New blogs

Leherensuge was replaced in October 2010 by two new blogs: For what they were... we are and For what we are... they will be. Check them out.

Sunday, June 1, 2008

Out of Africa but when?

I had been driven by my own amateurish excercises on haploid genetic timing to begin pondering that the OOA epysode may need to be significatively older than the usual 70-60 milennia BP if we are to account for the the old ages that well studied Y-DNA clades like R1b or
O3 are yielding in MRCA estimates.

But I just found some scholarly support and precisely for about the same ages I was pondering:

A Bokyo et al, Assessing the Evolutionary Impact of Amino Acid Mutations in the Human Genome. PLOS Genetics, 2008.

The paper is wider in scope, as it tries to find out how much of autosomal SNP diversity is adaptative and how much is neutral, an important issue on its own. But the part that most got my attention is the section of the Discussion about Inference of Demography. The best fit demographic models (table S1) for Africans and Europeans suggest that the former have been expanding since some 170,000 years BP (6800 generations x 25 years) and that Europeans probably had two expansions with a bottleneck (in between?). The first expansion was c. 130,000 BP and the second c. 14,500 BP. The duration of the bottleneck was of only some 2100 years (84 generations).

There are other two models that are theoretically somewhat fit for Europeans but both are simpler and not as well fit as the one above. They are:

One states that there was just one expansion (no previous bottlenecks) since c. 6500 BP (I'm sure creationists would love this one! - but it can't account for much older Africans, so forget it). From a rationalist viewpoint anyhow it makes not much sense as 6500 BP would be post-Neolithic, suggesting a total replacement of farmers by what? Indo-European hordes? Sumerian erratics? Megalith builders? Hard to make sense of. The corrsponding 4500 BCE date actually doesn't relate with any exapansion: it's rather the end of Neolithic expansion in Europe, save for some marginal areas, that were still hunting and gathering.

The other one suggests a single expansion since c. 22,000 BP (just before the LGM) and a brutal bottleneck of c. 193,000 years. So with this model Europeans would have been in the hiding since the very beginning of Humankind and would have only began expanding precisely in the coldest period of the Ice Age. Again hard to make sense of.

But the complex model makes some sense: expansion since after the LGM, as most agree with and a bottleneck that would not be too long: about half the LGM span. The problem comes with the other expansion date: c. 130,000 BP. It should be an OOA or post OOA expansion, I understand but normally the OOA age is dated to about half that age.

Yet I already mentioned that such an old OOA event is not imposible at all: not just there is presence of modern humans in West Asia and Northern Africa since about that age but also Indian paleolithic shows continuity before and after the Toba event and MP artifacts might be work of H. sapiens in the subcontinent since before 100,000 BP.

The OOA migration could certainly be older and I was already thinking on several grounds of dates that approach that one of c. 130,000 BP (more like 100-110,000 BP?).

Note: thanks to Razib for suggesting that this paper might be important, what got me to read it with some more attention.


terryt said...

Maju. "The first expansion was c. 130,000 BP". There is no reason at all to think that Y-chromosome Adam was part of this expansion. The oldest possible generally accepted date for Y-Adam is 70,000 years. I tend to believe there have been several expansions out of, and into, Africa since H. erectus.

Maju said...

But that's an estimate (an educated guess), and anyhow, I have seen more frequently dates of some 90-100,000 years in any case. 70 KY is (another good guess) for the OOA event.

The case is that the age estimates can't be very precise. Every estimate has a brutal error margin and is anyhow a modelling, a possibility. There's no hard science behind that and, considering that some rather low level haplogroups like O subclades are yielding TRMCA dates of c. 30,000 BP, I tend to think that all higher level guesstimates are very conservative. Otherwise, O subclades should be near-Epipaleolithic, and they are not apparently. The same problem we have with R subclades in the West, that are same level as the O ones.

Probably the whole Y-tree age estimates are understimated actually.


1. Effective population size is a mere guess that is needed to solve some equations.

2. The estimate of age distance between human and chimpanzee is used as base of the so-called molecular clock hypothesis. It's based in the fossil record, that is very sparse for the Pan genus, and recently I read that it falls quite short to explain the chimp/bonobo divergence if this has to agree with the geological formation of the river Congo, as it's generally agreed.

3. Often, like Karafet in her recent high-impact paper, a mere conventional assumption, like roughly 70 KY for CT (pre-OOA root clade) is made and the rest calculated based on that (and probably in constant mutation rates that may be another wrong assumption).

It's complicated but in the end common sense and archaeology probably have the last say and not feeble theoretical modelling.

terryt said...

I agree that both dates are guesstimates but they're pretty much based on the same assumptions, so probably correct in a relative manner.

To me there's no problem with accepting the dates at face value. After all Y-hap A must have had some ancestor and it's unlikely it derived from some Australopithecus who'd never left Africa. I think we can accept that modern human and Neanderthal mtDNA lines separated roughly half a million years ago so it's as likely mtEve's ancestor had come INTO Africa as Neanderthals' ancestors had come OUT of it.

To me there's also no problem with O and N separating as much as 30,000 years old. They're both fairly diverse today. And you have claimed elsewhere that the populations containing the clades probably remained in a small region before the haplotypes became fixed prior to their expansion.

Maju said...

I'm talking of O second order subclades like O3a or O2b. That would push the separation of O and N to c. 60,000, NO (and other brother clades) from K maybe to 105,000 BP, K 8and others) from F not much earlier (like 113,000 BP) and F and C from CF to something as brutal as c. 160,000 BP.

Of course it depends on how you count. The above dates (my own) are based on assumptions like constant mutation rates for SNPs, independent of population size, along best known Y-DNA branches. They approached reasonably well those of Karafet... before I decided to correct them to make these new age estimates for O3 and other O subclades, as well as for a putative Magdalenian post-LGM re-expansion of R1b1b2 (former R1b1c). It basically places Y-DNA Adam c. 200,000 years back, at the very origins of the H. Sapiens species.

It's just another guesstimate anyhow.

I would not get Neanders into this anyhow, as the coalescence of the H. sapiens branch surely took some more time after divergence. But it seems to affect the age of the OOA epysode, moving it back to ages that are double than the usual ones. This again would imply a rather long Eurasian coalescence prior to actual expansion out of South Asia and possible ages for this second expansion that may pre-date known AMH fossils by 20 or 30,000 years, probably just after the Toba event.

If I'm right (and well, I'm just an amateur: I may be wrong) the whole timeline should be pushed back in time: OOA would be long before Toba and only the large diversification and colonization of Eurasia (and suburbs) would be post-Toba in fact.

I have to think more about this but the possibility is clearly there, at least to my eyes.