0.0073823,0.025015,-0.0041461,-0.0038355,-0.0012098,0.0029797,-0.019447,0.011263,0.012846,-0.00027569,0.00060147,-0.00014312,0.0013503,-0.0046177,0.0027609,0.0011247,0.012914,-0.0083188
Ror_Ancestor_Other_Closer,0.099008475,0.038101245,-0.038602675,0.10272505,-0.037323775,0.0458789,-0.0014160875,0.0058117415,0.013693225,-0.0071311525,-0.00416823,-0.00062804,0.0015633925,-0.01657258,0.011491375,0.012470475,-0.0009549448,0.000430758,0,-0.0017414925,-0.003735709,0.0005784,0.0013402725,0.0082291175,-0.0051924925
The position of these looks consistent to be with Ror fitting as both with extra IA/MLBA steppe related (mainly) and AASI related (some) ancestry related to Swat/PAK_IA main cluster. I think this probably never really represents a single population but separate trends.
Same thing for Kalash:
Kalash_Ancestor_Other,0.10899,0.079041,-0.039772,0.059605,-0.061135,0.035013,0.0053254,-0.00017494,-0.058378,-0.046682,-0.0038811,-0.0056534,-0.002753,-0.01462,0.027761,0.0033641,-0.038812,0.0079102,0.0020945,-0.018389,-0.0061892,-0.0057847,0.0037805,-0.0022475,0.0020076
(Visual position of these on PCA: https://imgur.com/a/njruukK)
@vAsiSTha, one more criticism of Narasimhan 2019's approach to the SPGT samples, and why I think it may reach the results it does.
Consider their figure S 49: https://i.imgur.com/egvjiA2.png
This is part of the argument that justify arguing that SGPT_o (Loebanr_IA_o) and the PAK_IA+H samples fall on a particular cline, which points to a source close to Steppe_MLBA.
What's an issue with that fig?
Well first, the steppe cline they describe doesn't actually itself land on Steppe_MLBA, either Steppe_MLBA_BMAC or Western_Steppe_MLBA but points closer to Iran_N related populations than a pure Steppe_MLBA source: https://i.imgur.com/lFOy2fD.png
But what's another issue with this fig?
Well, whilst they keep the Indus_Periphery cline as separate samples, they essentially roll all of the huge PAK_IA+H+Medieval dataset up into 8 points, one of which is the SGPT_o outlier, and turn all of the SGPT samples into a single datapoint. This includes rolling all of those outliers within the IA set with more AASI ancestry up into their populations.
The problem with this is that would seem to be that basically all the internal structure in the SGPT samples is averaged out.
Where I'm making ghosts that are weighted mainly by PCA that represents the contrast within the majority of IA samples with little importance on any one outlier, Narasimhan 2019's method is probably mainly weighted by contrast between one particular IA sample, I12138, and the averages of sets of the rest of the samples, which includes historical and medieval samples.
You can also see this in their Fig4A and Fig4B, which both represent the PAK sequence through averages: https://imgur.com/a/J8n7nxE
They're placing a lot of weight on the relationships *between* LoebanrIA_o and other populations, many of which are historical, while deemphasing patterns *within* the rest of the Iron Age samples.
@kouros Rors are just jatt/khatri + Kangju. Theyre one caste which has a clear saka ancestry which likely entered between 0-400ce when kushanas ruled NW india.
@matt, i hear you. The whole methodology is flawed.
1. They went looking for a common source in all the agewise groups - SPGT, SPGT_o, H, MA, moderns etc.
2. the sample size is different in all the groups. 1 Loebanr outlier should not be a critical part of your decision.
3. Who knows what other ancestries entered post IA? The buddhist site with Swat_H label clearly has a different cline than SPGT itself, for eg. we are not concerned with those ancestries when answering the PIE question. the only population that matters is SPGT while answering that question, and also the outliers from BMAC bronze age.
4. Modeling using Onge as 3rd source for moderns and making conclusions based on that is just plain stupid when you know for certain that no pure AASI population existed in the north or even the south for that matter. Especially with the piss poor p values for North indians.
@kouros
Inpe_native = I8728, I11456, I11459, I11466. This averages out to close to Matts Swat ASI ghost with around 60% I8726 + 40% Irula like ancestry validated using qpAdm.
