July 02, 2010

Admixture in Uyghurs (again)

Gene Expression points me to this letter which suggests that Western Eurasian admixture in Uyghurs has been overestimated in previous studies. They base their claim on the alleged low population coverage of previous studies. I don't buy their argument, mainly because of Behar et al. (2010) which has as good a coverage of Eurasia as one might hope for, and finds (expectedly) Uyghurs to be about a 50-50 Caucasoid/Mongoloid mix. The same is also true for Hazara, shown by the authors as majority Mongoloid, but revealed by Behar et al. (2010) to be about 50-50 as well.

The authors make another claim:
STRUCTURE cannot distinguish recent admixture from a cline of other origin, and these analyses cannot prove admixture in the Uyghurs; however, historical records indicate that the present Uyghurs were formed by admixture between Tocharians from the west and Orkhon Uyghurs (Wugusi-Huihu, according to present Chinese pronunciation) from the east in the 8th century CE.14 The Uyghur Empire was originally located in Mongolia and conquered the Tocharian tribes in Xinjiang. Tocharians such as Kroran have been shown by archaeological findings to appear phenotypically similar to northern Europeans,15 whereas the Orkhon Uyghur people were clearly Mongolians. The two groups of people subsequently mixed in Xinjiang to become one population, the present Uyghurs. We do not know the genetic constitution of the Tocharians, but if they were similar to western Siberians, such as the Khanty, admixture would already be biased toward similarity with East Asian populations.
First of all, the authors forget about the eastern Iranian peoples (Sakas) who most surely were present among the ancestors of the Uyghurs. Second, there is no reason to think that Tocharians were similar to the Khanty, a population with a substantial presence of northern Eurasian Y-haplogroup N does not make a link with Uyghurs likely.

Am J Hum Genet. 2009 December 11; 85(6): 934–937.

Genetic Landscape of Eurasia and “Admixture” in Uyghurs

Hui Li et al.

Link

22 comments:

  1. Very interesting, thanks for posting on this.

    So now, thanks to this paper we know that there is probably a "Central Eurasian" or "NW Asian" distinct population component, which had laid undetected till the moment and that essentially takes over what used to be perceived as West Eurasian component.

    We also discover that East Asia has at least three quite neatly different populations: one inland, another coastal and yet another exclusive of SE Asia.

    It's nice to know. I hope they make more of these deep and well sampled (deeper and better sampled, if possible) cluster analysis studies. They are very informative.

    "First of all, the authors forget about the eastern Iranian peoples (Sakas) who most surely were present among the ancestors of the Uyghurs. Second, there is no reason to think that Tocharians were similar to the Khanty, a population with a substantial presence of northern Eurasian Y-haplogroup N does not make a link with Uyghurs likely".

    Well, Finns and Komis also have that substantial Y-DNA N and yet they cluster almost fully with Europeans. Various Y-DNA R1 populations (Indians and Europeans for example) cluster differently.

    Why? It's easy to explain from my viewpoint: Y-DNA is surely marking older migratory events (Paleolithic in most cases). Instead the autosomal clusters probably emerge from more recent demic episodes that I'd date from at least Late Paleolithic, when populations began expanding and were not anymore small bands moving around.

    Either the LGM or the Neolithic should be understood as lowest possible depth of cluster analysis because these clusters are formed by the regional homogenization of populations' genetic pools. Instead haploid DNA can point to older events (as well as to more recent ones in some cases maybe) because it does not represent just statistical affinity but true lineages fixated now and then again by drift and founder effects.

    So I would think that R and N have been present in the area for some time, maybe spread by the Indoeuropeans (R1 clades, J2, etc.) but within a local Central Eurasian gene pool. Actually, from the viewpoint of archaeology no major Europe to Central Asia migration is easy to discern, so probably Central Asian IEs were all the time specifically Central Asian rather than European.

    You don't have to agree of course. But I can't resist to provide my two cents.

    ReplyDelete
  2. Behar et al.'s study ought to be a turning point for population genetics studies in having a rich number, variety and covering of populations. Wish more populations had been analyzed in this study.

    ReplyDelete
  3. @ Maju : "So now, thanks to this paper we know that there is probably a "Central Eurasian" or "NW Asian" distinct population component, which had laid undetected till the moment and that essentially takes over what used to be perceived as West Eurasian component."

