January 1, 2017

Iberomaurusian ancient mtDNA

This is an issue that has lingered for a long time in the online anthropology circles. Once upon a time (2005) there was online a presentation in which a good deal of mitochondrial DNA (HVS-I) sequences from the key Iberomaurusian (or Oranian) culture site of Taforalt (North Morocco) were (not too formally) published. Eventually this presentation became almost impossible to find... however now Rym Kefi (the original author if I remember correctly) and colleagues have got back to resurrect that important data set from oblivion, plus sequences from Afalou cave as well.

Rym Kefi et al., On the origin of Iberomaurusians: new data based on ancient mitochondrial DNA and phylogenetic analysis of Afalou and Taforalt populations. Forsenic Sciences Research 2016. Freely available at the time & space of writing thisLINK [doi:10.1080/24701394.2016.1258406]

Abstract

The Western North African population was characterized by the presence of Iberomaurusian civilization at the Epiplaeolithic period (around 20,000 years before present (YBP) to 10,000 YBP). The origin of this population is still not clear: they may come from Europe, Near East, sub-Saharan Africa or they could have evolved in situ in North Africa. With the aim to contribute to a better knowledge of the settlement of North Africa we analysed the mitochondrial DNA extracted from Iberomaurusian skeletons exhumed from the archaeological site of Afalou (AFA) (15,000–11,000 YBP) in Algeria and from the archaeological site of Taforalt (TAF) (23,000–10,800 YBP) in Morocco. Then, we carried out a phylogenetic analysis relating these Iberomaurusians to 61 current Mediterranean populations.

The genetic structure of TAF and AFA specimens contains only North African and Eurasian maternal lineages. These finding demonstrate the presence of these haplotypes in North Africa from at least 20,000 YBP. The very low contribution of a Sub-Saharan African haplotype in the Iberomaurusian samples is confirmed. We also highlighted the existence of genetic flows between Southern and Northern coast of the Mediterranean.

The bulk of the data is this (tables 1 and 2):



Notice that all sequences are "old style", i.e. HVS-I only, and that's why haplogroup assessment is not always 100% certain, with particular emphasis on the "CRS" sequences, which have been proven in some ancient cases to belong not to the modern normal (H1, rarely H2 o other H) but to a very rare modern haplotype within U*, AFAIK only preserved in Asturias but somewhat common in the Magdalenian of South Germany. 

In any case, that they produce 100% safe H1, H6, H14, etc. and, even more surprisingly, J and T2 variants, should make people think about the hidden Paleolithic mtDNA diversity in the wider Mediterranean area. And by this I do not just mean North Africa but also all Southern Europe, including many areas that can't be considered part of the Mediterranean basin, such as most of the crucial Franco-Cantabrian Region, the most densely populated province, by far, of post-LGM Upper Paleolithic Europe, source of several cultural waves (including the Solutrean one, almost certainly at the origin of Iberomaurusian), yet outrageously neglected by research.

Table 5 and other materials in the paper also deal with which modern populations appear closest to the Iberomaurusian ancient mtDNA pool and these are:
  1. Tuscans (0.00090)
  2. Catalans (0.00134)
  3. Galicians (0.00223)
  4. Sicilians (0.00377)
  5. El Alia (0.00699)
  6. Valencian (0.00787)
  7. Matmata (0.00788)
  8. Slougia (0.00831)
  9. Jerba Berbers (0.00934)
Figures in brackets are FST distances, the smaller the closer the match, statistically speaking. Populations in cursive are from Northwest Africa.

That is I'd say quite surprising because we tend to think of those near matches as quite impacted by Neolithic and post-Neolithic inflows, at least judging on other recent research. It seems that the debate on the origin of modern Europeans and North Africans refuses to come to a close, as different aspects of the evidence available may be somewhat contradictory. What do you make up of all this?

See also:


UPDATE: single marker inferred lineages (not always the haplogroup is reported correctly):

Down in the comments' section, Capra questions with good sense the certainty of inferred haplogroups. The most clear one is TafV27, claimed to be H6a1a8, which cannot actually be that (would need another HVS-I marker defining H6, which is missing), however this means that it must be HV0 and possibly V, a lineage that appears (per the aDNA literature) in the European Neolithic, already quite towards the West, out of nowhere, and that "nowhere" should not be West Asia, where it has never been sequenced, unlike K. 

