Monday, January 7, 2013

East African mtDNA variation has implications on the origin of Afroasiatic

The Dienekes' Anthropology Blog shows a new paper on East African mtDNA with implications for the origin of Afroasiatic, namely with the citing: "making the hypothesis of a Levantine origin of AA unlikely",  unfortunately I do not have access to the paper, I would greatly appreciate if anyone has access to it to please send me a copy here:

Here is the abstract and the link:


East Africa (EA) has witnessed pivotal steps in the history of human evolution. Due to its high environmental and cultural variability, and to the long-term human presence there, the genetic structure of modern EA populations is one of the most complicated puzzles in human diversity worldwide. Similarly, the widespread Afro-Asiatic (AA) linguistic phylum reaches its highest levels of internal differentiation in EA. To disentangle this complex ethno-linguistic pattern, we studied mtDNA variability in 1,671 individuals (452 of which were newly typed) from 30 EA populations and compared our data with those from 40 populations (2970 individuals) from Central and Northern Africa and the Levant, affiliated to the AA phylum. The genetic structure of the studied populations—explored using spatial Principal Component Analysis and Model-based clustering—turned out to be composed of four clusters, each with different geographic distribution and/or linguistic affiliation, and signaling different population events in the history of the region. One cluster is widespread in Ethiopia, where it is associated with different AA-speaking populations, and shows shared ancestry with Semitic-speaking groups from Yemen and Egypt and AA-Chadic-speaking groups from Central Africa. Two clusters included populations from Southern Ethiopia, Kenya and Tanzania. Despite high and recent gene-flow (Bantu, Nilo-Saharan pastoralists), one of them is associated with a more ancient AA-Cushitic stratum. Most North-African and Levantine populations (AA-Berber, AA-Semitic) were grouped in a fourth and more differentiated cluster. We therefore conclude that EA genetic variability, although heavily influenced by migration processes, conserves traces of more ancient strata. Am J Phys Anthropol, 2013. © 2013 Wiley Periodicals, Inc.

mtDNA variation in East Africa unravels the history of afro-asiatic groups

UPDATE: Ok, got it, this was a nice little article to read, however with respect to the implications of East African mtDNA variation on the origin of Afroasiatic, it did not offer nothing really substantially new, in terms of material evidence, that any reasonable person that has read up on this subject a little bit would not have known beforehand, namely:

Concerning the third point, i.e., the place of origin of AA (EA or the Levant), our results do not allow us to make conclusive statements. Indeed, coalescent simulations of different genetic parameters (Supporting Information Fig. 4) according to the two mentioned hypotheses show that—even assuming complete correlation between languages and mtDNA variability—their confidence intervals largely overlap. Thus, we limit ourselves to the following observations. First, EA shows the highest levels of nucleotide diversity among the studied populations with a decreasing cline towards NA and the Levant (Supporting Information Fig. 1 and Supporting Information Table 1). This is true not only for the Ethiopian cluster A, but also, and especially, for groups belonging to clusters B1 and B2. Second, EA hosts the two deepest clades of AA, Omotic and Cushitic. These families are found exclusively in EA, while the presence of Semitic in this area is much more recent. Third, cluster C – collecting Berber- and Semitic-speaking populations from NA and the Levant – shows only modest signals of admixture with clusters A and B (Fig. 2, Supporting Information Table 1). None of these points,
taken by itself, is conclusive, but undoubtedly the hypothesis of origin of AA in EA is the most parsimonious one, if compared to the Levant.

It did also have some very nicely made contour maps for EA, as well as detailed mtDNA haplogroup assignments for some 30 or so East African groups, which I will make an interactive chart for within the next couple of days.

UPDATE2 (01/08/2013): mtDNA haplogroups (46) in 31 groups.

A note on the sources for the samples listed above:

The Dinka Samples are from Krings etal. (1999)
The Sudan and Ethiopia Samples are from Soares et al. (2011)
The Tigrai, Amhara, Gurage, Oromo and Yemeni1 Samples are from Kivisild et al. (2004)
The Beta Israel Samples are from Beharet al. (2008)
The Ethiopian Jewish Samples are from Non et al. (2011)
The Somali Samples are from Soares et al. (2011) and Watson et al. (1997)
The Daasanach and Nyangatom Samples are from Poloni et al. (2009)
The Turkana2 Samples are from Poloni et al. (2009) and Watson et al. (1997)
The Nairobi Samples are from Brandstatter et al. (2004)
The Kikuyu Samples are from Watson et al. (1997)
The Hutu Samples are from Castrì etal. (2009)
The Iraqw Samples are from Knight etal. (2003)
The Burunge and Turu Samples are from Tishkoff et al. (2007)
The Datoga and Sukuma Samples are from Tishkoff et al. (2007) and Knight etal. (2003)

All the remaining samples: Dawro Konta, Ongota, Hamer, Rendille, Elmolo, Luo, Maasai, Samburu and Turkana are new and sampled along with this study.


  1. The African (EA) origin of AA is pretty much clear for me and the only honest reluctance is the issue of the origin of Semitic, the only AA subfamily known to ever have been spoken in Asia (at least in any meaningful way, i.e. beyond the occasional Egyptian army or Berber pilgrim group).

    I think that this can be reasonably explained as pre-Neolithic Egyptian influences into Palestine, where the PPNA and very especially the Harifian (desert pastoralists) may have adopted it. These pre-Neolithic influences also explain E1b in the Levant and as far as Greece, where it arrived via West Asia almost for sure, causing a founder effect which may well be Mesolithic rather than Neolithic.

