October 16, 2013

Neolithic super-grandfathers of the Chinese

An interesting new paper on the arXiv. The title focuses on three star-like Neolithic expansions that account for ~40% of the modern Chinese. From ~3 men to ~270 million male descendants over ~6,000 years ain't too shabby.

Also of interest in the paper is the authors' work on the rest of the Y-chromosome tree (see Figure on the left). As always with age estimates, the details matter. From the paper:
It is worth to point out that
recently, Wei et al. published a similar study about Y chromosome sequencing of 36 individuals (mainly Haplogroup R1b and E1b), in which 3.15 or 8.83 Mbp range was sequenced 19, and they achieved a time of out-of-Africa at 57 – 74 kya using various methods, which is slightly older than our result  (54 kya), although the same mutation rate of 1×10-9 substitution/base/year were employed. The difference could be ascribed to the regions chosen for date estimation; we compared the regions that Wei et al. and we studied, and found that in their study, the SNP density in the region that was sequenced only in their study is significantly higher than that in the region that both studies have sequenced (P less than 0.005) (Table S3).

It thus seems that the time estimates may be lower than true. An interesting new finding from the paper is the near-simultaneous D/CF and C/F splits. The authors comment:
It remained mysterious that how many times the anatomically modern human migrated out of Africa, since that among the three superhaplogrous C, DE and F, Haplogroup F distributes in whole Eurasia, C in Asia and Austronesia, D exclusively in Asia, while D’s brother clade E distribute mainly in Africa 62, so there are two hypotheses, 1) haplogroups D and CF migrated out of Africa separately; 2) the single common ancestor of CF and DE migrated out of Africa followed by a back-migration of E to Africa. From this study, the short interval between CF/DE and C/F divergences weakens the possibility of multiple independent migrations (CF, D, and DE*) out of Africa, and thus supports the latter hypothesis 63 (Fig. S2 a).
I have argued for haplogroup E back-migration into Africa before in this blog, so it's nice to see that this idea is gaining some supporters.

arXiv:1310.3897v1 [q-bio.PE]

Y Chromosomes of 40% Chinese Are Descendants of Three Neolithic Super-grandfathers

Shi Yan et al.

Demographic change of human populations is one of the central questions for delving into the past of human beings. To identify major population expansions related to male lineages, we sequenced 78 East Asian Y chromosomes at 3.9 Mbp of the non-recombining region (NRY), discovered >4,000 new SNPs, and identified many new clades. The relative divergence dates can be estimated much more precisely using molecular clock. We found that all the Paleolithic divergences were binary; however, three strong star-like Neolithic expansions at ~6 kya (thousand years ago) (assuming a constant substitution rate of 1e-9/bp/year) indicates that ~40% of modern Chinese are patrilineal descendants of only three super-grandfathers at that time. This observation suggests that the main patrilineal expansion in China occurred in the Neolithic Era and might be related to the development of agriculture.

Link

39 comments:

Lank said...

Or CT and DE originated in Africa, but DE was the only lineage that survived in Africa, after the OOA migration.

I don't know why this possibility is so rarely discussed in the literature. Especially when the age of CT and CF have been shown to be roughly the same, making it highly likely that they originated in the same region of the world.

barakobama said...

I have always noticed a huge difference between how Y DNa and mtDNA haplogroups are distributed. In mtDNA for Europe and the Near east every haplogroup and alot of its deep subclades are everywhere. Even though U5 originated in Europe probably 35,000-50,000ybp its deep subclades of U5a and U5b are over 1% in most of the Near east and North Africa. But the only Y DNa haplogroup that originated in Europe in Palaeolithic is hg I and it is almost completely exclusive to Europe. mtDNA H is 40-45% in all of Europe(except Sami and Volga Russia) H1 takes up about 30% of that H and the majority of H1 is H1a and H1b and even deeper subclades than that like H1a1 and H1a2 are all over Europe. Almost all European J1 which takes up about 5% of mtDNA in Europe is J1c and deep subclades of J1c are all over Europe.

So many mtDNa haplogroups deep deep deep subclades are all over the place. But it is not like that at all for Y DNA. Y DNA seems to define people groups. Like R1b1a2a1a L11 which takes up about 50% of west European paternal lineages. It most likely was spread with Germanic and Italo Celtic languages starting 5,000ybp. That haplogroup defined Germanic and Italo Celtic tribes when they spread. Same for Y DNa R1a1a1b1 Z285 and Corded ware culture and today Balto Slavs. Same for R1a1a1b2 Z93 and Tocherian and Indo Iranian tribes.

It is hard to explain why constantly mtDNA haplogroups are distribted way differently than y DNA haplogroups. I think it might have to do with the difference between men and women. The reason why Y DNa R1b1a2a1a L11 is so dominate in western Europe is because the Germanic and Italo Celtic tribes conquered it. Same for why R1a1a1b1 Z283 is so popular in former areas of Corded ware culture and R1a1a1b2 Z93 in asia and specifically Indus valley and south central Siberia. Since men fight and die in wars the winners Y DNA becomes more popular and when you have kings like in ancient times who could have as many women as they wanted over 1,000 sometimes their haplogroup becomes much more popular in the future.

Look at England and Lowlands of Scotland Germanic R1b1a2a1a1 S21 is very popular and so is Y DNA I1 its because the Germanic Anglo Saxons conquered those areas of Britain.

Maybe why mtDNA hg's are more spread out is because when there is inter marriage between two tribes the wife moves to the husbands tribe. So Y DNA stays and new mtDNA comes in.

barakobama said...

These four super grandfathers of China cold have been kings or the patriarchs of future tribes like with Abraham and the Jews and maybe these tribes dominated China. 50% of western European men including myself direct male line goes back to the first R1b1a2a1a L11 man just 5,000-6,000ybp. The reason is his descendants conquered western Europe it could be the same thing for China. Over 50% of Slavic men(not southern slavs) direct male line goes back to the first R1a1a1b1a man probably only 5,000-6,000ybp. I know almost nothing about Chinese languages but maybe like how Germanic Italo Celts conquered western Europe and Corded wae culture conquered eastern Europe. Maybe Chinese speakers conquered modern eastern China.

terryt said...

