Monday, February 4, 2013

A speculative superimposition of E-M35 variants onto Afroasiatic.

Here is a speculative superimposition of the variants of YDNA E-M215/M35 (E1b1b/1) onto an Afroasiatic internal classification, Lionel Bender's (1997) classification. 


The red question marks represent a less unsure fit.

24 comments:

  1. I'm intrigued.

    How well does fit E-M35 with Afroasiatic languages among non-AA which are genetically and culturally akin populations, for example in the Sudans? Can we see a relatively sharp transition between AA and non-AA in this matter?

    I'm trying with this question to test the model in the most objective possible way. I'm not very sure but I have been suspecting that it was only E-M78 the one involved (or most involved) with AA expansion (assuming therefore many survivals of previous E-M35, and other layers) but I don't have the detailed knowledge you do.

    Another lineage that I suspect related to AA expansion would be J1, which I imagine pretty old also in Africa (it'd be a similar case to R1b-V88 but with a more general impact).

    ReplyDelete
    Replies
    1. E-M35 is prevalent in Northern Sudan but not in the south, some of the populations that have the highest amount of E-M35 in the North of Sudan are not even AA s, example the Borgu, Fur and Maasalit in North Western Sudan have high levels of E-M215/35 (Hassan et. Al (2008)), however they are all Nilo-Saharan speakers. The Anuak , Dinka, Shiluk, etc.. from South Sudan however, although Nilo-Saharan speakers have only low to moderate levels of E-M215/M35, as they are largely dominated by haplogroups A & B. In the North Eastern side of Sudan you also have high E-M35, but among AA/Cushitic speakers, Beja. Other populations that have relatively high E-M35 but don't speak AA languages, are in South Eastern Africa, namely the Maasai (Wood et. Al (2005)), but also the Datog (Tishkoff et. Al (2007)), these are both also Nilo-Saharan speakers. Some click speakers tend to have moderate levels of E-M35 as well, notably the Sandawe and Khoisan ( Henn et. Al (2008)), however most of the variants found in South Eastern Africa, for both Click and Nilo-Saharan speakers alike is E-M293, although moderate levels of E-M78 has been noted in the Maasai as well. Outside of Africa, as you know there are moderate levels of E-M35 in non-AA speaking South Eastern Europeans like Greeks for instance. Other than these populations, E-M35 is almost exclusively found amongst Afroasiatic speakers.

      It is hard to make the case that only E-M78 would have been involved, especially considering that E-M81 found in high levels in berbers has more recent common ancestry with other East African variants of E-M35 than it does with the main Egyptian variant (V68/M78), off-course, in the end it all depends which internal classification of AA you choose, however considering 3 of the main internal classifications of Bender, Ehret and Blench, the current known substructure of E-M35 fits better with Bender's classification, IMO.

      With respect to J-M267 and R-V88, as you can see above I have a question mark on them, as they are surely frequently represented in the respective branches of AA but I am not really sure how it all fits together.

      Delete
    2. Sorry I did not get back to you earlier, I forgot to click the feedback button and have been busy as well. I see that there could be some hints in the Sudan of discordance with your model but not many. Guess we could all imagine that they were "Nilo-Saharanized" later on but unsure.

      There's no reason (neither known historical languages nor toponymy) to imagine that early Neolithic Europeans of E-V13 partial patrilineage would be ever AA-speakers. We cannot discard it 100% either but it does not matter much because they would fall in any case within the E-M78 umbrella, so inside the AA group in my counter-hypothesis.

      "It is hard to make the case that only E-M78 would have been involved, especially considering that E-M81 found in high levels in berbers has more recent common ancestry with other East African variants of E-M35 than it does with the main Egyptian variant (V68/M78)"...

      The high presence among Berbers may just mean that it tends to accumulate in refuge areas like the mountains or in the West and that may already happened with the AA / Capsian (Gafsan) expansion, and then again being reinforced with the Phoenician colonization and the Arab conquest. It must have been in North Morocco at high frequencies already c. 7000 years ago, to take part (without J1) of the "Cardial bounce" in that area that probably helped to define the peculiar "Greek and Moroccan" genetic flavor of West (but not East nor Central) Iberia. I find that E-M81 is too frequent, too old (for the reason above, which is the most recent plausible scenario), too occidental (within NW Africa) and rather not too related to East African clades of E1b1b to be easily accepted as part of the Capsian-AA expansion.

