search this blog

Tuesday, February 28, 2017

European-specific mtDNA lineages on the Neo-Eneolithic Mongol Steppe


Interesting but not surprising:

The mitochondrial haplotypes U5a1 and H2a2a were identified from two upper Neolithic or early Eneolithic graves in Bayankhongor Province, Erdenetsogt Township, Shatar Chuluu. This is the earliest documented appearance of western mtDNA haplotypes on the Mongol Steppe, and the farthest east “western” mtDNA haplotypes have been recorded before the Iron Age. This evidence proves that individuals possessing western Eurasian mtDNA lived on the Mongol Steppe, east of the Altai Mountains, before the Bronze Age, and dispels the notion that the Altai Mountains were a significant barrier to gene flow.

...

When exactly western Eurasian steppe people migrated past the Altai to the Mongol Steppe is still not clear, however the identification of western mtDNA haplotypes from the Neolithic Mongol Steppe (this project) suggests that the migration predates the Eneolithic (~2900BCE). These first migrants most likely arrived along with the population from which the Afanasievo archaeological culture arose, or perhaps earlier with the first “western” people migrating onto the Mongol Steppe shortly after domestication of the horse, although no conclusive evidence of horse domestication dating before ~1300BCE has been found on the Mongol Steppe (Svyatko et al 2009; Vigne et al 2011; Taylor et al 2015). It seems more likely that the western migration onto the steppe was following sheep herds, which may have moved onto the Mongol Steppe around 3,300BCE (Lv et al 2015).

Rogers, Leland Liu, Understanding ancient human population genetics of the eastern Eurasian steppe through mitochondrial DNA analysis: Central Mongolian samples from the Neolithic, Bronze Age, Iron Age and Mongol Empire periods, Indiana University, ProQuest Dissertations Publishing, 2016. 10253175

See also...

R1b-M269 in Afanasievo

101 ancient Eurasian genomes (Allentoft et al. 2015)

A moment of clarity: PCA of ancient West Eurasia

154 comments:

Nirjhar007 said...

Surprise surprise lots of Indian Mtdnas like M3, M31, R7! .

Davidski said...

Yes, ASI markers, but in Mongolia and only 1,000BCE. That's well after PIE or even PII.

They should have migrated to the Western Steppe at least a thousand years earlier, then you'd have something to talk about.

Nirjhar007 said...

LOL

They are related to IE movements .

Davidski said...

I don't know much about Mongolia 1000BCE, but I suppose there may have been contacts between Vedic India and Eastern Steppe groups.

Maybe Buddhist monks and servants traveled up the steppe or something? Interesting, but so what?

Nirjhar007 said...

No its not Buddhist . I think its Iranian related folks out of India . Its possible even that they will show Indian L-657 .

Davidski said...

Yeah, Buddhist monks from Vedic India carrying L657 makes sense.

Nirjhar007 said...

Dave the dates are around 1000 BC , I don't know any Buddhists back then as Buddha appeared after 600 BC!!!.

jv said...

Wouldn't surprise me if an mtDNA H6 shows up in that area as mtDNA H6 was found in the Andronovo Culture & Okunev Culture.

Davidski said...

Dave the dates are around 1000 BC , I don't know any Buddhists back then as Buddha appeared after 600 BC!!!

As far as I can see, 1000BCE is the average date. Some of the samples are apparently from around 500BCE.

Nirjhar007 said...

Even if so , its not possible for them to be Buddhist .
If they were of around 300 Bc then a tiny possibility .

Davidski said...

Well they weren't early Iranians either.

Ted said...

The mtDNA was part of treasure that nomads brought home?

Nirjhar007 said...

Dave,

Possible , after de-urbanization , its not illogical to think that some groups from IVC area and surrounds went also North . They can be an example . Already from the early de-urbanization period , Harappans created a settlement in Northern Afghanistan i.e. Shortugai .

Also perhaps Mitanni , though we don't have their aDNA . And also they are very far from India and Mitanni can be qualified , as a related Indo-Iranian pop of Iran Instead.

Davidski said...

IVC was long gone at 1000BCE.

More like the steppe nomads traveled between India and the steppe regularly, taking South Asian women with them.

Nirjhar007 said...

No , its still living in traditions and culture .


No I don't think so . It happened during the Saka period, much later development .

Jijnasu said...

http://crossasia-journals.ub.uni-heidelberg.de/index.php/ejvs/article/view/323/312 possible evidence of early trade between Indians and steppe dwellers?

Matt said...

Not read this paper yet but surely if steppe nomads were traveling between South Asia and the steppe regularly, they were probably autosomally and mtdna wise South Asia already to a large extent and would just be taking women from their own culture (if any)?

Nirjhar007 said...

Jij,
Yes trade was on , even from the BMAC times .

Matt,

Steppe dudes , the rather huge ones came as Scythians (Saka) around 3rd century bc only .

This Mtdna is a clear evidence of late out of India movement . And by India , I am taking together also Pakistan and some parts of Afghanistan .

Davidski said...

Early Scythians had Siberian admixture. They weren't South Asian.

http://eurogenes.blogspot.com.au/2015/12/the-scythian.html

Nirjhar007 said...

But for your pleasure and if the Geneticists, that I don't know are also reading this . SC Asia already had contacts with Steppe, since 4th millennium itself!. The proto urban site of Tajikistan which was in contact with steppes had metallurgy and other crafts http://whc.unesco.org/en/list/1141

so the contacts with SC Asia and steppe was already there even before BMAC, Sintashta etc .

Please sample Sarazm.

Nirjhar007 said...

I didn't Say Scythians came fro India :P . But without any Indian aDNA to discuss at the table , I can't conclude...

Davidski said...

How come there's no South Asian admixture anywhere on the Early and Middle Bronze Age Steppe?

Nirjhar007 said...

I don't think calling something South Asian without aDNA , is wise .

But you are confident bloke .

postneo said...

I had commented last year that the supposed skewed sex ratio of yamnaya incursion to europe is an artifact. west Eurasian mtdna had already migrated to the steppe before the yamnaya incursion. So the yamnaya were bringing nothing new on the female side when they moved in to Europe

postneo said...

The recent q3 paper shoes proto yamnaya links between caucasus yamnaya south CAspian IVC and Altai. It's a mountain corridor connected by goats and sheep

Anonymous said...

I believe I read that Buddhism didn't penetrate Mongolia until either a century before or a century after the common era (year 0), though the religion only really made headway among the population many centuries on.

It's thought that Buddhism was introduced to Siberia's Buryat region a little earlier, in the 2nd century BCE, though Buddhism didn't actually take off until the 17th century CE in Siberia either.

For comparison, Buddhism reached China in the 1st or 2nd century AD.

Wikipedia has a page on Buddhism in Mongolia. There's an indirect hint at the date range for when Buddhism would have reached Mongolia, since the section on the "Early Introduction of Buddhism" in Mongolia says that "Many Buddhist terms of Sanskrit origin were adopted via the Sogdian language"

It actually took several centuries after Buddhism's inception for any streams of Buddhism to even adopt some form of the Sanskrit language. And the Iranian Sogdian language is dated to have existed from "100 BCE to 1000 CE". Buddhists were a minority in Zoroastrian Sogdia, and Buddhism's first appearance in Sogdia is dated to the 2nd or 1st century bc.

So it cannot have been Buddhists in Mongolia at 1000 to 500 BCE, as others here have said.

Kristiina said...

Out of curiosity, I checked the admixture graph in the Allentoft paper. North Indians, Kalash, Yamnaya and Afanasievo do share the greenish component. This component is probably the same as CHG whatever it really means. At a high K level, Afanasievo and Yamnaya also share the turquoise Kalash component. I do not know what this implies but one cannot say that Afanasievo and North India/Pakistan do not share any southern ancestry. Here the question is what is the relationship between ANI and CHG. I know that you, David, are against Iran ancestry in Yamnaya and maintain that it is only Caucasian ancestry.

Samuel Andrews said...

Unless they tested for M's coding region mutation we can't be confident any of these samples belonged to South Asian specific forms of M. I'm not experienced enough with Asian mtDNA to know whether the samples had any of the samples had South Asian specfic mtDNA. There's definitely a possibility they did.

Also this study referenced at least one study on ancient North Asian mtDNA none of us were aware of. One includes new Afanasievo mtDNA.

Anonymous said...

mtDNA M doesn't look all too common in Iran, going by the Derenko et al 2013 paper, which further found no R7.

Out of 353 samples, there are 13 instances of M. One of this was in a single Turkic speaking Qashqai, who was M4a. The remaining 12 samples of M are among ethnic Persians, and consist of single instances each of M18a, M2c, M3a1, M42b1b, M4b, M5a2a1a (from Kerman), and 6 persons with M5a2a4 of whom four were from Kerman province.

mtDNA M52, found in south Asians of the Himalayan region and in Bronze Age Novosvobodnaya aDNA, is not reported by the above paper as being present in the Iranian population.

Kristiina said...

Check Table 2: Bronze Age Mongolia Sequence Data!
Indian mtDNAs are R7, M31a2 x 2 and M3 x 2. One M is not clear.

Moreover, H14 is typical for Dravidians and the most basal haplotype is found there.

This means that 5(7) out of 26 samples are Indian. It is c. 20-25%.

Seven out of 26 samples are Siberian (A2,A2a1,D,Z,C4b8a(?)).

Two are East Asian (M8, B4).

Ten are Western Eurasian (N1a1a, K, H2a2a x2, H2a5b2, H5, H, HV0a1 x2)






Kristiina said...


This may be the first time that H14 appears in ancient samples anywhere.

The distribution is very exciting
http://www.ianlogan.co.uk/sequences_by_group/h14_genbank_sequences.htm

"The haplogroup H14 is uncommon in Europe but found mostly in the Near East and the Caucasus (Richards et al. 2000; Nasidze and Stoneking 2001; Roostalu et al. 2007; Al-Zahery et al. 2011) and so far only two subclades—H14a and H14b have been identified. However, based on new complete sequence analysis we identified an additional subclade, H14c, which has not been reported earlier (Fig. 1a). H14a and H14c lineages were found in Punjab, Uttar Pradesh, West Bengal, and Andhra Pradesh, whereas a single H14b2 haplotype was reported from Pakistan (GenBank KJ446337)."

Source: West Eurasian mtDNA lineages in India: an insight into the spread of the Dravidian language and the origins of the caste system

This what they suggest in the paper above:
"The autochthonous subhaplogroups—HV14a1 and U1a1a4 uniquely found in contemporary Dravidian speakers share their ancestry primarily with the Near East-Iran populations (Derenko et al. 2013). The coalescence times of HV14a1 and U1a1a4 were estimated to be ~10.5–17.9 kya. The shared ancestry of the Dravidian of South India and Iranian of Near East populations has been shown in the HV14 and U1a1 phylogeny (Fig. 1a) and their time estimates are consistent with the proto-Elamo-Dravidian language diffusion hypothesis which emphasized that the proto-Dravidian language evolved over 15 kya, specifically in western Asia before the beginning of agricultural development ~11 kya. This language was introduced by Neolithic pastoralists, and was thought to be associated with the spread of these west Eurasian-specific mtDNAs to peninsular India (Pagel et al. 2013)."

Samuel Andrews said...

This is the haplotype of the possibly South Asian M mtDNAs found in Bronze age Mongolia..

16126 16223

The 16223 mutation makes them non-R. mHG M3's single HVS1 mutation is 16126. South Asian M3 is more likely than anyother mHG.

This is the haplotype of the possibly R mtDNA found in Bronze age Mongolia.

16319

It could be R7 but it could also be countless other mHGs.

Samuel Andrews said...

One Bronze age individual (probably) has R0a. U3, U7, and HV1 have also been found in Bronze age Siberia. Maybe there were West Asians who migrated into Siberia.

Tesmos said...

@Samuel Andrews

Are you sure HV1 has been found in Bronze Age Siberia? I can't find anything about that matter.

capra internetensis said...

Bah, it's just HVS-I. Why are people still doing this?

What Krefter said, and also the purported M31a2 is missing 16195.

rozenblatt said...

@ capra internetensis

According to the paper, the samples were taken in 2009.

capra internetensis said...