"three strong star-like Neolithic expansions at ~6 kya (thousand years ago) (assuming a constant substitution rate of 1e-9/bp/year) indicates that ~40% of modern Chinese are patrilineal descendants of only three super-grandfathers at that time. This observation suggests that the main patrilineal expansion in China occurred in the Neolithic Era and might be related to the development of agriculture".

I have been suggesting that the evidence has consistently indicates as much for years. And have I had a huge amount of abuse for daring to make such a suggestion!! It is great to see yet more evidence supporting the idea.

"It is hard to explain why constantly mtDNA haplogroups are distribted way differently than y DNA haplogroups".

Y-DNA lines tend to spread across mt-DNA lines. In other words men don't always take wives with them but pick them up from local populations as they move by. It is a mistake to believe that a particular Y-DNA always carries a particular mt-DNA with it. The two lines a surprisingly independent.

Mason said...

Here's another upcoming paper by the same first autor: http://www.ashg.org/2012meeting/abstracts/fulltext/f120120220.htm

Paleolithic human migrations in East Eurasia by sequencing Y chromosomes. S. Yan1,2, C. C. Wang1, L. Jin1,2 1) MOE Key Lab of Contemporary Anthorpology, Fudan University, Shanghai, China; 2) Computational Genetics, CAS-MPG Partner Institute for Computational Biology, Shanghai, China.

Paleolithic human migrations in East Eurasia remains largely unknown due to the lack of sufficient markers derived from the mutations that occurred during that time frame. To tackle this problem, using the sequence capturing, barcoding technology and next-generation sequencing, we identified more than 4,000 new SNPs encompassing most single copy non-recombining region of human Y chromosome. New clades for haplogroups O, C, N, D, and Q could be geographically located. Especially, a few star-like expansions were unveiled, showing strong population growth. The phylogeny of Haplogroup N was radically rearranged, and all the N individuals could now be categorized into either a northern clade N1 or southern clade N2, revealing a Paleolithic migratory routes of the ancestors of Uralic speaking populations. Haplogroup C, especially the East Eurasia-dominant clade C3, could also be separated into at least two ancient clades, suggesting Paleolithic migrations in East Asia. Three major clades under O, M117+, M134xM117, and 002611+, each could be now further classified into several subclades. With these new findings, we proposed the modified the routes and dates for human populations’ migration, especially those in Paleolithic time. A few Y-chromosomal expansions could now be linked to certain prehistoric cultures or ancestors of language families.

You may contact the first author (during and after the meeting) at yanshi@picb.ac.cn


eurologist said...

Neolithic, or, perhaps much more likely, post LGM.

Mason said...

There's a news about ancient DNA results in Southeast China this year.

The German and Taiwanese researchers tested the mtDNA of two 8000-year-old skeletons which were excavated from Matsu islands along the coast of Fujian, Southeast China. The mtDNA of the two 8000-year-old samples belong to haplogroup E and R9, respectively.

The German and Taiwanese researchers are planning to test the Y-DNA of the two samples. I really wonder if their Y-DNA belong to haplogroup O or not.

http://www.ranhaer.com/thread-26011-1-1.html
http://www.chinadaily.com.cn/micro-reading/dzh/2013-07-17/content_9608308.html#blz-insite
http://en.wikipedia.org/wiki/Matsu_Islands

Jim said...

"These four super grandfathers of China cold have been kings or the patriarchs of future tribes like with Abraham and the Jews and maybe these tribes dominated China. "

barack, the legends of the Yellow Emperor and other rulers may reflect this.

The Yellow Emperor legend has a lot of myth woven into it, but it may also reflect some history. There has always been a sense that he was the founder of the Xia dynasty - to the extent that can ever be shown to be hisotrical.

It's pretty clear that there were seaprate cultures in that area, overlapping in time sometimes, fighting each other. There seems to be a linguistic discontinuity between the Shang, who came after the Xia, and the Zhou eras. That is not even taking into account the ancestors of the Hmong and Koereans, who were definitely in the area.

andrew said...

Even if the DE/CF split and the C/F split happened at about the same time, the notion that they were part of the same wave of migration, given their extremely different distribution and lack of thorough mixture of the clades in the same populations argues strong that they were different waves of migration perhaps over time frames too small to clearly differentiate with mutation based dating. Single migration events are going to present as a single more or less stable mix of founder haplogroups in descendant populations.

One can also imagine many possibilities other than those suggested.

Maybe D, E, C and F all arise in Africa near the Out of Africa point of departure and D, C and F are subsequently wiped out in the place where they arose. Maybe men in lineage E even exiled patrilineal clans of men in D, C and F from Africa over an extended dynasty.

Maybe DE arises in Africa, D leaves by a Southern Red Sea route, and CF arises elsewhere in Africa and leaves by a Northern Sinai route.

terryt said...

"I really wonder if their Y-DNA belong to haplogroup O or not".

At 8000 years probably not. O2a at the very most. More likely C*, F2 or some sort of K*.

"Neolithic, or, perhaps much more likely, post LGM".

O probably broke into is subgroups during the LGM but my guess is the expansions shown here are almost certainly Neolithic.

Lathdrinor said...

@Jim

But there is no linguistic discontinuity between the Zhou and the Shang. This has been shown by various linguistic studies of Paleo-Chinese as recorded in the oracle bone script and the Zhou bronze script. Whatever the Zhou were in the beginning, they had adopted the language of the Shang before they replaced them.

There is also no reason to believe that the ancestors of the Koreans were in the area. Very little O2b has been found in China and none of it ancient. I see no cause to believe that China was the source of O2b, the primary haplotype separating Koreans / Japanese and Chinese, and therefore the best candidate for explaining that difference.

The Hmongs were, however, in the area and I do see a deep connection between them and the spread of O3. Looking at the authors' other publications, they advance the argument that the ancestors of the Hmongs bore the rare haplotype O3-M7, today found primarily in Hmongs and in ancient times found in southwestern China. I'm not sure that I buy that, but I do know that the Hmong-Mien homeland overlapped with that of the Sino-Tibetan homeland, and linguistically Hmong-Mien used to be grouped with Sino-Tibetan.

Jim said...

"But there is no linguistic discontinuity between the Zhou and the Shang. This has been shown by various linguistic studies of Paleo-Chinese as recorded in the oracle bone script and the Zhou bronze script. Whatever the Zhou were in the beginning, they had adopted the language of the Shang before they replaced them."

