In my recent review of mtDNA L(xM,N), following mostly Behar 2008, I noticed that some of those lineages have distributions that either partly or totally do not seem to correspond with Africa South of the Sahara, but clearly penetrate since old times into West Asia, North Africa or both.
So I wonder: aren't some of them remnants of the migration Out of Africa, which never really made it to South Asia?
Let's see. The mutation count between "mitochondrial Eve" and the L3 node is of 23 (coding region only), to M is of 26 and to N of 28. This is roughly the "time frame" of the Out of Africa. By the "time" of 30 CR mutations the backflow of Eurasian mtDNA to West Asia was already beginning (mostly in form of R derivatives but also M1 and N1, and later some other N subclades too) and this became a clearly dominant flow. Also since M1 specially, scattered into East Africa, it's possible that there was a second flow from Africa in that period of colonization of West Asia that was, no doubt, coincident with the early Mousterian Puvial, when the region became less arid.
I find that the following lineages, normally tagged as "African", should have been involved in the Out of Africa episode:
L4b: at 24 mutations appears to have almost the same age as L3. It has two basal lineages: L4b1, which is only found in Yemen, and L4b2, typical of the Hadza and Sandawe, and found also among Ethiopians (with offshoots in Southern Africa and Arabia) and the Lisango of DR Congo and/or CAR.
L0f2: at 24 CR mutations looks only slightly younger than L3 and surely older than M. One of its sublineages (L0f2a) is only found at Oman, while the other is at Ethiopia (L0f2b).
L0a1b: at 26 CR mutations should have the same age as M. One of its sublineages is only found in Arabia Peninsula (L0a1b2), while the other is shared between Morocco and Southern Africa (L0a1b1).
L5c: at 28 CR mutations appears to be as old as N. One of its sublineages is found at Egypt (L5c2) and the other at Ethiopia (L5c1).
L3e2: at 29 CR mutations appears of the same age as R. L3e2a is shared by West and North Africa, while L3e2b has two lineages that are only found in Jordan (one of the L3e2b*) and Oman and Egypt (L3e2b2).
L6 shows 38 mutations, however the root of its very long stem is necessarily older than L3 (as it's the oldest branch of L3'4'6). Today it looks like Yemen harbors the highest diversity, with offshoots at Ethiopia and Egypt.
There are others but seem younger than the 30 mutations that I decided as limit and may represent other flows, maybe related to the arrival of M1 to East Africa.
It's quite striking, to say the least, that these lineages have been overlooked, they really appear to my eyes as participants of the OoA at a more modest scale, never reaching South Asia or beyond.
Also notice that Yemen and Oman appear insistently in this list. All mentioned lineages, except L5c and possibly L3e2 should represent local fixation processes when people took the coastal route via southern Arabia.
Update (Mar 14): corrected error re. L0f2 (nomenclature and mutation count).
I'd like also to add L3i to the list. It looks to me as centered at coastal Yemen and of similar age (slightly younger, 2 mutations downstream from L3, 25 from "Eve") as M.
Update (Mar 21): Browsing through the East Asian mtDNA data at Tanaka 2004, I just spotted 1/20 (5%) L2 among the Sakai (Semang Negritos) of southern Thailand (Trang province). Not sure what to think but it is odd indeed. Another unlikely survival?
Update (May 5): Map of a plausible Abbassia Pluvial (120-90,000 years ago) Out of Africa migration with the candidate mtDNA lineages to have participated in it (red: L3, purple: other):