@rozenfag

Ah, thanks. It's just a thesis anyway, so we can't expect too much, but it ends up just being a tease. Like that recent Moroccan aDNA paper.

I look forward to these samples getting the full treatment in the future.

Unknown said...

If Europeans come from the Steppe it is probably from both the paternal and maternal side, neolithic farmers probably have little influence except maybe in South Europe. I read the paper and it seems like Steppe populations had high frequencies of H while at the same time except in Bulgaria and Romania Middle Eastern and Central and West Europeans had little H.

Grey said...

@Davidski

"Because words related to metallurgy in Indo-European languages are of non-Indo-European
origin."

I'm not a linguist but...

*if* metallurgy 101 is pre-mining and pre-smelting and limited to cold working native metals then what words would you have?

1) The only native metals you'd have would be gold, silver, tin, copper and maybe meteoric iron but you wouldn't necessarily have the whole set in every language region.

So you might expect loan words to travel from the respective source regions i.e. region A might get its word for silver from region B and region B might get its word for gold from region A because each region only had one source each.

2) If cold hammering soft metals into simple shapes / jewelry was the limit of metallurgy 101 then what words might come from that?

Who knows but some possibilities might be

"ring" : from PIE "to turn, bend"

http://www.etymonline.com/index.php?allowed_in_frame=0&search=ring

"bead" : from PIE *gwhedh- "to ask, pray."

http://www.etymonline.com/index.php?allowed_in_frame=0&search=bead

#

there are probably other ancient words related to jewelry that i haven't thought of?

#

3) "bead" coming from "pray" echoes a bit with the etymology of iron in Germanic languages (eisen) which apparently comes from PIE "aes" for "holy"

http://www.etymonline.com/index.php?term=iron

also copper in latin was apparently "aes" before it changed to their word for Cyprus (because they got their copper from Cyprus)

http://www.etymonline.com/index.php?allowed_in_frame=0&search=copper

#

so i wonder

thing is - a people who developed early soft metal jewelry wouldn't have much of a military advantage but they might be (relatively) very rich

Grey said...

"there are probably other ancient words related to jewelry that i haven't thought of?"

just thought of another possible one

torque (n.) from PIE *torkw-eyo-, causative of *terkw- "to twist"

Davidski said...

@Kristiina

Out of curiosity, I checked the admixture graph in the Allentoft paper. I do not know what this implies but one cannot say that Afanasievo and North India/Pakistan do not share any southern ancestry.

You're confusing yourself and others like Jaydeep by misinterpreting the data. This leads to needless friction and arguments over matters that we should have all agreed on a long time ago.

I've already explained that Yamnaya and Afanasievo can't be modeled as part Kalash or any other South Asian pop with formal statistics. If you don't believe me, get in touch with Harvard and ask.

And I've also explained that Yamnaya, Afanasievo etc. can't have ancestry from the South Caspian because they lack South Caspian diagnostic markers that are found in ancient and modern Iranians, like mtDNA U7.

@Nirjhar

I don't think calling something South Asian without aDNA, is wise. But you are confident bloke.

But of course we do have aDNA from South Asia, from Harappa, and if you are aware of the results, then you should know that Bronze Age Europeans can't be part South Asian.

Rami said...

I am quite surprised a bit shocked at the considerable levels of South Asian Mtdna. Either its very old on the Steppe or its being picked by traders marrying women from South Central Asia.

Kristina you made excellent points!

David you need to get out of your Mein Kempf mindset , these Yamnaya/Afansievo people clearly do have some links with populations further South.If I recall there was one Yamnaya women with an M mtdna. Sometimes David you have your head so far up your ass , you cannot smell ur own bullshit anymore.

Davidski said...

@Rami

I'm not the one making up bullshit here. You are. Case in point.

If I recall there was one Yamnaya women with an M mtdna.

Imaginary results won't get you anywhere. You need real results.

postneo said...

@Davidski
"I've already explained that Yamnaya and Afanasievo can't be modeled as part Kalash or any other South Asian pop with formal statistics. If you don't believe me, get in touch with Harvard and ask."

Yes the modern Kalash may have a higher "ASI" component than needed, but an ancient population may work and may have just the right mix.

"And I've also explained that Yamnaya, Afanasievo etc. can't have ancestry from the South Caspian because they lack South Caspian diagnostic markers that are found in ancient and modern Iranians, like mtDNA U7."

A weak explanation. A single uniparental marker cannot be a vector for all ADNA.

Matt said...

The way I've come think of it (perhaps wrongly?) is that the ADMIXTURE algorithm has kind of different priorities compared to the formal tests; it's trying to fit lots of populations at once and on a panel of modern populations it's looking at lots of drift that has occured separately after the ancient samples we have and which is not captured by the formal statistics.

This means it will find it optimal to deprioritize relationships which are a faint part of the overall drift in the modern panel (e.g. the "ANE" connection) or limited to particular populations which are infrequent in the panel, like the steppe (HG and pastoralist) populations.

Some of the components it finds can be real, like the "Teal" population - which was doubted before the CHG paper, but seems more or less real in some form and in unsupervised ADMIXTURE terms is obviously more useful to find than a steppe / EHG / ANE population to explain genetic variation in its populations (since ADMIXTURE seems to always prefers to create it when given a choice). But also I guess it means that it can prioritise finding good proxies for recent populations which have experienced lots of drift, like Native Americans or Kalash, and then has to place these in ancients where they don't really quite fit, as a compromise. It's a real minefield.

Unknown said...

It's not just U7. If you folks paid attention to the paper you'd see several Y and mt markers that havre nothing to do with Yamnaya and European LN/ENA.

Unknown said...

Admixture is not to be taken literally, ever. It has its place, but should never be substituted in place of formal stats.

Davidski said...

@postneo

Yes the modern Kalash may have a higher "ASI" component than needed, but an ancient population may work and may have just the right mix.

Maybe in theory, but have you wondered where those Indian mtDNA HGs in Bronze Age Mongolia came from? Straight from South India, or from Pakistan and/or surrounds?

If from Pakistan and/or surrounds, then you need a new theory.

A weak explanation. A single uniparental marker cannot be a vector for all ADNA.

That was just an example.

By and large, Bronze Age Eastern Europeans do not show the haplogroups and subclades that we'd expect to see if a significant portion of their ancestry came from the South Caspian just before the Bronze Age.

So why bother with this angle?

bmdriver said...

David your white supremacist in genetic clothing.

South Asia is the source of your white supremacist genes, and all your holding onto is migrating Indian tribes that settled your homeland who returned to South Asia. Either way your people are nothing but white Indians.

And your whole opinion, attitude, has a long history which started with the White Christian supremacist attitude of your forefathers, who created a new God and placed themslves as gods chosen offspring, rewriting ancient history to their warped racist ideology, then invaded every known land on earth, wiping out native beliefs and replacing them with European Christians beliefs.

India is your kryptonite, and deep down you know it, sooner or later your world will come crashing down and the true history minus European Christian colonialism revisionist history with it, like they say repeat a lie a thousand times and it may appear to be the truth. India is starting to dismantle white supremacist revisionist history piece by piece, and I will rejoice when Indian history, language, ideology, religions, theology is presented by Indians and not white interpretors or cloned Indians still serving their colonial overlords, just as European history, culture, language, theology should not have Chinese point of view as the official position, nor would you accept it.

"We need to create Indian in colour but white Christian in taste and opinion"
-Thomas Macauley, ..........the days of loyal brown sahibs is over and Indian democracy taking root without the need of approval from a legacy of white supremacy.


One only needs to read history to understand racist, intolerant my absurd roots of aryanism.
Mice migration linked to spread of farming and ultimately IE is very revealing, spreading out from South Asia like a lotus.lol........


Davidski said...

Sure whatever you say. Never mind the facts.

Rami said...

Nobody is denying R1a is a product of Forrest Steppe so you can take comfort your wet dream of Steppe men has not been shattered. Hopefully when time travel is possible you can go back and have 4 steppe men Bukake on you and call it a day.

Also clearly the Yamnaya/Afansievo poplulations do have connections with populations in Iran and/or other places . You seem to be stuck on this CHG block because in that bird brain of yours they still qualify as Europeans. Neolithic Iranians live quite close these CHGs and are essentially similar peoples ( ANE + Basal)
Also there has not been any genome data from South Central/South Asia , so I do not know how u can make conclusions based of pulling statements out of your ass.

Davidski said...

I've just explained why Bronze Age Eastern Europeans can't have any significant ancestry from the South Caspian. No one, including you, has provided any plausible argument against what I said.

I also explained that ancient DNA from South Central Asia is already available and some people here, including me, are aware of some of the results, which do not support the idea of an migration from South Central Asia to the Eastern European steppe.

It seems like a lot of you aren't interested in what really happened, but in confirming your own fantasies. I don't understand why.

Matt said...

Re: ADMIXTURE, good ADMIXTURE panels will usually tell you more about how populations relate to each other than a given set of fits of ancients with formal stats will. Eurogenes K13 or K15 has more information about how similar the European populations are to each other, and at predicting which modern populations a person is from, than the fits of ancient ancestors using formal stats will. (Likewise the drifted Kalash clusters will tell you some things about their population history compared to other South Asians that are absent from the South Asian fits in Lazaridis et al 2016, and would help you fit whether and unknown person was a Kalash).

But the K components do not necessarily imply real ancestors. ADMIXTURE (and PCA) are designed to fit relatedness between their samples to a lower dimension space / set of variables and only find ancestral populations, when they do, as a co-incidence of that

Re: CHG vs Iran, haven't we've been through this before recently with Jaydeep (and he is at least a polite and rational person to speak with)? There are fairly good reasons to prefer a combination of Anatolian/Levant Farmer/CHG (Caucasus Neolithic) over Iran_Chalcolithic, or any populations further towards South Central Asia. It seems a bit hypocritical for someone to accuse Davidski of having an irrational preference for Caucasus Neolithic populations, while they themselves have an even more irrational and more unsupported preference for Iranian and South Central Asian Neolithic populations, just because in their mind it's closer to India.

Samuel Andrews said...

@bmdriver,

There's absolutely no evidence of South Asian or SouthCentral Asian or whatever migration to Europe. None. Ancient DNA traces European's origins to Meso/Neo Europe, Near East, and Caucasus. That's fact not opinion. Even if there is ancestry from Iran, Iran isn't India.

The only racist here is you. You're a freak dude. You're full of hate. You need to talk to someone about your hate for white people, it's unhealthy. Please understand I don't think European imperialism can be justified. White people thought aren't an evil race. A handful of nations in Europe had the needed talents to create powerful imperial nations. Big deal. Other non-Euro nations joined the imperialism trend later; Japan, Turkey, United States, Russia. Imperialism had more to do with hardcore human wants and needs; money, dominance over rivals, power. than it did with a belief in racial superiority.

Taymas said...

Why are a handful of South Asia-centered mtdna samples from millenia too late (bc at this point we already have centuries of writing in divergent IE daughters) being treated as a such a big deal? This is a millenium after the steppe hypothesis would place interchange between IE and BMAC and centuries after Indo-Aryans were showing up as far as Syria.

We can also go back about this far in west Asian (Cimmerians, Scythians) and east Asian (Wusun, Yuezhi) writing illustrating that steppe groups were pushing each other around over huge distances, they were already centuries into full equestrian nomadism.

Finally R7 and M31 tend eastern in India, at least according to wiki. Given that we also have East Asian mtdna at this time/place, not clear they even came by the same route as the M3 or H14. Seems to me this result is interesting, but the interpretation is very unclear yet, and not relevant at all to the first IE expansion.

What am I missing? I can't access the full paper right now, seems to be down.

Finally, if such weak evidence is being used to sling Nazi accusations, what should we call Out of India proponents given the complete lack of any unequivocal South Asian ancestry northwest of the IAMC? I really resent anyone getting called a Nazi unless they're advocating for industrialized murder of identity groups.

Unknown said...

The following paper looks like it is coming out tomorrow in Nature Scientific Report:
http://www.unica.it/pub/print.jsp?id=34799&iso=96&is=7
Fascinating stuff--J2b1 and I3 in Mesolithic Su Carroppu Sardinia and no U's (mtDNA). Su Carroppu is important in that a few hundred years later, it becomes fully cardial-impressed Neolithic and like (Y chromosome) G2a's immigrate. My question is what is the corresponding Y lineage in Mesolithic Sardinia? (Hint--not I2's like the lack of U's there at that time). Is it R1b-V88 or R1b-L389 or a partial R1b-M269 like in the Baltic?