      I may well be wrong however but I tend to think of E-M81 as what was pushed Westward against the ocean and against the mountains by the AA E-M78 and J1 expansion in North Africa. If I am wrong then I cannot explain well the duality E-M81 vs J1 & E-M78 in NW Africa, in which the latter seem to be pushing the former against the ropes, so to say.

      I just saw your new entry on Sudan, which I bet is quite interesting.

      Delete
    3. “I find that E-M81 is too frequent, too old (for the reason above, which is the most recent plausible scenario), too occidental (within NW Africa) and rather not too related to East African clades of E1b1b to be easily accepted as part of the Capsian-AA expansion. “

      That was what I was trying to explain to you however, E-M81 is more phylogenetically related with East African variants of E-M215/M35 before it is related with the Egyptian/Sudanese variant (E-V68/M78). Specifically, as the diagram above shows, E-M81 is related with E-V42, E-M293, both of which are restricted to the southern and Eastern parts of Africa, and E-M123, which is primarily found in East Africa and the Levant only. Look at the latest ISOGG phylogeny and you will see it there. You will find no papers published on these new SNPs yet.

      The other problem with E-M78 solely being involved in the expansion of Afroasiatic is that Omotic speakers, the earliest branch that putatively diverged from AA, do not seem to have much of E-M78, this can be exemplified by Plaster's data-set of the Maale, whom are southern Omotic speakers, the variants of E-M215 they carry are V6, M34, M281 and M35* for the most part, they have very little E-M78, they also have a significant amount of E-M329 which is a 'cousin' to M35 and represents a more ancient presence of the E-P2 lineage in East Africa . Now this is not enough data to make a conclusion and we definitely need more samples, but it gives us a glimpse of what is to come. E-M78 could be implicated for the expansion of Cushitic speakers, but even then, we do not have enough detailed data from many different types of Cushitic speakers, we mostly have it for what are deemed 'Lowland East Cushitic speakers', which is just one sub type of Cushitic. Take the Afar for instance, they have < 10% E-M78.

      “I cannot explain well the duality E-M81 vs J1 & E-M78 in NW Africa, in which the latter seem to be pushing the former against the ropes, so to say. “

      I don't understand? There is hardly any significant amount (>10%) of E-M78 in North West Africa, especially in Berbers, the little E-M78 that there is, is found in arabs, if I recall correctly, same with J1 there is hardly that much J1 in NW Africa, they are by a large fraction E-M81.

      Delete
    4. Judging on Semino 2004, E-M78 does not seem at all rare in NW Africa and at least partly follows the same distribution pattern of J1 (i.e. it is generally more common to the East than the West).

      "the little E-M78 that there is, is found in arabs"

      For all I can tell "Arab" is meaningless word in population genetics: it just means that at some point they dropped their ancient languages and adopted Arabic instead. In a few cases this may be related with visible immigrant elements but in most cases not. I don't think that there is any "Arab" marker in NW Africa - in fact I do not think there is a single true Arab there either, as in 'original from Arabia Peninsula in any sizable fraction of their ancestry', not even in Y-DNA. The "Arab" invasion of North Africa, as well as Egypt and other areas earlier, was a mere military occupation cum alliances and gradual religious propaganda (later more extreme, Taliban-style even). Arabs in North Africa are those who for this or that reason adopted Arabic language early on, in some cases replacing Latin, which had previously replaced Phoenician, (coastal urban zones especially) and in others directly Berber (or even Vandal Germanic in Tunisia I guess). "Arabs" from Carthage, "Arabs" from Rome, local aristocrats who found more useful to become "Arab", local plebs who followed their lords' choice, the occasional nation-less mercenary who spoke Arabic and whatever was his original language (typically Berber but also "Slavs" or Black Africans here and there), and then of course there was the occasional aristocrat who had some genuine origin in Arabia like Abd-al-Rhaman III but were so wildly mixed generation after generation that we can easily claim him more as Basque than as Arab (his grandmother and mother were Basques, one a slave the other a princess).