It's inaccurate to say there was no linguistic discontinuity. Lnaguage shift is never - as in no recorded case - that clean or simple. Think of Englsih in Ireland. Clearly there is great disconinuity between those two languages, and clearly there is a load of Irish features in Hiberno-English. And some of these are obscured by the load of Celtic features in English grammar. So language shift is never a neat process, especially in the verb phrase. Mexican Spanish is another familiar example.

"There is also no reason to believe that the ancestors of the Koreans were in the area. "

I was vague. When I said "area" I meant Northern China and the areas that held the cultures that fused into China. And I didn't mean the lineal ancestors of modern-day Koreans, but some related group. They may have shared a culture and lnaguge but originally been separate populations earlier. Speculation. That genetic disconnect you cite is significant.

Speaking of Hmongs, they share their legendary ancestor, Hou Yi, with the Koreans or some gropup that was off in that direction.

"and linguistically Hmong-Mien used to be grouped with Sino-Tibetan.'

That was abandoned 50 years ago. It was based on nothing more than the presence of tonal systems. Phonologicla similarities almost never indicate any kind of genetic relationship between languages, and in the particular case of these two groups, the actual tonal systesm don't resemble each other at all anyway.

On the other hand, and here is more speculation, Korean is a lnaguge isolate and Hmongic is a small isolated langugae family. This kind of thing is pretty common around the north Pacific, but there may be a link we can't see any longer. The advanced state of consonant decay in Hmongic may well have obliterated any evidence, and the lack of historical depth in Korean has the same effect on the Korean side.

"but I do know that the Hmong-Mien homeland overlapped with that of the Sino-Tibetan homeland.'

I don't know about the Hmongic homeland, but the Sino-Tibetan homeland is about as slippery as the PIE homeland. And bear in mind, Sino-Tibetan is not a done deal as a valid linguisitic grouping. After all these years it is still just a working hypothesis, but with a lot of very lose ends.

Whever the Hmongic homeland was, it's clear from evidence in Chinese dialects that the Chu state was originally Hmongic-speaking with a Sinified elite, with Chinese replacing Hmongic through elite dominance over an extended period. The Chu state was had some pretty fluid boundaries, but it was centered in Hubei along the Huai River, right up just south of the Qinling, on the edge of the North China Plain. That puts them up near those Dongyi I mentioned above, possibly para-Koreanic. Nothing solid, but not implausible.

terryt said...

"Very little O2b has been found in China and none of it ancient".

A synthesis of this paper and the recently posted other one on much the same subject suggests that the three main O3 haplogroups expanded with the Neolithic. What's more the expansion was not a simple single episode but rather a series of expansions from the Yellow River region. Because O2a tens to be found to the south of where the expansion started and O2b to the north of it my guess is that O2 represents the first of the series with later O3 expansions separating the two branches of O2.

terryt said...

Sorry. Jim's comment came through after my previous comment:

"Speaking of Hmongs, they share their legendary ancestor, Hou Yi, with the Koreans or some gropup that was off in that direction".

The old paper on Chinese Neolithic haplogroups shows a proportion of O2a in the Daxi, presumed ancestors of the Hmong-Mien. The paper also shows two thirds of the Wucheng as O2a. If the split between O2a and O2b is at all 'recent' that would account for the similarity of myths.

"the Chu state was originally Hmongic-speaking with a Sinified elite, with Chinese replacing Hmongic through elite dominance over an extended period. The Chu state was had some pretty fluid boundaries, but it was centered in Hubei along the Huai River, right up just south of the Qinling, on the edge of the North China Plain".

I think that encompasses the Wucheng region.

Lathdrinor said...

@Jim speculation is fine and well, but given the linguistic and genetic profile of Koreans I see no cause at all to place their forebears in China except in the very generic sense of all O descendants being from China. I simply see no need to do it. Their language is not connected to Chinese. The portion of their genes that differ them from Chinese are largely absent in China. There's no evidence of mass migration from China in the late Neolithic in the archaeological record. At best they're the result of admixture between peninsular / southern Manchurian natives and the expanding O3 star cluster migrants described in this article - though see the absence of archaeological support above - but in that case it's rather inaccurate to call the O3 side of the admixture Korean because that side is not what made them different.

All in all, I don't see the connection, and out there in eastern Manchuria there was plenty of room for cultures and peoples to develop, and Korea had been populated from the early Neolithic, and not by a Jomon people ala Japan. In case the only cause for seeking a connection is via the generic Dongyi exonym, then I rather ignore it because it's known that the Chinese rulers used Dongyi to refer to practically everybody to the east of them and who was not under their control. Even the English were called Yi when they first made contact.

@terryt I think so too. O2 expanded earlier. But O2b strikes me to have expanded mainly after it reached southern Manchuria / the Korean peninsula / the Japanese islands. That is to say, Korean and Japanese O2b were in situ developments and not the result of migrations from elsewhere. What I'm less decided about is whether O2* expanded through a now disappeared land bridge, because its absence in Neolithic sites in western Manchuria challenges the idea of a land route. But then again, there's still the shot that O2* strictly hugged the coast on its northward journey. Not enough data to tell.

Ebizur said...

terryt wrote,

"The old paper on Chinese Neolithic haplogroups shows a proportion of O2a in the Daxi, presumed ancestors of the Hmong-Mien. The paper also shows two thirds of the Wucheng as O2a. If the split between O2a and O2b is at all 'recent' that would account for the similarity of myths."

However, the split between O2a and O2b is not at all what I would call "recent." Although the estimates of different researchers have varied somewhat, both this study and the previous one with a similar phylogenetic tree have placed the split between O2a and O2b at about the same point in time as the split between Q and R.

Would you claim that most Native American and Indo-European populations share common Neolithic ancestry? If not, you should not claim that carriers of O2b share common "Chinese Neolithic" ancestry with carriers of O2a, let alone carriers of any other subclade of O-M175.

eurologist said...

Terry,

Yeah, I did not read carefully - I didn't realize they meant 4-6 branches downstream of O3a-M324. Still, the Chinese neolithic saw major expansions ~8,500 to 9,500 ya - so their ~6,000 ya doesn't match by a factor of ~1.5 if the star-like branching took place in the early neolithic vs. the middle-to-late neolithic.

