April 26, 2014

Diversity and legacy of ancient European farmers

Two new Swedish papers, published almost simultaneously, add important extra information and, of course, also some more questions to the previous analysis of European ancestry by Lazaridis et al. (see here, here and here). They are:

Evangelia Dasakali, A late Neolithic Iberian farmer exhibits genetic affinity to Neolithic Scandinavian farmers and a Bronze Age central European farmer. Paper IV of the greater collection Archaeological Genetics - Approaching Human History through DNA Analysis. Acta Universitatis Upsaliensis, 2014. Freely accessible PDF → LINK [ISBN 978-91-554-8816-1]

Pontus Skoglund, Helena Malström et al., Genomic Diversity and Admixture Differs for Stone-Age Scandinavian Foragers and Farmers. Science 2014. Pay per view (supp. info. freely accessible)LINK [doi:10.1126/science.1253448]

I must say that it seems a bit odd that, being colleagues of the same university, Dasakali and Skoglund/Malström have decided to publish such two strongly related studies separately. But, well, whatever...

The first study briefly compares the genome of a Chalcolithic Iberian farmer from El Portalón (Atapuerca) to modern Europeans and some previously published ancient genomes, mostly Neolithic but also one from Epipaleolithic Iberia (La Braña).

The study is just a very short synthesis and lacks detail but provides this PCA graph:

On it's own, it already provides evidence of ancient European farmers not being as homogeneous as the  Lazaridis sample (n=3 but all from near the Alps) may suggest. The Iberian sequence deviates more towards modern Italy than the Alpine ones (here represented only by Ötzi), who tend to cluster towards modern Sardinians. Also the Gökheim sequence from Southern Sweden, the only clearly Megalithic sample so far to be studied, clusters with Basques. But Gökeim is better addressed in the Skoglund paper in fact.

This is because Skoglund & Malström use more samples from Neolithic Scandinavia: four from the Gökheim site (Megalithic and Funnelbeaker farmers) and six from Götland's Pitted Ware late semi-foragers of the Baltic (probably with partial Eastern European roots).

The results are more directly visualized and comparable with those Dasakali and Lazaridis in the Principal Component Analysis:

Fig. 1. Population structure and genetic affinities of ancient Eurasians. (A) Ancient individuals were projected on principal components computed on 57 modern-day Western Eurasian groups. (...)
Notice that ancient samples are projected on the modern PCA, what explains especially the anomalous position of Mal'ta 1 (MA1), which is actually much more divergent from all modern Europeans. SF11 = Stora Förvar 11.

Obviously not all is just a simple PCA, together we find these heatmaps that help us to better understand the affinities of the ancient farmers:

Fig. 1. Population structure and genetic affinities of ancient Eurasians. (...)
(C) Shared genetic drift between Western Eurasian populations and Ajvide58, a Neolithic hunter-gatherer from Scandinavia.
(D) Shared genetic drift between Western Eurasian populations and Gökhem2, an early farmer from Scandinavia.

In fact Gökheim 2 shares a bit more genetic drift with Sardinians than with Basques, an that's probably the reason why she is located rather towards Spain and Italy in the PCA, even if they don't share greater genetic drift with these than with Basques and Sardinians (PCAs can be misleading). However the shared drift is probably less than among Alpine farmers, owing largely to Gökheim's greater aboriginal hunter-gatherer admixture.

In general, like other early European farmers, Gökheim people seem to align best with Western Mediterranean peoples, particularly those with low paleo-Siberian (MA1) affinity, which are Basques and Sardinians. 

Judging on the PCA, there seems to be some diversity among them, with one being more extremely "Atlantic". It is very possible that in the Neolithic and Chalcolithic Atlantic Europe there were others like her (most Gökheim samples are female although I'm uncertain about this unnumbered one).

On the other hand, Pitted Ware sequences share greatest drift with modern Baltic peoples, including the oversampled Swedes but also with Eastern Baltics (Estonian, Finnish, Lithuanians). This last surely owes to the origins of this culture in Eastern Europe but in what regards to Sweden, it must imply that their kind of Eastern-like genetics were very strongly favored when the Kurgan macro-culture (Single Burials, part of the wider Corded Ware culture) took over the region c. 2500 BCE. Instead the ancient megalithic farmers quite obviously suffered a genocide in this part of the world.  