      J1 is pretty big in Tunisia and Algeria, and even in West Sahara (cf. Adams 2008 → map). Actually J1 being about 20% in North Africa and nearly 0% in Iberia, and, inversely, J2 being around 15% in Southern and SW Iberia and nearly 0% in North Africa, indicate that the E-M81 and other Ibero-Berber exchanges of the past must have happened before J1 (and E-M78) arrived to NW Africa but after E-M81. They can still be two different flows but at least the NWA→WIb one clearly implies E-M81 but not J1 nor E-M78 (except E-V13 but this one comes from Greece via Italy, being one of the main markers of European Neolithic).

      Delete
    5. Re. the Omotic, I would not put too much relevance there because it is likely that the main AA expansion post-dates the formation of Omotic and may have been starred by totally different lineages. The one language family = one lineage model seems extremely over-simplistic to me, ignoring the fact that ethno-cultural expansions happen in often unrelated bouts, and in the hiatuses there is lots of admixture and therefore the new waves can start with any sort of genetic constitution. For example imagine that in the 22nd century or whatever there are massive Jamaican, Nigerian and Indian demographic expansion with English dialects as main language, what conclusions would a hypothetical ignorant population geneticst from later on reach from that: that the expansion of "Atlanto-Indian" (English-derived) languages has something to do with
      West African and Indian lineages and would look at whatever remnant of England as anomalous and hard to fit in.

      Same with Omotic: because it is old, it actually belongs to a layer that is not really related to the main expansion.

      Delete
    6. "Judging on Semino 2004, E-M78 does not seem at all rare in NW Africa"

      The high E-M78 found in Semino (2004) was in the Morrocan Arabs, who had 43% E-M78, All the other North Western Africans in that dataset had E-M78 < 15%, including the Tunisians. With respect to J-M267, All of them had it < 14%, except for the Saharawis, Algerians and Tunisians who had it at 17, 35 and 30.1% respectively, which the authors interpret as mostly being of a more recent expansion. Although a minority of it could be neolithic as well. Remember also that there are datasets showing some Tunisians with 100% E-M81. The above you mention is just one dataset, you need to look at it as a whole, and when you do look at North West Africa as a whole, E-M78 and J-M267 are quite minor, ~ 5 and 13% respectively, with E-M81 ~ 62% , if you take out the cosmopolitans and those that are labeled as 'Arabs' in studies then E-M78 and J-M267 are ~4 and 7% respectively, with E-M81 ~70%. You are making a supposition with very little data, please do the math first.

      “Actually J1 being about 20% in North Africa and nearly 0% in Iberia, and, inversely, J2 being around 15% in Southern and SW Iberia and nearly 0% in North Africa, indicate that the E-M81 and other Ibero-Berber exchanges of the past must have happened before J1 (and E-M78) arrived to NW Africa but after E-M81.”

      Unless of-course most of the J1 in North Africa came into North Africa in much later times, as Semino et al. implicate.

      “The one language family = one lineage model seems extremely over-simplistic to me, ”

      Depends how old both the lineage and the language family are, an old a diverse lineage like E-M215/M35 can accommodate many, if not all, the branches of Afroasiatic.

      Delete
    7. Just to emphasize that "Arab" in North Africa is most people. In Morocco there are three larger Berber communities but in Algeria, Tunisia, Libya, West Sahara... Berbers are very fragmented and minoritary. So "Arab" is not any peculiar community but almost everyone, even the grandson of a Berber whose father migrated to a city and has forgotten his mountain home and his ancestor's language. It just implies a less well-defined (refuge) and more cosmopolitan population. There are even cases of Berber tribes wholly arabized in recent times, if my memory is correct Gaddafi's tribe was one of those.

      We also have a decent historical knowledge of the Muslim or Arab invasion of North Africa and what implied (or rather not implied at all) re. demographic movements. Essentially, except to very localized military stations, there was never any migration but a process of intra-Berber struggles in which some groups sided with the invaders and others resisted but later yielded. Many of those became soon Arabized, beginning by the Romanized populations of the coast, but many tribes retained their traditional social structures and some of them that way retained their ancestral language.