I'm therefore reviewing here all the single-marker inferred haplogroups as carefully as possible, please double-check them and report if I seem to be committing any error:
  • AF22B - 16126C (reported as JT or H14b1) - JT(xJ,xT) or H14b1 indeed, notice that JT* was probably also one of the lineages described (also by the HVS-I method) for Nerja cave (Solutrean, same time frame), just across Alboran Sea (cf. Fernández-Domínguez 2005). Whether is one or the other or even something else, it's almost certain that the lineage was shared across the sea between Europe and Africa, what just makes total sense for this culture.
  • Taf55-IB - 16239T (reported as H1) - I actually get specific H1 variants (H1bf'bg'bh'ch specifically) or H17c. H sublineages in any case, unless it is an extinct or unreported R* lineage, most unlikely.
  • TafV27 - 16298C (reported as H6a1a8) - must be HV0, possibly even V, a distributed Euro-African lineage with three hotspots: Kabylia, Catalonia and Lappland. Per Caramelli 2003 and successor studies, this lineage was already present in Italy (Pagilicci cave) in the Gravettian era.
  • TafVIII - 16223T (reported as U4a2b) - It cannot be U4 at all because it lacks the HVS-I mutation 16356C. In fact the reverse 16223C mutation defines macro-haplogroup R, and I can't find any other matches within R, so it must be L3(xR), maybe N(xR) but maybe M or other L3. The possibilities are many, for example: N1, N2, N9, N11, etc. within N, a lot of possibilities within M, and also several within L3e'i'k'x(xL3x). The only thing we should be certain here is that it is not R and also not upstream of L3 (that's how good as it gets with HVS-I methodology, really). 
  • TafXXV3 - 16126C (reported as H14b1) - as with AF22B, it can be JT(xJ,xT) or H14b1 and again I must emphasize that there is an extremely good chance that this matrilineage had relatives in Iberia (Nerja) within the Solutrean culture.

In case you want to re-check, the best resource is of course PhyloTree.



IMPORTANT CAVEAT: the above update was done assuming they had fully sequenced the HVS-I, which is not that long, but it seems that in many cases (second column of the tables above) they have only sequenced up to the 16317 locus, what makes prediction of haplogroup even harder. That would explain their H6a1a8 prediction, although there's no way they can exclude the much more common HV0 (H6 is unheard of in North Africa). Thanks again to Capra for pointing that out.

7 comments:

  1. Those 100%s cannot be taken seriously, they seem to come from modern mtDNA distributions which can hardly be projected back 20 000 years.

    H6a1a8 is predicted due to the mutation 16298C. But 16298C is also defining for HV0. (Of course it occurs in other branches as well but not by itself.) Most of its subclades, including V, have only this mutation in CRS until you get down to specific sub-sub-branches. Why should this sample not be HV0 or V? That seems more likely to me; in any case the probability of H6 is certainly not 100% in reality.

    Another example is H14b1 at 100%; again it is only 16126C, which also defines JT and R0a, so it can be some branch of those.

    The 100% U4a2b is so because it is *missing* the 16223 mutation characteristic of haplogroup R, and there is no coverage for 16356 which is found in U4. For all we know this can be N or L3.

    And so on, there is hardly any result without a similar problem. It is safe to say that this is mostly or entirely macrohaplogroup R, and there is probably a good deal of R0(xR0a)/HV/H (all CRS at root) though in principle it can be some kind of U or even R* (unlikely). The further back in time we go the closer to CRS in HVS I we will get. At Afalou it seems J and T were definitely present.

    ReplyDelete
    Replies
    1. Nope. Enafaa does not report H6 in North Africa at all. The link is in the "see also" section of the entry but anyhow only the samples from Iberia and the Near East carried it at all. So finding H6 in UP North Africa cannot correspond to any translation of modern lineages of any sort.