Overall, it seems obvious to me by now that Eastern European genetics distributed by the long-term Kurgan expansion (Indoeuropeans) must have been somewhat similar to aboriginal Western European hunter-gatherers but with greater paleo-Siberian affinities (MA1-like), which they also distributed by Europe for the first time.

For a wider comparison with World genomes, take a look at figures S3 and S4 please. In them we can see that Ajv58 is comparatively closer also to South Asians, Siberians and Native Americans, while Gök2 is instead relatively closer to Palestinians, Arabians and slightly to Egyptians too (not apparent in the above graph but certainly in the global one).

This is quite interesting because it directly relates to the mysterious "Basal Eurasian" element that Lazaridis et al. detected in their ground-breaking study, which is more clearly perceived as something "Palestinian", with whatever NE African and/or residual out-of-Africa element in it, which partly detaches the ancient farmers from the main Eurasian ("out-of-India") branch.

Some of all this is also quite apparent in the attempts of Skoglund & Malström to infer the ancestry of the various paleo-populations:

Fig. 2. (A) Sample locations of ancient human remains that are included in the population history model. (B) Admixture graph of population history fitted to ancient genomes showing more hunter-gatherer admixture in Neolithic Scandinavian farmers than in central European farmers (table S15).

The above "admixture graph" is identical in essence to fig. S6. However I find even more very interesting the fig. S7:

Fig. S7. A) Admixture graph of population history fitted to ancient genomes and two modern-day genomes from Europe. (...)

This alternative result has been produced by merely adding modern French and Sardinians to the analyzed pool and allowing for two extra admixture axes. The result is quite different:

Essentially the overall unity of all Europeans, ancient and modern is restored as a single derived branch distinct from the circum-Pacific populations. The deviation to "Basal Eurasianness" of Neolithic Europeans and Sardinians is instead largely explained by an admixture axis from the Dinka, which act as a proxy for whatever Red Sea area input in the early European Farmers. This is just as I expected and have tried to explain before, so I guess you understand why I do prefer this second admixture graph over the other.

In this sense, I feel that Skoglund & Malström could have done a bit better by testing more alternative models (different samples, different pre-definition of the admixture axes, etc.) but I guess it's good enough for their Swedish focus. Not so much for the pan-European one though. 

It is also very striking that the French sample appears unrelated to any of the ancient samples, be them farmers or not. Instead they show up as a branch of their own with proto-Sardinian admixture. This probably owes to the fact that neither of the sampled ancient populations is directly ancestral to the French, unlike what happens with Lochsbour, for which the French show the longest IBD segments, indicating a more direct ancestry. So I guess that we can take the French branch as being partly Lochsbour-like, plus whatever Eastern European (but not Scandinavian) extra ancestry the Celts, Romans and Germanic tribes brought with them (and of course the strong Neolithic element indicated as paleo-Sardinian admixture). Sadly the analysis strategy is not designed to provide us with such information here, but it can still be inferred if we consider all the available data.

Of course most modern samples show less drift because they are more cosmopolitan populations than the ancient ones, whose demographic base should be more like that of Australian aborigines, the only modern sample that shows a similarly high drift parameter. 

As concluding remarks, I wish to underline the greater aboriginal European (WHG or blue color) admixture in the Megalithic farmers of Gökheim, which resemble modern Basque genetics. It is very likely that more populations like those existed in the Megalithic axis from Southern Iberia to Southern Sweden, although naturally each one had its own peculiarities and we should not take Gökheim as any sort of ideal representative of such ancient Atlantic population but rather as one of its many likely subpopulations. 

Before the Indoeuropean consolidation of the Corded Ware period, these Atlantic farmers must have greatly contributed to at least a preliminary stage of "modernization" of the Central Europan (and possibly also others') genetic pool, as discussed in this entry. Notice also that the demographic expansion was much greater with Megalithism than with Corded Ware and Bell Beaker and that this applies in general terms to nearly all studied regions, with the main exception being Jutland. However in Scania (Southern Sweden), there is a clearly second demographic expansion with the arrival of Corded Ware. 