      Some geneticists on light of this evidence of W/E clinality in North Africa, may postulate massive Arab migrations but neither the Arabia Peninsula could ever produce enough such people nor known history allows to imagine any such migration. Actually written history would more easily suggest massive flows from Iberia upon the conquest of Granada and the alleged expulsion of the Moriscos but we see no such signature anywhere either. Apparently tens of thousands of alleged refugees vanished in the night of time, so did the invading Arab armies which were not larger and unable to justify the high frequencies of Eastern (J1, E-M78) haplogroups in NW Africa.

      PS- Was looking for a percentage of "Berber vs Arab" figure in Morocco and all I found is "99% are Berber-Arab", in the article Moroccan genetics, it says:

      "Recent studies make clear no significant genetic differences exist between Arabic and non-Arabic speaking populations, HLA DNA data suggest that most Moroccans are of a Berber origin and that Arabs who invaded North Africa and Spain in the 7th century did not substantially contribute to the gene pool".

      References are both from 2000 (Ballais and Gómez-Casado) but I have never read anything that made me think otherwise. I think that E-M78 and J1 must be markers of Afroasiatic (Berber) expansion and that E-M81 has been pushed against the ropes a bit, being clearly older in that area.

      Trying to infer the genetics of primitive Berbers from moder Berbers only is trying to reconstruct primitive Celts from the Irish or primitive English from Jamaicans. It does not work that way: you can change language (and most people do without much fuss for practical reasons) and therefore ethnic identity but you cannot change genes.

      Delete
    8. I meant the WP article 'Moroccan genetics', sorry it was left a bit confusing: http://en.wikipedia.org/wiki/Moroccan_genetics

      Delete
    9. I don't disagree with the rather 'loose' terminology of 'Arab' in North Africa, but at the same time to think that the Islamic Expansion had a simply negligible genetic impact would be rather naïve also.

      “Some geneticists on light of this evidence of W/E clinality in North Africa, may postulate massive Arab migrations but neither the Arabia Peninsula could ever produce enough such people nor known history allows to imagine any such migration.”

      They don't postulate it only because of the W/E clinailty of J-M267, they postulate it due to microsatellite diversity and patterns, within J-M267 , (and even within what is known as J-P58 now) there is a pattern described by YCAIIa=22-22, this pattern is prevalent in >90% of North African J-M267 chromosomes, while it is much less found in Ethiopian and European J-M267 chromosomes, in addition the haplotypes that are found within this pattern in North Africa have very limited internal diversity, hence why they are thought to be relatively very young, probably within the Islamic Expansion period.
      Look at this image here , it shows you the J-M267 haplotypes, those that are shaded in gray harbor that pattern or motif.


      “I meant the WP article 'Moroccan genetics', sorry it was left a bit confusing: http://en.wikipedia.org/wiki/Moroccan_genetics”

      Right, take a look at it carefully, and take out the Egyptian berbers and Egyptians in that table and you get the general figures I say above for NW Africa on Average which is , E-M78 : 4 -5 % , J-M267 : 7 -13% and E-M81: 62-70%. This is with and without including the 'Arabs' and cosmopolitans. These figures simply are not significant enough to fuel your hypothesis is what I am saying, and not that your hypothesis itself maybe impossible.

      Delete
    10. My interpretation would instead be that J1 (which anyhow still needs a deep look at phylogeny and distribution) had only one (and "late") expansion to NW Africa with AA. Also Ethiopia and other places may have got indeed "recent" secondary Semitic inputs from West Asia which can well have increased the diversity of J1. It'd need a careful analysis.

      "... you get the general figures I say above for NW Africa on Average which is , E-M78 : 4 -5 % , J-M267 : 7 -13% and E-M81: 62-70%".

      The figures of Morocco should not be extrapolated to all NW Africa. If J is 10% here it is more than double in Tunisia, almost 25%, while E-M81 is <50% instead. You may wish to attribute that only to Arab colonization but I would say that Phoenician colonization was longer and deeper and we see no "Lebanon-like" gene pool either. In fact we don't see any typical influence of any modern West Asian Y-DNA pool: only J1 and that is suspicious of being old, not recent. Otherwise we'd see more J2 and other lineages typical of West Asia or Egypt. We do not.

      Using Abu Amero's Arabian Y-DNA pool, and assuming all migrants were from Arabia peninsula (what makes no sense but whatever), we should see for each 100% J1 (roughly):
      → 30% J2
      → 30% E1b1
      → 13% T
      → 13% R1a1
      → 6% G

      So for each 8 J1 we should find 3 J2 and instead we find one J2 per each 10 J1! E1b subclades seem to be also (each one) independent of J1. R1a1 is also extremely rare (it should be at 3% all around but just less than 1% in Tunisia, 0% elsewhere).