G, IJ, and P branches then occur around 50 kya (not 33 - 36 kya), when we know that major movements started towards W Asia (end eventually, Europe). And the Q/R split falls at ~37 kya, when people started to settle S Siberia.

Nevertheless, this is of course an important paper - even though it contains gems like: "differing from the case in Europe, where immigrant farmers from the Middle East contributed to the majority of modern Y chromosomes." Yes, the majority of modern European y-haplogroups are G, J, and E. ;)

Bronze said...

I have been saying this for a long time on various anthropological forums. Its very likely that DE* and E* have an euroasian origin and not african. Its hard to swallow for many people of african descent because it means lots of asians came there and took their women from them, resulting in majority of black male lineages being pushed aside and going extinct.
This also explains caucasoid traits in east africa and northern africa, as it came from the middle east. Modern africans are however mostly native african in mtdna and in autsomal dna, so the euroasian male lines where probably spread further into africa by mixed individuals, probably from east africa where the initial mixing took place.

Lathdrinor said...

@terryt @jim the coalescent age of O2a is on the order of tens of thousands of years - see Kumar 2007 and Sharma 2012 for the age of O2a in SEA and South Asia. What 'shared myths' there are between the Koreans and the Hmongs are not liable to be due to any sort of primordial oral tradition. Both peoples did, however, make heavy use of Chinese records, both fictional and historical, to narrate their own past, and it's easy to see how 'shared myths' emerged from the self-other dynamic whereby both groups read the Chinese 'other' into themselves.

Personally, the entire field of trying to read haplotypes into myths is bunk. These myths exhibit obvious features of a late historical environment and thus coalescent date. They're also heavily areal and diffuse readily via culture. Trying to match them up with haplotypes is an exercise in futility.

terryt said...

A few comments re. the paper:

"We found that all the Paleolithic divergences were binary; however,
three strong star-like Neolithic expansions"

With our current knowledge we do find Paleolithic star-like expansions. For example C (although the authors only look at C3 in the paper), MNOPS and even F, when it finally got going.

"After diverged from Haplogroup C, no major split was observed in F
for 18 thousand years, suggesting a strong bottleneck of F lineage".

Hmmm. But not in the C lineage. Surely that indicates that C, inlike F, was not held up significantly before its expansion. To me it is therefore very unlikely that both haplogroups followed the same route east.

"The most surprising discovery in the tree is the three star-like expansions in
Haplogroup O3-M324"

That should not have been a suprise. Some of the same authors had already discovered this to be so:

http://www.ncbi.nlm.nih.gov/pubmed/21505448

According to that paper just O3a1c, O3a2c1 and O3a2c1a make up a substantial proportion of Chinese Y-DNA.

"It would be interesting to know the detailed subclade of the
Austronesian P201*(xM134,M7) clade 23,83, in order to know their relationship to the
continental peoples".

Very much so. Turning to Lathdrinor:

"But then again, there's still the shot that O2* strictly hugged the coast on its northward journey".

Unlikely. One thing that the authors along with Jim and I would agree with is that either the ancestor of N and O, or the haplogroups themselves, entered East asia through southwest China. Almost certainly via where the modern countries of India, China, Burma and Tibet meet. Not along the coast. I thing you are still seduced by 'the great southern coastal migration'. Jim's comment:

"Whever the Hmongic homeland was, it's clear from evidence in Chinese dialects that the Chu state was originally Hmongic-speaking with a Sinified elite, with Chinese replacing Hmongic through elite dominance over an extended period".

What is really interesting about East Asian languages and haplogroups is that they can be connected to a remarkable extent. It is productive to refer to this paper:

http://www.humpopgenfudan.cn/p/E/E3.pdf

Firstly: O1, associated with Austronesian and Tai-Kadai. Both these language groups' expansion is definitely 'Neolithic'. O1 makes up three quarters of the early Neolithic near the mouth of the Yangtze. Makes sense that they would move south along the coast and out to Taiwan and the Philippines.

Furhter up the Yangtze we find the Wucheng, two thirds of which is O2a. That haplogroup is associated with Austro-Asiatic languages including Munda in India. Its closest relation is O2b to the north of China. O2a's expansion south could well be early Neolithic. As Lathdrinor said, 'Korean and Japanese O2b were in situ developments and not the result of migrations from elsewhere'.

Yet further up the Yangtze, at the Three Gorges, we find O3a2b making up a third of the Y-DNA in the Daxi. Presumably, then, O3a2b is associated with Hmong-Mien languages.

North of the Yangtze we find entirely O3a2c1 and some unidentified O3 group. The last is almost certainly the new O3a2c1(xO3a2c1a) from the paper. These two haplogroups are almost certainly associated with Sino-Tibetan languages.

To have such a close association between language and haplogroup implies, very strongly, recent expansions, not Paleolithic ones.

terryt said...

"the split between O2a and O2b is not at all what I would call "recent."

But 'tens of thousands of years' is certainly an exaggeration.

"see Kumar 2007 and Sharma 2012 for the age of O2a in SEA and South Asia".

The aim of both papers seems to be to 'prove' an ancient Indian origin for O2a. That view is very difficult to support when we consider the phylogeny. Indian O2a is mostly O2a1a. Kumar's paper claims 60-70,000 years for O2a. The Sharma paper is a bit more realistic but also seems to ancient.

The most convincing time line I have found for the separation of O2 from both O1 and O3 is 27,000 years. Given that the actual time may be half as long again we still run into a huge problem. If we accept that O2a is associated with Austro-Asiatic, as both authors do, the expansion of either cannot be more than around 10,000 years ago. Otherwise we would not be able to discern an 'Austo-Asiatc family' in the first place. That date would fit well with an early Neolithic expansion from the Yellow/Yangtze River basin as suggested in the paper Dienekes linked to:

"The earliest agriculture in
North China emerged before 10 kya"

Kristiina said...

”Hmmm. But not in the C lineage. Surely that indicates that C, inlike F, was not held up significantly before its expansion. To me it is therefore very unlikely that both haplogroups followed the same route east.”

