New analysis of African mtDNA

There is a new paper on 'African' mtDNA L(xM,N) with a focus on trans-Saharan flow:

Nourdin Harich et al., The trans-Saharan slave trade - clues from interpolation analyses and high-resolution characterization of mitochondrial DNA lineages. BMC-Evolutionary Biology 2010. Open access.

The authors conclude that all or most of this mtDNA flow happened in recent times within the trans-Saharan slave trade, mostly in the last few centuries of Arab domination. I think that this position is simply untenable (and I'd dare say falling into 'romantic' Orientalism per many of the reasonings like: some harems could be enormous, reaching even the extravagating number of 14,000 concubines) but there is interest anyhow in the data managed (from many other different studies mostly) and specially in the many nice distribution maps provided.

They do not only fall into pseudo-romantic visions of the slave trade and its importance but also fall in another pool of moving sands: excessive reliance on the molecular clock age estimation methods, when they draw their conclusions.

I really fail to see how MC age estimates of c. 10-15,000 years ago (L3) can justify a recent shared ancestry within the slave trade. They would actually point, if anything, to late Paleolithic or Epipaleolithic migrations (Capsian culture?) There is only one case with a more recent date in all the L3 comparisons but it's not between an Tropical and Mediterranean Africans but between one Jordanian and one North American (Afro-American?) Instead there are many L3 cases with older age estimates: up to 37,000 years ago.

They do not have age estimates for the other haplogroups, specially some L0 and L2 sublineages that also permeate the Sahara-Red Sea barrier.

As I have said before, some of these lineages look very old North of the Sahara (see here) and may have been pouring towards the North and NE since around the out-of-Africa migrational episode, which may be c. 120,000 BP, quite roughly.

They fail to analyze two particular haplogroups that are apparently more common North of the Sahara and East of the Red Sea: L3k and L3i respectively (L3i is plotted on a map but North African specific L3k is totally ignored).

As I said above, the best is the distribution maps, which allow for a quick visual impression of the distribution of the L(xM,N) lineages. Form these I think that:

• L0a shows a clear South-East Africa-West Asia distribution, with some very high concentrations in this last region. I previously mentioned that I think that L0a1b2 and L0a1c (?) are clear candidates for a Middle Paleolithic flow to West Asia related to the OOA migration.
  
• L1b surely had a Middle Sahelian origin but also clearly permeates across the Sahara. Within it, L1b1a2 is typical of the Red Sea area, while L1b1a5 is found in Mauritania and Cyprus (per Behar 2008). I think that its expansion happened some time after the L3 one (29 CR mutations to the root vs the 23 of L3) but may well belong to the late phase of the Abbassia Pluvial.
  
• L2a is so ubiquitous that cannot be analyzed as a whole. The timeline of its expansion is surely the same as that of L1b (again 29 CR mutations to the root) but it's much more impressive in its dimensions. Some of the high level sublineages (L2a1j) are exclusive of the Mediterranean, while some others are shared across the desert and surely deserve specific attention if we want to clarify anything.
  
• L2d. The map concentrates this lineage in West Sahara but (from Behar 2008) it's actually also present in Ethiopia, Algeria and Yemen. It's another candidate for an old lineage beyond the Sahara.
• L3b also looks pretty much North African/West Asian but has also presence in West and East Africa.
• L3d is an even more clear case of a North African/West Asian lineage: rare south of the Sahara and very common north of it.
  
• L3e looks by the map as split in a North African and a Bushman clades, however by Behar's data the situation is much more complex: scattered in many distinct African and West Asian populations, all very suggestive of an old spread within the L3 explosion (OOA related).
  
• L3f looks like spread from Sudan in all directions early on as well.
  
• L3i is shown as the Red Sea-South Arabia lineage it is and is a very strong candidate for OOA migration remnant in Southern Arabia. Can never be attributed to 'slave trade' but it's most probably the "little sister" (niece in fact) of M and N. It is actually closer to the L3 root than M and N by CR mutation count, so it's surely older, just that it did not reach South Asia.
  
• L3x is a very similar case, however it should be slightly more recent than L3i, with an age similar to that of the N and R explosions in further Asia.
• I'm wondering if L3w is just another name for L3k. It does not exist in PhyloTree but the distribution (Algeria, Ethiopia) is reminiscent of the one reported for L3k elsewhere.

There's no mention to L6 and L4b, both of which are also candidates for participation in the OOA flow to South Arabia.

In brief: it seems that the preconceptions of the authors are preventing them to reach to rational conclusions. The paper has some interest but this lays mostly on the illustrations (maps). Hopefully, as it happened with the similarly confused Balaresque paper, someone else (Morelli in this case) will take the bull by the horns and do something better with all that data.

Related posts:

• The early expansion of H. sapiens in Africa (mtDNA)
• Are we overlooking the signature of the Out of Africa?
• Reviewing the mtDNA lineages (notes): L0, L1, L2 and L5, L3'4'6