July 11, 2010

mtDNA of Tibet

American Journal of Physical Anthropology doi:10.1002/ajpa.21350

A mitochondrial revelation of early human migrations to the Tibetan Plateau before and after the last glacial maximum

Zhendong Qin et al.


As the highest plateau surrounded by towering mountain ranges, the Tibetan Plateau was once considered to be one of the last populated areas of modern humans. However, this view has been tremendously changed by archeological, linguistic, and genetic findings in the past 60 years. Nevertheless, the timing and routes of entry of modern humans into the Tibetan Plateau is still unclear. To make these problems clear, we carried out high-resolution mitochondrial-DNA (mtDNA) analyses on 562 Tibeto-Burman inhabitants from nine different regions across the plateau. By examining the mtDNA haplogroup distributions and their principal components, we demonstrated that maternal diversity on the plateau reflects mostly a northern East Asian ancestry. Furthermore, phylogeographic analysis of plateau-specific sublineages based on 31 complete mtDNA sequences revealed two primary components: pre-last glacial maximum (LGM) inhabitants and post-LGM immigrants. Also, the analysis of one major pre-LGM sublineage A10 showed a strong signal of post-LGM population expansion (about 15,000 years ago) and greater diversity in the southern part of the Tibetan Plateau, indicating the southern plateau as a refuge place when climate dramatically changed during LGM.



onur said...

I wonder if they tested the westernmost Tibeto-Burman language family speakers (especially Tibetan subfamily speakers), who live in India, Nepal, Pakistan, Bhutan and Bangladesh. Also I wonder the west to east and north to south distributions and ratios of Caucasoid lineages and also Subcontinental-specific lineages among the subjects tested.

Natsuya said...

This is the whole paper:


Ebizur said...

Western Eurasian MtDNA in samples of Qin et al. 2010

1/59 = 1.7% Tibetan/Shigatse, Xizang

2/46 = 4.3% Tibetan/Ngari, Xizang
2/59 = 3.4% Tibetan/Lhasa, Xizang [recorded erroneously as 3.6% in Table 1]

1/61 = 1.6% Tibetan/Chamdo, Xizang

2/58 = 3.4% Tibetan/Nagqu, Xizang
2/59 = 3.4% Tibetan/Shigatse, Xizang
1/59 = 1.7% Tibetan/Lhasa, Xizang
1/51 = 2.0% Monba/Nyingchi, Xizang

1/44 = 2.3% Tibetan/Yushu, Qinghai

Ethnic Tibetan total
1/491 = 0.2% J1b1
4/491 = 0.8% T1
1/491 = 0.2% U2
5/491 = 1.0% U7
1/491 = 0.2% U5
12/491 = 2.4% Western Eurasian total

3/59 = 5.1% Lhasa, Xizang [2 T1 + 1 U7]
3/59 = 5.1% Shigatse, Xizang [2 U7 + 1 J1b1]
2/46 = 4.3% Ngari, Xizang [2 T1]
2/58 = 3.4% Nagqu, Xizang [2 U7]
1/44 = 2.3% Yushu, Qinghai [1 U5]
1/61 = 1.6% Chamdo, Xizang [1 U2]
0/53 Nyingchi, Xizang
0/55 Garze, Sichuan
0/56 Shannan, Xizang

MtDNA haplogroups U7 and T1 seem to occur quite regularly among Tibetans in southwestern Tibet. However, the overall percentage of mtDNA of Western Eurasian affinity in the pooled ethnic Tibetan sample is only 2.4%, which is about the same as the percentages of mtDNA of Western Eurasian affinity that have been found in samples of other East Asian populations (Han Chinese, Koreans, Japanese, etc.). These data may be sufficient to establish the existence of a longitudinal cline in the distribution of Western Eurasian mtDNA lineages (or at least U7 and T1) among Tibetan populations, with higher frequencies toward the west and lower frequencies toward the east. I believe the sampling is inadequate for determining whether there is any latitudinal cline in the distribution of Western Eurasian mtDNA lineages among Tibetans.

Monba total
1/51 = 2.0% U7 Monba/Nyingchi, Xizang

Lhoba total
0/20 Lhoba/Shannan, Xizang

terryt said...

"I wonder if they tested the westernmost Tibeto-Burman language family speakers (especially Tibetan subfamily speakers), who live in India, Nepal, Pakistan, Bhutan and Bangladesh".

Or Burmese. Natsuya's link shows they did consider other Tibeto-Burman speakers. From that link:

"It is noteworthy that sublineage A10b has been found outside the Tibetan Pla-teau (mostly in Tibeto-Burman speaking populations residing in Yunnan China), suggesting that these people migrated southward from the plateau".