Target:SPGT_o
Inpe_Native: 40.4 +- 4.8
Sintashta_MLBA: 38.9 +- 2.6
ShahrISokhtaBA1: 20.7 +- 5
tailprob: 0.419
InPe_Native: 38.6 +- 5
Kashkarchi_BA: 42.8 +- 3
ShahrISokhtaBA1: 18.6 +- 5
tailprob: 0.937
InPe_Native: 27.6 +-5
Dali_MLBA: 42.5 +- 3
ShahrISokhta: 29.9 +- 5
tailprob: 0.9401
Sapalli_Tepe Outlier I7493
Russia_MLBA_Sintashta: 38 +-3
Iran_C_SehGabi: 62 +-3
tailprob: 0.0001 fail due to less wshg & eshg as compared to wehg & eehg
Kazakhstan_MLBA_Zevakinskiy: 40 +-3
Iran_C_SehGabi: 60 +- 3
tailprob: 0.0544 pass
Dali_MLBA: 40 +- 3
Iran_C_SehGabi: 60 +- 3
tailprob: 0.044
@Vashishta, Merci. SIS1/Jiroft is of a different epoch , further Loebanr outlier has miniscule AASI , hence he should have even much less Indus ancestry, on PCA he shift toward Tajik, like other SPGT and Matt's SPGT Steppe ghost he has significant Maykopish ancestry, consistent with Central Steppe MLBA deme.
Target: PAK_Loebanr_IA_o:I12138
Distance: 1.8952% / 0.01895184
34.2 RUS_Krasnoyarsk_MLBA
31.4 UZB_Dzharkutan1_BA
25.4 IRN_Shahr_I_Sokhta_BA2
9.0 KAZ_Kumsay_EBA
Target: PAK_Loebanr_IA_o:I12138
Distance: 2.5827% / 0.02582732
43.6 RUS_Sintashta_MLBA
35.8 IRN_Shahr_I_Sokhta_BA2
20.6 IRN_Shahr_I_Sokhta_BA1
Well that kumsay eba is nothing but excess wshg on top of krasnoyarsk consistent with eastern steppe pops.
Krasnoyarsk itself is eastern steppe.
@vAsiSTha
Kumsay EBA is like Steppe Maykop. Krasnoyarsk MLBA is like Sintashta with a little bit of Siberian.
And, by the way, Krasnoyarsk is located in South Siberia not on any steppe.
@kouros
qpAdm for SPGT_o With dzharkutan bmac instead of SiS. no need to reject the whole thing because of SiS lol. For all we know SiS pops existed till bronze age in east iran.
SPGT_o
Inpe_Native: 39.5 +- 4.7
Kashkarchi_BA: 41.3 +- 3.2
Dzharkutan1: 19.2 +- 5.6
tailprob: 0.867
Inpe_Native: 29.2 +- 4.7
Dali_MLBA: 40 +- 3.1
Dzharkutan1: 30.8 +- 5.2
tailprob: 0.896
Inpe_Native: 41.2 +- 4.6
Sintashta_MLBA: 37.3 +- 2.8
Dzharkutan1: 21.5 +- 5.4
tailprob: 0.349
Inpe_Native: 38.4 +- 4.6
Krasnoyarsk_MLBA: 38 +- 2.9
Dzharkutan1: 23.6 +- 5
tailprob: 0.735
You really dont get tailprobs better than these, a 4th source is absolutely not needed and you will end up overfitting.
@davidski, the samples labeled krasnoyarsk are not actually from krasnoyarsk town, but a few hundred kilomteres southwest of the town in sharypovsky district. They fall neatly on the border of the region archaeologist Frachetti, who is an expert on the region, calls the eastern steppe. This region is where the Ob & Irtysh rivers end. His map can be seen here https://www.jstor.org/stable/10.1086/663692?seq=3#metadata_info_tab_contents
Technically, the Eastern Steppe is in Mongolia and Manchuria, and it's actually also known as the Mongolian-Manchurian steppe.
@vAsiSTha, that seems like a fairly sensible approach to modelling, using "InPeNative".
Q: SPGT_o is I12138 right?
