One of those never ending discussions I have with some readers who dislike the coastal migration model is about my apparent finding, based on Behar 2008, that some L(xM,N) lineages in the Arabian Peninsula are maybe extremely old there. Admittedly I have all kind of doubts but these are of different nature than those of my usual opponent, Terry T.
While my reserves are about the size of samples, specially in Africa and the depth of lineage description, Terry argues that these lineages appear to be younger in Arabia than those arrived from South Asia, notably R0a (R0a1 actually in this area).
The lineages I feel most confident, after due revision, to represent an ancient flow out of Africa across the Red Sea at nearly the same time as the flow that seeded Eurasia with modern humankind (lineages M and N) are L0a1b2, L0f2a, L6, L4b1 and L3e2b2 (in red below).
In the following scheme the “>” signs represent one coding region mutation each (using
PhyloTree, build 12). Count begins at the MRCA of all humans, “mtDNA Eve”.
Purple color used to mark x10 CR mutations from MRCA for easier count.
>>>>>>>>>L0
>>L0’a’b’f’k
>>>>>L0a’b’f
>>>L0a’b
>>>>L0a
>L0a1
>>L0a1b
>>L0a1b2 (Arabia Pen.)
>>>>>>>>>L0f
>>>>L0f2
>>>>L0f2a (Oman)
>>>>>L1-6
>>>>L2-6
>>>>>>>L2’3’4’6
>>L3’4’6
>>>>>>>>>>>>>>>>>>>L6 (Yemen)
>>>L3’4
>L4
>>L4b
·······> L4b1 (infinite line, Yemen)
>>L3
>L3e’i’k’x
>>L3e
>L3e2
>L3e2b2 (Oman, Egypt)
>>>M
>M1’20’51
>>>>M1
>>>>>N
>N1’5
>>N1
>N1a’c’d’e’l
>>N1a’e’l
>>>>N1a
>R
>R0
>R0a’b
>>R0a
>R0a1
>HV
>>H (for reference only)
>R2’JT
>>JT
>>>J
>J1
>J1b
>>>U
>U2’3’4’7’8’9
>U8
>>>U8b
>>>K
The blue clades are not necessarily only found in Arabia but they are common enough to help us discern the matter and, in any case, did not coalesce before the backflow from Southern Asia took place, maybe c. 48,000 years ago.
What can we discern? That at least two of the suspect lineages appear to be older than any backflow from Asia, which could not have happened before the 30th C.R. mutational step. These two lineages (L0a1b2 and L3e2b2) coalesced, it seems, at the 28th mutational step and are therefore of the same estimated age as N, the ancestor of R0a, which probably lived in SE Asia.
L0f2a also coalesced before R0a1 (1 mutational step earlier). L6 however appears younger but it is the best researched case of all these lineages, with the haplotype structure pointing to a coalescence in Yemen (and later migration to Semitic Ethiopia). So what it lacks in age, it has in certainty.
My only claim is that these lineages may be remnants of a once maybe steady flow across the Red Sea into Arabia Peninsula (evidence for the Fertile Crescent seems weaker), survivors of bottlenecks produced by periods of aridity and the backflow from South Asia and further North in the West Asian region.
There may be more, looking at
Amero 2007 there is a clear diversity of L(xM,N) lineages in the area but no academic effort has been made to discern which of these L(xM,N) lineages might be specific of Arabia (or North Africa also) with deep local roots. In general the assumption has been that they are recent historical arrivals but that assumption probably does not hold. L(xM,N) lineages in Yemen (the most fertile part of Arabia) are as much as 37%.
Another complaint by Terry is that there is not much L3 in the region, what makes these lineages less likely (??) to be part of an Out-of-Africa migration led precisely by L3 subclades (M and N). I have admittedly not found too many specific L3 sublineages that can be claimed to be part of such old OoA flow into Arabia but the possibility remains as L3 makes up 11% of Yemeni mtDNA pool and L3d is surprisingly common (4%, more than in Ethiopia), hinting at the possibility of finding other Arabian-specific lineages within this clade.
The full development of this line of research obviously beats my means, as I only work with data mined by academic researchers. I can just hope that someone finds this preliminary exploration interesting and develops it further in the future.