Yes, it’s possible.

“Fst is prone to sample size bias. IMO, a better method would be ASD.”

I’d love to see how different the results of Fst and ASD are in Africa, with a Hadza sample included. A global comparison of Fst vs. ASD distance methods applied would also be good to see. Do you have any paper, etc. in mind that did Fst next to ASD for Africa or the world? What I know is that it’s typical for small, isolated populations to have huge Fst. Hadza’s Ne is much smaller than that of South African Khoisans.

PCA suggests the Hadza are closer to Eurasians than the South Khoisan on the first dimension, but Fst tells us they are not closer. However, Fst is prone to sample size bias. IMO, a better method would be ASD.

What does SSA stand for? Sub-Saharan African? If so, then I think South Khoisan developed their specificity after breaking off from the rest of Sub-Saharan Africans who in turn retained their affinity with Eurasians. That’s what put Khoisan in a more distant position from Eurasians. It’s unlikely that the Khoisan remained underived and isolated during the past 200,000 years. They are fully modern linguistically, culturally and biologically. Also, in the recent paper that onur and I discussed ad infinitum, Hadza exceeded South Khoisan in Fst from Europeans (Tuscans). Hence, it’s likely that this is going to vary between different African and non-African groups.

The available complete genome sequence of a Southern Khoisan man suggests they are by far the most distant African population compared to Eurasians. The split between proto-SSA (x S-Khoisan) and -OOA likely occurred after the split between proto-S-Khoisan and proto-SSA (x S-Khoisan)+OOA.

Following your logic, if we are able to detect the continent of origin, then we should at the same time also be able to detect the exact geographical coordinates of origin on the scale of a village(!), which is of course nonsensical. So it is completely a matter of scales. As new data accumulate, we get to finer scales on the issue of modern human origins.

Following the same logic, we can’t pinpoint on which continent humans originated. If genetics is a tool that tells us that humans originated in Africa, then this very tool should be able to tell us which part of Africa they originated in. If the principle of decreasing diversity and serial bottlenecks works on the global scale, it should work on the regional scale. Maybe we should just admit that we must have come from terra incognita.

In the case of South African Khoisans, more distant because more basal than East Africans – that’s one if the tenets of the out-of-Africa theory. There was time when mtDNA M1 found only in East Africa was considered to be the basal branch within the non-African macrohaplogroup M. Then M1 was demonstrated to be the result of a back-migration. Now, at least from the point of view of Y-DNA the A and B branches found among Khoisan are thought to be derived from East Africa. Hence, another “back-migration” into – now – South Africa. We could of course think of East Africa as a place from which all migrations take place at different times, but, under a different demographic scenario, Africa is not a refuge for archaic lineages but a sink for all kinds of lineages, and South Africa is even more of a sink – more distant from Eurasia because more derived than East Africa.

No, it says nothing about the direction of the migration between Africa and Eurasia. It has long been thought by many Out-of-Africa supporters that East Africa is the homeland of modern humans and by even more of them that West Africans (including Bantus from anywhere) and South Africans are genetically more distant to proto-Eurasians and consequently to modern Eurasians than non-Bantu East Africans (even when we exclude Semitic and even Afro-Asiatic speakers among them) are.

BTW, the genetic position of Hadza isn’t clear, there are many conflicting results. Their extremely small population size and bottlenecks complicate the matter even more.

But since indigenous South Africans are likely derived from indigenous East Africans (http://mbe.oxfordjournals.org/content/early/2011/04/04/molbev.msr089.short), the proximity of the latter to Eurasians likely suggests that the migration was into Africa and not out of Africa. Eurasians > East Africans (Maasai, Hadza, Sandawe) > South Africans (San).

Naturally the indigenous (pre-Bantu) populations of East Africa (such as the Sandawe) should be genetically closer to Eurasians compared to South- (Khoisan) or West (Niger-Congo) Africans , as we know that the Out-of-Africa migrants branched off from archaic East Africans (L3m and L3n).

The Bantu expansion only reached Kenya roughly by the 1st century AD. That’s relatively recent. The Bantus never penetrated further than Lake Turkana and the Tana River. Thus Horn African populations were geographically well separated from Bantus for thousands of years. Bantu expansion associated haplogroups are also exceedingly rare in the Horn of Africa (north of the Lake Turkana – Tana river boundary).

The Sandawe began to settle and adopt agriculture only beginning from the colonial times, during the late 19th century to be specific (they were all fulltime hunter-gatherers before then), thus through Western influence directly or indirectly and their transition to agriculture and settled life completed as recently as the 1970s and only with the compulsion of the Tanzanian government over them to settle and adopt agriculture. Despite that, they still preserve much of their hunter-gatherer ways.

http://areainfo.asafas.kyoto-u.ac.jp/english/activities/fsta/17_yatsuka/root.html

While there are mtDNA haplogroups in Ethiopia that are attributed to ancient West African gene flow (for example, mtDNA haplogroup L2b), these are ancient enough that they can be distinguished from West African lineages. The reason for the lack of West African/Bantu autosomal affinities in the Horn of Africa is that the ancient admixture from other parts of Africa is already adequately described by the Sandawe/Nilotic clusters, populations who have higher ancient West African contributions if uniparental markers are to be believed.