      Anyhow, I think that the J1-J2 comparison is more than enough to show alone that the expansion of J1 in NW Africa (and hence in other parts of Africa) is prehistorical: it cannot be Phoenician nor can be Arab, unless we postulate a huge hyper-recent (late Medieval or Modern) wave of J2 southwards. The lack of J2 actually indicates that J1 arrived before the admixture of both lineages in West Asia whenever it happened but probably in the Neolithic.

      Why in the Neolithic (my best guess)? Because European and North African J flows are mutually opposed: one is almost exclusively J2 and the other almost only J1, suggesting very strongly that when those arrived J2 and J1 were much more neatly segregated than today in West Asia and that was probably before the PPNB flow southwards from Anatolia/Kurdistan.

      J1 can still represent a flow of Neolithic influences from (PPNA) West Asia into Africa, probably with AA language (which ones) and probably at least partly in NW Africa as part of late Capsian (Capsian Neolithic). I can't imagine any later possible scenario for the arrival of J1.

      Delete
    11. "The figures of Morocco should not be extrapolated to all NW Africa."

      It wasn't. I used this table from WP : http://en.wikipedia.org/wiki/Y-DNA_haplogroups_by_populations_of_North_Africa after taking out the North East Africans (Egypt) and do a weighted average of the NW Africans, which include Algerians, Tunisians,Morrocans, etc... you get the numbers I showed above, which is a representation for general NW Africa.

      Here also is an interpolated map for J1-M267 spread in NAfrica and the Near East :http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2986692/figure/fig1/ from Toffanelli (2009), it doesn't show the % but you can get a general feel. BTW, Toffanelli (2009), like you, reject the 'Islamic expansion' scenario and postulate a mid holocene expansion for the 'Arabic' or low diversity type of J-M267 haplotypes found in NAfrica and the Near East, but I think this is simply a mutation rate issue or an issue related with the conversion of hapoltype diversity to a timeline, but the bottom line is that the majority of J-M267 haplotypes in NAfrica (including Sudan) have a much lower diversity than other J-M267 haplotypes. I however, tend to buy the scenario where a genetic impact of the islamic expansion being responsible for the slight (<10%) representation of
      J-M267 haplotypes in general in NW Africa, possibly accompanied with other minor ones, E-M78, J2, R1b, and G.

      Delete
    12. The list of Y-DNA frequencies you mention should be handy, thanks.

      "I however, tend to buy the scenario where a genetic impact of the islamic expansion being responsible for the slight (<10%) representation of
      J-M267 haplotypes in general in NW Africa"...

      In most large populations it is >20% and > 30%! And there is no proportionality with J2 as would be expected from any "Arab" flow from West Asia, be it Peninsular Arab (8/3 J1/J2 ratio) or Fertile Crescent (roughly 1-1 J1/J2 ratio).

      Even Kabyles (Berbers related to ancient Numides) have 16% J1! But then 0% J2. Compare with nearby "Arab" Algeria or Oran: 23% J1 vs 5-6 J2. What is Arab in this? Easy: all 5-6% J2 and the equivalent (x1 or up to x2.5) fraction of J1, i.e. 5-15%. 23%-16%=7%, so that extra J1 of Algiers and Oran and all or nearly all J2 may be Arab (or partly Phoenician).

      Take Tunisia: <4% J2 means that at most 10% of J1 can be of Arab or otherwise West Asian (Phoenician) origin, most likely not more than 6%. That means that maybe 25-30 of those 32-35 percentile points of J1 in Tunis capital are pre-Semitic, not just pre-Arab but also pre-Jewish, pre-Phoenician...

      They could still be Neolithic but that would still be (AFAIK) a process within Capsian culture and therefore of ancient Berber origin. That's what happens with most of NW African J1: we cannot imagine a Semitic migration without their share of J2, which should be between 2/5 and 1/1 of the frequencies of J1.

