Friday, 29 April 2011

Not quite Egyptians?

So continuing my search for Native North African components, I use the Dogon as the West African pole. The Dogon are a people of Mali, living in the Sahel region not far from the southern "shore" of the Sahara. They speak the Dogon language, presumed but not universally accepted as, a Niger-Congo language. So they are likely related, perhaps central to, the West African Neolithic Revolution Core.
Any source for slaves carried by caravans to the North must have come from here.
For the first run I used 24 Dogon as the pole, all Masai and all Basques:
"Dogon24" to Basque 168 (to "Yoruba24" previously 166)
Basque to Masai 97 (98 in "Yoruba24" run)
Masai to "Dogon24" 42 (to "Yoruba24" 35)
So results are very similar to the "Yoruba24" run using an otherwise identical set-up. Suggestive of great similarity of the Dogon and Yoruba, due to common West African Neolithic descent. Interestingly, "Dogon24" seems more distant to Masai and Basques than "Yoruba24".

Now using only 5 Dogon individuals:
"Dogon5" to Basque 179 (to "Yoruba5" 180)
Basque to Masai 113 (112 in "Yoruba5" run)
Masai to "Dogon5" 55 (to "Yoruba5" 53)
Results are somewhat similar as well. Dogon and Yoruba are interchangeable as poles. Reducing individual numbers in either pole leads to major changes in components. I don't think the "West-African" in North Africans is derived from West African Neolithic populations. It must be native to North Africa.

So how can more be found about it?
Previously I used population isolates, with little or no relation to populations analysed, as "childless" poles to "fish out" unknown components.
Now I'm going to be more straightforward. I'll use restricted poles, each comprised of only 5 individuals, from perhaps slightly more related populations, together with full-set ones just a bit more distant. Here are the poles used:
1) "Dogon5" Five random Dogon individuals - I named the component "Nortwest Africans". Dogon are West Africans, but I believe the restricted pole captured instead such a hidden component
2) All 150 or so Yoruba - "West African". Yoruba serve to "focus" the "Dogon5" pole into capturing the hidden element since the 150+ Yoruba will pull the actual West African I think.
3) "Sandawe5" similarly to "Dogon5" - named "Nile Core Component" or NC. I don't know what's the relation of Sandawe to possible Green Saharan populations, but it doesn't matter since it's intended as a pole for NC anyway. They seem to be just a Masai-like East African population, but I want them to pick something else here, since the much more numerous Masai pole is meant to pull the East African elementf
4) All 100+ Masai - named "East African". They serve the same function vs "Sandawe5" as Yoruba vs "Dogon5": focusing the lens into the desired independent component if it exists (some of the poles in previous blindly supervised runs got nothing)
5) All Basques plus all Lithuanians: named "Mesopotamian Core" or MPC. They are meant to be representative of Anatolian/Mesopotamian/Levantine Core expansion into the region.
6) All San-Namibia: to pick out Subsaharan African Forager elements.

I included some control pops and individuals. FRNA has French and North African ancestry. Tigrinya is ancestraly from Eritrea. Assyria1 and 2 are Assyrian Christians.
Results are very similar to previous blindly supervised runs with different set-ups.
Unsupervised runs tend to pick more composed components centred in Bedouin/Saudis and Mozabite/Tunisians instead, but with lots of MPC as well.
Bedouin and Palestinians may have high NC elements due to early spread of NE Green Saharan incipient pastoralists into the Arabian Peninsula. Pastoralist lifestyles seem to have been important in the evolution of the Green Saharan Neolithic cores from a much earlier stage compared to the MPC Neolithic perhaps giving them from an early stage more of a comparative advantage for marginal soils.
I suspect all four populations derive at least partially from the Green Sahara incipient Neolithic.
1) The Nile Core matured earlier and fused with Neolithic intruders from the North, and would later expand into the same region.
2) Native Northwest Africans lived in a Mediterranean climate but held the Fertile Crescent onslaught well and thus may not have been much behind. They fused with the MPC element and possibly then had a minor expansion into Iberia. I suspect from some recent runs that these are really a Western wing of the Northern Green Sahara pastoralists/incipient Neolithics. There may be an exotic forager element here too, early divergent but sufficiently distinct from the Pygmies and San (whose pole seems to have picked up Pygmy elements quite well). Both elements may be lumped together here.
3) West Africans were protected by a difficult tropical climate from Fertile Crescenters whose crops weren't adapted. They took a longer time to reach a Neolithic critical expansion point maybe for the same reason.
4) East Africans saw some small Fertile Crescent intrusion and admixture, but were also somewhat protected by envirnmental factors
These groups may also have unseen important Forager admixture. This would explain why Pygmies cluster so tightly in PCA and yet have a homogeneous large "West African" element.

E African to W African 28
E African to NC 57
E African to NW African 50
W African to NW African 64
W African to NC 93
NW African to NC 81
E African to MPC 123
W African to MPC 187
NW African to MPC 150
NC to MPC 125
African F to MPC 280
African F to NC 179
African F to E African 116
African F to NW African 168
African F to W African 116 (same as for E African)

Fst is somewhat distorted by seeding individuals, but since they're only 5, it should have some vague validity. It's interesting NW Africans are seemingly both considerably less like the San/Pygmies and the MPC than E Africans.
Note that E African's fst values are being affected by the very many (100+) Masai who seeded the pole, and thus the smallish Northern element in the Masai is pushing fst closer to MPC and away from African F, so that it seems to be closer to MPC than the NC. The high number of individuals shouldn't be a problem for the Yoruba pole since it is more representative of the actual corresponding core population.
NC is considerably closer to MPC than W African, and fst is including some "East Asian" in the MPC pole populations used, as well as NC being based on the Sandawe pole. If corrected for genetic distance "Subsaharan African" contributions to Middle Easterners are not that dissimilar to those reported by various papers.

A thought: Why do Druze and Assyrians have less of this NC component? Is it because they have been more recently isolated due to their different culture and religion? Or have they from ancient times been more isolated from outside influences due to environmental (eg mountains) or cultural idiosyncrasies and thus have kept distinctive cultures/religions?


Next I'll expand this into Europe and rest of ME to see if it holds together.


  1. It is interesting but I dont understand very well why dogon5 is representative of NW Africa.

  2. "Dogon5" is not representative of NW Africa, the Dogon are West Africans just as the Yoruba. But just as in previous runs Beijing-Chinese might be picking Amerindian in Mexicans; and Maasai West African in African-Americans, if the pole is sufficiently small; so does "Dogon5" capture what may be a distinctive component here, not directly related to them.

  3. ok I see and the results looks plausible. Have you read these sudies about North Africa:

    1) Ancient Local Evolution of African mtDNA Haplogroups in Tunisian Berber Populations

    2)Biocultural Emergence of the Amazigh (Berbers) in Africa: Comment on Frigi et al. (2010)

  4. All this is highly experimental, but I wouldn't be surprised if "NW African" turned out to be itself composed of an ancient forager substrate and a Green Saharan melting pot population. This would later suffer major superimposition by "MPC" elements.
    Surprisingly, the fst distances provided by ADMIXTURE for NW African ("Dogon5" but clearly dominated by sequences in North Africans) are quite large versus MPC, NC and E African...
    Don't know if this is real, or why it's so large.
    Amerindian large fst distances, if they are related to "Native Europeans" are suggestive however...

  5. It is nice to see someone taking a deeper look at the components that make up Africans both North and Sub-Saharan.