Sadly there is no direct comparison of modern Swedes as hypothetical admixture between Pitted Ware Chalcolithic foragers and Megalithic/Funnelbeaker farmers. It would have been interesting to know more details about that, especially if a third element from Germany, Poland or Russia is needed to explain their modern genetic pool, as I suspect, and in which apportions.

As for the Neolithic farmers, they were clearly somewhat more diverse than the Lazaridis study suggested, varying at least in their amount of "Basal Eurasian" and Western hunter-gatherer ancestry, and possibly also in other subtle aspects we still do not understand well, as suggested by El Portalón sample.
Their most direct legacy is in any case concentrated towards Southwestern Europe.


  1. http://en.wikipedia.org/wiki/File:Floristic_regions_in_Europe_(english).png

    European climactic zones.

    I think - hopefully provable - that the Boreal zone shown in that map extended further south during the LGM and retreated back over time. If the southern boundary was in northern Germany / Poland then you have a (rough) distribution map of
    - cardium
    - megalithism
    - LBK
    - corded ware
    - yamnaya

    1. The floristic regions. I agree with Grey. I think that ecology and archaeology interact with one another far more than many would expect.

      Ötzi had an unusually high amount of Neanderthal ancestry didn't he? Do any of the other samples in here show that? Are there any suggestions as to where that admixture came from (ie "basal Eurasians" or somewhere else)?

    2. "Ötzi had an unusually high amount of Neanderthal ancestry didn't he?"

      That's what John Hawks informally published once in his blog (and I mirrored here, I believe) but it looked a bit preliminary and I have never seen it confirmed elsewhere.

      "Do any of the other samples in here show that?"

      Not that I know.

      "Are there any suggestions as to where that admixture came from (ie "basal Eurasians" or somewhere else)?"

      Same: no info.

      First of all it would have to be confirmed that Ötzi actually had that extra Neanderthal admixture. I've seen some of Hawks' estimates be contradicted in other studies, so I don't know what to think in this particular case. Same for whatever other ancient sample. So far I think nobody even bothered checking that.

    3. Unfortunate that no one has bothered to check. The ubiquity of Neanderthal admixture suggests it must have started very early, but I doubt it was only a single event.

      On pages 37 and 38 they have the residuals of their admixture model, which seems to show high values for Denisovans vs both Ma-1 and Saqqaq. That seems to potentially show some Denisovan admixture into both groups.

      There's also a fairly large remainder for Australian vs Ma-1, as well as between Ajvide58 and Anzick (though the last one is probably less surprising).

      Any thoughts? Would it be crazy to suggest a genuine albeit small contribution from Oceanian-like coastal migrants making their way inland to Lake Baikal? And further contributions both via ANE and perhaps the coast to the Saqqaq expansion?

    4. Can't say much. The extra admixture seemed not strong enough to be a clear signal but, in case it's real, I guess that they admixed at low frequencies with Central Asian Neanderthals, who had some Denisova admixture (this time literally Denisovan and not just proxy-Denisovan, i.e. H. ergaster).

      This kind of second admixture with Neanderthals may have also happened elsewhere in West Eurasia but, as the area supported larger populations, it got diluted easily.

    5. Hrm. From the looks of the paper, I don't think they actually modeled each possible migration edge. Am I mistaken in reading it that way?

      What I still can't understand is La Brana's Y-DNA haplogroup. It couldn't have got there through ANE peoples as he doesn't have that affinity.

      "This kind of second admixture with Neanderthals may have also happened elsewhere in West Eurasia but, as the area supported larger populations, it got diluted easily."

      I think you're right on the money there.

    6. AFAIK, you tell TreeMix how many migration axes you want and it produces a tree according to that and also a "residual" table that is used to estimate how good is the result. Both trees used looked reasonably good: any anomalous residual did not affect Europeans ancient nor modern.

      C-V20, which is also found in Europe at very low frequencies (and nowhere else that we know of) almost surely arrived here with the overall westward migration of the early UP from Asia.

      True that it is most closely related an almost equally rare Japanese lineage (former C1) but both are within the same clade as Wallacean and South Asian C sublineages (former C2, C5 and C*), and maybe also with Australian Aboriginal ex-C4. Only the former C3, now C2, is separated, and this one is precisely the NE Asian and Native American kind of lineage that more easily could be related to Ma-1 affinities.