D lineage seems to be the most bottle-necked of all by a factor of two. It seems that they hardly survived before arriving to the East. Perhaps F and D lineages were fighting the Neanderthals on the northern route. On the basis of the time estimations and distribution maps, it seems to me that the Mode 4 technological complex arrived to China with NO or D lineage 38.000-34.000 years ago . It would also be only natural that it left a strong genetic mark on the area. According to this paper, the QR split took place 24.000 years ago, and my guess is that it happened somewhere in Pakistan. Even if the QR split happened earlier, e.g. 35.000 years ago in Pakistan, that means that NO would have arisen c. 40.000 kya and it would still be in a good position to be the lineage that introduced Mode 4 into China. Moreover, by looking at Fig S2 of this paper, Q is quite far from China and going to a completely wrong direction.

”Modern africans are however mostly native african in mtdna and in autsomal dna”

In Africa there are three autosomal components: East-African, Palaeo-African and West-African. If E back-migrated to Africa 60.000-50.000 years ago, it means in my opinion, that Africans and Eurasians are less different from each other and the Eurasian ancestry is well present in at least East-African and West-African components.

terryt said...

"Would you claim that most Native American and Indo-European populations share common Neolithic ancestry?"

Of course not.

"If not, you should not claim that carriers of O2b share common 'Chinese Neolithic' ancestry with carriers of O2a, let alone carriers of any other subclade of O-M175".

I'm not actually claiming O2a and O2b 'split' during the Neolithic. What I'm suggesting is that they 'spread' during the Neolithic, although probably very early in it. Until their expansions I'm presuming they lived somewhat close to each other and were probably separated by the later O3 expansion.

Jim said...

"@Jim speculation is fine and well, but given the linguistic and genetic profile of Koreans I see no cause at all to place their forebears in China except in the very generic sense of all O descendants being from China. I simply see no need to do it. Their language is not connected to Chinese. The portion of their genes that differ them from Chinese are largely absent in China. There's no evidence of mass migration from China in the late Neolithic in the archaeological record"

Lots of langauges spoken in china have no connection to Chinee other than contact phenomena, and in nay case the territory i am referring to was not part of china at the time.

In any case I never said anything about the Koreans coming from China. I was referring to them as a relic population. that is often how these lnaguage isolates arise. Not only is Korean not genetically related to Chinese, it's not related to any othr language either.

"In case the only cause for seeking a connection is via the generic Dongyi exonym, then I rather ignore it because it's known that the Chinese rulers used Dongyi to refer to practically everybody to the east of them and who was not under their control. Even the English were called Yi when they first made contact."

Actually that's not the only link, as I mentioned. Secondly, it's not very sound practice to think a term is going to retain its identical semantic load for 3,000 years. Also we do not know if the Yi were unitary or a collection of peoples, so it's circular to say that the application tof the term to a collection of peoples means the Koreans were not related or the same as the rest of the Yi. It's question begging.


terryt,
"Firstly: O1, associated with Austronesian and Tai-Kadai. Both these language groups' expansion is definitely 'Neolithic'.

Tai-Kadai has been shown to be a subgroup of Austronesian. This was recent, but there doesn't seem to be much dissent. It is co-ordinate with Malayo-Polynesian and the Taiwanese groups.

"Presumably, then, O3a2b is associated with Hmong-Mien languages. "

If we want to associate anything with the proto-Hmongic speakers, we should look at the people whose language has the most basal split in the group, the She over east in Fujian.

If that haplotype isn't represented at the same rate in all those populations, all that means is that one or more populations language shifted at some pointed. We don't fly apart because Indians and europeans have diffenret haplotype structures in their various popualtions.

terryt said...

"Perhaps F and D lineages were fighting the Neanderthals on the northern route".

I think F's phylogeny argues fairly convincingly that it moved east through India. The delay in its expansion may indicate South Asia was populated by some other group that prevented F's entry.

"it seems to me that the Mode 4 technological complex arrived to China with NO or D lineage 38.000-34.000 years ago"

Isn't the Upper Paleolithic in East Asia northern in origin? That would mean Q introduced it and various NO haplogroups adopted it.

"Tai-Kadai has been shown to be a subgroup of Austronesian. This was recent, but there doesn't seem to be much dissent. It is co-ordinate with Malayo-Polynesian and the Taiwanese groups".

Exactly. And O1 is associated with both language groups. Where Tai-Kadai speakers are O2 it is most easily expalined as the result of language shift.

"If that haplotype isn't represented at the same rate in all those populations, all that means is that one or more populations language shifted at some pointed".

Yes. The pattern of the disjointed language distribution in Southern China suggests Austro-Asiatic expanded first, followed by Hmong-Mien and then Austronesian/Tai-Kadai. Sino-Tibetan was last cab off the rank in South China, although it may have entered Burma reasonably early.

http://en.wikipedia.org/wiki/Austroasiatic_languages

http://en.wikipedia.org/wiki/Hmong%E2%80%93Mien_languages

http://en.wikipedia.org/wiki/Austro-Tai_languages

Lathdrinor said...

@terryt First of all, I want to know what exactly is your methodological issue with the papers I cited. Simple disagreement with their conclusions is not sufficient to counter hard data. I agree that O2a is not 'Indian' - and observe that the papers also call O2a Austroasiatic - but that is not equivalent to agreeing that their coalescent testing gave the wrong KYA.

Also, phylogenetically speaking O2a and O2b both split from O2*, that is to say, O2-P31, M268 xM95, xSRY465. They did not split from each other, and there is thus no cause for them to have existed in the same region prior to the split. It is easy to see how O2a-M95 emerged from O2* in the south while O2b-SRY465 emerged from O2* in the north without overlapping ranges. Indeed this is the parsimonious model but for the existence of a small minority of O2b in SEA, which unfortunately have not been tested for their age and STR diversity. The jury is still out, but all the available tests I have seen place O2a's split from O2* around the late Paleolithic. It is way too early for myths.

@Jim I addressed both of your links - the myth and the exonym. Further you didn't specify what concept of China you were talking about except that the 'Koreans were in the area,' which invokes a geographic concept, and geographically China = China proper and is time independent. Thus it is illogical to say the territory was not part of China at the time, lest your concept of China is a political one. Yet such a concept is irrelevant to the paper, because the paper itself is talking about a geographical concept of China ie the Yellow River valley.