      Delete
    13. I've been studying the table in more detail and I see that it is characteristic of NW Africa, Upper Egypt and (for what I've seen elsewhere in this blog) of The Horn (therefore Afroasiatic in most of Africa, most of AA by diversity) that J1 is much greater than J2 or even that it appears with no J2 whatsoever.

      There are very few populations lacking J1 and all look like highly endogamic and very isolated (Tunisian Berbers that have been shown elsewhere to be extremely endogamous, inbred), the very remote and peculiar Amizmiz Valley pop. of Morocco (not representative of Moroccan Berbers in general) and the Tuaregs (remote like them alone and with very strong founder effects - only pop. on Earth with >70% mtDNA H1 for example). Regular Berbers like Kabyles, Sened, Morocco's Middle Atlas or Asni Berbers... all show plenty or at least some significant J1. As said before it seems more concentrated towards Tunisia/Algeria and rarer in Morocco.

      What happens with these Berber populations with J1 is that, unlike their cosmopolitan "Arab" neighbors they usually lack any J2. "Arabs" anyhow have only a small apportion of J2 (highest is 6% in Algiers, while highest J1 is 34% in Tunis, hardly comparable at all).

      The only mainland North African population to have J1 and J2 more or less even is Lower Egypt, what is a more clear evidence for Asian inflows after Neolithic.

      The only population that has more J2 than J1 is Canary Islands, which was never arabized in any way and must track all that J2 to Spain. They still retain quite more J1 (4%) than any place in all Iberia, an obvious signal that J1 was at least to some extent present in Westernmost Berbers before the Arab/Muslim invasion. Most of Canarian Y-DNA genetic pool is nowadays Iberian but they still retain almost 9% of E-M81 and 4% of E-M78 (some of these two might be Iberian but not that much probably, at least not E-M81). Today Morocco (Onofri's general sample) has almost three times more E-M81 than J1 (but zero J2!!!), what is consistent with what we see in Canary Is. (would be c. 10% Y-DNA surviving from the Guanches, more in mtDNA and autosomal, the other 90% or almost is very similar to today's Iberian Y-DNA pool, there you have a clear recent unarguable replacement with everything fitting expectations almost perfectly).

      Leaving "the lucky islands", there are two kinds of populations in regard to J:
      → (1) North Africans, Upper Egyptians and Horners: with lots of J1 and very little to zero J2.
      → (2) West Asians and Lower Egyptians: with near parity of J1 and J2 (at worst slightly under 2/1 ratio). There are differences in West Asia with Northerners having little J1 and Southerners (and East Caucasians) relatively low J2 but never to the near-zero levels of Africa (excepted Lower Egypt).

      Hence if NW African J1 lacks diversity, as you sustain (I believe you in that), it cannot be attributed to Arabs but to some NE African source (Upper Egypt, Horn, maybe Sudan), which I can only imagine to fit with Capsian or Neolithic flows but not at all with any Semitic flows, be them Phoenician or Arab.

      Delete
    14. Sorry, I had been busy with other engagements to reply back any sooner, but let us step back and take a look at what you are arguing, in general you are claiming, very boldly if I may add, that Afroasaitic in North Western Africa was likely brought by E-M78 and J-M267 carriers, and NOT by those who brought the E-M81 paternal lineage into North West Africa.

      I on the other hand oppose your claim by arguing that the introduction of the E-M81 lineage into NW Africa likely brought with it the Afroasiatic tongue, maybe also in association with other minor lineages that may have been picked up along the phylogeographic path traced by the E-M215/M35 lineage from East Africa, because:

      a) E-M81 is the most frequent lineage in NW Africa.
      b) E-M81 has a more recent common ancestry with East African variants of E-M215/M35 than it does with the Sudano-Egyptian variant of E-M215/M35, i.e. E-M78, thus hinting at the latter's independent movement from the former, or vise versa.
      c) There is simply not enough frequency of E-M78 and J-M267 as a whole in NW Africa to justify your supposition.
      d)Most J-M267 haplotypes in North Africa have limited diversity, thus pointing to too shallow a time depth for their introduction to be congruent with the age of Afroasiatic in the region, perhaps even as shallow a time depth as the Islamic Expansion of the early middle ages.
      e) If J-M267 was responsible for bringing AA into NW Africa, then we would see a more even distribution of the lineage throughout NW Africa and not pockets of heavy concentration only in Algeria and Tunisia, mostly only in the cosmopolitan areas to boot.