Rob said...

@ Roy
What's the bet it's R1b ?
:)

Davidski said...

Since you're dropping hints Roy, do you know if the paper has Y-DNA data?

Ryan said...

Links not opening for me right now. Where/when exactly are the South Asian mtDNA samples from? Can we tie them to a specific group/culture? The Wusun or Yuezhi maybe?

20-25% of mtDNA isn't a small amount of contact. I think it either means ASI mtDNA was already present in Central Asia in large amounts (BMAC?) when Indo-Europeans entered the scene, or else at least one of the Indo-European groups in China migrated en mass from fairly deep in South Asia. Male-only migration wouldn't explain the mtDNA results either.

Samuel Andrews said...

@Roy,

There's no yHG called I3. Maybe the article is referring to a type of I2.

One study claimed to have found mHG HV1 in Mesolithic Sicily. J2b1 in Mesolithic Sardinia points to a connection with West Asia.

There's an unlimited amount of possible explanations for those strange uniparental markers in Mesolithic Sardinia. I don't know what to expect.

@Rob,
"What's the bet it's R1b ?
:)"

According to that article the first two results are; I3(probably a type of I2) and J2b1. Even if some had R1b the chances they had the R1b-U152 which takes up most R1b in Sardinia today is tiny. R1b M269 is a late arrival in Sardinia and Western Europe.

Unknown said...

@Davidski,
I doubt the paper has Y chromosome data since the Barbujani lab does mostly ancient mtDNA. Just guessing with the Y and the possibility that a variety of R1b lineages spread from Central Asia through North Africa and Eastern and Western Europe with the trapezoidal/geometric microlith Mesolithic culture. Too early of course for full M269, but earlier lineages like V88, P297, and partial M269 could have partaken in these migrations slightly before the Neolithic Era in Europe. I've actually been to a Neolithic Cave about 10 km from Su Carroppu.

Davidski said...

@Taymas and Ryan

Here's the paper.

https://drive.google.com/file/d/0B9o3EYTdM8lQMTZaVFgxZEQtak0/view?usp=sharing

@Samuel

mtDNA J2b1 and I3

Seinundzeit said...

I've been following David's blog for a while.

I can say this with some confidence: David isn't racist.

I mean, racists tend to be total douchebags, and are also usually quite stupid.

David is extremely nice/kind, as demonstrated by the comments stream here (he's obviously very liberal, when it comes to letting people hurl accusations at him), and he is exceedingly intelligent, so he just doesn't fit the "type" (lol).

So, you dudes should take it down a notch, because these results are actually very interesting.

Anyway, I think these results do show that later eastern steppe populations could have had substantial South Asian admixture, due to the linkages that existed between greater India and the eastern steppe.

I mean, the Pamiri peoples are all direct descendants of "Saka" (apparently, Sythians), and they are 50%-60% LNBA European, but the rest of that 50%-40% is mostly West Eurasian ancestry that is shared with North Indians (Iran_Neolithic-related stuff, but with more ANE).

So, some of those later Eastern Iranian populations, on the eastern portion of the Eurasian steppes, could turn out to be very similar to modern Pamiri Tajiks.

Still, lets be honest though, when it comes to Yamnaya/Corded Ware/Sintashta/etc, none of those people seem to have had any South Central Asian ancestry.

Yamnaya's "southern" ancestry is too "western" (it doesn't have an excess of ANE affinity, or an excess of ENA affinity). And the "southern" ancestry in Corded Ware/Srubnaya/Sintashta/etc is even more "western". Matt has already made this point.

At the moment, I think what we see in this thesis is (perhaps?) a reflection of later dynamics.

Also, perhaps Indian mtDNA M lineages had a much wider distribution in ancient West Eurasian populations?

I mean, we've seen ancient Indian mtDNA M in Maikop, in ancient Syria, in ancient Iraq (I think, if my memory serves me right?), and now in ancient Mongolia.

We now know that mtDNA M isn't a "East Eurasian" haplogroup, since many Paleolithic West Eurasians from Europe had it (again, if my memory serves me right), so what if many M lineages aren't as "Indian" as we thought, based on present distributions?

I'm just spit-balling here, thinking about possibilities. Who really knows.

As always, the more aDNA we see, the more interesting/weird/cool prehistory becomes.

Romulus the I2a L233+ Proto Balto-Slav, layer of Corded Ware Women said...

Formation of WHG involves increased affinity to Near Easterners and the arrival of R1b so makes sense we see some Near Eastern mtDNA show up too.

Samuel Andrews said...

@Davidski,
"mtDNA J2b1 and I3"

That changes everything. J2b1 appears in both EEF and Steppe. mHG I appears in Steppe, Neolithic Levant, and Chalcolithic Iran.

Unknown said...

@Samuel Andrews
I agree that the Mesolithic Y in Sardinia cannot be a U152 or even L51--too early. I think the likely candidate is V88-M18. Su Carroppu is in SW Sardinia and V88/M18 is found in South Sardinia in modern populations. U152 is much more frequent in Corsica and Northern Italy than in Sardinia and is likely a later migration (Celto-Ligurian?) I also think that I2-M26, which is common in Sardinia is a later migration from Iberia/S France. V88 is found in North Africa and could reflect the Capsian trapezoidal Mesolithic culture found there, concurrent with France and Sardinia.
@Romulus
Yes, when the paper is published, it looks like J2b1 and T2/(?T3) will be from seafaring Mesolithic people from further east.

Rob said...

@ Sam A

" Even if some had R1b the chances they had the R1b-U152 which takes up most R1b in Sardinia today is tiny. R1b M269 is a late arrival in Sardinia and Western Europe"

Well, Yes, that's rather obvious one would think as U1523 isn;t mesolithic. .
Just like modern R1a in Europe is almost all Z645+ as CWC erased other R1a lines, and later Balto-Slavic Z280 and M458 also expanded over other lines.

But the point is people like Gio have long argued that there is a lot of old diversity in Italy, often ridiculed, but he might be right as teh evidence accumulates. Of course, this doesn;t exclude the possibility that U152 represents a move back into Italy, but the issue is that even at the L51 level, maximum basal clades seem to be in Italy (and that is under-represented compared to western European countries/ Anglo-celtic Americans).

@ Roy

" I also think that I2-M26, which is common in Sardinia is a later migration from Iberia/S France"

Yes, i think so.

Ryan said...

@David - Thanks for the mirror.

Looks like Slab Grave Culture or its close relatives? They point to Afanasievo, but wouldn't Adronovo make more sense if we need a vector for ASI lineages? M31 is a really, really ASI haplogroup.

Spanked said...

@bmdriver
Long time lurker here, but your psychotic, delusional rant finally forced me to post.

You really need to put down your conspiracy pamphlets and talk to a psychiatrist... and then an actual historian.

Davidski said...

@Ryan

Where does the author point to Afanasievo to explain the M lineages?

Samuel Andrews said...

1/3 of the mtDNAs from Eneolithic and Bronze age Mongolia can't be placed to a mtDNA haplogroup. Most unknowns were R and therefore maybe H. Confirmed West Eurasian mHGs are; T1a, J, N1a1a, HV17, R0a, H2a2b, H5, and U5a1b1.

There really isn't a lot of West Eurasian mtDNA in the Bronze age samples, unless all the Rs are H or HV. About 15% of the Bronze age samples could be placed under West Eurasian mHGs. If we assume some of the Rs are H, then maybe something like 20-30% had West Eurasian mtDNA.

Ryan said...

@David - not for the M lineages. Afanasievo is where the point for the western lineages. They just consider the M lineages "East Asian" and leave it at that. M31a2 is from the Andaman Islands and some tribal groups in India...

Davidski said...

From memory the author mentions both Afanasievo and Andronovo as potential sources of the West Eurasian lineages on the Mongol Steppe.

Afanasievo is a likely source for the Eneolithic West Eurasian lineages, but also many of the later ones, because many of them could have survived there since the Eneolithic.

Ryan said...

Agreed, but even for Andronovo, either ASI ranged pretty far north or there was a secondary expansion/migration of Indo-Iranians out of India to Mongolia. Any thoughts as to which is likely?

Alternatively (and much less likely) this is a relic from however the hell Indigenous Americans got their Onge-like ancestry.

Davidski said...

Seems more like a migration, perhaps mostly of women, than an expansion.

Nirjhar007 said...

Dave,
But of course we do have aDNA from South Asia, from Harappa, and if you are aware of the results, then you should know that Bronze Age Europeans can't be part South Asian.
Stop being a schmuck again , you and I very well know, that there is no such results .

capra internetensis said...

Alternatively, it isn't actually M31.

There is a remarkably high frequency of A2 for anywhere west of Chukotka. These guys would be not only part ASI, but part Eskimo.

Ryukendo K said...

This is really not surprising at all. The spread of S Asian and W Asian mtDNA lineages into central asia and the steppes from the bronze to iron age transition is well documented, so much so that post-iron age mitochondrial variation in Kazakhstan and the Yenisei valley clusters with present-day south-central Asians.

Iron Age Central Asian mtDNA
Map of sites

The bronze age mitochondrial pool of the same two regions, prior to the change, is very 'Northern', very EHG-shifted.

Pre-Iron Age Central Asian mtDNA
Source

So some time after the initial IE incursions into the Steppe, a large amount of mtDNA, and presumably some Y-DNA and autosomal ancestry, from SC and West Asia introgressed deeply north. This matches the historical pattern with the Eastern steppes as well, where East Asian mtDNA, and later YDNA, continuously introgressed into the steppe, in this case with non-IE ethnolinguistic groups, after the initial introduction of the IE pastoralist package into East Asia--a process that extended well into the historical period.

More interesting to me is the similarity of the Eastern elements of the mtDNA pools of these Steppe inhabitants with the mtDNA recovered from the steppe-sown frontier cultures of China, such as those from the Lajjia site, the Tianshanbeilu culture, Miaodigou and Ordos cultures. These sedentary cultures all disappeared, replaced by a full-fledged pastoralist culture, during the late metal ages and coincident with the rise of the Shang dynasty. They may perhaps be identified with the ethnolinguistic groups that replaced IE in the eastern steppe by the historical period. It is also interesting that some of these cultures are derivatives of the Neolithic of the Yellow River valley, which similarly appears unusually "northern" in mtDNA when compared to the Han inhabitants there today for whatever reason.

Taymas said...

Thanks Davidski.

I wish they had which samples were from which cultures (unless I just missed it somewhere).

Noticed that there's zero M3/M31/R7 (some unspecified H but no specified H14) in the Xiongnu samples. Mongolian Empire samples had one R7 and an "M3/G1b", and definitely no H14. Weird.

Ryukendo K said...

@ Capra
Interesting observation. The Okunev culture auDNA, and the mtDNA from lokomotiv in the paper cited above, seem to show that, prior to the introduction of 'South Siberian' ancestry with the Turkis and Mongolics (whether or not that ancestry is native to South Siberia is debatable to me) the area was quite 'Paleosiberian-like', with mtDNA and auDNA far more 'Northern' than it is today.

Davidski said...

@Nirjhar

Stop being a schmuck again, you and I very well know, that there is no such results.

The Harappan samples from Rakhigarhi have a lot of ASI admixture, and therefore they cannot be ancestral to any ancient European group.

Not my problem if you don't know this yet. But you'll see it spelled out when the paper comes out.

Nirjhar007 said...

Dave,

Do you want to destroy this discussion too by rambling all these lies again?.

Why don't you send an email to this person :

The director of the whole study . Dr. Vasant Shinde :
vasant.shinde@dcpune.ac.in

Davidski said...

He probably doesn't even know what ASI is yet. But he will when the paper is put together.

Nirjhar007 said...

Figures , you know 0 of whats actually going on .

But don't worry , its going perfectly , stay prepared for an early retirement ;) ....

Aram said...

Mtdna I and J2a2 was found in Levant. PPNC and Natufian. This mean that Mesolithic Sardinia will show strong contacts with Levant.