    So, despite the fact that we have south Siberian results that points towards R1a1a + typical west eurasian MtDNA hgs + light-pigmented population with skeletons considered Europoids (and well different than their neighbours) + an archeological culture who appears suddenly around 3,500 BC with several resemblances with the Yamnaya culture (the date would also fit well with the fact that the Tocharian was centum (in the Kurgan theory, that is)) we should consider there is no _direct_ ("recent", so to speak) population link with the west?

    To me, there are signals frequently pointing to the west. Even some where we wouldn't suspect them.

    For instance, in some Japanese dialects and in the Ainu language, the word for "cattle" and "cow" are respectively "beko" and "peko" which are pretty similar to the indo-european stem *peku for cattle (think of latin pecus) while the Japanese cattle have been shown to be partially related to European cattle :

    http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1460404/

    "Unlike Africa, half of Japanese cattle sequences are topologically intermingled with the European variants. This suggests an interchange of variants that may be ancient, perhaps a legacy of the first introduction of domesticates to East Asia."

    Maybe it is to be linked to the population at the origin of the Yayoi culture, apparently arriving between 900 BC to 300 BC in Japan, probably from Jiangsu, the Chinese coastal region just south of the Shandong region.
    The Linzi (capital of Shandong) aDNA samples from 500 BC, were categorized twice (studies of 2,000 and 2,006) as much closer to west Eurasian than east Asian (especially since there were tracks of mixing with East Asians which means the results should have been closer to west Eurasian/European populations "originally").

    http://mbe.oxfordjournals.org/cgi/content/full/17/9/1396

    http://findarticles.com/p/articles/mi_qa3659/is_200608/ai_n17171074/?tag=content;col1

    The Jiangsu peoples could have had links with pastoralist IE populations (where they got some cattle and their word for it, and the savoir-faire?)) just north of them before migrating in Japan, I guess.

    ReplyDelete
  4. And BTW, There is also the fact that Udege population from south-eastern Siberia with several west Eurasian mtDNA hgs (H, H5 H11a, U2e and T2) - despite the rather low size of the samples - that are probably not related to Russian colonization.

    Don't you think there are some insistant tracks/hints deep into Asia from after the paleolithic (haplogroups, language, technology and culture considered as coming coming from way west)?

    ReplyDelete
  5. I am not interested in haplogroups as they are much older than those ethnic groups those haplogroups are found. Plus there is the effects of elite dominance by a small group who become the mother's and father's of many.

    I have seen SNP Structure studies which show that the Caucasoid element in the Uyghurs is complex. The same can be said of the Hazara. There is a Caucasoid element in Uyghurs and Hazara which is more like that found in South Asians and Middle Easterners than in Europeans. Essentially there are a number of Caucasoid founder groups: the Mesolithic survivors in Europe, the Middle Eastern group which also colonized Europe, SW and CS Asia, and an ancient CS Asian Caucasoid group, who also contributed to Europe and the Middle East. Uyghur and Hazara Caucasoid is mostly a mix of Middle Eastern and CS Asian Caucasoid, there is very little of the Mesolithic European element in them which probably came with the I.E speakers in that part of the world in the Bronze Age.

    Anyone can download Structure, and various ethnic group data and run their own admix runs.

    ReplyDelete
  6. "To me, there are signals frequently pointing to the west. Even some where we wouldn't suspect them.

    For instance, in some Japanese dialects and in the Ainu language, the word for "cattle" and "cow" are respectively "beko" and "peko" which are pretty similar to the indo-european stem *peku for cattle (think of latin pecus) while the Japanese cattle have been shown to be partially related to European cattle"

    But that may be a mere matter of sprachbund from when the earliest Neolithic spread (in either direction or both) through the steppe corridors. It's not in any case the word we'd use in Basque for instance: abere, clearly related to Sumerian aba (or something like that - just lost all bookmarks again).

    ReplyDelete
  7. "And BTW, There is also the fact that Udege population from south-eastern Siberia with several west Eurasian mtDNA hgs (H, H5 H11a, U2e and T2) - despite the rather low size of the samples - that are probably not related to Russian colonization".

    Sure. And H8, U2 and others in Central Asia. As I said in Leherensuge, IMO this data should be read as the formation of a distinct hybrid population (i.e. mixed from the West and East Eurasian branches) in the early Upper Paleolithic, then coalescing into a distinct population on their own right.