The idea that Koreans are a relic population is agreeable but for the prickly issue of where their source population was. I happen to think that there is no cause to believe that the proto-Koreans were outside of their current range in the Korean peninsula and southern Manchuria during the Neolithic.

A process of elimination during the Neolithic suffices. To the northwest, in the Liao River Valley, we find a population whose tested haplotypes - N primary - are a mismatch for Koreans, while further north we find hunter gathering populations with low genetic diversity belonging to N and C3 haplotypes, again a mismatch for Koreans. To the immediate west, across the sea in eastern China, we do find a source of O3 but practically no O2b whether modern / ancient. To the east is the ocean and to the south the Japanese islands. The choice of where the proto-Korean paternal line came from is then obvious - it was in an area spanning southern Manchuria, the Korean peninsula, and the Japanese islands. I don't know enough about the Japanese Neolithic to say whether it's valid to rule out the Japanese islands for the source of O2b, but in the case that it is, then the choice is even easier.

eurologist said...

According to this paper, the QR split took place 24.000 years ago, and my guess is that it happened somewhere in Pakistan.

Kristiina,

Well, it appears we now have some R in S/C Siberia 24,000 ya. So, while I agree on a NW subcontinent (~Pakistan) origin of all F-descendants including P, the Q-R split likely happened earlier, as I suggested above.

It is amazing to see how the reality of old haplogroup dates in ancient DNA never fails to upset the hopelessly young ages from most molecular clock estimates.

Jim said...

Lathdrinor,

"The choice of where the proto-Korean paternal line came from is then obvious - it was in an area spanning southern Manchuria, the Korean peninsula, and the Japanese islands. I don't know enough about the Japanese Neolithic to say whether it's valid to rule out the Japanese islands for the source of O2b, but in the case that it is, then the choice is even easier."

Well this makes sense in another question. There is a persistent meme that Japonic and Korean are gentically rleated. On the other hand there is some actuual lexical evidence that not Korean but the Goguryeo langauge, not directly ancestal to Korean, and Old Japanese are related.

So this is ironic. Recently some Japanese textbook authors dared suggest a Korean migration event as the basis of the original Japanese polity and were met with death threats from hyper-nationalists. I don't imagine Korean nationalists would be any happier with this infromation.

Yes, i was vague when I said "China". it's not very clearly defined half the time anyway.

terryt said...

"First of all, I want to know what exactly is your methodological issue with the papers I cited".

For a start both papers are in complete dissagreement as to the age of the haplogroup. The authors also make the assumption that all the diversification occurred in India whereas an already diverse population would provide an earlier date. Neither acknowledges that Aautro-Asiatic is an eastern language, not Indian.

"phylogenetically speaking O2a and O2b both split from O2*, that is to say, O2-P31, M268 xM95, xSRY465. They did not split from each other"

Agreed. But you know what I meant. You're getting like Maju now.

"They did not split from each other, and there is thus no cause for them to have existed in the same region prior to the split".

The two haplogroups obviously formed somewhere within the region O2* had occupied. That is unlikely to have been a huge region and so the haplogroups coalesced reasonably close to each other.

"all the available tests I have seen place O2a's split from O2* around the late Paleolithic. It is way too early for myths".

But their spread from the region O2* had occupied need not be anywhere near as ancient as the time of their formation from that haplogroup. I strongly suspect the haplogroup was split by O3's expansion which may have displaced haplogroups we would call O2c, O2d etc. if they had survived.

"I agree on a NW subcontinent (~Pakistan) origin of all F-descendants including P"

I think that the F-derived haplogroups formed along the route F took after it coalesced. G, to me, cannot possibly be of Pakistani origin. Nor IJK. And probably not F3. H, F1 and F4 are obviously South Asian while F2 is East Asian. IJK similarly dropped off haplogroups a it moved east. If the star-like MNOPS expansion is correct than P cannot be Pakistani either. However I strongly suspect the star expansion is not fully resolved and P did originate in Pakistan.

Jim said...

"all the available tests I have seen place O2a's split from O2* around the late Paleolithic. It is way too early for myths".

Actually, no. The Garden of Eden myth at its deepest level is about the shift from Paleolithic to Neolithic, for foraging to agriculture.

. The meme persists in the US as

Lathdrinor said...

@terryt the argument against such a scenario for the scattering of O2*, O2a, and O2b would be that there is so little O2a in O2b areas and vice versa. The idea of fragmentation from a central cluster of O2*, O2a, and O2b ought to have resulted in populations with less skewed distributions between the three. Of course, subsequent drift is an explanation for the disparity, but such neatly regionalized drift effects are not parsimonious and further, we have evidence that O2a was established early in South Asia, whether you believe it came from there / not.

Ebizur said...

Lathdrinor wrote,

"the argument against such a scenario for the scattering of O2*, O2a, and O2b would be that there is so little O2a in O2b areas and vice versa."

I agree. I think that many people have been misled by the nomenclature, and fail to realize that a so-called "subclade" with a name like O2b-M176 may be older in reality than a so-called "primary haplogroup" with its own alphabetic symbol, such as N-M231.

According to the present paper, there may not be any extant O2*: "All the sequenced O2-M268 samples other than O2b-M176 form a monophyletic clade, labeled by F1462, and the SNP PK4 lies inside this clade."

A previous work by the same authors has found O2a*-PK4 and O2a1-M95(xM88) throughout China, but with a South > North cline, and O2*-P31/M268(xO2a-PK4, O2b-M176) also throughout China, but with a North > South cline. However, they also have found one O2b-M176 individual in a sample of 65 males from "Southern China," defined as "provinces of which the capitals [are located south of] the Line of Qinling Mountains-Huai River."

Anyway, it is quite possible that most or all of the O2*-P31 Y-DNA of previous studies, including that Y-DNA line that has been associated with Cao Cao (Emperor Wu of Wei) and his descendants, may belong to O-F1462 (or a redefined O2a-F1462, with O-PK4 demoted to O2a1 and O-M95 demoted to O2a1a). The independent history of this O2a*-F1462 clade, separate from O2a1-PK4, seems to date back to a time just after the separation of R1-M173 and R2-M479, before the separation of R1a-M513 and R1b-M343. That time appears to have been before the MRCA of all extant representatives of N-M231 lived; in other words, both O2a*-F1462 and O2a1-PK4 are older than the entire haplogroup N as we know it.