Davidski said...

It makes no difference what's actually going on. The ancient data will do the talking in the end.

The idea that North India was populated by a fully West Eurasian population, and ancestral to Yamnaya, during the Copper Age just looks insane without having to see any ancient DNA from North India.

But don't worry, see it we will, and then maybe, just maybe you'll realize how crazy you are.

Nirjhar007 said...

just maybe you'll realize how crazy you are

No, you will find out how crazy YOU were! :D , I have no problems though, with people gong crazy or whatever , just don't throw lies around , it doesn't matter how attractive a lie is , it will at the end betray you badly . You will find out ...

Aram said...

Chemurchek Megalithic culture in west Mongolia ha mtdna H and T. 2500bc
Kovalev proposed that Chemurchek is the homeland of Tocharians.

Ryukendo K said...

@ Sein

Thanks for posting the nMonte results with Andronovo. This is old, but just to raise it again, when using single formal stats, the populations that are most differentially related to the available high-low caste population pairs from the same region in India are always Andronovo, Sinthastha, and Srubnaya. If we favour modern genomes, i.e. the highest quality comparisons, by sorting by Z score instead, the populations most favoured end up being EEF-rich West Europeans:

GujaraitA vs D, sorted by value
GujaraitA vs D, sorted by D score

Brahmin UP vs Chenchu, sorted by value
Brahmin UP vs Chenchu, sorted by D score

This pattern is quite difficult to explain if the IAr ancestry brought in was Yamnaya-like.

I've always thought that the preference for Yamnaya in modelling (modelling, not comparisons of high vs low caste groups) is an artifact of mixtures of Iran_Chl and excess Central Asian ANE ancestry picked up by the time IAr entered India producing a 'Caucasus-like' mix that biases the overall Steppic input to look like Yamnaya. In fact, the West Eurasian group that shares the highest amount of drift with both GujaratiA and GujaratiD is Yamnaya, followed by Poltavka, which means the overall West Eurasian ancestry of S Asians plus excess crown Eurasian ENA ancestry tends to produce an overall drift signature similar to Yamnaya, while the population most distinguishing the high and low caste populations are EEF rich populations like Andronovo, Basque, and Icelandic.

differential relationship of West Eurasians to GujaratiA and D

Furthermore the only IE ancient populations with high IBD with both S Asians and Europeans is Andronovo, Sintashta, and Srubnaya, though other factors, like age, influences this as well. When picking freely from a panel of ancients, the steppic group most often picked by S Asians is Scythian_IA, which, other than being quite similar to Sintashta and Andronovo, accounts for a greater proportion of the ancestry of S Asians than when Andronovo or Sintashta is used alone. So I think there is still some evidence that the contribution is from latter steppic groups with substantial EEF ancestry, though this can only be conclusively solved with S Asian aDNA.

Nirjhar007 said...

Aram,

But I agree with Mallory , that Xiaohe makes good sense .
http://sino-platonic.org/complete/spp259_tocharian_origins.pdf

Ryukendo K said...

For the above links, the second link in the first two sets of images are sorted by Z score, not D.

Nirjhar007 said...

S Asians is Scythian_IA, which, other than being quite similar to Sintashta and Andronovo, accounts for a greater proportion of the ancestry of S Asians than when Andronovo or Sintashta is used alone. So I think there is still some evidence that the contribution is from latter steppic groups with substantial EEF ancestry

Interesting , if I am not wrong , I think Gujarat A,B,C s , do have such affinity , wch also goes to the fact that those areas were ruled by Indo-Scytians .

postneo said...

@David
"The idea that North India was populated by a fully West Eurasian population, and ancestral to Yamnaya..."

Does not need to be "fully west eurasian" whatever that means ,....just related. And yes it could be an ancestral component if the sample is older that yamnaya. If not then it will be derived.

ASI/ANI cannot be ancestral to any archaic population, they are components derived from modern populations.

finds like Ganj dareh, are not fully representative of the south caspian. Its 200 miles away, Neither is Hotu, its too old.

We see that M seems to spread in an arc from Maykop to the Altai perhaps from the chalcolithic. Today it seems to survive best in south Asia. Q3e seems to show a similar pattern as M.

Unknown said...

I am putting the finishing touches on some charts and tables from a month long investigation of IBD sharing between various N & E Eurasian populations and W/SC/S Asians, and the results indicate that ADMIXTURE severely underestimates N & E Eurasian gene flow to many W and S Asian groups such as Kurds, Pashtuns, and Punjabis.

I should be publishing the results within the next few days on my website, EurasianDNA.com. Anyways, as an aside I decided to look at IBD sharing between W/SC/S Asian and Lithuanians. I did this using BEAGLE and setting thresholds to only include sharing of rare haplotypes, such that when combined with segment length thresholds, the results would indicate actual IBD.

I realize that many S Asians are still skeptical of the impact of steppe populations on the genetics of W or S Asia, for some it is because they don't have much faith in formal analysis or uniparental data, for others it is hard for them to let go of preconceived notions.

A little surprising, but consistent with proto IE migrations from the Eurasian steppe, some of my Kurd samples and Sein had larger IBD matches with Lithuanians than with most of the other Asians in the run. You can imagine what the results may have been if I had included more NE European populations, instead of a couple of Lithuanian samples.

BTW, the runs were at 404K overlapping SNPs, using genomes from the Estonian Biocentre.

I just can't see how Sein's and the Kurd results can be explained by other than a relatively recent event where a steppe pop contributed to Lithuanians, Kurds, and Sein. I mean the reverse, where there were simultaneous migrations from Kurdistan and Pashtunistan into Lithuania seems rather far fetched I think most will agree.

The full outputs will be available within a few days on my website.

Nirjhar007 said...

Cool looking website :) .

Ryan said...

@David - Why a random migration of women from India to Mongolia then though?

@Capra - Why does that exclude M31 though? I agree re: there being old Beringia stuff there as well. There's tonnes of weird connections to Beringia/Americas going on. A2a is closely associated with the Arctic Small Tools Tradition and is probably a good candidate for a Dene-Yenisien marker - which I suppose isn't that surprise for Mongolia since Kets were living in northern Mongolia pretty recently and the Yenisei river has its source in Mongolia.

There's also C4b - its relative (~20kya back) C4c is found in northwestern North America and in a few patches in south/central America. C4 is also found in some kurgan burials in the Ukraine though, so it could be Siberian/Beringian by way of the Ukraine.

Also on the Y-chromosome side, C-M217 is found in the Americas with a similar distribution to mtdna C4c (ie NW North America + some weird hotspots in places like Ecuador and Columbia). Most of it is C-P39, but not all.

Davidski said...

@postneo

Does not need to be "fully west eurasian" whatever that means ,....just related. And yes it could be an ancestral component if the sample is older that yamnaya. If not then it will be derived.

Anything ancestral to Yamnaya can't have any South Asian ancestry.

So if you're positing that a population from North India was ancestral to Yamnaya, then that means you're claiming that North India at some point, I'm guessing the Neolithic and/or Copper Age, lacked South Asian ancestry.

See how idiotic that sounds?

finds like Ganj dareh, are not fully representative of the south caspian. Its 200 miles away, Neither is Hotu, its too old.

All ancient and modern Iranians and nearby populations show South Caspian specific uniparental markers that Bronze Age Eastern Europeans lack.

So if a population from Iran contributed to the ancestry of Bronze Age Eastern Europeans, it was rather special.

Why keep pushing this angle despite the lack of evidence? Why not just choose the more parsimonious solution that Yamnaya is a mixture of Eastern and Caucasus foragers with some steppe farmer ancestry thrown in?

Is this really so painful to accept?

capra internetensis said...

@Ryan

It might really be M31, it's just not a very solid haplogroup prediction (two common mutations + missing one mutation).

Karelia_HG's C1 is also very Beringian. I think there was a substantial back-migration from that part of the world.

Al Bundy said...

@RK You think there were IE incursions into the steppe?

Kristiina said...

@David ”By and large, Bronze Age Eastern Europeans do not show the haplogroups and subclades that we'd expect to see if a significant portion of their ancestry came from the South Caspian just before the Bronze Age.”

Why Proto-IE or any IE language should have been spoken in Bronze Age Mongolia or even in Afanasievo? On the basis of linguistic analysis, many elements of proto-IE point to Caucasus.

@Matt There are fairly good reasons to prefer a combination of Anatolian/Levant Farmer/CHG (Caucasus Neolithic) over Iran_Chalcolithic, or any populations further towards South Central Asia.

This again is an indication that these early Mongolian nomads did not speak an IE language.

@Ryukendo So I think there is still some evidence that the contribution is from latter steppic groups with substantial EEF ancestry, though this can only be conclusively solved with S Asian aDNA.

@ Ryukendo So some time after the initial IE incursions into the Steppe, a large amount of mtDNA, and presumably some Y-DNA and autosomal ancestry, from SC and West Asia introgressed deeply north.

Andronovo samples are dated 1800-1400 BC, 1400-800 BC and 800-100 AD. Andronovo samples in Allentoft paper have EEF/Anatolia Neolitichic. Andronovo mtDNA (incl. U4, U2e, and T1, T4 in the oldest period) is indeed different from this Mongolian mtDNA. A good case can be made that Andronovoans spoke an IE language, but IMO ancient Mongolians did not and I also doubt that Afanasievo spoke an IE language. I make my ”judgment” when we get Afanasievo yDNA. Neolithic mtDNA from the Lake Baikal showed us that the original mtDNA in the Baikal area was Siberian and not Western Eurasian.

Davidski said...

@Kristiina

Afanasievo is identical to Yamnaya and Poltavka.

Apart from that, not sure what to say, because I don't understand what you're trying to convey in your post.

Samuel Andrews said...

@Aram
"Chemurchek Megalithic culture in west Mongolia ha mtdna H and T. 2500bc
Kovalev proposed that Chemurchek is the homeland of Tocharians."

Can you post the source for those results?

postneo said...

Actually its unlikely south asian populations are ancestral to yamnaya or vice versa. They are too far apart

They have shared ancestral compoents. A population similar to iran chalc + ANE/EHG + ASI for south asia
and CHG + EHG for Yamnaya. These makes it seem Yamnaya is ancentral.

@Ryu why are you comparing UP Brahmin and Chenchu? Its a contrived test. If you are comparing caste do it for the same region. Obviously one will show greater affinity to andronovo just by geographical distance.


Davidski said...

@postneo

Actually its unlikely south asian populations are ancestral to yamnaya or vice versa. They are too far apart.

Idiotic statement, considering the presence of R1a-Z93 and related subclades in ancient Eastern Europeans and their direct descendants thousands of miles away on the Asian steppes.

Are you brain damaged in some way?

Aram said...

Samuel

This is a student work in Russian. Look at the page 15. I got it from the Russian Wikipedia about Chemurchek.

https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=2&cad=rja&uact=8&ved=0ahUKEwi-saOz6LTSAhXCE5oKHQvMAgUQFggdMAE&url=http%3A%2F%2Fdlib.rsl.ru%2Floader%2Fview%2F01002814763%3Fget%3Dpdf&usg=AFQjCNFETYSJUGK-KWzQMcKAFhDlBESrEA&sig2=ashc0yVGFSjvCte3Hkg5aQ

Nirjhar

If You agree with Mallory then it's OK. I have no favourite theory on Tocharians currently.
Btw it would be good idea if some Indian users do a Google map with ancient DNA haplogroup M. Would be easy to keep the track of their movements.

John Thomas said...

Ill-mannered unhinged ranting chauvinists.

Ryukendo K said...

@ Postneo

My apologies. These are comparisons of UP Brahmin vs Chamar, not of UP Brahmin vs chenchu, I mixed them up. Since Chamar and UP Brahmin are both from UP, this is a valid upper-lower-caste pair.

@ Kurd

Kurd, could you extract the IBD segments between SC Asians and E Europeans, and check if they are more closely related to either the same loci in Sardinians or Dravidian Indians? If the segments shared between Lithuanians and SC Europeans are European-like instead of S Asian-like, this should show that the shared ancestry between Europeans and SC Asians came from a European, as opposed to a S Asian-like population.

@ Al Bundy

From Pontic steppe to the Eurasian steppes more generally.

