November 1, 2011

Minimal 'Denisovan' admixture in SE Asians

In outright contradiction with a recent paper by Reich, Skoglund and Jakobson now propose that there is tiny bits (c. 0.67%)  of Denisovan admixture among SE Asians, notably South China populations.

Pontus Skoglund & Mattias Jakobsson, Archaic human ancestry in East Asia. PNAS 2011. Open access. [doi: 10.1073/pnas.1108181108]

I must say that I really hate these highly complex papers that use ill-explained statistical modeling: they make almost impossible to make a criticism of any sort and make me suspect, right or wrong, that they could be burying, consciously or not, inconsistencies under the complex modeling.

But they could be right and I'm just to shallow to understand.

In any case, it could make some sense with a pattern I did detect in Reich 2010:
In (...)  table S8.2 French, Han and Cambodians (and only them) also appear to show some admixture with Denisovans, though maybe a third or fourth of that of Melanesians.

I almost nailed it according to this paper. 

It would also be consistent with the alleged 'Denisovan' introgression of HLA*A1. But this introgression was called to question, as the allele is common in Uganda however, what is inconsistent with any introgression in Eurasia and rather suggest random founder effects (or selection) from an African source.

So your take, I have no opinion - at least not a clear one. 

Thanks to Neanderfollia[cat] for calling my attention on the matter.

Update: take a look at John Hawks' opinion if you wish. It's interesting.

16 comments:

  1. "In (...) table S8.2 French, Han and Cambodians (and only them) also appear to show some admixture with Denisovans, though maybe a third or fourth of that of Melanesians."

    But they did not found any denisovan admixture in Europeans this time.

    "It would also be consistent with the alleged 'Denisovan' introgression of HLA*A1. But this introgression was called to question, as the allele is common in Uganda however, what is inconsistent with any introgression in Eurasia and rather suggest random founder effects (or selection) from an African source. "

    In fact, Africans appear to be closer to denisovans than west Eurasians (according to maps).

    I don't understand either why do native Americans in one map appear to have the same level of denisovan admixture as East Asians, nor why is only the northern part of Australia marked in orange like SE Asia (see maps).

    They managed to "separate" the Denisovan admixture from the neandertal one. They say that's why denisovan admixture was not detectable in SE Asians, because they were biased towards neandertals (?)

    ReplyDelete
  2. As I was mentioning in your blog a moment ago, French in table S3 appear more "Denisovan" than Han or Cambodians and are still the more "Denisovan" population of West Eurasia at least by what they call "average frequency of the Denisova allele at SNPs where Denisova differs from both Neandertal and chimpanzee".

    But then they make other computations in table S4 that do not include all populations (they average across regions). Also what's the difference between 0.534 (Papuans) and 0.519 (Brahui), which are the extremes of their list? 1.5 percentage points or a mere fluke?

    I do not understand the maps either. It was my first idea to post one of the maps as illustration here but eventually I gave up, as they are more confusing than informative.

    ReplyDelete
  3. "take a look at John Hawks' opinion if you wish. It's interesting".

    As usual:

    "south China differs from north China in essentially the same way that the Oceanian people do from other regions. And the Oceanian populations (here, Papua New Guinea and Bougainville) differ from other regions because of their Denisovan ancestry. So Skoglund and Jakobsson infer that the north/south comparison reflects Denisovan ancestry as well".

    Makes sense. Presumably the Denisova influence through much of East and Central Eurasia has been diluted by the subsequent expansion of a population that was not admixed with Denisova. The final comment in the abstract sums it all up:

    "the history of anatomically modern and archaic humans might be more complex than previously proposed".

    ReplyDelete
  4. "Presumably the Denisova influence through much of East and Central Eurasia has been diluted by the subsequent expansion of a population that was not admixed with Denisova".

    You have no evidence of that presumption of yours. In fact all the genetic evidence supports a South to North flow (or several). Even if there have been North to South flows also these have been late and minor.

    So quit presuming and begin looking at the facts, not as something you can twist to fit your preconceptions but as something per se, something to meditate rather than just rant about.

    ReplyDelete
  5. I think the pattern can be understood if you propose that the Denisovan admixture took place before the Toba eruption (and the data seem to indicate as much). The eruption itself, the ensuing change in climate (and much later sea level changes) all would have contributed to an ancient, admixed population survival in the extreme Southeast.

    On the flip side, there were another 30,000 years or so for later arrivals to dilute the admixture in South China and surroundings to the ~1% estimated here. And much of that could be functional for that climate and region, so it is not that far-fetched that it is not present farther north even if many migrations took that route. However, I still believe that some migrations took place inland (Myanmar/adjacent China, and/or southern Siberia), anyway.

    As to the color strangeness, some of it is surely just the sampling. For example, if there is no Australian data point, that would explain the color transition there. Same for Africa.

    ReplyDelete
  6. I don't understand what you mean, Eurologist. First the continuity of the admixture in SEA is not fully demonstrated (two recent papers say opposite things in fact). Second, even if it would be, it is a matter of a tiny slice of admixture, so thin and subtle that is extremely elusive.