    But, well, I think that all those lineages, or at least most of them spread in the early UP and not later. From my viewpoint it's easy to assimilate this third population with Western and Easter haploid lineages but a distinctive unique diploid cluster. Because for me the lineages spread first and the clusters formed later, by founder effect too, but specially by local homogenization of the genetic pool across the millennia.

    ReplyDelete
  8. "We also discover that East Asia has at least three quite neatly different populations: one inland, another coastal and yet another exclusive of SE Asia".

    Yes, 'exclusive of SE Asia'. And interestingly the pink East Asian population seems to have expanded over it to some extent. The SE Asian population seems to have no connection to the purple SE Asian one. And yet to many people the two groups look much the same.

    "Instead haploid DNA can point to older events (as well as to more recent ones in some cases maybe)"

    To me that doesn't really make sense. I'll quote from your own blog on the subject:

    "Y-DNA or even mtDNA, surely scattered in the early Upper Paleolithic when Eurasians were still very much undifferentiated, while these components are surely shallower and represent regional homogeneization processes that happened surely only after the LGM".

    I agree that haplogroups of some sort must have spread in the Paleolithic, but you seem to be saying that autosomal genetic spread occurs without any corresponding movement of either mtDNA or Y-chromosome haplogroups. Surely that's impossible. Consequently modern haplogroup distribution must be more recent than autosomal genetic distribution, at least at a relatively local level. It's easy to visualise how it happens. As men (or less often, women) move into a new region they tend to pick up locals and have children with them. Thus the haplogroup can move through a population without hugely altering the autosomal genetic makeup. You even admit as much at your blog:

    "Buriats and Finns share Y-DNA lineage, as do Polish and many Indians or West Europeans and some Central Africans but it's obvious that these don't correlate too well with autosomal genetic clustering"

    In these cases obviously the local haplogroup is younger than the local autosomal DNA. I agree completely with your comment that the Central Asian population 'was probably created by the admixture of a West Eurasian and East Eurasian migration converging in that peripheral area and then coalescing into a rather homogeneous cluster on its own right'.

    ReplyDelete
  9. So now, thanks to this paper we know that there is probably a "Central Eurasian" or "NW Asian" distinct population component, which had laid undetected till the moment and that essentially takes over what used to be perceived as West Eurasian component.

    That big "Central Eurasian" component or cluster exists probably because of the relative earliness (several thousand years ago) of the Caucasoid-Mongoloid admixture that gave rise to that component.

    ReplyDelete
  10. In the STRUCTURE analysis Subcontinentals (Indians/Pakistanis/Bangladeshis) are clearly much closer to West Eurasians (especially West Asians) than Central and East Eurasians. But that isn't clear on the maps and the genetic distances diagram. The authors should have been more careful in their coloring and shading of the Subcontinent area.

    If the STRUCTURE analysis continued further, I think Subcontinentals would show their own component (a component that is genetically much closer to West Eurasians than Central and East Eurasians).

    ReplyDelete
  11. What I say about Subcontinentals is also corroborated by what we know about Subcontinentals from previous genetic studies.

    ReplyDelete
  12. onur, I told you not to triple post.

    ReplyDelete
  13. If the blog rules are so strict, then I have no chance but to obey them. But I should state that it was much better when there was no comment moderation.

    ReplyDelete
  14. "That big "Central Eurasian" component or cluster exists probably because of the relative earliness (several thousand years ago) of the Caucasoid-Mongoloid admixture that gave rise to that component".

    It would not show up. It would not be even a cluster most probably but in any case it would not show up so early in the Ks.

    "Several thousand years ago" most of the populations sampled were already different, yet they do not show up as such. Russians and Irish appear the same, so do Han, Koreans and Japanese or the diverse SE Asian or Inner China/Mongolia populations. Additionally the recent East Asian (Turco-Mongol) component still shows up as something distinct.

    ReplyDelete
  15. "And yet to many people the two groups look much the same".

    Take a look at the genetic distances mini-tree: West and East Eurasians are in any case close to each other.

    "It's easy to visualise how it happens. As men (or less often, women) move into a new region they tend to pick up locals and have children with them. Thus the haplogroup can move through a population without hugely altering the autosomal genetic makeup".