@ Kristiina

The Okunevo and Karasuk genomes seem like they retained the Afanesievo contribution in addition to ENA input, instead of one from Andronovo or Sintashta.

Ryukendo K said...

@ Matt

Matt, quite a while back you mentioned Turchin's work in the context of his usage of a admittedly arbitrary intervention in his model, where the West Russian area is modeled as receiving agriculture at a very late date and therefore struck off as a region of early state formation. It seems like intensive, settled agriculture only reached Russia with the penetration of the East Slavs, which occurred well into the historical period around the 6-8th centuries, if I'm not wrong. Do you still think this is an issue in his modelling?

@ bmdriver
You really should not think that there is some kind of vast conspiracy in the academic sphere to serve the needs of one group after another. The scientists involved in aDNA may be a small group, but they often have vehement, even vitriolic disagreements with each other, even if thats not publicly known. If there were a crack in the analytical framework, it would have been exploited by now.

Ryan said...

@Kristiina - I don't think anyone is suggesting Proto-IE here, just a daughter language - either Iranian or Tocharian in extraction. Who else do you think Afanasievo would be? The oldest accounts we have for the Tarim and Turpan basins have Indo-European groups living there, and likely living there for quite a while according those accounts and the remains and material goods (ie jade) left behind. The area where the samples were taken from is closer to the Turpan basin than it is to Lake Baikal. There's a pretty reasonable narrative here of Indo-European nomads migrating to the steppe, and then slowly being marginalized and pushed out by the people living in more forested/mountainous areas - namely the Xiongnu. And the Xiongnu knew this based on their high-status graves only being made in the steppe/mountain areas, and never the steppe.

Shaikorth said...

@rk, does Karasuk retaining Afanasievo-like western mtDNA actually look likely?

http://www.nature.com/article-assets/npg/srep/2016/160916/srep33446/extref/srep33446-s1.xls

Table S5. There's been Karasuk remains with H, I, K, T1 and T2.

On the Y-DNA side, since Afanasievo was so similar to Samara Yamnaya autosomally we could speculate about them having R1b which would not really be present in Karasuk (R1a, Q) either.

Kristiina said...

How come nobody has noticed this! This is huge!
Peuplement du sud de la Sibérie et de l'Altaï à l'âge du Bronze : apport de la paléogénétique
https://tel.archives-ouvertes.fr/tel-01296484/

Afanasievo is R1b-M269, R1b-P297 (Yamnaya line I presume) and Q-M242 (Maf)
Okunevo is R1b1a2-M269, Q1a3 and N
Elunino is Q1a3
Chemurchek is C-M130
Sagsai is R1a1-Z93, Q1a3a, C-M130 (looks Turkic)
Munkh-Khairkhan is N

Therefore, I think that Afanasievo could be Yamnaya Samara IE language, Okunevo a Siberian language, Chemurchek could be Proto-Tungusic, Sagsai Proto-Turkic, Munkh-Khairkan proto-Samoyedic.

Shaikorth said...

Proto-Tungusic homeland is considered by recent studies to be in Manchuria.
http://www.marumaru.eu/Martine_Robbeets/Publications_files/Robbeets2014_review%20Malchukov%20%26%20Whaley.pdf

Making associations based on very low resolution Y-DNA isn't going to produce solid results.

Afanasievo being linguistically grouped with Yamnaya is plausible but there were always factors other than Y-DNA linking them.

Gioiello said...

@ Roy King
"Just guessing with the Y and the possibility that a variety of R1b lineages spread from Central Asia through North Africa and Eastern and Western Europe with the trapezoidal/geometric microlith Mesolithic culture. Too early of course for full M269, but earlier lineages like V88, P297, and partial M269 could have partaken in these migrations slightly before the Neolithic Era in Europe. I've actually been to a Neolithic Cave about 10 km from Su Carroppu".

You are ridiculous. Villabruna R1b1 was in Italy 14000 years ago. Read what I demonstrated about R-V88: all the oldest subclades are in Italy and Western Europe and African and Middle Eastern ones are very young: not more than 5000 years ago.

Ryukendo K said...

@ Kristiina

Thanks for this paper! I have to say though, all of their results come from a rather small area around the Altai mountains at the corner where Kazakhstan, Russia, China and Mongolia meet, and its unlikely that so many languages could have their urheimats thickly clustered around a single area.

Also interesting that the site 1 at bolshemysskaya North of Mongolia should return R1b-P297 at the 4th millenium BC, makes you wonder if a continuum of R1b existed in Siberia far to the east.

Matt said...

@ Ryukendo, OT but re: Turchin, IRC (been a while I said I read this paper, though I have been discussing it recently elsewhere funnily enoug) it's not so much Russia that I was concerned about, as that there would still be deviations of his model's absence of agricultural seeding in East Central Europe, where we do have early agricultural societies (Funnelbeaker, Lengyel, etc.).

IRC in his model there are not, and then if they were in the model, the parameters are such that it would predict early state formation there, which didn't happen.

I also have doubts as to whether effective agriculture spread to Southeast and South Asia along the lines of his model, and whether an archaeobotanist of those regions would recognize the early and very effective agriculture it places there. My understanding is agriculture had similar problems expanding outside the regions of South Asia where we have early civilizations, similar problems of adaptations of Near Eastern crops to climates unsuited to them.

These two together help the fit of his models of steppe "Miltech" frontier + agriculture - no empires / states in East Central Europe, despite steppe frontier because of no agriculture, and no states in South and Southeast Asia despite effective agriculture because no steppe frontier.

Particularly without the latter, I was doubtful whether his model had any better fit than simple agriculture+time models, and then the steppe recedes from importance in state formation.

Rob said...

@ Kristiina

You gem. Really interesting, even if its lower throughput. The Y haplogroup information alone is invaluable, and it can be correlated with what we already know from GW data, which was missing some Y DNA markers.

In terms of overall numbers:

Afansievo: R1b-M269 x 2, R1b - P297 x 1, Q - M242
Okunevo: R1b- M269 x 1, No x 3, Q1a3 x 2
Sagsai: R1a - Z93 x 4, Q1a3 x 3, C - M130 x1.

I've not listed a couple of others.

What does this suggest about where R1a & R1b came from ?
R1b has been missing so far from all Altaic Meso-Neolithic samples to date, but appears in the Bronze Age.
All 4 R1a, otherwise always a minority in the Altai, seems to be Z-93. And the interesting thing is, they were likely there already in the Neolithic.

Ryukendo K said...

@ Matt

Matt, thanks for the explanation, it does make sense. I noted that the spread of agriculture in his model is quite unrealistic as well. It does seem to me, however, that there was no establishment of a steppe-sown frontier of the type seen in China or the Middle East (i.e. sedentary agriculture geographically close to full-fledged nomadism) in Eastern Europe until the one between the Khazars and the East Slavs, so in broad terms he was right to intervene in some cases at least. Perhaps the underspecification of things like soil types and so on creates weaknesses which he is forced to correct using such interventions. Given all the simplifications of modelling, it is still quite surprising how well his model does.

@ All

With the recent goldberg and rosenberg paper of 17:1 ratio of migrating males to females from the steppe, there is some controversy between different aDNA specialists as to the validity of the results, and e.g. Reich's reception of the paper is poor. I can honestly see why: note that in the mathematical model within that paper, (further described in reference no. 17) after migration creates a mixed population, internal mating within the mixed population is random; for each recursive modelling of % ancestry on autosome and X chromosome for each generation, the only parameters that influence the next generation are the size of further influx from source populations and the sex bias of the influx; the mixed population inherited from the previous iteration is assumed to reproduce panmictically. This is quite unrealistic. The model is unable to distinguish, say, a scenario where 17 males move into CEur for every 1 female continuously over multiple generations and another one where 1/10 of all females each generation in the invading group is of local Neolithic ancestry, such that the average quantity of female-derived neolithic ancestry rises over time in the invading group but then the mixed population continues to practice sex-biased incorporation of local females. The statistic Q is not informative in distinguishing such scenarios either. So the claim of 17:1 migration rates should not be taken at face value.

Samuel Andrews said...

I'll make a post at my blog this week comparing ancient and modern Siberian mtDNA.

Davidski said...

@rk

Yeah, that ratio in the Goldberg paper looks too high. I'm sort of having a discussion about that with Greg Cochran here. Hehe.

https://westhunt.wordpress.com/2017/02/27/charm-offensive/

Rob said...

@ Dave

I think you're closer to the mark there in the discussion. The said changes were protracted over 2000 years, at least, and haplogroup G didn't just disappear after the Late Neolithic, we can bank on that. AMR..

Gioiello said...

@ Sam Andrews, who thinks to be a researcher of mt haplogroups, not Roy King

Not only mt J2b1 is above all European, but also I3, if found in Sardinia 11000 years ago, demonstrates the "Italian Refugium". These are the oldest samples found so far, all of I3a and not older than 5000 years:
Poltavka Russia Lopatino II, Sok River, Samara [I0440 / SVP 53] M 2885-2665 BC 1,058,753 R1b1a2a2 CTS1078, Z2105 I3a Mathieson 2015
Unetice Germany Esperstedt [ESP 29 / I0117] F 2272-2039 calBCE (3743±25 BP, MAMS-21496) 874,673 I3a T16086C T16187C C16189T G16230A T16278C C16311T G16391A Adler 2012; Brandt 2013; Haak 2015; Lazaridis 2016
Unetice Germany Esperstedt [I0114/ESP 2] M 2131-1979 BC I2a2b L368+, L181+, P218+, P217+, M438+, L34-, P223-, M223-, [L39+] I3a T16086C T16187C C16189T G16230A T16278C C16311T G16391A Adler 2012; Brandt 2013; Haak 2015; Mathieson 2015; [Additional info on SNP L39 from Hans De Beule and Atanas Kumbarov]
Unetice Germany Benzingerode-Heimburg [BZH 18]
1653-1627 BC I3a T16086C T16187C C16189T G16230A T16278C C16311T G16391A Brandt 2013


Ryan said...

@ Kristiina - Thanks for that paper.

I think Afansievo is probably specifically Tocharian. There seems to be a pretty decent mapping of satem languages to R1a and centum languages to R1b...

Kristiina said...

@Ryukendo “I have to say though, all of their results come from a rather small area around the Altai mountains at the corner where Kazakhstan, Russia, China and Mongolia meet, and its unlikely that so many languages could have their urheimats thickly clustered around a single area.”

Look at this map of the indigenous North American languages. Look at the amount of languages on the West Coast for example.
https://en.wikipedia.org/wiki/Indigenous_languages_of_the_Americas#/media/File:Langs_N.Amer.png

The Afanasievo “Yamnaya Samara” language was probably the first western language in the area and it seems to have covered quite a large area. Okunevo is a later culture located north of Altai on the Yenisei River in the Krasnoyarsk area. Chemurchek is equally old and located in the Mongolian part of Altai. Elunino is also from the same period but it is located northwest of Altai on the Ob River. The distance from Barnaul on Ob to Tsenkher in Eastern Altai is 1200 km. The distance from Tsenkher in Eastern Altai to Krasnoyarsk is about the same. Therefore, these three cultures all have their own area of settlement. In general, I think that many cultures spread along the rivers and a similar language was spoken along the same river.

Sagsai and Munkh-Khairkhan are from 1400-900 and 1700-1400 respectively, i.e. one thousand years later. We know that today Altai is Turkic-speaking, so I find this connection between Sagsai and Turkic very plausible taking into account their yDNA. Today yDNA N in Altai speaks Turkic and I made the presumption that Kurgan N in Altai (if it turns out to be N1b) spoke previously Eastern Uralic language (related to Ob-Ugric and Samoyed branches) but turned to Turkic spakears.

Ryan said...

Re: that pre-contact map, I find the eastern half of North America just as interesting. There are language families there that are relatively young (~3000-4000 years old) that cover an area the size of Europe, and we know virtually nothing about how and why they expanded. A lot of within-group diversity occurs on the coast too - the deepest divisions in Algic and Dene are between coastal and interior populations.

Kristiina said...

@Shaikorth
This is what the Chinese Ranhaer site suggested with respect to the ancient Amur DNA:

"Unfortunately, two women, with no ydna data, have provided strong evidence for the origin of the paternal C3c marker of the Tungus in the Amur River, and it is clear that the C3c should mark the age of iron migration from Siberia to the Amur River Siberian germline group. Rather than C3c spread from the Amur to Siberia."