      So I would rather relate this main C1 (new name) subclade with people living between South and SE Asia after the C2 (ex-C3) clan moved northwards. The subsequent expansion in various directions is similar to that of K (former MNOPS or Kx(LT)), so this European C almost certainly arrived in the same wave as P, had a very limited founder effect, barely enough to survive drift, and that's all.

    7. That UP migration from Asia would still likely be separate from the one that introduced I to Europe though would it not? Mixed origins for WHGs would make sense IMHO though.

      Do the molecular clocks from the breakup of C1a (the subclade both C-V20 and C-M8 belong to) for this scenario?

      Even if they do, P and its descendants seem very much to be the ANE-affiliated. In order for La Brana to not have any strong ANE affiliation, I suppose that means C1a would have had to split up before ANE differentiated itself.

      I guess that's pretty plausible really. It'd be interesting to look at the phylogeny of Y DNA Q in Japan.

    8. For me the West Eurasian colonization in the early UP is a single process, regardless of whatever complexities involved at local level. On the Y-DNA side there were haplogroups R1, Q, IJ, G and T (plus rarities like K4, P* or C-V20) and on the mtDNA side R0, U, JT, N1, N2, X and M1. Naturally we can detect some geographical patterns affecting some of these lineages in the periphery of the expansion such as yDNA T and mtDNA M1 heading across the Red Sea or yDNA Q1 to Siberia with some mtDNA X2, or yDNA G being initially restricted to West Asia or mtDNA X1 to Egypt. But overall it is extremely difficult to discern without much more aDNA evidence than we have now other patterns, especially in the core areas of West Asia and Europe.

      Said that, what we see in Iberia are lineages less common elsewhere like mtDNA H. It's probable that this also happened in the northern margins of Eastern Europe (Sunghir, later Karelia). Why? A possible explanation is that the intrusive nature of Gravettian, which seems essentially unrelated to Aurignacian and rather a secondary wave from West Asia, may have displaced the first H. sapiens settlers to the margins, later experiencing some recovery with the Magdalenian (which is in essence a refined Aurignacian). But all this is very difficult to discern with any clarity, so I'm just saying very tentatively. It would explain the star-like nature of mtDNA H and yet its rarity in pre-Neolithic DNA (expanded massively and was then pushed to the edges).

      As for molecular clock-o-logy, I don't have enough data to judge yDNA C. C as such must be older than 55 Ka BP IMO but this is a very minimal age based only on the 8 1000GP samples.

      "P and its descendants seem very much to be the ANE-affiliated"...

      I don't see why. Some descendants of P were clearly present in UP Siberia but that doesn't imply necessarily any sort of "affiliation" in general terms, much less in relation to a much more recent role as conjectural vectors of pseudo-ANE (Eastern European) genetics in Chalcolithic and Metal Ages' Europe.

      "C1a would have had to split up before ANE differentiated itself".

      It surely did, although I don't think your logic is necessary for that. On the other hand, in pure theory, it's perfectly possible that C1a was introduced to either Japan or Europe or both from Altai - although I do not think it is the case and rather think that C1 was associated as "minor partner" to the expansion of K (MNOPS)being often found in similar areas (East Asia, Australasia and now it seems that Europe as well).

    9. Given the amount of other Siberian-derived Y-DNA and mtDNA in Japan, I think C1a1 and C1a2 following path via India to Siberia seems most plausible to me at least. The median joining network in this paper seems to suggest that as well: http://www.nature.com/jhg/journal/v51/n1/pdf/jhg20068a.pdf

      Also, from Ray Banks' Facebook post:

      "The haplogroup C tree at ISOGG has been considerably revised. The C1 group primarily from Japan is found to share a common origin with the European V20 subgroup, and these both share a common origin farther back with the southern Asian M356 group which is most common in far western India. Together these comprise the new C1 group."

    10. But India was probably also involved in the Eastern Asia genesis, judging on the convoluted zig-zags (SA-SEA) of mtDNA N/R and Y-DNA F/K, among other less notorious lineages. Even the closest relative of NO under K seems to be not P nor Australasian K-derived lineages but a mysterious Indian "X" lineage.