Another reason is that Bantu influence, which came to other parts of East Africa relatively recently, is virtually nonexistent. No uniparental markers in Ethiopia have been linked to the Bantu expansion, and the mtDNA haplogroups that are linked with West Africa are not associated with Bantus. If I recall correctly, one single exception to this rule has been found among the many Ethiopians that have been sampled; a Y-DNA E1b1a sample in a miscellaneous southern Ethiopian group of samples. Not a surprise that these exceptions would exist there due to Kenyan proximity.

* * *

Another big picture observation worth making is that Bantu, unsurprisingly given Bantu origins, has an Fst distance from West African (0.05) that is lower than any other in the genetic distance chart and West African admixture is shown everywhere Bantu ancestry is known to exist even though we know that the sole source of West African admixture in the non-West African populations is almost exclusively Bantu.

If one merges the Bantu and West African components as one category, you are left with three black African components that are roughly equidistant – West African, Nilotic and Sandawe, and the pruned K chart looks a lot more straightforward.

* * *

There Fulani are a population that I expect from experience to be pretty atypical of non-Bantu Niger-Congo language speakers. This is unsurprising because the Fulani pretty much span across the entire West African boundary between Afro-Asiatic linguistic or Nilo-Saharan linguistic areas and Niger-Congo linguistic areas across the African Sahel, buffering the rest of West Africa from direct contract with outside population influences from the North. To use a crude analogy, they are the welcome mat at the front door that the keeps the rest of the Niger-Congo homeland from getting muddy. Unraveling the demographic histories of populations in boundary areas is generally going to be harder than doing it for core areas.

Other populations and geographic features buffer most non-Bantu Niger-Congo language speakers from much direct interactions with populations to the East. The Congo jungle and Bantu populations seem to be buffer to the South.

There is also a quite recent modern trend of ethnic identity fusion between the Afro-Asiatic Chadic language speaking peoples such as the Hasua and the Niger-Congo language speaking Fulani due to their common interests as pastoralists of the Sahel that is leading to higher rates of admixture than one might otherwise expect.

* * *

The near total absence of West African/Bantu components in Ethiopian populations is surprising. While the combined SW Asian/Mozabite share is about what I would expect in Ethiopia, the fact that all of the Subsaharan African contribution can be attributed to Nilotic and Sandawe components is not. Low resolution uniparental markers do not so unequivocally rule out West African/Bantu contributions in Ethiopia, and usually autosomal measures make populations look more rather than less admixed than uniparental ones do.

There is no component that strongly distinguishes Chadic language speaking peoples (such as the Mada) from a generalized SW Asian/Mozabite or Nilotic or West African component.

This seems to suggest that the West African/Nilotic/SW Asian and Mozabite divide may have roots in deep African population structure; but that the Afro-Asiatic language family does not have a single coherent population structure. There is no unified single component that unites the Semitic, Berber, Chadic and Cushitic language families, unlike Khoisan, Sandawe, Niger-Congo and Nilotic language famillies which each have a clearly identificable common genetic thread that unifies almost all speakers of those languages, at least that the K=12 level.

Echoing onur’s comment no. 3, the genetics seem to support a theory of Afro-Asiatic language expansion in which the connections between its major linguistic subfamilies (which have long been controversial) are cultural rather than resulting from demic expansions (although demic expansion is visible within some of the subfamilies). There is very little genetic overlap between the historically Semitic language Saudis, the historic Berber language Mozabites, and the Chadic language speaking Mada.

The genetic contribution from SW Asian associated with Ethio-Semitic in Ethiopia is pretty distinctive, although intra-Ethiopian breakdowns seems to show a residual SW Asian element in Cushitic populations apart from the Ethio-Semitic SW Asian element. So, Ethiopia is bit trickier to parse, as is the historically Coptic language family area.

Why not? Joking aside, I am much more open to different possibilities than I might have seemed to be in my discussions on this blog.

the hausa are only marginally modified from yoruba and igbo. that seems some sort of elite diffusion of dialect, the inversion of the fulani, who seem *more* extra-sub-saharan than the hausa, but speak a niger-congo language. i guess i lean to a north african/west asian origin for the group, but my confidence is pretty shaky at this point.

My general impression is that Afro-Asiatic Blacks are a genetically pretty heterogeneous bunch, they have no unity (even without the Semitic speaking ones).