"The ancient germplasm of the northeastern region of the Neolithic Age belongs to the northeastern or Arctic type, distinct from the Gosibirian type and the North Asian type, and the Tungusic group of the modern Amur River basin is the Neolithic type of the ancient Northeastern and the Iron Age The eastward migration of the ancient Siberian type."

(Google translation from Chinese!)

Azarov Dmitry said...

@Kristiina
Afanasievo is R1b-M269, R1b-P297 (Yamnaya line I presume) and Q-M242 (Maf)
Okunevo is R1b1a2-M269, Q1a3 and N
Elunino is Q1a3
Chemurchek is C-M130
Sagsai is R1a1-Z93, Q1a3a, C-M130 (looks Turkic)
Munkh-Khairkhan is N


And once again we have no traces of relic subclades of R1a(R1a-SRY10831.2*) in aDNA from the Eurasian Steppe and influx of exclusively R1a-Z93 subclades in the Bronze Age. Present distribution of relic subclades of R1a (Iran – Eastern Anatolia) and absence of relic subclades of R1a in aDNA from the Eurasian Steppe are absolutely inconsistent with claims that R1a folks came from steppe.

Shaikorth said...

@Kristiina

Amur River basin covers vast areas east of Baikal in both Russia and China, and there's no doubt the area's been rich in C from the Mesolithic to modern times. Don't see how remains of two women(?) counter the linguists' proposals about Tungusic urheimat.

Remember that Turkic and Tungusic are younger and better studied than Native American families. If their urheimats were in the same location we should see more similarities in the early stages of their development than now, but it is the opposite.

I'll quote Asya Pereltsvaig here:

"If Altaic were a legitimate family, then we would expect that the further back into the past we go, the more similarities we would find. This is certainly true for Indo-European languages: there is more much common ground between Latin and Sanskrit than between Italian and Hindi. But the same does not seem to work for Altaic languages: an examination of the earliest written records of Turkic and Mongolic languages, for example, reveals fewer rather than more similarities. In other words, we can observe convergence rather than divergence between Turkic and Mongolic – a pattern that is easily explainable by borrowing and diffusion, but which makes little sense if these languages descended from a common ancestor."

Early and important words, like "horse" *(x)at in proto-Turkic and *morï in proto-Mongolic are clearly different. Tungusic "horse" is borrowed from Mongolic so they likely were even more separated from Turkic, and perhaps also from horses (and thus the steppe) in the proto-stage. Proto-Tungusic from Southern Altai is way more unlikely than proto-Turkic.

Ryukendo K said...

@ Kristiina
I can actually read Chinese, would you like me to translate it for you? You'll have to send me the source.

Ryan said...

I think that Chinese source is wrong. If C3 spread to NE Asia as late as the iron age, how the hell did it make it all the way to Ecuador?

Kristiina said...

@Ryan “If C3 spread to NE Asia as late as the iron age, how the hell did it make it all the way to Ecuador”

IMO, they suggested that C3c spread from Baikal area to Amur in the Iron Age, but the typical American C3 is a different haplotype, C-P39 (C2b1a1a). According to Wikipedia: C-M48 (C2b1b; previously C3c) is found at high frequencies in northern Tungusic peoples, Kazakhs, Oirats, Kalmyks, Mongolians, Yukaghirs, Nivkhs, Koryaks, Itelmens, and Udegeys, with a moderate distribution among southern Tungusic peoples, Buryats, Tuvinians, Yakuts, Chukchi, Kyrgyz, Uyghurs, Uzbeks, Karakalpaks, and Tajiks.

Therefore, C-P39 may have been in Amur and Okhotsk and C-M48 in Baikal/Mongolia.

@ Shaikorth I have not once used the word Altaic protolanguage. I was only talking about Proto-Tungusic. IMO, Turkic is of different origin, but Mongolian and Tungusic could be daughter languages of the same ancestor language and related to C2b1b-M48 and C2b1c-F4002/M504.

@Ryukendo Ranhaer is not available at the moment

On second thoughts, if that kurgan N is N1c, it is probably the Yakut line, as we know that Yakuts have “Han” like mtDNA and Munkh-Kharkhain site N. 8 is close to Xinjiang.

Ryukendo K said...

@ Kristiina
I really do wonder if they were talking about genetic evidence from the study at all. Its clear that the Armur river samples in no way cluster with 'Northeastern Arctic' peoples, unless one is referring to physical anthropology, in which case the narrow-headed 'baikal-beringian' type paleosiberian peoples are indeed separate from the more classic altaic 'Tungids'--whether or not this reflects phylogenetic realities or no.

I don't disagree that the Y chromosomal clades may have arrived from further west, since C, N and so forth is found in the Altai prior to the Iron Age already, but from the analysis of the Devil's Gate samples, its clear that the Eastern ancestry of all 'Altaic' Siberian and Central Asian groups, even Turkish from Anatolia, is more Southern than the eastern ancestry of Karasuk, Okunev and Iron Age Altai and is approximately as Southern as the Iron Age Russian, meaning that at least one episode of East-To-West gene flow must have occurred prior to the exit of the ancestors of the Anatolian Turks from Russia.

Two interesting observations about that paper I have not brought up yet: the Eastern ancestry of the Karasuk is seen to peak in Central Siberians, which is interesting considering the presence of Yhap Q among them, and the Okunev genomes, for some reason, tend not to demonstrate shared drift with Beringians but with Amerindians and Central Siberians only. The Eastern half of the 'South Siberian Cluster/Altaic cluster' seems to show equal parts Central Siberian, 'Devil's Gate' or Northeast Asian ancestry, and East Asian ancestry proper.

Unknown said...

@ Nirjhar

Thanks.....I wish I had more time to spend there...


@ RK

The shared haplotype segments are large, rare, and specific to specific individuals, not small and common segments which could have been inferred by a distant common ancestor to a large population.

For example, here is a small snapshot of the sorted output for Sein with various individuals. Notice that Lithuanian 3 tops his list. The guidelines are any segment with a fastIBD score < 10 -10 provides strong evidence that the shared haplotype is IBD if the length of the shared haplotype length is ≥ 1 cM.

SAMPLE SEGMENT START SEGMENT END IBD SCORE SNPs
Lithuanians3 79114 79889 9.61E-19 775
Pashtun_Afg1 270973 271579 6.97E-36 606
Kyrgyz_Tajikstan2 225871 226422 4.53E-17 551
Punjabi_Jatt 249078 249595 2.88E-27 517
India_Ho 148680 149172 2.31E-14 492
.Kurd_SE 403597 404085 3.78E-31 488
Punjabi_Sikh 395809 396267 7.18E-23 458
Kets1 156039 156484 3E-22 445
Buryats13 84425 84860 1.1E-13 435
Tajik11 106057 106491 9.24E-14 434
Pashtun_Afg1 140306 140740 2.67E-21 434
Selkup3 18502 18931 9.41E-16 429
Pashtun_Afg2 20532 20961 9.09E-15 429
Pashtun_Afghan1 22171 22594 9.56E-21 423
Pashtun_Afghan5 75131 75542 1.06E-15 411
Iranians3 378356 378765 5.06E-26 409
Tajik10 159572 159980 8.46E-12 408
Buryats14 155912 156315 9.95E-17 403
Swedes2 57409 57808 1.18E-14 399
Pashtun_Afghan2 205961 206349 6.68E-19 388
Evens_Magadan5 40648 41031 1.27E-16 383
Tajik_Shugnan 387042 387414 3.33E-14 372
Lezgins1 9233 9603 3.63E-13 370
Pashtun_Afghan3 184162 184530 9.9E-15 368
Selkup2 156135 156502 5.44E-19 367
Cossacks2 54557 54922 7.61E-12 365
Pashtun_Afghan3 362684 363047 4.55E-17 363
Kyrgyz4 119001 119358 3E-18 357
Pashtun_Afg1 160061 160416 1.87E-16 355
Pashtun_Afghan2 8880 9233 1E-13 353
Tajik11 60226 60574 8.22E-16 348
Germans3 304960 305302 1.78E-12 342
.Kurd_C1 147290 147631 3.75E-17 341
Cossacks_Kuban2 155673 156014 1.47E-12 341
Lezgins2 161332 161672 7.11E-17 340
Azerbaijanis2 206530 206870 2.32E-13 340
Pashtun_Afghan3 387097 387431 2.61E-13 334
Forest-Nenets2 156195 156521 3.58E-13 326
Pashtun_Afghan2 329854 330178 1.54E-13 324
India_Thakur 368239 368561 4.24E-13 322
Pashtun_Afghan2 162148 162469 1.27E-14 321
.Kurd_F8 315939 316260 5.73E-18 321
Kazakhs2 119028 119348 8E-17 320
India_Mpradesh 264969 265283 3.14E-12 314
Pashtun_Afg2 40313 40621 1.89E-13 308
Pashtun_Afghan4 98164 98467 3.32E-16 303
Pashtun_Afghan1 278270 278572 7.45E-14 302
Pashtun_Afghan4 220478 220776 2.47E-12 298
Kabardins2 44366 44662 2.79E-14 296
Tajik9 249138 249432 1.7E-17 294

Unknown said...

@ RK

Here is the output for Iraqi Kurd, again showing Lithuanian samples close to the top of the list inspite of other W Asians in the run.

SAMPLE SEGMENT START SEGMENT END IBD SCORE SNPs
Kabardins1 194851 195356 1.89E-22 505
Lithuanians1 349344 349793 7.11E-15 449
Russia_Vepsas2 285964 286362 5.94E-13 398
Karelians2 74464 74857 1.13E-17 393
Azerbaijanis2 4756 5129 2.71E-12 373
.Kurd_F3 368103 368473 7.02E-16 370
Russia_Vepsas1 349424 349780 5.80E-15 356
Assyrians1 126904 127252 7.13E-17 348
Tajik_Rushan2 10476 10823 6.66E-13 347
Tatars1 91591 91930 3.79E-16 339
Circassians1 207496 207824 1.17E-13 328
Buryats10 161361 161680 1.32E-14 319
Balkar2 74252 74570 7.08E-15 318
Ingrians1 286453 286770 2.96E-16 317
Tajik2 322225 322535 7.98E-14 310
Assyrians3 133403 133708 8.67E-12 305
.Kurd_SE 285280 285582 6.37E-24 302
Cossacks_Kuban2 111583 111882 6.06E-13 299
Buryats13 122409 122703 2.53E-13 294
Germans3 387495 387789 6.83E-12 294
Kumyks3 149371 149656 2.09E-17 285
Tajik11 88606 88884 7.78E-13 278
Tatars3 46587 46854 5.15E-21 267
Kryashen-Tatars5 74326 74586 5.32E-14 260
Koryaks12 122150 122408 2.09E-14 258
Tundra-Nenets2 115442 115696 4.84E-16 254
Tundra-Nenets3 138351 138585 5.03E-15 234
.Kurd_F6 218334 218563 4.26E-14 229
.Punjabi_Gujjar 390304 390531 3.28E-13 227
Kryashen-Tatars4 86103 86324 1.58E-12 221
Punjabi_Sikh 43100 43320 8.97E-17 220
Bashkir3 229497 229713 7.15E-12 216
Azerbaijanis1 28340 28555 9.26E-12 215
Mongolians5 122260 122471 7.03E-12 211
Abkhazians1 85209 85420 1.19E-13 211
Balochi1 108501 108711 3.62E-12 210
Tajik7 91591 91800 3.70E-15 209
Shor1 71293 71501 4.44E-13 208
North-Ossetians1 8971 9178 6.17E-12 207
Montenegran3 303323 303524 3.44E-12 201
Armenians3 282747 282948 2.28E-12 201
Forest-Nenets1 373058 373258 5.34E-13 200

Kristiina said...