      Delete
    15. "... you are claiming, very boldly if I may add, that Afroasaitic in North Western Africa was likely brought by E-M78 and J-M267 carriers, and NOT by those who brought the E-M81 paternal lineage into North West Africa".

      I'm proposing tentatively, alternatively to your model. I do not have a well finished theory on the matter.

      By the moment I can say with great certainty that E-M81 and J1 have complementary distributions in NW Africa, that J1 looks like a later arrival that must have displaced E-M81 westwards and towards the refuge pockets ("Berber", mountains) and that most of that J1 does not look at all Arab (neither peninsular nor from the Fertile Crescent, nor otherwise Semitic) because it lacks J2 (and other clades) in the apportions necessary.

      Whether E-M78 in NW Africa is related to that J1 inflow (as I thought first) or not, I'm not so sure but it looks plausible on first look.

      Whether of the two inflows only the latter is related to the AA/proto-Berber cultural input or both are, I am in doubt as well - but, even if both are AA-related, they are two different waves, probably separated by a lot of time.

      "There is simply not enough frequency of E-M78 and J-M267 as a whole in NW Africa to justify your supposition".

      Sorry but we are talking figures of 30% J1 alone in many places and more than 20% all around and even more than 15% among some "Berber" refuge areas like Kabylia. Do you think that for a language to become dominant an almost total replacement is necessary? Then we are in completely different concepts of reality and I am positive that it can be proven with facts that it is not necessary at all.

      For example Irish and Basque have similar frequencies of R1b (more precisely of R1b1a2a1a1b-P312/S116) yet ones speak a pre-Indoeuropean language and the others an Indoeuropean one since history began some 2000+ years ago. Obviously either one (surely the Irish) changed language with only minimal incorporation of foreign Y-DNA (they have some minimal R1a that Basques lack for example) but most men continued there with little to no change other than language and eventually also identity. There are many more examples: Turks used to be Indoeuropean Anatolians, Spaniards used to be Celtic and Iberian peoples, Southern Iraqis used to speak Sumerian, Kurds were Hurrian, South Iranians were Elamites, Azeris were Iranians, etc.

      So a 20% frequency (or even much less) is pretty much more than enough to cause a language shift if other social and political conditions are favorable.

      (continues)

      Delete
    16. ...

      "Most J-M267 haplotypes in North Africa have limited diversity, thus pointing to too shallow a time depth for their introduction"...

      Not really: diversity does not tell us that so linearly (for example probably Brazil has greater R1b, or general genetic diversity than Portugal but that's because it has incorporated peoples from diverse origins), instead it'd seem to talk of a marked bottleneck (founder effect) at the origins of that migration, what is plausible.

      Anyhow an Epipaleolithic to Neolithic time frame is not that long ago and I fail to see any other plausible phenomenon for the introduction of J1 in NW Africa, once I had to discard West Asia (at least West Asia after the earliest Neolithic) as plausible source.

      In any case this can be tested because there must be a hidden phylogeny within J1 that we are not really accessing here (the lineage is not too well researched). My bet is that there is an African specific branch or branches of J1 and that the North African main one hangs downstream of the NE African (Upper Egyptian, Sudanese or Horner) branch(-es) and not the Asian-specific ones (of course the residual Arab J1 should hang from these but that, as I expose above, is not the bulk of it).

      "If J-M267 was responsible for bringing AA into NW Africa, then we would see a more even distribution of the lineage throughout NW Africa"...

      Not necessarily. You can imagine that the immigrants conquered some of the best and more accessible lands and there proceeded to mix with the locals and extend their power and language without much further demographic expansion. Possibly historical polities like Numidia or Mauretania or their speculative predecessors of the Metal Ages, etc. played that role of spreading not just proto-Berber but then also Berber over it (as you may know Berber appears too young and homogeneous suggesting a second linguistic expansion within NW Africa itself). You don't need to replace a people to impose your language if you have the time and the power/prestige in your favor. And from Capsian/Neolithic to historical times there were many millennia, enough time for all sorts of intra-regional rearrangements.

      Whatever the case, you are not providing an alternative valid frame for the spread of J1 (without J2, etc.). If you could provide an acceptable scenario, then I would change my mind.