In summary, even if the presence of O2a*-F1462 in (mostly northern) China does tip the scales one point in favor of a northern East Asian origin of the O2-P31 clade, the separation of the O2a-F1462 and O2b-M176 clades is clearly of deep Palaeolithic time depth. Trying to find cultural links among the modern representatives of these clades is at least as foolhardy as looking for cultural links among all representatives of Y-DNA haplogroup R, and quite a bit more foolhardy than looking for cultural links among all representatives of Y-DNA haplogroup N.

terryt said...

"The idea of fragmentation from a central cluster of O2*, O2a, and O2b ought to have resulted in populations with less skewed distributions between the three".

I disagree absolutely. If members of O2* became separated we would expect two separate haplogroups to develop, especially if that 'original' region had been later over-run by another haplogroup. But that early separation need not have been geographically huge.

"we have evidence that O2a was established early in South Asia, whether you believe it came from there / not".

How early are you proposing? Surely thae fact that O2a in South Asia is reasonably closely related with Austro-Asiatic/Munda languages indicates an arrival more recently than 10,000 years, and quite probably more recent.

"A previous work by the same authors has found O2a*-PK4 and O2a1-M95(xM88) throughout China, but with a South > North cline"

That is not surprising because we know Y-DNA O3 has expanded from the north more recently than O2. That would certainly progressively limit the proportion of O2a in the more northerly regions.

"O2*-P31/M268(xO2a-PK4, O2b-M176) also throughout China, but with a North > South cline".

Now that fits perfectly with a more northern origin for O2*.

"However, they also have found one O2b-M176 individual in a sample of 65 males from 'Southern China,' defined as 'provinces of which the capitals [are located south of] the Line of Qinling Mountains-Huai River'."

One individual is hardly surprising. Humans have ben moving round China considerably in more recent times.

"both O2a*-F1462 and O2a1-PK4 are older than the entire haplogroup N as we know it".

That is entirely possible and I see no problem with the idea. N's origin and expansion does appear to be relatively recent. But its ancestors spent some time in some isolated region after forming from NO and diversifying from the ancestors of O1, O2 and O3.

"the separation of the O2a-F1462 and O2b-M176 clades is clearly of deep Palaeolithic time depth".

But that separation could considerably predate their individual expansions.

"Trying to find cultural links among the modern representatives of these clades is at least as foolhardy as looking for cultural links among all representatives of Y-DNA haplogroup R"

I actually agree with that comment now. It was a wild postulation when I first made the comment.

eurologist said...

Terry,

As to my comment about F-descendants, of course in each line you will eventually find examples that moved away from the source region to far distances; in some lines earlier then in others.

As to G and IJK, let's agree to disagree. As I have often stated, I don't think there was any substantial region in Eurasia west of Pakistan, that was not exclusively populated by Neanderthals but was inhabitable between ~70 and 50 kya. There is no evidence for that, but conversely plenty climatic evidence that rules out human inhabitation for most of the southern portion of west Asia.

As such, I believe that G, IJ, LT, and certain subgroups of K(xLT) all made their first appearance north or west of Pakistan at or after 50 ka, some sooner (perhaps IJ) than others (G, R), and some first more northerly (P, R), others more southerly (LT), and yet others in between (perhaps G and IJ).

terryt said...

"in each line you will eventually find examples that moved away from the source region to far distances"

True, but more so in more recent times. During the Paleolithic expansion was likely to be slow and involve whole families rather than individuals. As a result I believe we can assume that various haplogroups first appear within the region of their 'parent's' geographic range. Perhaps at the margin of that range as selection is likely to be greater at the margins than in the middle.

"As I have often stated, I don't think there was any substantial region in Eurasia west of Pakistan, that was not exclusively populated by Neanderthals but was inhabitable between ~70 and 50 kya. There is no evidence for that, but conversely plenty climatic evidence that rules out human inhabitation for most of the southern portion of west Asia".

That sounds reasonable until we follow the link:

http://arxiv.org/ftp/arxiv/papers/1310/1310.3897.pdf

Quote:

"After diverged from Haplogroup C, no major split was observed in F
for 18 thousand years, suggesting a strong bottleneck of F lineage".

From that I think we can assume F remained isolated and unable to expand anywhere for some time. The region of its isolation is very unlikely to have been anywhere in South Asia because the haplogroup's diversification, and presumably expansion, seems to have been quite rapid once it reached that region. What's more we can see in the diagram at Dienekes' blog that the haplogroup diversifies in a very straightforward, and perhaps revealing, order: from west to east. The first to form is the most westerly haplogroup: G. But it too was apparently unable to expand until long after its ancestral form had split from F. That may indicate it remained behind in the region F had first coalesced in. We also see from the diagram that the next haplogroup to form is IJK, with IJ being basically a non-Indian haplogroup, although probably not as westerly as G. Although the diagram in this paper doesn't show it we know from other sources that F1, F3 and H are all South Asian haplogroups and demonstrate that F's expansion and diversification was quite rapid once the haplogroup had reached that region. LT was the next to diverge from basal K. I agree absolutely that LT is a South Asian haplogroup. But it seems very unlikely to me that 'F' is originally a 'South Asian haplogroup'.

eurologist said...

Terry,

You could make the same argument for C and D, which show similar bottlenecks - yet that does not tell us much of their precise location during that long bottleneck time. Or do you propose they also held out for several to many 10 ka in West Asia and only expanded once in Central/ East Asia? That seems extremely unlikely, given their current distribution.

Apparently, all humans were bottlenecked until around 50 kya - and that could have simply been in different locations within the subcontinent. I prefer the NW for F because of the location of its descendants.

That "peeling off" theory is not so straightforward, since HIJK is mostly not West Asian, and neither are the other Fs. So, F seems to be local to the NW subcontinent when all of this takes place; one group leaves west and becomes G, another almost at the same time (but two branches down) does the same thing (IJ), while P a little later moves north. Almost everyone else moves south or east. Again, the NW of the subcontinent (perhaps the upper Indus River) seems a likely central location for this star-like spread.

The only male haplogroup I can picture surviving in a Gulf oasis or similar is E.

terryt said...