@Ryukendo “it is clear that the Eastern ancestry of all 'Altaic' Siberian and Central Asian groups, even Turkish from Anatolia, is more Southern than the eastern ancestry of Karasuk, Okunevo”

Sure, but IMO there is a difference between South Siberian Ice Age refuge yDNA and China/Tibet yDNA and Amur/Okhots yDNA from the Mesolithic period. When The Silk Road started, there was intense geneflow between China and Central Asia. It was clear from the Allentoft paper that the southern geneflow started only in the Iron Age. IMO, there was surely yDNA N and C in the South Siberian Ice Age refuge and they lacked Southeast Asian component.

C is one of the oldest haplogroups and extends from Spain to America/Australia.

Nirjhar007 said...

Indeed Pakths and Lithuanians are close ;) .

Shaikorth said...

@Kristiina

It doesn't have to do with the Altaic hypothesis - the principle is the same without presuming a shared predecessor language, if proto-Tungusic was spoken in southern Altai we should see increasing similarity to Turkic in the earlier stages, yet we don't.

Also that N is recent enough to be Y16323, the subclade of L1026 that made its way to eastern steppe and Beringia. https://www.yfull.com/tree/N-L1026/

Kristiina said...

@Shaikorth “Also that N is recent enough to be Y16323, the subclade of L1026 that made its way to eastern steppe and Beringia. https://www.yfull.com/tree/N-L1026/”

Sure, NO(xO) is pretty vague!

@Shaikorth “if proto-Tungusic was spoken in southern Altai we should see increasing similarity to Turkic in the earlier stages, yet we don't.”

Not if the Proto-Turkic came from Kazakhstan and established itself in Altai only with Sagsai (1400-1100 BC) and Proto-Tungusic developed in Eastern/Central Mongolia 2000-1500 BC.

jv said...

postneo said...
I had commented last year that the supposed skewed sex ratio of yamnaya incursion to europe is an artifact. west Eurasian mtdna had already migrated to the steppe before the yamnaya incursion. So the yamnaya were bringing nothing new on the female side when they moved in to Europe.......................................................... Not true. Just pointing this little bit of mtDNA info into the comment arena. Specific clades of mtDNA H, like mtDNA H6, came in to Europe with the Yamnaya(Catacomb) Culture migrations. This lineage was not found in Central/Northern Europe Neolithic Cultures and is associated with Corded Ware Culture (Espersedt Germany approx. 2400 BCE)

Shaikorth said...

You'll have hard time finding linguists who agree with that placement for proto-Tungusic. Just because some site had Y-DNA C doesn't make it Tungusic, otherwise half of northern Eurasia would be their urheimat.

Languages that developed near the steppe during the metal ages should also have their own words for important things like horses. Turkic, Mongolic and even Ket do have their own words but in Chinese, Tungusic, Korean and so on it all derives from a steppe language (Mongolic).

Kristiina said...

"You'll have hard time finding linguists who agree with that placement for proto-Tungusic."

Well, maybe the origin is in the Baikal forests. However, paleo-languages were spoken before in Amur/Sakhalin, such as Nivkh and Ainu. Tungusic languages seem more modern compared to them.

In any case, only linguistic and genetic evidence counts. Theories change with time. I do not mind if I am wrong or right. It is not important.

Seinundzeit said...

Kurd,

Thanks for posting that wrora, exceedingly interesting!

I'm seeing a lot of Buryat/Ket/Selkup/Even/Kyrgyz matches, which is pretty damn awesome/cool, considering that I always show a very strong signal for this sort of ancestry, with nMonte.

And I've noticed that those Afghan Pashtuns always show strong linkages with me, in all the IBD analyses I've seen of myself.

That's definitely strong evidence for the folklore that claims that contemporary northeastern Pashtun tribes like the Daudzai, Khalil, Mohmand, etc, stayed for many generations in contemporary Paktika/Ghazni, which is from where these Afghan Pashtun samples ultimately hail.

All in all, a really neat analysis for me, one which gives me some clarity concerning my personal ancestral connections, but which also has some relevance in regard to Pashtun tribal history. Again, thanks for running my data wrora, I really appreciate it.

Also, I do agree, the Lithuanian, Swede, German, and Cossack matches make sense in terms of shared steppe ancestry (be it Yamnaya-related, or Sintashta/Andronovo-related).

I do have a lot of Polish, Ukrainian, Russian, Bulgarian, and Macedonian "cousins" at 23andMe, (but oddly, none from Spain, and very few from the British Isles, if my memory serves me right. Also, lots of Americans of generic European descent, but that's a product of their database being very, very skewed towards those people).

In fact, Poland used to be my top "country of ancestry", on that old feature they discontinued at 23andMe, along with Afghanistan, Iran, and India.

I'll be running to work now, so I'll catch up with you later.

Matt said...

@ Ryukendo :

"however, that there was no establishment of a steppe-sown frontier of the type seen in China or the Middle East (i.e. sedentary agriculture geographically close to full-fledged nomadism) in Eastern Europe until the one between the Khazars and the East Slavs, so in broad terms he was right to intervene in some cases at least. Perhaps the underspecification of things like soil types and so on creates weaknesses which he is forced to correct using such interventions."

I see what you mean, but to some degree that also just seems to take us back to the question of whether all areas designated as agricultural by Turchin really are the same. (I guess what I'm saying is if we're saying it has to the right type of agricultural society with the right soil, etc, placed up against a steppe frontier, then it seems we are at least obliged to then test again with just using that type of agricultural society vs others as a variable on its own, and test how much, if anything steppe "miltech" then adds).

Thanks for explaining the Goldberg and Rosenberg lab model some more. There seems no doubt that there was fairly strong sex assymetry with steppe vs european MN ancestry at least between the LNBA Bell Beaker and Unetice cultures. Later cultures and present day are not modelled in the paper, for whatever reason and Corded Ware seems tentatively less so. But I had no understanding of how they got to the 17:1 ratio and a model like 10% Neolithic females into LNBA steppe ancestry per generation seems intuitively more possible.

@ Kurd Dgk, very cool stuff with the segment matching. Also though, I don't doubt connections between Sintashta / Andronovo and South and South Central Asia (it's the mainstream archaeological and linguistic model!) and thus Lithuanians, but would like to see in the context of what the background level of matching segments is with other populations in the Near East.

One thing I am very aware of with haplotypes is although they show direct descent (and that's why I am *very* interested in using them for ancient populations), you do have the thing of a small population admixing into a larger one and experiencing a population increase tends to sometimes share more IBD than local people who have more ancestry in common (e.g. Coop et al - The Geography of Recent Genetic Ancestry across Europe, where British on average share more IBD blocks with Irish than with other British, because haplotypes are more diverse in British).

So something like the relative ranking of populations in their haplotype sharing with Lithuanians can be more informative than, "Does Pop X share more haplotypes with Lithuanians than Pop Y?", which is more heavily influenced by whether Lithuanians are a historically small population with a lot more common haplotypes which are likely to be captured in a small sample.

(Not hugely on topic but I made some graphs of IBS stats vs the haplotype sharing stats, normalised to a max on 1, from Cassidy et al 2016 - http://i.imgur.com/VjtFNJp.png).

Ryan said...

@Kristiina - most is C2-P39 in the Americas, but there's been C-M217 found in Ecuador that is C2-P39 negative too. Not sure which other markers they tested for there though.

C-M48 and C-P39 are both pretty young too I believe? Either way I think there's a pretty good chance that C-M217 has a pretty long history in Beiringia / NE Asia, and so I think there's a good chance that C-48 was already there in the Bronze Age or earlier.

Shaikorth said...

@Matt

Looks like Tvd-based fits get around the issue of small populations' concentrated haplotypes well. In Broushaki 2016 fits using the HO set Lithuanians are only donors for European populations which had relatively recent and direct contact with Balts (Belarusians, Estonians, Eastern Ukrainians) and weren't representing Slavic expansions despite the very high IBD sharing with, for instance, South Slavs seen in Ralph & Coop.

Unknown said...

@ Matt

Unlike IBS, only samples that met the thresholds of 330 SNPs. amd fastIBD score <10-12 made the list for the 3 Lithuanian samples. 404K SNPs.

None of my Iranian, Syrian or Jordanian samples made the lists for the 3 Lithuanian samples. Kurds did, and some Caucasus samples. Kurds and some SC Asians having larger matching IBD segments with Lithuanians than Europeans is thought provoking, suggesting much more recently than late Bronze Age interactions between Kurds / SC Asians and Lithuanian ancestors.

I checked out your normalized graphs. Pretty interesting....although I am not sure I completely understood how samples score 0 on the IBS axis, and non 0 on the haplotype axis.

Here is the output for Lithuanian 1

SAMPLE SEGMENT START SEGMENT END IBD SCORE SNPs
Russians-Central2 149869 150516 4.92E-28 647
Lithuanians3 259920 260529 1.45E-29 609
Kyrgyz2 48347 48931 9.46E-21 584
Russia_Vepsas1 85317 85872 1.63E-19 555
Kryashen-Tatars6 79081 79625 2.07E-14 544
Russia_Komi1 59851 60370 1.29E-23 519
Russia_Vepsas2 303761 304238 7.39E-14 477
Kryashen-Tatars5 129840 130311 1.83E-17 471
Russians-West2 186493 186961 2.15E-15 468
Kyrgyz2 340497 340958 8.86E-14 461
Swedes2 85616 86072 4.30E-17 456
Lithuanians3 251167 251618 1.62E-21 451
Russians-West2 81522 81973 1.16E-16 451
Chuvashes3 372672 373122 6.94E-31 450
.Kurd_C3 349344 349793 7.11E-15 449
.Kurd_C1 225396 225844 2.92E-14 448
Ingrians3 80762 81208 1.60E-17 446
Bashkir1 160555 160992 5.30E-12 437
Germans3 251208 251643 4.45E-14 435
Uygurs 81291 81721 2.24E-13 430
Russia_Vepsas1 79687 80113 2.26E-13 426
Turkmens3 279122 279547 2.64E-12 425
Russia_Mansi1 259943 260363 3.94E-18 420
Swedes2 284272 284686 1.12E-24 414
Cossacks1 386119 386532 8.63E-12 413
Russia_Vepsas3 103957 104361 1.35E-17 404
Russia_Mansi1 251081 251484 3.58E-18 403
Cossacks_Kuban1 333763 334164 3.38E-13 401
Kryashen-Tatars6 81841 82239 3.23E-17 398
Russia_Mansi1 84603 85001 2.30E-16 398
Lithuanians3 400852 401245 4.15E-15 393
Russians-West1 112257 112649 2.16E-15 392
Albanians1 112260 112649 1.69E-13 389
Ingrians2 239728 240116 7.07E-21 388
Russians-Central2 179500 179887 9.14E-14 387
Germans1 141269 141655 8.71E-14 386
Bashkir5 219972 220358 1.57E-19 386
Kryashen-Tatars4 183997 184382 1.61E-15 385
Russia_Mari1 184112 184496 2.41E-12 384
.Parasar 21107 21490 1.98E-13 383
Russians-North2 266243 266626 5.69E-14 383
Russians-North1 280561 280942 5.24E-15 381
Russia_Vepsas1 369973 370350 5.86E-13 377
Montenegran5 101649 102022 2.01E-22 373
Russians-North2 395803 396175 1.27E-13 372
Russia_Vepsas3 244237 244608 3.33E-17 371
Kyrgyz_Tajikstan2 225460 225830 3.30E-14 370
Ingrians1 210503 210873 9.61E-17 370
Russia_Vepsas1 349424 349793 3.07E-13 369
Russians-West2 40571 40938 5.04E-13 367
Karelians3 156430 156796 5.93E-13 366
Russia_Vepsas2 316417 316783 6.47E-15 366
Russia_Mari3 371897 372261 2.40E-16 364
.Greek_AT 81291 81654 2.24E-13 363
Chuvashes3 168165 168528 6.76E-17 363
Tundra-Nenets2 16451 16814 4.52E-16 363
Bashkir3 124473 124834 6.93E-13 361
Tatars2 53770 54131 8.03E-14 361
Lithuanians2 13279 13638 6.92E-15 359
Karelians1 38770 39128 5.90E-25 358
Cossacks2 242323 242679 1.69E-13 356
Kyrgyz_Tajikstan1 270710 271065 1.22E-13 355
Lezgins4 155922 156275 1.65E-12 353
Chuvashes2 155524 155875 6.43E-12 351
Buryats17 92894 93244 5.40E-17 350
Russia_Vepsas2 289778 290128 9.97E-12 350
Bashkir3 237109 237458 6.81E-23 349
Russia_Vepsas1 80138 80487 4.09E-17 349
Russia_Vepsas1 83005 83354 3.90E-16 349
Russians-North1 216966 217312 9.93E-14 346
.Kurd_F3 105481 105825 1.39E-13 344
Russians-West2 332646 332990 7.68E-12 344
Karelians2 279120 279463 5.16E-15 343
Russians-West1 274429 274769 1.24E-21 340
Swedes2 300379 300717 4.85E-12 338
Russia_Mansi2 222421 222756 2.58E-12 335
Russians-Central2 83840 84174 2.11E-12 334
Altaians6 292060 292392 3.03E-13 332
Hungarians2 224031 224361 5.79E-13 330
Tajik7 238682 239010 2.32E-14 328

Unknown said...

