Out of Africa-Out of Asia-Out of Africa

Roberta Estes recenty posted about the fact that Spencer & I will be revisiting The Journey of Man, fifteen years after he wrote the book and made the documentary.

As part of the preparation, I’m revisiting some uniparental results, in particular, recent ones. This preprint in bioRxiv Carriers of mitochondrial DNA macrohaplogroup L3 basic lineages migrated back to Africa from Asia around 70,000 years ago., caught my attention:

Background: After three decades of mtDNA studies on human evolution the only incontrovertible main result is the African origin of all extant modern humans. In addition, a southern coastal route has been relentlessly imposed to explain the Eurasian colonization of these African pioneers. Based on the age of macrohaplogroup L3, from which all maternal Eurasian and the majority of African lineages originated, that out-of-Africa event has been dated around 60-70 kya. On the opposite side, we have proposed a northern route through Central Asia across the Levant for that expansion. Consistent with the fossil record, we have dated it around 125 kya. To help bridge differences between the molecular and fossil record ages, in this article we assess the possibility that mtDNA macrohaplogroup L3 matured in Eurasia and returned to Africa as basic L3 lineages around 70 kya. Results: The coalescence ages of all Eurasian (M,N) and African L3 lineages, both around 71 kya, are not significantly different. The oldest M and N Eurasian clades are found in southeastern Asia instead near of Africa as expected by the southern route hypothesis. The split of the Y-chromosome composite DE haplogroup is very similar to the age of mtDNA L3. A Eurasian origin and back migration to Africa has been proposed for the African Y-chromosome haplogroup E. Inside Africa, frequency distributions of maternal L3 and paternal E lineages are positively correlated. This correlation is not fully explained by geographic or ethnic affinities. It seems better to be the result of a joint and global replacement of the old autochthonous male and female African lineages by the new Eurasian incomers. Conclusions: These results are congruent with a model proposing an out-of-Africa of early anatomically modern humans around 125 kya. A return to Africa of Eurasian fully modern humans around 70 kya, and a second Eurasian global expansion by 60 kya. Climatic conditions and the presence of Neanderthals played key roles in these human movements.

It was really hard for me to parse all the coalescene and phylogenies here. I know some readers are pretty well versed in this area so I’m curious about critiques and reactions to the above.

4 thoughts on “Out of Africa-Out of Asia-Out of Africa

  1. The paper, from what I can tell, seems to have several questionable features and seems not to be especially strong or convincing (or to present much in the way of decisive new evidence).

    Also, some commenters at forums on the subject have discussed it, and I have across some critiques there which seem interesting (some of which seem to express some similar problems in the paper to those I suspected, better than I could)

    One point (among others made at the link below) by the commenter Lank, I also found interesting:

    “The correlation between Y-DNA DE and mtDNA L3 in Africa has been obvious for many years. There is no a priori reason to assume Y-DNA DE originates outside Africa, especially when it has roughly the same age as mtDNA L3 (although, if it was part of a back migration, it certainly would have brought some mtDNA L3). There was mtDNA M and even pre-N, as well as a lot of Y-DNA C even in Paleolithic Europe, so the modern concentration of Y-DNA/mtDNA diversity in eastern rather than western parts of Eurasia is not representative of the distribution going back tens of thousands of years.

    Nothing new to see here as far as I can see.”


    One other (seemingly important) observation is that the (OOA) ancestry/origin (apparently all or the the vast majority) in modern Eurasians dates from the approximately 70 ka BC wave (the second major wave) of migration of modern humans from Africa, and not from an earlier one dating to 125 ka BC ., as the authors seem to suggest (which instead seems to have left little to no legacy in modern Eurasians overall). Much less (it would seems to me) would said earlier migration (with generally little to no autosomal legacy) be likely to be the source of such major/dominant modern Eurasian uni-parental lineages (as the paper discuses)—particularly when concerning the maternal lineages ancestral to all/nearly all those of living Eurasian populations.

    (The earlier migration ca 125 ks bc, before the major one ca. 70 ka BC, having as far as I know, left at most—according to some evidence at least—a small relict approximately 2% of ancestry only in certain isolated groups of southern Eurasians such as Papuans and Aboriginal Australians.)