    Even if men would only have a minor genetic impact, which I agree it can be the case, this impact is not nil. And if it was, then the new Y-DNA could not become dominant, certainly not in Neolithic conditions of rather large growing populations.

    So I would expect such "minor" male-only inputs to have an impact of the order of 10-30%. We do not see that in this case.

    Additionally there are issues of mtDNA as well (Central Asian mtDNA is more Western than Eastern), so it's even harder to explain this in the way you say.

    "In these cases obviously the local haplogroup is younger than the local autosomal DNA".

    It's not "obvious" at all to me. I'm more and more inclined to explain Y-DNA and mtDNA distributions in Eurasia essentially on Paleolithic terms. On early UP ones probably.

    ReplyDelete
  16. It would not show up. It would not be even a cluster most probably but in any case it would not show up so early in the Ks.

    "Several thousand years ago" most of the populations sampled were already different, yet they do not show up as such. Russians and Irish appear the same, so do Han, Koreans and Japanese or the diverse SE Asian or Inner China/Mongolia populations. Additionally the recent East Asian (Turco-Mongol) component still shows up as something distinct.


    Early appearance of a big cluster in the Ks doesn't mean it is very old, especially when it appears at the boundary of Caucasoid and Mongoloid populations, which we know usually has been sparsely populated and thus has been open to big genetic changes by migrations.

    For instance, in some Japanese dialects and in the Ainu language, the word for "cattle" and "cow" are respectively "beko" and "peko" which are pretty similar to the indo-european stem *peku for cattle (think of latin pecus) while the Japanese cattle have been shown to be partially related to European cattle

    Japanese, even if it isn't an Altaic language, has some typically Altaic features, so we can say with confidence (I am assuming that the Altaic homeland is somewhere around Mongolia, thus in East Asia) that it is an East Asian originated language. But I think Sino-Tibetan language family came to East Asia somewhere from the west, from the Caucasoid area. As Sino-Tibetan languages moved to East Asia and spread there further, they tended to become tonal and lose consonant clusters and many of their consonants. Long dead Sino-Tibetan languages like Old Chinese and Classical Tibetan evidently didn't have tones (they only had simple stress as in English) and they had consonant clusters (some of them very long) both at the beginning and end of the syllables (we don't have any knowledge about Sino-Tibetan languages so old other than Old Chinese and Classical Tibetan). Some modern western Tibetan languages still preserve these archaic features (in line with my theory that Sino-Tibetan languages became tonal and less consonantal as more and more they spread among Mongoloids, as the western Tibetan speakers are less Mongoloid than the eastern ones).

    ReplyDelete
  17. onur said,

    "Japanese, even if it isn't an Altaic language, has some typically Altaic features, so we can say with confidence (I am assuming that the Altaic homeland is somewhere around Mongolia, thus in East Asia) that it is an East Asian originated language. But I think Sino-Tibetan language family came to East Asia somewhere from the west, from the Caucasoid area. As Sino-Tibetan languages moved to East Asia and spread there further, they tended to become tonal and lose consonant clusters and many of their consonants. Long dead Sino-Tibetan languages like Old Chinese and Classical Tibetan evidently didn't have tones (they only had simple stress as in English) and they had consonant clusters (some of them very long) both at the beginning and end of the syllables (we don't have any knowledge about Sino-Tibetan languages so old other than Old Chinese and Classical Tibetan). Some modern western Tibetan languages still preserve these archaic features (in line with my theory that Sino-Tibetan languages became tonal and less consonantal as more and more they spread among Mongoloids, as the western Tibetan speakers are less Mongoloid than the eastern ones)."

    Wow.

    First, you assume that Japanese is somehow related to "Altaic" languages. Second, you assume that "Altaic" languages must have spread from the vicinity of modern Mongolia. Third, you assume that Sino-Tibetan languages were originally spread by Caucasoids, despite the fact that the earliest textual references to speakers of these languages identify them indisputably as stereotypical Mongoloids, whereas the earliest references to peoples who have been tentatively identified with Altaic speakers (e.g. Hu, Xiongnu, Xianbei, Dingling) describe them as having exhibited some typically Caucasoid features ("high nose," "long nose," "golden hair," "deep eyes," "jade eyes," "curly hair," etc.) that appeared to be exotic from a Chinese point-of-view. Furthermore, the Altaic language family is often considered to be ultimately related to the Uralic and the Indo-European language families, the majority of whose speakers are mainly Caucasoid. Sino-Tibetan is considered to be an ancient isolated family, or else related to one or another language family of Southeast Asia (e.g. Kradai, Austronesian, Hmong-Mien, Austro-Asiatic), Siberia (Yeniseian), South-Central Asia (Burushaski), or some language families spoken by present inhabitants of the Caucasus region; all these populations exhibit greater or lesser degrees of Mongoloid affinity or admixture.