    Finally, when we quit looking at regions and look again at samples (table S3, S.I.), we have the French quite high in the ranking. Actually, assuming that a number of samples in the 0.52 range imply zero admixture and Papuan (0.535) implies 2.4% admixture (with Erectus, double with Denisovan but half is redundant Neanderthal IMO), then we can infer the possible Erectus admixture in each sample: each percent point change in table S3 equals 1.6 actual percent points of Erectus admixture (roughly). So we get:

    Pima, Hazara, FIN, Druze Balochi, Brahui 0%
    JPT, LSFIN, GIH: 0.2%
    Makrani, Burusho, Oroqen, Uyghur, Kalash, TSI: 0.3%
    Japanese, Bedouin, Palestinian, Pathan, Maya: 0.4%
    Lahu, Sardinian, CEU: 0.5%
    Adygei, Basque, Cambodian, Russian, Hezhen: 0.6%
    N. Italian, Mongola, Xibo 0.7%
    Han, CHB, Tuscan, Daur 0.8%
    Dai, Tu, French 0.9%
    Tujia, Yakut, Naxi 1.0%
    Miaozhu, She 1.1%
    Karitiana 1.2%
    Colombian 1.3%
    Melanesian 1.4%
    Yizu 1.6%
    Papuan 2.3%

    Very roughly but that's a more detailed result than the regionally-averaged ones the paper offers.

    So we do not have, if we follow this paper, an admixed and a non-admixed population, at least not so clearly cut, but we do have a diversity of tiny levels of admixture, somewhat greater in Eastern Eurasia (Native America included), but with lots of variation (Japanese or Pima rank very low for example, French rank relatively high instead).

    Europeans in general, excepting the Finnish, seem relatively high (0.5-0.9%), which may be attributed tentatively to the fact that the main lineages Y-DNA R1 and mtDNA R probably derive from ones original from SE Asia (MNOPS and N respectively).

    But I fail to see how the Toba explosion fits anywhere in all this.

    ReplyDelete
  7. Well, if we look at sites where modern humans are known to vary, and where simultaneously Denisovans carry the derived allele, and then we see that about half the time we do, too, to me that indicates that the split of humans was relatively close to that between Denisovans and Neanderthal (very roughly speaking). But there was always a mixed human population, so, one would expect some of the variation to the ~0.52... to be just chance (and by selection). But the authors see that the variation is not random but consistent with more Denisovan admixture not only in Papuans/ Australians and some of Oceania, but South China, as well.

    Close to zero admixture may very well be in the .525 range, with statistical fluctuations to either side. At any rate, I agree, it is a weak signal - but not the only one in the paper.

    As to Toba, I just brought it up because some timing estimates (of Denisovan admixture) agree, and it can account for the isolation of the population that soon later would become Papuan and Australian - while the remainder of the (now shrunken) population in SE Asia had a very long time to have that initially same amount of Denisovan admixture diluted by newcomers from the Indian subcontinent.

    ReplyDelete
  8. Papuans, Chinese and Europeans share the same lineages such as Y-DNA MNOPS and mtDNA R: their timing must be similar, and their origins as well: the higher concentration of "Denisovan" (Erectus) admixture, if real, must have other origins, such as admixture in Sundaland plus some random drift, very evident in the diverse apportions even if in the same region.

    "Close to zero admixture may very well be in the .525 range, with statistical fluctuations to either side".

    Statistical fluctuation would then be +/- 0.05 and that means no Denisovan admixture almost anywhere (only Papuans). Also I did spot the French as anomalously high in the Reich 2010 paper and it does seem like Europeans in general have a small slice of that admixture, we like it or not.

    Ideally I would have compared with Yorubas or other Africans but they are not listed, quite mysteriously.

    ReplyDelete
  9. their timing must be similar

    So what? That still leaves a perhaps 80,000 to 70,000 ya or even 50,000 ya years time frame. We all know that y-DNA and mt-DNA is not a good indicator of autosomal DNA, anyway.

    IMO, so-called Denisovan admixture is either local and thus erectus, or it is from the northern part of the sub-continent (or even farther north), and thus more likely heidelbergensis-like.

    I have more faith in the second scenario (i) because of the timing, (ii) because of the published Denisova data, and (iii) because even in this paper, the about 50% global rate of modern humans sharing the derived Denisovan allele, as far as I understand, brings the human split off the line very close to the Neanderthal-Denisova split (and thus much later than erectus).

    ReplyDelete
  10. So what? You are suggesting a storyline that is NOT what the haploid lineages tell us at all. Papuans are by haploid lineages very similar to East Asians and Europeans and that joint node must be from AFTER Toba.

    "IMO, so-called Denisovan admixture is either local and thus erectus, or it is from the northern part of the sub-continent (or even farther north), and thus more likely heidelbergensis-like".