    After the expansion of PPNB from Kurdistan/Turkey (languages related to Caucasus surely), the so-called Circum-Arabian Pastoralist Complex (CAPC) probably inherited the language and traditions from Harifian (even if generically absorbed into PPNB, it retains Harifian roots as far as I know).

    Then in the 4th millennium BCE, at the edge of History, they expanded quite suddenly into the agricultural regions around their semi-desertic econiche (the mythical "flood", which is probably a wordplay in Sumerian between amaru=flood and a-maru=semites, also known as amurru). I can only imagine that climatic conditions were at play but whatever the case this is pretty much documented archaeologically and, in the case of Sumer, also in text (only "after the flood" Semitic names begin to appear).

    That should explain it. However it'd be nice if we'd also have good archaeological information from Arabia, very especially the Hedjaz, which may well have played a role in the formation of the CAPC or in the backflow of early Semitic branches into Africa (Ethiopia).

    Rather than AA languages, I'd like if someone studied Y-DNA J1 in depth. It's often not even tested for but rather described as J(xJ2) but it is key to understand the interactions between North and East Africa with West Asia. Of course it's possible that J1 did expand from West Asia with the Neolithic but the matter must be researched on its own right. It should be much more informative re. languages than mtDNA, after all in most cases the ethnicity, and hence the language, is patrilocal.

    1. yeah, good synopsis overall, one thing I would like to add though is that the internal classification of Afroasiatic has not reached consensus, for instance the 'BoreAfrasan' group which includes Chadic, berber, ancient egyptian and semtic, and that is championed by Ehret is contrasted with Lionel Bender's 'Macro-Cushitic' which groups berber, cushitic and semitic together, incidentally, if you take some of the E-M35 lineage frequency distributions at face value, and make the following rough oeverlaps: E-M81 - Berber, E-M78 -Ancient Egyptian, E-M123 -Semitic, E-V42/M293 -Cushitic/Southern cushitic then the SNP E-Z827 that unites M123, M81, V42 and M293 but excludes M78 would better mirror Bender's classification than Ehret's, but again these things are changing fast, but the 'SNP dust' should settle within the next couple of years where we will get the final picture, hopefully.

    2. "After the expansion of PPNB from Kurdistan/Turkey (languages related to Caucasus surely),"

      One thing I forgot to add with respect to this statement you made above is that using Bayesian techniques the Turkey or Asia minor region is also now largely thought to have been the earliest area/center where Indo-European (rather than Caucasian) type languages were first spoken:
      Bouckaert et. al(2012) -

      So maybe the PPNB people were speaking some form or another of proto-Indo European ? I admit however, I don't know much about the history of IE languages.....

    3. I am not in agreement at all with the Anatolian hypothesis for Indoeuropean, I find it impossible to sustain, very especially in (1) the steppary migration of Indo-Iranian, (2) the fact that we know many many non-IE languages from Ancient Anatolia and surroundings, as well as other areas that would have been influenced by the alleged IE Neolithic migrations (notably Iberian and Basque, which may well descend from whatever Cardium-Impressed Pottery peoples spoke) and (3) the fact that the alternate (and much more mainstream) model known as "Kurgan", which roots IE in early Chalcolithic Samara Valley (Russia between the Volga and Ural rivers), works nearly perfectly to explain every single IE branch from Hittite and Tocharian to Celtic and Indo-Iranian, going through the various Balcanic branches.

      I must write a page at my blog about it. Needs some work however.

      So I don't think that PPNB would be IE speakers but rather of one or several of the Caucasian families. Also I think that Sumerian (another example of non-IE non-AA West Asian "Neolithic" language) correlates best with NE Caucasian (Chechen, many Daghestani languages) and therefore with its probably relatives of what would be Kurdistan and Armenia, the Hurro-Urartean (sub-)family.

      As for the list of non-IE languages that relate to Neolithic expansion these are: Basque and Iberian (plausibly derived from Cardium Pottery language, possibly related to whatever other European Neolithic peoples spoke), Etruscan-Lemnian (original from Anatolia almost for sure), Eteocretan (original from Anatolia on archaeological grounds), Etecypriot, Hattic (rel. to NW Caucasian), Hurro-Urartean (rel. to NE Caucasian), Sumerian, Elamite (arguably related to Dravidian, probably spoken at IVC) and then of course the Afroasiatic family. It's very difficult to imagine that the main impulse for the expansion of Indoeuropean was the West Asian Neolithic and then so many non-IE languages survived that can or often must also be related to the same process.

      But then of course there are two reasons that yield support the Anatolian hypothesis:

      (1) that much of the IE diversity originated at the very waves of expansion (hence the Balcans, where that expansion caused complex mixtures in a rough topography, retained high historical diversity: at least Greek, Albanian and Traco-Phrygio-Armenian, possibly more as we know little of Dacian, Illyrian, etc.). Instead in the Northern European Plain and the Eurasian Steppe, the differences were flattened out more easily producing Western and Eastern IE almost only (add the remote Tocharian language to that). This is admittedly confusing and may also apply to Afroasiatic, which retains its highest diversity in and around the abrupt regions of Ethiopia but which may well be an illusory effect of "diversity flattening" in other lands, hence the archaeological record is also very important. Luckily for the IE case, that one is rather well known and totally supports the Kurgan model.

      (2) Most IE speakers prefer to imagine themselves as cultural descendants of industrious and semi-civilized farmers (Anatolian model) rather than barbaric horse raiders (Kurgan model). The perception of this difference of behavior has been smoothed now (it was very sharp when Gimbutas first articulated the Kurgan model but now we know that the farmers were not such nice people either and often went to war and probably styled some sort of Patriarchy also) but the subjective preferences remain and bias the debate.