Saturday, January 16, 2010

290. Aftermath 5

A debate has been raging in the field of population genetics, not over a Toba bottleneck, which has been on the back burner for some time, but the meaning of the genetic evidence in South Asia generally. Things came to a head back in 2003, with an exchange of letters-to-the-editor in the American Journal of Human Genetics. The exchange was initiated by Richard Cordaux and Mark Stoneking, who begin by challenging a recent paper by Endicott et al. on the origins of the Andaman Islanders, which “support[s] the growing evidence of an early movement of humans through southern Asia.” Reconsidering both the mtDNA and Y evidence, Cordaux and Stoneking reject this idea: "In our opinion, Endicott and colleagues’ results do not support any relationship between the present Andamanese population and the hypothesized early southern migration."

They then move to a consideration of the distribution of mtDNA haplogroups M and U, which "has been taken as a genetic signature for an 'early' (i.e., Middle Paleolithic) colonization of South Asia by modern humans and, consequently, as a confirmation of the 'southern route' hypothesis." They dispute this evidence as well, first on archaeological grounds, since "the earliest evidence of modern human industries and remains is dated to ∼30,000 years," then on the basis of the genetic evidence. For example:
  • Alu insertions data are interpreted as supporting an ancient African-PNG relationship, but India is not a part of this relationship (Stoneking et al. 1997).
  • Y-chromosome and mtDNA data suggest a connection between the Indian subcontinent and Australia, which is, however, dated to less than 5,000 years ago.
  • mtDNA haplotypes in South Asian ethnic groups are most closely related to east Eurasians and do not show any particular ties to African or PNG populations (Kivisild et al. 2003; Cordaux et al. 2003).
  • An mtDNA control region motif proposed by Forster et al. (2001) to represent a signature of an early migration from Africa to Sahul through the southern route is not found in South Asia (Cordaux et al. 2003).
"In summary, there is no convincing support to date for a Middle Paleolithic genetic contribution to South Asia by migrants from Africa to Sahul along the southern route." Their letter is consistent with the conclusion I've already quoted, in a paper by Cordaux et al from the same year, which questions the implied continuity of the genetic evidence for a southern route migration:
[A]lthough they show close affinities, the east Asian and Indian mtDNA gene pools are fairly distinct. This result is consistent with the suggestion that the east Asian and Indian mtDNA pools have been separated from each other for about 30,000 years (Mitochondrial DNA analysis reveals diverse histories of tribal populations from India).
This is in accord with Stephen Oppenheimer's positing of a "deep genetic rift" between South Asia and Southeast Asia -- for him this is due to the effects of the Toba eruption, a topic neither party of this debate is willing to touch.

In their reply, Phillip Endicott, Vincent Macaulay, Toomas Kivisild, Chris Stringer and Alan Cooper reject the above critique, offering detailed point by point refutations, followed by the following concluding statements:
On the basis of the mtDNA and Y-chromosome data presented here, we see no need to accept the view of Cordaux and Stoneking regarding the settlement of South Asia. . . We must infer an early dispersal of AMH with non–Upper Paleolithic technology through Asia to explain the early Australian [archaeological] evidence (Stringer 2000), although we agree with Cordaux and Stoneking that the precise route(s) taken is still unclear (Stringer 2002). But we see no requirement for the South Asian mtDNA gene pool to demonstrate close affinities with either PNG or Africa to discuss an early settlement of this region. Given the continuity of the archeological record within India, from the Middle Paleolithic onward, and the range of estimated dates for Indian haplogroup M, there is no clear reason to preclude the presence of modern humans in this region prior to 30,000 years ago.
The debate could be summarized as follows: on the one hand, the genetic evidence reveals a gap between Africa and the Sahul, and a disconnect between South Asia and East Asia, both inconsistent with the theory that early Africans migrated through Asia via a southern route; on the other hand, since archaeological evidence from Australia clearly implies a middle Paleolithic migration from Africa to the Sahul, and the genetic evidence is clearly inconsistent with a northern route, a southern route would seem the only logical possibility -- for Endicott et al, genetic evidence of discontinuity along this route is unclear and open to dispute.

While some evidence presented by Cordaux and Stoneking dates from several years prior to their letter, and may no longer be valid, a paper from the following year, entitled Most of the extant mtDNA boundaries in South and Southwest Asia were likely shaped during the initial settlement of Eurasia by anatomically modern humans, by Mait Metspalu et al (including the same Phillip Endicott who responded to Cordaux and Stoneking), provides striking support for some of the same discontinuities noted in their critique. What's more, it confirms the strange bias in the distribution of the M haplogroup I've already mentioned, as noted by Oppenheimer.