Presumably the Burmese themselves entered Burma from the north too. And:

"Notably, some of these pre-LGM Paleolithic sites were found in the northern part of the plateau,where there are very few current inhabitants living with harsh climate, indicating that most regions of the pla-teau was inhabited in the pre-LGM period".

So much for the region being uninhabitable during the Paleolithic.

And from the abstract:

"the analysis of one major pre-LGM sublineage A10 ..."

Supports the theory that mtDNA A is ancient in northern East Asia.

German Dziebel said...

"However, unlike the northern Asian populations, the Tibetan populations contents very low frequencies of western Eurasian genetic components, indicating that the Tibetan Plateau is more like a ‘‘genetic capsulate’’ area."

Unlike other Old World refugia such as the Caucasus and Papua New Guinea, Tibet is very uniform linguistically, with all populations under study speaking the languages of the Tibetan branch of the Sino-Tibetan family.

"The distribution of subhaplogroup D2b (‘‘D2a’’ in (Derenko et al., 2007)) in Asian populations indicated a southern Siberian rather than Beringian origin of haplogroup D2 lineages."

Derenko et al. 2007 based their conclusion of the South Siberia origin of D2 lineages on older dates obtained for D2b (D2a in Derenko dated at 12K) in South Siberia over D2a (D2b in Derenko dated at 7K) in Aleuts, Eskimos and Chukchis. The D2a in Na-Dene (all the way to the Apaches) weren't counted. Neither the D2 (I assume D2a) in Tarahumara (Penaloza-Espinosa et al. 2007. “Characterization of mtDNA Haplogroups in 14 Mexican Indigenous Populations.” Hum Bio 79 (3): 313-320) were taken in consideration. Hence, it's not a fact that D2 originated in South Siberia.

Be it as it may, the close relationship between Arctic/Subarctic and Tibetan D2s is remarkable. It gives some credence to the Na-Dene-Ket-Sino-Tibetan linguistic link. The Na-Dene-Ket linguistic link is now accepted, and the population expansion out of a Beringian refugium, accounting for a wide geographic range of the Ket-Dene languages, is possible. (It also weakens the claims by Sagart and others for a Sino-Tibetan-Austronesian connection.) The authors admit, "In this study, time estimates of several haplogroups resulted in similar old ages of around 22,000 years, making it less possible to be an over estimate of the colonization time. Giving the inhospitable environment of the Tibetan Plateau and the population decreasing during LGM, the effective population size on the plateau might be very small at least in part of the history; therefore, the colonization time of the plateau could be older than our estimates." This may concern the age estimate for D2 in Tibet, which may be closer to the upper limit of the 5-10K range offered in this study and therefore a good fit for the post-glacial expansion that is responsible for the current distribution of the Na-Dene-Ket-Sino-Tibetan languages. This would push the origin of the Sino-Tibetan language family beyond van Driem's Neolithic time frame and would reserve it for the later stage of "expansion" of this family into its presently found locations.

Annie Mouse said...

Nice data.

Looks like the Tibetans might represent the population of Eastern Asia prior to the the south east asians migrating north into China.

eurologist said...

The diversity in the mtDNA seems quite extraordinary. One of my pet theories has been that NE Asia was populated by two migrations: a coastal one, from SE Asia and the (now) islands region, and an inland migration via NE India, the Brahmaputra river, and the adjacent northern Myanmar and Sichuan.

The lack of B in Tibet, and the PC analysis would at least not contradict this - however, I really don't know enough about Asian mtDNA distributions. The authors claim all of Tibet's haplogroups are highly derived and seem to favor the standard, single, counter-clockwise population route.

However, I don't know that one can come to that conclusion without testing NE Indian and adjacent regions, which could have harbored fairly large populations for significant time, so that now Tibetan's appear highly derived. The more important question is, are they more derived than, say, Sichuan and northern China?

German Dziebel said...

"Looks like the Tibetans might represent the population of Eastern Asia prior to the the south east asians migrating north into China."

If for most of Eurasia it's customary to talk about a "western" vs. an "eastern" component, for Tibetans it's more about a "northern" (earlier) vs. a "southern" (later) component.

Ebizur said...

German, quoting the authors of the present paper, said,

"'However, unlike the northern Asian populations, the Tibetan populations contents very low frequencies of western Eurasian genetic components, indicating that the Tibetan Plateau is more like a ‘‘genetic capsulate’’ area.'"