      In fact, based on Hammer's haplotype network, the only thing I can gather is that Australian ex-C4 is probably closely related to Wallacean ex-C2, all the rest being most unclear. This extreme has not been clarified yet however. There's not even ex-C3 (now C2) in that graph for what I can see.

      I was privately asking Ray Banks for explanation on the new C haplogroup reorganization and he kindly explained me that former names C1, C2, C5 and C6 make up the new C1, under which C1a includes former C1 and C6. C3 is apart (now C2), while the affiliation of Australian C4 is unclear. Not sure where Indian C* goes anyhow (probably Banks doesn't know either). Regardless of whether it hangs from the basal C node or from the new C1 one, it would support a "loosely Southern Asian" origin for C or C1 respectively.

      Whatever the case, my skepticism about C1a having arrived Japan via Altai is actually based on the lack of Q in the islands. If Q did not make it, why would C1a?

      Anyhow, it is so tiny in both regions that it is not really too important.

    11. Re: the residual. here's what the software's authors have to say:

      "Positive residuals indicate pairs of populations where the model underestimates the observed covariance, and thus populations where the fit might be improved by adding additional edges. Negative residuals indicate pairs of populations where the model overestimates the observed covariance; these are a necessary outcome of having positive residuals, but can also sometimes be interpreted as populations that are forced too close together due to unmodeled migration elsewhere in the graph."

      "After finding the single migration edge that most increases the likelihood, we attempt a series of local changes to the graph structure (Methods). We then iterate over this procedure to add additional migration edges. In principle, migration edges could be added until they are no longer statistically significant (see the following paragraph). In our experience, however, we prefer to stop adding migration events well before this point so that the resulting graph remains interpretable."

      "We have made this assumption to simplify the search for the maximum likelihood graph; additionally, we speculate that in graphs with complex structure, there will be many graphs that lead to identical covariance matrices, and thus several different histories will be compatible with the data. That said, improvements to the search algorithm could allow the assumption of approximate treeness to be somewhat relaxed. Currently, if the number of admixed populations is large relative to the number of unadmixed populations, this assumption breaks down. For example, in the human data, note that we see no evidence of the documented gene flow from Neandertals to all non-African populations"
      So yes, the residuals are certainly fine, but that doesn't mean they're meaningless either.

      Source: http://www.plosgenetics.org/article/info%3Adoi%2F10.1371%2Fjournal.pgen.1002967#pgen.1002967.s009 Figure S15 is interesting btw, with a migration edge from the San to Eurasia. It also shows the Mbuti as closer to Denisovans than other African groups, which is shown in the Skoglund paper.

      Re: Q in Japan - it did reach Japan - though only at 0.5%.

      I wouldn't call a haplogroup that makes up ~5% of Japan's Y-DNA pool unimportant.

      FYI, I'm not objecting to C or C1 having broken up in SE Asia or South Asia. Just C1a. It seems a bit unlikely to me that C1a would survive at such low levels that it just barely persists today in Japan and Europe and nowhere in between. I'm skeptical of it surviving for a huge length of time over such a huge distance . I realize founder effects and whatnot make human genetics a very stochastic process, but still. Rather, it seems more plausible that it was at high levels in a small group, with that group contributing to both Europe and Japan's hunter-gatherers (or perhaps a later entry in Japan's case).

      It's like flipping a coin 100 times, and ending up with Heads, Tails, Heads, Tails and so on. That's just as unlikely as flipping a coin 100 times and only getting heads or only getting tails. It's possible, but it feels contrived.

    12. But the residual is low, excepted for some non-European pairs, which are pretty much irrelevant here.

      "I wouldn't call a haplogroup that makes up ~5% of Japan's Y-DNA pool unimportant".

      I thought it was more like 1-3%. Whatever the case it's clearly a founder effect and its absence elsewhere makes it nearly impossible to track its pre-Japan paleohistory.

      Unlike Q, which probably reached Japan via Korea, i.e. with Y-DNA O, C1a1 almost certainly reached it earlier causing a very limited founder effect. It is precisely the fact that it is not found elsewhere, what suggests its extreme antiquity and makes it likely related to the arrival of D2, whose uniqueness to the archipelago is similar (although quite more common).