    “But the third paper, by a team led by Mait Metspalu of the Estonian Biocentre in Tartu, makes a different claim. Analyzing 379 new genomes from 125 populations worldwide, the group concludes that at least 2% of the genomes of people from Papua New Guinea comes from an early dispersal of modern humans, who left Africa perhaps 120,000 years ago. Their paper proposes that Homo sapiens left Africa in at least two waves.

    Reich questions that result, but says that his and Willerslev’s studies can’t rule out a contribution of only 1% or 2% from an earlier H. sapiens migration. Akey says: “As population geneticists, we could spend the next decade arguing about that 2%, but in practical terms it doesn’t matter.” The most recent migration “explains more than 90% of the ancestry of living people.”

    And one source cited by the authors on the early modern migration does not appear to support the argument it is claimed to, (as discussed below):

    “The authors cite a paper that found DNA of a modern human ghost population in Neanderthals. This Neanderthal-ghost population admixture event was dated to 100ky ago. Cabrera et al use this admixture event as evidence that the ancestors of all living Eurasians must have already been in Eurasia by 100ky ago. I checked what their source (Kuhlwilm et al 2016) actually says, and it has nothing to do with living Eurasians, nor with the Africans who supposedly back migrated to Africa.

    If this new OOA model were accurate, we’d expect the ghost human DNA in these Neanderthals to be closely related to living Eurasians and the Africans who supposedly back migrated. That is, the last populations we’d expect to have an affinity with this ghost population’s DNA, is Pygmies and Khoisan. However, the actual paper says that Khoisan are closest to this ghost population, not the supposed ‘backmigrants’ or Eurasians:

    “”Because there is fairly weak information in the data to support such inference, the uncertainty in the inferred values is quite high, and different values are obtained in the four runs (lowest in the ‘Chinese’ analysis and highest in the ‘San’ analysis). However, if we take the union of the 95% Bayesian credible intervals for the four runs, we can conclude that the source population likely diverged from present-day humans between 138,000 and 433,000 years ago, which is consistent with divergence either before or slightly after the divergence of the San from other present-day populations.””


    Note that this is not a case of a self-defeating citation that just happens to disagree with Cabrera et al. They can’t fix this by removing the citation from the preprint and pretending Kuhlwilm et al just had a different opinion. Cabrera’s whole 120ky OOA migration doesn’t even involve the ancestors of living populations, but some extinct human population that’s older than the Khoisan.

    Cabrera et al are going to regret they wrote this. We’re in the aDNA age now. This is not the early 2000s where you can write a specious paper and thrive because there is no aDNA to falsify it conclusively. Cabrera et al are one aDNA sample away from getting debunked with relevant aDNA. They might get debunked with new aDNA before this comes out of preprint.

    The best part of this paper is the supplementary data. Looking forward to them making good on their promise to include their trove of unpublished mtDNA and Y-DNA data.”


    A somewhat earlier, perhaps relevant, 2015 discussion re: Ydna DE, CT and E

    But one obvious issue of course seems to be a dearth of ancient DNA (to reconstruct earlier genetic distributions before later confounding population movements, etc.), which I certainly hope will be recovered from relevant populations in the fairly near future (an reduce some of the existing ambiguities in the evidence).

  2. Cont:
    To my understanding at least, an African (likely East or perhaps North East African) origin, of CT, DE, and E, seems at least as likely as a Eurasian one (from what is now known), if not more so—though still not yet resolved—(and an African—likely Eastern African—origin yet more likely in the case of L3, a greater number of whose major branches are, as far as I know, African). As some have suggested, it seems plausible that the population associated with them may have been native to around East Africa, and related to, but perhaps distinct from, that which (left Africa and) became the OOA population. One group expanding out of Africa ca. 70 ka bc, and the other (group/cluster of related groups) within Africa from around the same time and gradually absorbing much more divergent local African homo sapiens and/or proto-sapiens populations as it expanded out from the East of the continent.

  3. Edit to first post:

    “And one source cited by the preprint authors Cabrera et al on the early migration does not appear to support the argument it is claimed to”

    (apologies for the many edits. I worried my phrasing of this section may have originally been somewhat sloppy or ambiguous)

Leave a Reply

Your email address will not be published. Required fields are marked *