I'll begin with a comment on the M haplogroup:
Our results indicate that the frequency distribution of haplogroup M varies across different Indian regions by a significant cline towards the south and the east . . .

The very interesting distributions of some of the most important M haplogroups are mapped in Figure 1 (click on the figure to enlarge it):


Note especially the distributions of M6, M6b and M2b, all found, for the most part, in the east and south, where Oppenheimer also noted the prevalance of this extremely old and important haplogroup. There is an equally interesting presence of M6 and M6b in the northernmost reaches of the Indus valley, the only region of South Asia where tone languages are spoken. More on this presently.

Especially significant are the maps presented in Figure 11,
"illustrating the spatial frequency distribution of mtDNA haplogroups native to West Eurasia (panel A), South Asia (panel B) and East Eurasia(panel C)":


I'll have a lot to say about these maps, and what they tell us, in the next installment.

5 comments:

Maju said...

The main Andamanese haplogroups are not M2 or M4, but distinct M31 and M32. As such they appear to be yet another distinct offshoot of the huge star-like M expansion, that IMO defines the main Eurasian expansion.

It may be that in 2003 they had not yet described these Andamanese clades, creating some confusion.

Remember that we are in 2010 and a very big deal of research has happened in between these dates, very specially in what regards to mtDNA M.

This timestamp probably explains also why the strange emphasis on haplogroup U, that only seems to matter in regard to West Eurasia (and South Asia, of course).

Alu insertions data are interpreted as supporting an ancient African-PNG relationship...

This is not phylogenetically valid... or at least accepted by the vast majority of geneticists (on whose consensus reference trees are drawn). Nothing in the Y-DNA tree (nor the mtDNA one) suggests any particular connection between Africa and New Guinea (i.e. distinct from the general Eurasia-Africa ancestral connection).

An mtDNA control region motif proposed by Forster et al. (2001)...

Same as above.

Phillip Endicott, Vincent Macaulay, Toomas Kivisild...

Big names here.

But we see no requirement for the South Asian mtDNA gene pool to demonstrate close affinities with either PNG or Africa to discuss an early settlement of this region.

Ditto.

The debate could be summarized as follows: on the one hand, the genetic evidence reveals a gap between Africa and the Sahul, and a disconnect between South Asia and East Asia...

It's not "evidence" more like a fringe argumentation on irrelevant issues such as insertions (normally not considered SNPs), control region markers (that vary wildly and have only very limited phylogenetic capability) and other variants of "searching for the three legs of the cat".

It's probably best understood as a historical debate, now essentially closed.


...

Metspalu's paper is a classical (another would be Thangarai'07) but again it's relatively "old". Sadly I don't know of any paper that sports that kind of nice mtDNA maps with a more modern and extensive approach.

Note especially the distributions of M6, M6b and M2b, all found, for the most part, in the east and south, where Oppenheimer also noted the prevalance of this extremely old and important haplogroup. There is an equally interesting presence of M6 and M6b in the northernmost reaches of the Indus valley, the only region of South Asia where tone languages are spoken. More on this presently.

Actually I was giving some reason to Oppenheimer on the grounds that M6 looks to me slightly younger than other M sublineages. This is because M6 is defined by 6 mutations (4 in the control region) in contrast to most other M subclades that only have one or two mutations at their basal stems.

Please, check this map of M sublineage distribution and tell me if it doesn't compare well with Atkinson's reconstruction of ancient Eurasian populations.

Manju Edangam said...

Maju:
The main Andamanese haplogroups are not M2 or M4, but distinct M31 and M32.

Why did you bring M4 here? What are your thoughts on M2 being the oldest is found in eastern Indian regions in high frequencies? These regions do not have native Dravidian or IE tribes.

Maju said...

Because if you browse through the paper you find it also mentions it:

The authors identified three different mtDNA haplotypes in 11 Andaman islanders, two belonging to haplogroup M2 and one belonging to M4. These haplogroups had previously been reported only in the Indian subcontinent (Kivisild et al. 1999b; Bamshad et al. 2001). The Andaman M4 haplotype has been found previously in mainland India (Kivisild et al. 1999b), whereas the two Andaman M2 haplotypes are (so far) unique to the Andamanese.

Manju Edangam said...

Okay. I didn't check the papers I was going thro' the blog post and didn't find it. I suppose M2 and M4 were latter identified with M32 and M31. However, M31 is also observed in eastern tribals.

Maju said...

It's possible. I have warned more than once here against relying too much on old papers, very specially in regards to mtDNA M, which was still very poorly known in 2006.