In which case, can you have a go at modelling I6893 (Ror-like Saidu_Sharif_H) and I6548+I6546 (most West Eurasian shifted SGPT, from Barikot_IA) the same way? I6893 is of course later, but I'm interested to see what happens and if the G25 patterns replicate.
The cline from the most InPeNative shifted SGPT samples (I12981, I12446, I12460, I2470, I5399, I6888, I12134, I12472) to the most West Eurasian shift SGPT seems subtly different in G25 from where Loebanr_IA_o I12138 is (Loebanr_IA points to a more steppe rich place).
(Before posting this was about to go on a long rant about the shortcomings of the qpAdm modelling in Narasimhan's supplement, but won't bother for now).
@ Vashishta , Merci, SIS is similar to rest of Central Asian farmer but has far more Indus ancestry it seems, but even then SPGT and Loebanr outlier prefer BMAC individuels, from French observations, these people use similar amphora and adza axes with horse or eagle handle, and bury in same tradition as BMAC, suppine flexe, with similar figurines along with Androvan motifs.The resulte reflects this interactions.
The Steppe MLBA source in Northern Pakistanais groups, esp Androvan shifted South Asian demes, seem to have Maykopish like ancestry, not Botai. Maykop Steppe and Botai are very different and even on PCA, Botai has 0 Eneolithic Steppe, Maykop/Kumsay Steppe have 50% almost! In Matt's ghost both for Kalash and SPGT, it is this Maykopish ancestry which comes through not Botai or Okunevo. We see that Usotovo sample David posted to that early Androvan sample David had in earlier posts, it this deme which primarily intermingle with Corded ware like groups at least for Aryans who arrive before mixing with Central Asian farmer groups. It would be interesting to hear Matt's opinions on this.
@Matt SPGT_o is Loebanr outlier I12138 and Udegram_o I1992. But I1992 only has 46k snps as compared to 666k for I12138. So, SPGT_o is mainly I12138. I will run your other tests in some time.
@kouros, there is no need for any extra steppe maykop or steppe eneolithic ancestry is this model. I checked by adding steppe enolithic to the right populations in qpAdm, there is no fall in tail probability, in fact it improves. If there was an extra steppe_en input, the tailprob should have fallen drastically and model should have failed. so a steppe eneolithic source as input is not required at all.
SPGT_o
InPe_Native: 29.2 +- 4.7
Dzharkutan1: 30.8 +- 5.2
Dali_MLBA: 40 +- 3.8
tailprob: 0.94
output file https://pastebin.com/39GDAU6N
Added Kumsay_EBA to right pops instead of steppe_eneolithic. Tailprob remains same at 0.93.
This is formal proof that Kumsay_EBA or steppe_eneolithic is not required as a source when the 3 sources in left pops are present.
output file https://pastebin.com/BrUaLkQB
@Matt
qpAdm for target: Pakistan_H_SaiduSharif_I6893
InPe_Native: 41.7 +- 4
Kazakhstan_Kangju.SG: 58.3 +-4
tailprob: 0.21
output: https://pastebin.com/gCEvqq61
Saidu-Sharif_H 10 samples
InPe_Native: 64 +- 2
Kazakhstan_Kangju.SG: 36 +- 2
tailprob: 0.22
output file https://pastebin.com/P9dXngt5
2 implications:
1 - Swat native like population existed till Iron age without steppe or bmac or SiS ancestry.
2 - A kangju like saka source from IAMC admixed with the native pop.
For I6546 + I6548, these are the models i have tested so far
Pakistan_IA_Barikot_test
InPe_Native: 26.2 +- 4.5
Uzbekistan_BA_Dzharkutan1: 43.3 +- 4.4
Kazakhstan_MLBA_Zevakinskiy: 30.5 +- 3.4
tailprob: 0.0545 pass
Pakistan_IA_Barikot_test
InPe_Native: 37.6
Uzbekistan_BA_Dzharkutan1: 36.2
Kazakhstan_MLBA_Dali: 26.2
tailprob: 0.0208 fail
Pakistan_IA_Barikot_test
InPe_Native: 45.1
Uzbekistan_BA_Dzharkutan1: 32.2
Russia_MLBA_Sintashta: 22.7
tailprob: 0.0038 fail
@vAsiSTha, cheers. Thanks for running those off.