The sorted list for Lithuanian 3. I will use many more NE European samples in future runs.

SAMPLE SEGMENT START SEGMENT END IBD SCORE SNPs
Russia_Mari1 148733 149788 1.56E-39 1055
Montenegran1 148889 149788 6.67E-27 899
Albanians3 78722 79608 3.42E-18 886
Kryashen-Tatars6 264849 265695 2.78E-24 846
Bashkir1 148961 149788 1.51E-24 827
Cossacks_Kuban2 148782 149593 1.26E-37 811
.Sein 79114 79889 9.61E-19 775
Cossacks_Kuban2 167395 168041 8.12E-27 646
Albanians1 304734 305355 4.98E-21 621
Lithuanians1 259920 260529 1.45E-29 609
Kryashen-Tatars5 264996 265587 3.93E-12 591
Swedes1 264967 265548 2.47E-12 581
Montenegran3 153543 154094 1.09E-23 551
Russians-North1 318735 319278 3.19E-13 543
Russia_Vepsas1 48077 48616 2.97E-18 539
Kryashen-Tatars6 153618 154143 1.28E-19 525
Albanians2 148964 149487 2.82E-13 523
Chuvashes2 248853 249372 3.01E-18 519
Cossacks1 149109 149628 1.27E-18 519
Lithuanians2 264996 265513 7.04E-12 517
Montenegran5 78457 78974 4.75E-20 517
Russians-Central1 201096 201597 2.23E-24 501
Russians-North1 227577 228073 7.95E-14 496
Russians-North1 113111 113604 2.61E-19 493
Bashkir3 357664 358156 1.93E-20 492
Russians-West2 268311 268793 3.09E-20 482
Kryashen-Tatars4 232492 232965 2.72E-22 473
Russia_Mansi1 251167 251632 1.61E-17 465
Russians-West2 245842 246298 1.01E-15 456
Russia_Vepsas2 231848 232302 1.30E-17 454
Lithuanians1 251167 251618 1.62E-21 451
Russia_Komi1 44353 44803 1.01E-18 450
Karelians2 207040 207484 2.21E-15 444
Hungarians1 84783 85225 5.30E-22 442
Pashtun_Afghan3 225503 225937 5.59E-16 434
Russia_Vepsas4 322155 322586 1.49E-17 431
Tatars1 79589 80018 1.34E-12 429
Russia_Komi1 10891 11314 1.73E-13 423
Mongolians3 222408 222827 3.60E-17 419
Russia_Komi2 225089 225502 2.92E-18 413
.Marathi1 20911 21323 2.28E-16 412
Russia_Komi1 248765 249175 5.60E-12 410
Russians-West2 65595 66002 2.68E-18 407
Cossacks_Kuban1 327335 327741 1.92E-12 406
Montenegran1 5602 6007 1.51E-14 405
Russia_Mansi1 105588 105993 1.05E-14 405
Russians-West2 41251 41649 5.41E-14 398
Kets3 158088 158485 1.28E-20 397
Tajik4 232668 233064 3.57E-13 396
Russia_Mansi1 259969 260363 6.50E-17 394
Lithuanians1 400852 401245 4.15E-15 393
Russia_Vepsas3 368533 368926 1.03E-19 393
Kryashen-Tatars6 304966 305355 1.55E-12 389
Circassians2 239937 240321 1.36E-12 384
Kyrgyz_Tajikstan1 304973 305355 1.55E-12 382
Tajik11 304973 305355 1.55E-12 382
Tatars3 63208 63588 1.31E-15 380
Swedes1 217860 218238 3.65E-28 378
Russians-West1 292984 293360 4.68E-14 376
Russia_Shor1 139407 139780 6.62E-14 373
Russia_Komi1 174043 174411 1.19E-19 368
Russia_Vepsas3 54029 54397 2.69E-14 368
Tajik9 213154 213522 4.43E-14 368
Turkmens3 315826 316189 1.81E-15 363
Cossacks1 6198 6559 2.79E-12 361
Cossacks_Kuban1 217501 217861 3.86E-15 360
Chukchis3 10997 11356 7.20E-12 359
Russians-West2 232547 232903 4.78E-21 356
Cossacks_Kuban1 165700 166054 2.32E-15 354
Russia_Mari4 2624 2978 7.53E-15 354
Russians-Central1 12986 13340 1.29E-12 354
Hungarians2 97104 97456 5.04E-16 352
Swedes1 264243 264595 5.09E-12 352
Mongolians1 185890 186239 2.51E-19 349
Russia_Mari3 207040 207388 3.14E-17 348
Uzbek1 20968 21316 1.56E-16 348
Chuvashes1 7387 7733 2.05E-20 346
Buryats5 302775 303120 2.01E-13 345
India_Kurmi 20968 21313 5.19E-16 345
Cossacks1 183814 184158 1.68E-15 344
Russia_Vepsas3 204968 205312 5.21E-17 344
Russians-North1 9786 10130 4.74E-16 344
Germans3 82850 83193 2.51E-14 343
Cossacks1 369643 369985 3.51E-13 342
Tundra-Nenets3 129759 130101 9.77E-21 342
Russia_Vepsas3 250299 250640 2.49E-17 341
Russians-West2 174184 174524 4.10E-15 340
Russia_Mansi1 245959 246298 1.99E-13 339
Russia_Shor2 134528 134864 1.18E-12 336
Montenegran5 195940 196274 9.83E-18 334
Karelians2 275262 275592 4.46E-18 330

Unknown said...

Matt,

It just occurred to me that in addition to Iranians, Syrians, and Jordanians, Armenians and Georgians also did not make the list for IBD with Lithuanians. This deepens the mystery a little as to how the Kurds got matches with Lithuanians. Perhaps the NE Caucasus or C Asians route?

Matt said...

@ Kurd Dgk, thanks for that, very nice confirmation it isn't just a general bloat of Lithuanian haplotype sharing, and even more interesting than before.

Btw here's the median haplotype sharing from Cassidy 2016 against IBS (or closest equivalent I had a stat for), without that scaling of everything to 1 which I had added and which was maybe confusing - http://i.imgur.com/Mio3LWa.png

Matt said...

@ Shaikorth, to be honest, I don't totally know about the details of Broushaki's fit work. I don't doubt them. Kind of more trying to talk about that the method isn't going to be as simple as we naively read it to be "Most haplotypes, most ancestry" more than I'm saying haplotypes are not good for fitting (actually the opposite should be true if you've got high enough quality samples).

Btw, bit off topic but as we've had discussions about the utility of Fst (vs TVD, etc) in the past I thought you might be interested in this.

I found when I was trying to use PCA on Fst that they produced horrible plots with no relationship to the normal genetic PCA.

What I've found recently is that this is totally the wrong method, and what I should have been using was Principal Coordinates Analysis, another mathematical technique and feature in Past3, and treating the input data as a supplied distance matrix, not euclidean dimensions.

(Here's a couple of pages explaining the differences and why PCoA is better than PCA for a distance matrix:
http://occamstypewriter.org/boboh/2012/01/17/pca_and_pcoa_explained/ and https://folk.uio.no/ohammer/past/multivar.html. Non-metric MDS scaling which I was trying to use before to deal with Fst data, was also the wrong technique :( ).

When I used PCoA on the Fst matrix of modern West Eurasians from Lazaridis 2016, in this manner, it really reproduced a neat version of the classic West Eurasia PCA:

Dimensions 1v2 and 1v3: http://i.imgur.com/yrbKhPQ.png

(Dimension 1: North Europe vs Southern ME, Dimension 2: Sardinian vs Caucasus, Dimension 3: Sardinian+Caucasus vs East Europe and Southern ME)

Dimensions 1v4 and 1v5: http://i.imgur.com/OJZQbay.png

(Dimension 4: Caucasus and Basque / NW Europe vs East Europe, Dimension 5: Basque vs Sardinian)

Dimensions 1v6: http://i.imgur.com/aELUnMr.png

(Dimension 6: Northwest European vs Northeast European and Basque).

Neighbour joining on the dimensions produces an almost perfect duplicate of neighbour joining on Fst http://i.imgur.com/Yx1Hvnw.png (so the dimensions and Fst scores are very much congruent).

(Note I've removed some heavily drifted subpopulations here who messed with things a little (not by breaking distances, but by dominating some of the low dimensions themselves).

Works nice and logically for ancients: http://i.imgur.com/aCiyXRH.png (same plus a few moderns who fit where they should - http://i.imgur.com/kHeNJmC.png), and also for modern world populations: http://i.imgur.com/n7Fpfpm.png (zoom on West Eurasia for 1v2 and 1v3 - http://i.imgur.com/hMJOAuR.png).

I would like to try this with TVD distances, if you've found any nice numerical matrixs of TVD distances in any online papers.

Shaikorth said...


Alright, a quick search got some sets of Tvd's. POBI paper has two different tables, between-population and between-inferred-ancestry profile-of-population, and Fst to compare with (UK only)

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4632200/

Then there's a Tvd set with numerous SSA samples and 1000genomes pops
https://elife-publishing-cdn.s3.amazonaws.com/15266/elife-15266-fig2-data3-v1-download.xlsx
https://elife-publishing-cdn.s3.amazonaws.com/15266/elife-15266-fig2-data4-v1-download.xlsx

Chromopainter is apparently computationally (time)intensive and Tvd's first bigger utilization in this way was in POBI IIRC, so there shouldn't be many of these tables around yet.

Matt said...

OK, trying out the African TVD scores first as they seemed easiest to grab and use:

Coordinate 1v2: http://i.imgur.com/s9UI5UL.png (1 - looks like West vs Southeast Africa; 2 - looks like deviation by West Eurasian influenced populations in Northeast Africa).

Coordinate 1v3: http://i.imgur.com/euCm5bz.png (3 - looks like Khoi-San populations from South Africa breaking off from the others).

Coordinate 1v4: http://i.imgur.com/YLtpbun.png and Coordinate 1v5: http://i.imgur.com/zswLvEN.png (At this point, I'd have to go back to the paper to *try* and intrepret what these dimensions mean).

Neighbour joining trees: http://i.imgur.com/AkxMBUP.png. Top two of three in link are based on the PCoA output, while the third is just based on the matrix. Structurally I'm not so sure if this PCoA is as successful in mirroring the matrix. (I've noticed while experimenting that PCoA does seem to have some drawback in terms of seeming to sometimes produce spurious higher dimensions which don't actually change relative position of populations but accumulate to inflate drifts from each other. Slight drawback. There are some settings in PAST3 with a "transformation exponent" that may be able to limit these, but I'm not so confident how to select the right one)

(Couldn't include the 1000G here as you need a full matrix with both row and column, and columns were absent for 1000G).

IRC they had a TVD matrix in the West Eurasian admixture equivalent of the African paper you've linked, but only in heatmap format.

Shaikorth said...

There are rows and columns for 1000genomes Eurasian pops between themselves, and from the African sheet you can add individual SSA pops representing different regions, I checked the SSA-Eurasan Tvd's in both sheets and they're compatible.

Matt said...

Ah, didn't see there were two spreadsheets that could be joined that way, OK:

1000 G samples - http://i.imgur.com/Icld76R.png (PCoA) http://i.imgur.com/1aYbmHj.png (neighbour joining)

African samples plus 1000 G - http://i.imgur.com/SBLfby5.png

Anonymous said...

Interesting. Exact maps would be nice for posts like this.