    The mtDNA lineage of Denisovans is NOT Homo ergaster (the common ancestor of Neanderthals and us), much less H. heidelbergensis. It is from a much older node (H. erectus/georgicus necessarily). The autosomal DNA is indeed closer to Neandederthals but that is VERY STRANGE and can only be explained reasonably because they were a hybrid population (what makes total sense as both Neanderthals and H. erectus converged at that location).

    Solved the mystery (it is indeed a signal of H. erectus, let's face it) the location also seems obvious: SE Asia, where some H. erectus lived until recently (H. floresiensis, maybe others).

    "brings the human split off the line very close to the Neanderthal-Denisova split"....

    Only at autosomal level and anyone who has read Cavalli-Sforza (for example) knows that trees of autosomal blocks can be extremely misleading when the population is admixed. Mixed populations tend to appear as very high branches inside one of the components.

    Autosomal DNA is not a line but an ocean of data: a line or even a plane or 3D space can only be approximations, sometimes very rough ones. Also no H. erectus DNA has been studied to date; hopefully this will be solved soon.

    The passing of time has proven me right in previous cases and I'm 99% certain that it will vindicate me in this one as well.

    ReplyDelete
  11. "In fact all the genetic evidence supports a South to North flow (or several)".

    Only because that's how you've chosen to interpret the data.

    "Papuans, Chinese and Europeans share the same lineages such as Y-DNA MNOPS and mtDNA R: their timing must be similar, and their origins as well"

    But Y-MNOPS and R-mtDNA are not the only haplogroups in each region. These two haplogroups look to have expanded through an already-occupied world.

    "the higher concentration of 'Denisovan' (Erectus) admixture, if real, must have other origins"

    Yes. Those haplogroups were not the main carriers of the Denisove gene complex. They had mixed with SE Asian Homo erectus. But the Denisova genes had already become widespread, with the earlier expansion of C and F Y-haps and M and N mtDNAs.

    "IMO, so-called Denisovan admixture is either local and thus erectus, or it is from the northern part of the sub-continent (or even farther north), and thus more likely heidelbergensis-like".

    I agree. We should accept all evidence at face value until proved otherwise. Those who find it necessary to manipulate the data do so as a result of their pre-conceived beliefs.

    "You are suggesting a storyline that is NOT what the haploid lineages tell us at all".

    I agree with Maju here. I'm fairly sure that the Neanderthal/Human mtDNA line branched off the Denisovan line. Autosomal DNA indicates later mixture between Neanderthal and Denisova, but that may have taken place after a Denisova/Human admixture event.

    "The autosomal DNA is indeed closer to Neandederthals but that is VERY STRANGE and can only be explained reasonably because they were a hybrid population (what makes total sense as both Neanderthals and H. erectus converged at that location)".

    Surely the fact is that they 'became a hybrid population'.

    "Solved the mystery (it is indeed a signal of H. erectus, let's face it"

    If you're prepared to call the Altain Denisovans 'Homo erectus' I'm prepared to agree.

    ReplyDelete
  12. No, I'm not prepared to agree with you, Terry. No.

    But I'm not willing to discuss all your statements point by point again either.

    ReplyDelete
  13. Well, according to my recent NG Geome test results, I have all the usual Northern European and Mediterrianian percentages which I expected, 11% American Indian, also expected, plut 3.2 Denisovan which I didn't expect! Everything that I have read about this indicates that the American Indians are more closely related to the Neanderthal (where my 2.6% comes in) and that the Denisovans are almost exclusively traced back today to the Melanensians or Papuans. I then remember reading something about my Native American tribe (extinct Maidu) which mentioned that a few hundred years ago some of the members inter-bred with some native Hawaiians who had been brought in from the Islands to work the fields. That's my story and I'm sticking to it!

    ReplyDelete
    Replies
    1. The Denisovan admixture is also found among Australian Aborigines, and then, more diluted, in Polynesians and Filipino Negritos. So it's basically a Native Oceanian trait. Some has been claimed in SE China but, if real, it'd be at very low levels anyhow. I'm not aware of any being detected among Native Americans yet, much less at the levels you mention, which are on the Filipino Negrito range (about half of Papuan/AA).

      I'm a bit puzzled by your results, admittedly. What company did you use?

      Delete
    2. National Geographic Human Geome Project 2.0. My mother's mixed heritage was Swede, Welsh, Scott, and American Indian

      Delete
  14. Well, according to my recent NG Geome test results, I have all the usual Northern European and Mediterrianian percentages which I expected, 11% American Indian, also expected, plut 3.2 Denisovan which I didn't expect! Everything that I have read about this indicates that the American Indians are more closely related to the Neanderthal (where my 2.6% comes in) and that the Denisovans are almost exclusively traced back today to the Melanensians or Papuans. I then remember reading something about my Native American tribe (extinct Maidu) which mentioned that a few hundred years ago some of the members inter-bred with some native Hawaiians who had been brought in from the Islands to work the fields. That's my story and I'm sticking to it!

    ReplyDelete

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).