    13. I'm getting the frequencies of C-M8 from this paper:

      Supplementary materials:

      I think I found it free through Google scholar too, but I've got to go to sleep for work in 8 hours some I'm afraid I can't dig it up now. They have it at 0% for their "Kyushu", so you may be right about it being just 1-3% overall in Japan, with the higher result in the paper being due to oversampling of coastal regions (Aomori and Shizuoka are both actually on Kyushu's coast, so I'm not sure on the distinction).

      It's apparently absent in the Ainu as well, which is curious. http://link.springer.com/article/10.1007/s10038-004-0131-x

      This paper suggests C-M8 is from an intermediate migration wave, after the initial Jomon settlement, but prior to agriculture: http://www.scirp.org/Journal/PaperInformation.aspx?paperID=8687#.U2h5HvldX_E

      They date it at ~8460 - 18,690 years (which I believe actually comes from Hammer et al 2006).

      "But the residual is low, excepted for some non-European pairs, which are pretty much irrelevant here."

      Ma-1 and Avjide-58 have some decent sized residuals, and I think they're fairly relevant to Europe.

      Does something have to be relevant to be interesting? At the very least, I think a pattern found in one part of the world can yield insight into the history of another.

      It certainly shows hints of things suggested elsewhere. And perhaps hints of things not suggested elsewhere. Why do Ötzi and Avjide have a higher residual than Gökeim and Avjide? It can't just be WHG+EEF since the residuals from La Brana show the opposite pattern. If I'm reading the scale right, those residuals are big enough to be interesting.

    14. "Ma-1 and Avjide-58 have some decent sized residuals"...

      In both graphs Ma1/Aj58 have residuals colored as bright yellow: not outstandingly odd. According to what you said: the model overestimates the observed covariance; these are a necessary outcome of having positive residuals...

      So it can be just a by-effect of missing an admixture edge Denisova-Ma1 or whatever.

      Maybe you mean that Anzick/Aj58 "blue"?

      "Does something have to be relevant to be interesting?"

      No... just to be on topic.

      But especially this model attempts to reconstruct European formation, so it's just normal that there are some oddities at the edges, i.e. in the non-European zone. For example it seems apparent that Denisova has some issues in those graphs, probably because it affected at low levels the Siberian ANE pop. but that's best discussed in another study focused on those populations.

      In fact I think that Skoglund & Mälstrom could have done it without Denisova in the equation. I would have personally also reduced the number of African samples and very probably removed one of the ancient NA samples. On the other hand I might have included some more modern Europeans in order to test the model.

      Particularly the authors claim that modern Swedes are PWC-derived but that is not tested by Tree-Mix. It's quite possible that they end up more or less like the French, deriving in fact from other unrepresented populations, which just happen to be somewhat similar to Götland's PWC population but not quite the same.

    15. Sorry, just to be clear, when I say Avjide and Ma-1 have decent residuals, I don't mean their residuals (or lack thereof) with each other, but rather that they both have residuals with other external groups.

      Yah, I agree that omitting Denisovans would probably have been helpful. I wonder if adding Han would improve the model. And yah, removing Saqqaq would seem sensible.

      I think the differences between the residuals for Ötzi and Gökeim may be useful for answering your comment at Eurogenes re: the levels of ANE in Basque populations though. EEFs may not have been a homogenous group, and some EEF groups may have had some ANE affinities to begin with.

      In terms of why they didn't include more modern European groups, from the TreeMix paper seems to suggest the software has a harder and harder time modeling early admixture events the more admixed populations you add to it. Instead of adding Swedes and other populations all together, they'd have to do each population one at a time. Which would still be very interesting of course.

      Here's the relevant passage on TreeMix's limitations:

      "Currently, if the number of admixed populations is large relative to the number of unadmixed populations, this assumption breaks down. For example, in the human data, note that we see no evidence of the documented gene flow from Neandertals to all non-African populations"

    16. By the way, the more I look at this paper and the others I mentioned above, the more it seems to support your scenario for C-M8 and C-V20, or at the very least it seems to exclude any sort of Siberian connection. Consider me convinced. I do think there's an interesting story to be told when it comes to exactly how and when C1a split up and got to its present locations of course, but that data doesn't seem to be out there yet.