    I think I am going to have a hard time taking anything you say seriously for a while, onur.

    ReplyDelete
  18. @ Ponto : "There is a Caucasoid element in Uyghurs and Hazara which is more like that found in South Asians and Middle Easterners than in Europeans (...) Uyghur and Hazara Caucasoid is mostly a mix of Middle Eastern and CS Asian Caucasoid, there is very little of the Mesolithic European element in them which probably came with the I.E speakers in that part of the world in the Bronze Age"

    Even though I think there were some Europoid pastoralist R1a that I assume came from the north of the black sea during chalcolithic and bronze age and went deep into Asia (the aDNA seems to confirm it even in Xinjiang), it's still possible y-dna J2 made its way at the same time or even before R1a deep into China.

    The percentage of J2 is rather high in Xinjiang and is found even easter in China if I'm not mistaken. Yao et al 2002 also found out that the Analyses of present-day Han mtDNA sequences from different regions in China detect a residual presence (less than 5%) of west eurasian haplotypes in a few regions.
    These include Qinghai (east to Xinjiang and Tibet) and Yunnan, as well as some coastal regions The only two west Eurasian mtDNA lineages found in modern Chinese correspond to T1 and HV haplogroups according to Yao et al. 2003 - which are also found among the bronze age Kazakhs (lalueza-fox et al. 2004).
    I assume mtDNA T1 and HV could fit well with Y-DNA J2.

    ReplyDelete
  19. PS : but mtDNA hg HV and T1 were also found with the bronze age south Siberian R1a1 (Keyser et al, 2009).

    ReplyDelete
  20. @ Maju : IMO this data should be read as the formation of a distinct hybrid population (i.e. mixed from the West and East Eurasian branches) in the early Upper Paleolithic, then coalescing into a distinct population on their own right.

    As I said on your blog, personally I doubt it, as the bronze age aDNA in Kazakhstan (lalueza-fox et al 2004)) and south Siberia (Keyser et al 2009) seems to reveal a different situation. During bronze age the aDNA in these places was almost exclusively west Eurasian, haplogroup-wise, then later appear more and more east Asian haplogroups (in Bronze age south Siberia : 90% during bronze age, then 67% during Iron age).
    Also the fact that the Afanasevo culture population's skeletons seem Europoid and different than their neighbors don't go in the sense of an admixed west eurasian-east asian population forming as early as the paleolithic time, IMO.

    ReplyDelete
  21. Wagg:

    "As I said on your blog, personally I doubt it"...

    I have extensively replied to you there, so no need to be redundant.

    ReplyDelete
  22. Ebizur,

    I was talking about a theory of mine, not my convictions, maybe I should have been more careful in drawing the line. "Altaic" (I use " because it is a disputed language family) languages are generally thought to have originated and/or spread from somewhere around Mongolia (Chinese physical descriptions of some ancient peoples who are "Altaic" or supposedly "Altaic" speakers all come from historical times, thus late, and they are very Sino-centric, exaggerating small and/or rare physical deviations from the Chinese type. Japanese shows marks of "Altaic" influence, so it probably originated somewhere not very far from the "Altaic" homeland(s).

    Sino-Tibetan-like properties in some Southeast Asian languages are in general explained as results of borrowings from Chinese or areal influnces that affected both Southeast Asian languages and Sino-Tibetan languages as a result of spreading in nearby territories (both geographically and genetically). Very few scholars think that Sino-Tibetan languages have any relation to any other language family or isolate, and none presents any conclusive proof for it.

    So the homeland of Sino-Tibetan languages is very far from clear (even when we take into account early borrowings from other language families and isolates). I admit that my theory about Sino-Tibetan language origins is highly speculative, but still I think it is worth dealing with.

    ReplyDelete

Stay on topic. Be polite. Use facts and arguments. Be Brief. Do not post back to back comments in the same thread, unless you absolutely have to. Don't quote excessively. Google before you ask.