"You could make the same argument for C and D, which show similar bottlenecks"

Exactly, but actually neither suffered such an extreme bottleneck as did Y-DNA F. And, with the information we have at present, their expansions are far more starlike than is that of either F or KMNOPS. According to ISOGG (which obviously may not include all mutations) C has just fewer than half the number (12) of basal mutations F has while D has just two mutations. This difference between C/D and F suggests strongly that a different migration route east was used by C/D and F. Especially if you're going to place F in:

"the NW of the subcontinent (perhaps the upper Indus River) seems a likely central location for this star-like spread".

But the spread is not actually 'star-like'. You even admit as much when you write:

"one group leaves west and becomes G, another almost at the same time (but two branches down) does the same thing (IJ), while P a little later moves north".

The diversification is sequential, not starlike.

"that does not tell us much of their precise location during that long bottleneck time".

But it would do if we could discover the sequence in which the haplogroups formed rather than just the starlike sequence we are currently saddled with.

"Or do you propose they also held out for several to many 10 ka in West Asia and only expanded once in Central/ East Asia? That seems extremely unlikely, given their current distribution".

With the evidence we have we could postulate that the expansion to Central/ East Asia was considerably earlier than that of F.

"That 'peeling off' theory is not so straightforward, since HIJK is mostly not West Asian, and neither are the other Fs. So, F seems to be local to the NW subcontinent when all of this takes place"

You appear to be suffering what, in Maju, I called 'the Garden of Eden Syndrome'. That is the desire to believe in some single point of origin for all humans. That viewpoint is also strongly apparent in your comments in Dienekes' D4500 blog. Besides which your comment doesn't really make sense compared to what you just wrote:

"all humans were bottlenecked until around 50 kya - and that could have simply been in different locations within the subcontinent".

Yes. Different locations. But why the need to confine all the locations solely to 'the subcontinent'? For example Y-DNA D had almost certainly never set foot anywhere in South Asia until some members of the haplogroup entered from northeast of the subcontinent. The distribution of D and E suggests haplogroup DE had a rather wide distribution before haplogroups in the intervening region were wiped out for some reason. And Wallacean C2 haplogroup is just one mutation from basal C and so must have formed there quite rapidly after C's expansion. Even Australian C4 is just two mutations from basal C.

eurologist said...

Exactly, but actually neither suffered such an extreme bottleneck as did Y-DNA F. And, with the information we have at present, their expansions are far more starlike than is that of either F or KMNOPS. According to ISOGG (which obviously may not include all mutations) C has just fewer than half the number (12) of basal mutations F has while D has just two mutations.

Terry,

Because of the differences in interest and financial backing, you can't use anything but complete y-DNA analyses to judge the number of mutations and associated approximate time interval. Otherwise, all the letters of the alphabet would be found under C and O, and O would (now) be called F5a1a2, instead. ;)

Shi Yan et al. put the first C-split just barely before the G-IJK split off F, and the first D-split (in their data set) at the same time as the R-Q split off P; and Poznik et al. put the first E-split (in their data set) after the H-IJK slpit off F.

The diversification is sequential, not starlike.

It occurs during an extremely small time interval, measured by full y-DNA analysis. F had been isolated a long time. For whatever reason, a lot of area and several distinct niches became available in a short time frame (my guess: within a few thousand years after ~55 kya between F --> G-HIJK and the K(xLT) splits M-NO-P-S). That is star-like, in my books. I mean, this covered everything from Afghanistan and Balochistan to extreme SE Asia within a few 1,000 years...

But why the need to confine all the locations solely to 'the subcontinent'?

I have repeatedly stated my reasons, based on both archaeology and uniparental markers, but also matching autosomal DNA results.

terryt said...

"Because of the differences in interest and financial backing, you can't use anything but complete y-DNA analyses"

I'm aware of that. I did say, 'with the information we have at present'.

"Shi Yan et al. put the first C-split just barely before the G-IJK split off F"

That is not actually correct. The only C haplogroup they consider is C3. So what they show is that C3 formed 'just barely before the G-IJK split'. The diagram says nothing about C's basal divergence.

"and the first D-split (in their data set) at the same time as the R-Q split off P"

Their diagram actually suggests the D/E split occurred at that time. If that is correct it would be a very interesting conclusion.

"That 'peeling off' theory is not so straightforward, since HIJK is mostly not West Asian"

The fact that most HIJK haplogroups are not West Asian is not relevant. That argument over the number of haplogroups does not hold up to any real scrutiny, and I'll explain why. H is not part of IJK as I understand the situation so let's confine the discussion for now to IJK. Of the subgroups within that haplogroup neither I nor J are South Asian and so it is doubtful that IJ was. And IJ haplogroup formed before any other IJK-derived haplogroup, if only a short time before. Next to form from K(xIJ) was LT, almost certainly a South Asian haplogroup, or very nearby,
so K(xIJ) must have entered the region by the time LT formed. Then we have what does appear to be a starlike expansion, that of K(xLT). This is where your problems start. If you're going to argue from the majority of haplogroups are found you cannot possibly claim any South Asian source for K(xLT)'s starlike diversification. Of the 7 haplogroups within that clade (M, NO, P, S, K1, K2 and K3) 5 are from southeast Asia and even beyond.

"It occurs during an extremely small time interval, measured by full y-DNA analysis".

How do you see that as a problem? Presumably it indicates that F's expansion, once it escaped from where-ever it had coalesced, was very rapid. That makes sense. Once it had overcome whatever obstacle there was it was able to expand rapidly. That situation seems an unlikely fit to anywhere in South Asia.

"F had been isolated a long time. For whatever reason, a lot of area and several distinct niches became available in a short time frame"

I agree 100%. But its expansion in its region of origin was still restricted. Hence G has a long mutation tail as well and obviously remained isolated a long time further.

"That is star-like, in my books".

Well, no. Starlike implies a complete inability to discern any sequence yet for both F and IJK we see a distinct sequence of formation. If all the sub-haplogroups had expanded rapidly from a single region we would indeed see a real starlike diversification.

"I mean, this covered everything from Afghanistan and Balochistan to extreme SE Asia within a few 1,000 years..."

It's indicative of a rapid expansion, not a starlike one. And my guess is it started a little further east than Balochistan.