    17. Haplogroup C-M8/M105 has been found throughout Japan (including Okinawa) with low frequency. I have estimated that it should be found in approximately 3.1% of the total male population of Japan according to an average of published frequencies of this haplogroup in Japanese subpopulations, weighted by population size.

      It has been found outside of Japan in one individual from the island of Jeju, which is located south of the Korean Peninsula and west of Kyushu, Tsushima, and the Goto Islands of Japan.

      Tokushima (Hammer et al.)
      7/70 = 0.100 C-M8

      Aomori (Hammer et al.)
      2/26 = 0.077 C-M8

      Kansai total (Nonaka et al.)
      1/19 = 0.053 C-M105

      Tohoku total (Nonaka et al.)
      1/19 = 0.053 C-M105

      Shizuoka (Hammer et al.)
      3/61 = 0.049 C-M8

      Japanese/Honshu (Tajima et al. 2004)
      4/82 = 0.049 C-M8

      Okinawa (Hammer et al.)
      2/45 = 0.044 C-M8

      Chubu total (Nonaka et al.)
      2/46 = 0.043 C-M105

      Japanese/Kyushu (Tajima et al. 2004)
      4/104 = 0.038 C-M8

      Kanto(xTokyo, Chiba) total (Nonaka et al.)
      1/41 = 0.024 C-M105

      Chiba (Nonaka et al.)
      1/45 = 0.022 C-M105

      Jeju (Kim et al. 2011)
      1/87 = 0.011 C-M105

      Shikoku(xTokushima) total (Nonaka et al.)
      0/9 C-M105

      Kyushu total (Nonaka et al.)
      0/13 C-M105

      Chugoku total (Nonaka et al.)
      0/16 C-M105

      Tokyo (Nonaka et al.)
      0/51 C-M105

      Kyushu (Hammer et al.)
      0/53 C-M8

      Korea(xJeju) (Kim et al. 2011)
      0/419 C-M105

  2. There is a BBC program where they reconstruct a Neanderthal from skeletal data and chat with anthropologists about scenarios of late Neanderthal adaptations. The idea presented is that Sapens filled in the valleys, and Neanderthal got stuck in isolated mountain populations that didn't support enough population size to be sustained over many generations.

    Think Kalash or Appalachian "hill billies." Hard to displace by outsiders but they pay the price of a small gene pool. Orzi was up in the mountains. The Tyrolean Alps could be a refuge spot for late Neanderthal in Europe (as it was for the Rhaetians apparently), but idk if that's where predecessors of Oetzi picked up the extra % (if Hawks was right in the preliminary blog study).

    Ust-Ishim might give us more to go in. We might find ourselves talking about "Neandrthal Wave 1 (Africa)", "2 (Western)," and "3 (Eastern)" when more data come out.

    1. While there may be some evidence of "last Neanderthals" to maybe as late as 20 Ka BP (a remote location in Cantabria), there's absolutely nothing afterwards. Extremely isolate and inbred tiny populations such as the one you speculate with wouldn't survive for long: inbreeding kills but most importantly causes inviable fetuses and reduces fertility. Extreme inbreeding is effectively inviable, just ask Charles II.

      I'm quite skeptic to say the least. Even if it'd be confirmed that Ötzi has more Neanderthal ancestry than usual, it should be attributed to other mechanisms, i.e. H. sapiens populations with more Neanderthal admixture than usual.

    2. I'd be skeptical of "pure" Neanderthal stragglers that late as well, but I wouldn't discount the possibility that Neanderthal admixture levels were less homogenous in the past.

    3. Initial admixture must be c. 50% by definition but the rate of decline could vary widely. If Neanderthals were cold adapted that 50% might reduce to 2% quite rapidly in hot regions but much more slowly in colder latitudes or higher altitudes leading to, for example populations that were 20% Neanderthal in the hills but only 2% Neanderthal in the adjacent valleys.

    4. I wouldn't be surprised if population density had a lot to do with it as well, with more heavily admixed populations being overwhelmed and mixed into the less admixed populations.


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).