      I also do not think that U4 is common at all in modern North Africa, on the other hand haplogroups like V and K should be present (they are today at significant frequencies), as well as a much larger fraction of L(xM,N), whose striking absence in the sample is a highlight in the paper. Within H, we should expect (Enafaa again) that most would be H1 and H3 (for Morocco) and (in other cases) also including significant fractions of H4 and H7, as these four are the North African H tetrad (>90% of all North African H), all them derived from SW Europe per Cherni 2009. That is not at all what we see in the Taforalt and Afalou samples at all.

      "H6a1a8 is predicted due to the mutation 16298C. But 16298C is also defining for HV0".

      Excellent point. I double-checked and you are right. Do you think they have actually found the oldest known HV0? A lineage that otherwise seems to appear only in the Neolithic, literally out of nowhere -- because it has not been spotted in West Asia either, unlike what happens with K, right?

      I think you are right on this issue: in order to be H6a1a8, it should also have the marker 16362C, which defines H6 (when you sequence HVS-I, you sequence it all, absences by error are not to be expected). So very well spotted, thank you. I'll write an update on this.

      "Another example is H14b1 at 100%; again it is only 16126C, which also defines JT and R0a, so it can be some branch of those."

      Fair enough. I'll double check that too.

      "The 100% U4a2b is so because it is *missing* the 16223 mutation characteristic of haplogroup R, and there is no coverage for 16356 which is found in U4. For all we know this can be N or L3".

      Sounds credible. I'll double-check that as well.

      "The further back in time we go the closer to CRS in HVS I we will get".


      That may be the case within R but with many many exceptions. In fact the U*-CRS is not caused by that but because it repeats independently exactly the same three CRS-wise mutations that U as such includes in the opposite direction, what is a most strange. In fact most "CRS" must be under HV and exclude the other two common Western R sublineages (U and JT), although extremely rare and unexpected exceptions (which rather look like further Asian and not West Eurasian at all) cannot be totally discarded without proper coding region or some other sort of testing (for instance Alu+ means H, etc., as defended by Hervella et al.)

      Delete
    2. I updated the entry with a double check of the five sequences that are defined by a single HVS-I mutation:

      → two are JT*/H14b1, which is quite interesting because the same HVS-I sequence was reported for the Solutrean of Nerja cave (Andalusia) by Fernández 2005, what suggests that we are before a marker of this Solutrean→Iberomaurusian migration, whatever the actual named lineage.
      → One they report as H1, should be specific H1 subclades or H17c, no other options I can see.
      → As mentioned before, the one reported as H6, should actually be HV0, what is also most interesting potentially.
      → The one reported as U4a2b must be L3(xR), because 16223C defines R in the HVS-I. It might be some N but also some M or L3(xM,N). I'd like if it'd be L3k, which IMO should have been in NW Africa since very old times but just a wish.

      Delete
  2. Actually the sequencing only goes up to 16317 for most of the samples, so there is no read for 16356 or 16362 in the case of the supposed U4 or H6 examples, unfortunately.

    Anyway you are quite right there is no H6 in North Africa now while HV0/V is rather common, so it is not based on modern frequencies in that area but just some generic prediction.

    ReplyDelete
    Replies
    1. Are you sure about that? I mean: it would be an awful methodology, how do you know that they did not sequence above 16317? Do they say that in the paper somewhere or is just something you infer from reported mutations? Not quite the same thing: they may not report them because they just don't happen to find them.

      "so it is not based on modern frequencies in that area but just some generic prediction."

      Yeah, that's clear. They still may get the prediction wrong but it's not because of some sort of regionalist bias, unless it's the generic Eurocentric one that is much more common across the literature.

      Delete
    2. OK, I see it now: it is where they report the "position" in the second column of the tables. That's most weird, really!

      I must write a caveat therefore.

      Delete

Please, be reasonably respectful when making comments. I do not tolerate in particular sexism, racism nor homophobia. Personal attacks, manipulation and trolling are also very much unwelcome here.The author reserves the right to delete any abusive comment.

Preliminary comment moderation is... ON (your comment may take some time, maybe days or weeks to appear).