I'm not 100% about survival of a InPe_Native like population producing I6893. It seems more likely to me that there is further admixture from a population that is more AASI rich than InPe_Native into SPGT, as well as an LBA steppe source. I think you'd get a higher tailprob if we had the samples and that was possible. But I think what you do find is an intuitive result in the context of the G25 position, given the constraints of the available population samples. (You might or might not get a more fitting model with AASI rich Saidu_Sharif outlier I7722 and some LBA population).
Models for I6546+I6548, despite not being very strong, do look like they have a tendency to prefer a different BMAC:Steppe_MLBA related ratio than SGPT_o. Discarding InPe_Native, for BMAC related source:SteppeMLBA_related source:, SPGT_o = 43:57, while your models for I6546+I6548 are all at a rock solid reversed 58:42 BMAC_rel+SteppeMLBA_rel. Again that's intuitive with the G25 data. If you find a more solid model that says the same thing, I
Its also possible that I7722 like pops were present as natives of saidu sharif. Its basically a choice between the white cline or orange cline in the PCA. Sources for PCA are Iran_N, sintashta, Okunevo, irula, rest are projections. Link - https://imgur.com/TPzuFAT. Visually, somehow i prefer the kangju model pasted above.
Saidu sharif is from an imp region of the Maurya era, and magadha was the capital of Mauryan empire (modern Bihar, ganga region). So its possible that I7722 was a migrant from deeper in India.
As far as SPGT_o and your barikot samples are concerned, both are different.
SPGT_o rejects Zevakinskiy_MLBA as 3rd source, whereas Barikot samples reject sintashta as 3rd source.
So Barikot(and rest of SPGT) leans towards LBA or eastern steppe/iamc and rejects sintashta like ancestry whereas SPGT_o accepts sintashta, krasnoyarsk, dali, kashkarchi etc as source. and rejects lba type ancestry.
Watch from the 20 min mark
Archaeologist Michael Frechetti
https://www.youtube.com/watch?v=lP16l1PFoLA&t=1342s
Mayuresh M. Kelkar
Target: TKM_Geoksyur_En
Distance: 3.7864% / 0.03786404
84.4 TKM_Geoksyur_N_OSS
11.0 RUS_Progress_En
4.6 GEO_CHG
0.0 KAZ_Kumsay_EBA
0.0 RUS_Steppe_Maykop
0.0 RUS_Tyumen_HG
Target: TKM_Geoksyur_En
Distance: 2.7688% / 0.02768837
90.0 TKM_Geoksyur_N_CO
7.4 RUS_Progress_En
2.2 GEO_CHG
0.4 KAZ_Kumsay_EBA
0.0 RUS_Steppe_Maykop
0.0 RUS_Tyumen_HG
So what is that extra steppe-related ancestry in Geoksyur_En? Something Kelteminar-related?
Target: TKM_Geoksyur_N:644_Oss
Distance: 3.3727% / 0.03372715
65.8 IRN_Ganj_Dareh_N
13.2 GEO_CHG
8.2 IRN_HotuIIIb_Meso
6.6 Anatolia_Tepecik_Ciftlik_N
6.2 RUS_Tyumen_HG
0.0 Anatolia_Barcin_N
0.0 IRN_Seh_Gabi_LN
0.0 IRN_Wezmeh_N
0.0 KAZ_Kumsay_EBA
0.0 Levant_PPNB
0.0 RUS_Progress_En
0.0 RUS_Sosonivoy_HG
Target: TKM_Geoksyur_N:644_Co
Distance: 3.3905% / 0.03390517
55.2 IRN_Wezmeh_N
13.8 IRN_HotuIIIb_Meso
13.4 GEO_CHG
7.4 RUS_Progress_En
6.8 KAZ_Kumsay_EBA
3.4 Anatolia_Tepecik_Ciftlik_N
0.0 Anatolia_Barcin_N
0.0 IRN_Ganj_Dareh_N
0.0 IRN_Seh_Gabi_LN
0.0 Levant_PPNB
0.0 RUS_Sosonivoy_HG
0.0 RUS_Tyumen_HG
And why is the _Oss sample so different from _Co? Particularly why does _Co pick Progress and Kumsay while _Oss only Tyumen?
_Oss makes it seem like CHG had a greater presence outside of the Caucasus, perhaps even further South closer to Belt Cave.
Post a Comment