This assertion is way off base. MtDNA of Western Eurasian affinity has been found with much greater frequency in "North Asian" populations sensu stricto, and it also has been found with varying frequency in other populations of East Asia besides Tibetans (especially Mongols, but also in some samples of other populations in northern China, Korea, etc.) For example, Han-Jun Jin et al. 2010 have reported finding the following examples of typically Western Eurasian mtDNA haplogroups in a sample of Udegeys (Southeast/"Amurian" Tungusic speakers from southeasternmost Siberia near the Sea of Japan):

3 H11a
1 H5
1 H
1 T2
1 U2e

5/51 = 9.8% H total
1/51 = 2.0% T total
1/51 = 2.0% U total

7/51 = 13.7% Western Eurasian total

German Dziebel said...

Ebizur, I think the authors simply compared the degree of West Eurasian admixture in Tibetans vs. North Asians. I don't think they meant to deny the presence of West Eurasian lineages in East Asia. If the majority of Tibetans are of Holocene or even Neolithic origin and they are very low on West Eurasian lineages, then this gives us relative chronology for the entry of West Eurasian lineages into North and East Asia. Unless I misunderstood your comment. Good data, though. Thanks.

German Dziebel said...

Further on D2: Volodko et al. "Mitochondrial Genome Diversity in Arctic Siberians, with Particular Reference to the Evolutionary History of Beringia and Pleistocenic Peopling of the Americas" (2008) identified more D2 lineages in Arctic populations (e.g., D2* in the Yukaghir) and pushed the origin of this haplogroup back to 26KY (+/-8). If I'm not mistaken, the Tibetan D2b is Volodko's D2a1b, which diverged from the rest much later: "The ages for the D2a (21.9 ± 8.1 kya) and D2a1 (12.0 ± 5.8 kya) falls within the range of late Pleistocene to early Holocene, but the split between the Chukchi-Eskimo-Aleut-Na-Dene D2a1a and Evenki-Buryat-Yakut-Mongolian D2a1b seems to occurr later, at 6.9 ± 4.1 kya."

Ebizur said...

Oops! Please ignore the complaint I have made in my previous post.

The intended meaning of the passage that German has quoted in his previous post is much clearer in context:

Common ancestry of the Northern Asians and the Tibetan populations
Analyses of 562 mtDNA sequences from 11 Tibeto-Burman populations residing in different regions of the Tibetan Plateau showed that the maternal variation on plateau was largely contributed by northern Asian-prevalent
haplogroups, thus testifying a common maternal ancestry of the Tibetan populations and northern Asian populations (Torroni et al., 1994; Gayden et al., 2009).
This notion was confirmed by both PC analysis and MDA plots, where the Tibetan populations clustered
closely to the northern Asian groups and separated from central and southern Asian populations. However, unlike the northern Asian populations, the Tibetan populations
contents very low frequencies of western Eurasian genetic components, indicating that the Tibetan Plateau is more like a ‘‘genetic capsulate’’ area. Moreover, extremely low frequencies of the South Asian lineages on the plateau reflected the strong ‘‘barrier effect’’ of the Himalayas between the Indian subcontinent and the Tibetan Plateau (Gayden et al., 2007; Kang et al., 2010)."

They simply have noted that the mtDNA pool of the Tibetans closely resembles that of northern Asian populations minus most of the Western Eurasian component(s) that have been found in modern northern Asians. This makes much more sense. In fact, most of the Western Eurasian mtDNA in Tibet seems to belong to either U7, which is usually associated with Indo-Iranian populations or with South Asians in general, or T1, to which a Near Eastern Neolithic origin is usually ascribed, so the origin of the small amount of Western Eurasian mtDNA that is found in modern Tibetans is probably different (and of a more Southwest Asian flavor) from the origin of the Western Eurasian mtDNA that is found with greater frequency in modern North Asian populations.

eurologist said...


Since you seem to have a good grasp of the Asian mtDNA distribution, can you comment on the question whether all non-Western-Eurasian Tibetan mtDNA is highly derived and of (far) SE and central East origin?

Is there presently enough information to either exclude or perhaps suggest a pre-LGM direct route from NE India (for part of the population) instead of the traditional counterclockwise model?

ren said...

Ebizur, where has there been W. Eurasian lineages found in Japan?

Josh said...

Excuse me Ebizur, could you please post the websites or sources that said there are West Eurasian mtDNA Haplogroups found in Korea?

Josh said...

Excuse me Ebizur, I was wondering if you could please post the website(s) or source(s) that you found that said that Koreans have some West Eurasian mtDNA Haplogroups. Thank you.