      Delete
  2. The devil lies in knowing the archeogenetic record... Only then will we be able to tell.

    ReplyDelete
  3. Certainly a good archaeogenetic record would tell a lot but the first isolate findings can be quite confusing. I hope that there is something of that for Africa and West Asia soon. By the moment it is all very fragmentary and localized.

    ReplyDelete
  4. I wouldn't hold my breath for any ancient DNA extraction from Africa just yet, maybe some mtDNA (because of its abundance) but as for YDNA, it takes a lot more resources

    ReplyDelete
  5. Just some more data about J1 in Morocco from Alvarez 2014 "Y-chromosome analysis in a Northwest Iberian population: unraveling the impact of Northern African lineages.". Interesting as usually Moroccan samples studied are endogamic and isolated groups.

    Samples below are from costal cities (except Figuig oasis). Alvarez used haplogroup predictor here for these samples from other studies but nevertherless quite interesting:

    J1 frequencies:
    Rabat (n=267) : 21.3%
    Casablanca (n=166) : 15.7%
    Figuig Oasis (n=96) : 29.2%
    El Jadida (n=49) : 8.2%

    ReplyDelete
    Replies
    1. But notice that E1b1b in NW Africa has two variants: M81, which is specific of the region and most concentrated towards the West and M78, which is articulated around Egypt/Nubia and has a Wester subclade (V65) with core in Tunisia. The latter is most closely associated, I understand to J1, which also has a Nile origin for this region and surely represents surely a secondary or tertiary settlement wave (possibly Capsian or Neolithic), while M81 is in NW Africa like R1b in West Europe: a bit mysterious re. its origins but clearly concentrated towards the Atlantic.

      Any analysis of Y-DNA in NW Africa should consider these two different E1b subclades as separate but, lacking this, I think it's quite safe to assume that E1b in the areas where J1 is more prominent will likely be less M81 and more M78.

      Otherwise I'd think "normal" that the more cosmopolitan populations (Arabs) have more "Tunisian" genetics, while the more isolated ones (Berbers) have less of them and keep a greater residue of R1b (and I2a!), which were presumably more important in the distant past (Oranian, also rather high I frequencies among Guanche mummies, which may be a local founder effect).

      Delete
    2. *Corrected berber results*
      Interesting dataset, the total Moroccan data had the following composition according to my calculations:
      N =647

      E1b1b 52.1%
      J1 18.4%
      E1b1a 6.8%
      R1b 4.8%
      J2a4b 4.2%
      G2a 3.7%
      R1a 2.6%
      I2a 1.5%
      J2a4 1.4%
      C3 1.1%
      O3 0.8%
      E1a 0.6%
      I2a1 0.3%
      H 0.3%
      E3b1)-M215 0.3%
      J2a4h 0.3%
      T 0.3%
      J2b 0.2%
      I1 0.2%
      L 0.2%

      While the Morocco Berber only data (Rabat, Figuig, Azghout and Khenifra) had;
      N =190

      E1b1b 54.7%
      J1 13.2%
      E1b1a 10.5%
      R1b 9.5%
      J2a4b 3.7%
      G2a 2.6%
      I2a 1.6%
      E1a 1.1%
      E3b1)-M215 1.1%
      O3 1.1%
      I1 0.5%
      J2a4 0.5%

      And the Morocco Arab data (figuig, jadida and Rabat) had:
      N =223

      E1b1b 47.1%
      J1 23.8%
      R1a 7.2%
      G2a 5.4%
      J2a4b 3.1%
      E1b1a 2.2%
      J2a4 1.8%
      R1b 1.8%
      C3 1.3%
      I2a 1.3%
      O3 1.3%
      E1a 0.9%
      I2a1 0.9%
      J2b 0.4%
      H 0.4%
      T 0.4%
      J2a4h 0.4%

      The Biggest differences I see between the berber-arab data are in J1,R1a and G2a being more significant in the Arabs, whereas R1b and E1b1a were more significant in the berbers. While Slightly more higher E1b1b frequencies were observed amongst the berbers.

      Delete
    3. Maju, these results posted are predicted from str haplotypes, meaning the ability to discern particular E1b1b variants can not be trusted, at best, below E-P2, only the discernment between E1b1a and E1b1b can be trusted.

      Delete