April 10, 2012

Claim that Japanese are 60-72% Neolithic

Jomon clay head
An open access letter claims that modern Japanese are 2/3 of Neolithic ancestry (except Ryukyuans, who'd be 2/3 Paleolithic instead).

The explanation is however not really clear for me and, looking at their own data, I can't really accept such conclusions easily:

Only the yellow component (at K=4), almost totally absent in the Ryukyuans and dominant among North Chinese and Koreans, the likely parent populations of the Yayoi farmers, can be considered to inform the input of such immigrants to Japan. The exact apportions are not detailed anywhere in the letter but it seems to be c. 35% among Koreans (KR-KR) and Shanghai Chinese (CN-SH) and slightly above 20% among North Han (NHan, CHB). 

By comparison mainland Japanese (Japanese, JPT, JP-ML) show c. 10% in most cases. IF the parent proto-Yayoi population would be Koreans, then Japanese would have less than 1/3 Yayoi blood, while if the proto-Yayoi is equated to Northern Han instead, then the result would be at most 50%. 

In the case of Ryukyans, the Yayoi input would be negligible, almost zero. They'd be almost 100% Jomon, assuming this concept applies to the Ryukyu islands at all.

In truth I do not know what to think of this article other than it seems confusing and inconsistent with its own data.


  1. It is quite possible that the proto-Yayoi were admixed themselves. One can perhaps equate the roughly similar amounts of red and yellow components in N Chinese or Koreans to ancient people related to today's Sinitic and Altaic speakers, respectively. The S Chinese have much less of the yellow component as expected, while the Ryukyuans have none. The latter might have acquired the red component alone through gene flow from Taiwan or adjacent southern China.

  2. Components are affinity indexes, nothing else. In some clear-cut cases they do represent ancestral populations but not always.

    In this case I find the orange ("red") component (in K=4) to be a generic East Asian affinity component, unless you can reasonably argue that there was more migration from South China (where the component is strongest) to Japan than from North China/Korea.

    Of course, East Asian Neolithic does seem to have originated in South China c. 7000 BCE but by the time farming arrived to Japan c. 300 BCE that was not anymore relevant: all China (but the West and parts of Manchuria maybe) was Neolithic by c. 6000 BCE and Korea at least since c. 3500 BCE.

    It's also likely that the analysis might benefit from greater depth in terms of K values. It is indeed rare that such a shallow level is best, normally the best levels are in the "teens" K values for what I have seen.

  3. "normally the best levels are in the "teens" K values for what I have seen."

    It depends on the nature of the dataset, the relative homogeneity/heterogeneity of the dataset, the number of samples, geographical spread and so forth, ADMIXTURE for example gives you a Cross Validation error computing method to come up with an appropriate K value, which I discussed here .

    STRUCTURE, also has applicable methods for determining an optimized K value which Tishkoff (2009) discusses in the supplemental material:

    “The maximum K value was determined on the basis of: (1) the K value at which the likelihood distribution reached a maximum and began to plateau or decrease; (2) high stability of clustering patterns between runs (the primary mode was observed in at least 60% of the 25 runs) and; (3) from the Kmax value at which Kmax + 1 no longer refines the clusters (i.e. Kmax + 1 no longer splits the cluster distinguished at Kmax).”

    1. Indeed (and thanks for the clarification) but I have no means to check these for the present case. I'm talking based on other cases where I have seen explicit cross-validation data (typically 13 and the like).

      Here the dataset is quasi-continental in character (it includes much of East Asia) so I would really expect it to still split meaningfully at higher K values.

      We can actually test it with the 1000 genomes dataset and ADMIXTURE, I'm just not feeling like working these days.

  4. I wrote a bit about Ainu, Basque, Albanian related to "Week as social construct" at my blog: http://the-arc-ddeden.blogspot.com/2012/04/week.html


    1. A bunch of amateurish pseudoscientific junk, sorry. The first thing you could do is to check etymologies or in the case of astelehena, realize that it's a composite word from 'aste' (week) and 'lehen(a)' ((the) first): the first of the week, ironically the last of the week (asteazkena) is wednesday and nobody knows why (I speculate that the weekend was four-days long but who knows).

    2. "A bunch of amateurish pseudoscientific junk, sorry."

      First, online I'm always rushed, (like right now!)so its a mess, with info from the book and also from my own research mixed up. In that sense, yes, amateurish junk. Certainly not intended as psuedoscientific. And possibly correct & revealing.

      "The first thing you could do is to check etymologies"

      Etymologies are often subjective and incomplete.

      " or in the case of astelehena, realize that it's a composite word from 'aste' (week)"

      aste is astral/planet related, planets "wander", English-Dutch "week" ~ wander/vika/weave back & forth/via/vector/vaya.

      "and 'lehen(a)' ((the) first): the first of the week,"

      likely cognate to isnin/ithnen/eka (first)

      "ironically the last of the week (asteazkena) is wednesday and nobody knows why (I speculate that the weekend was four-days long but who knows)."

      nekaz = azkena?

    3. Do you speak Basque even a little bit? Obviously not. Hence shut up.

      "Aste" is week. Astelehena is literally week-first, asteartea (tuesday) means "week-midle" and asteazkena means "week-last". Then come osteguna (the day of Ost, the Sky, later Jupiter and the Judeo-Christian Yaveh-God), etc. all with special meanings.

      Astral comes from aster (star, see the common etymology?). Star in Basque is izar, which may be related to the IE word but does not seem related with aste at all (not in any obvious way certainly). Whatever connection of these words is in any case not with Ainu but with a geographically closer language family: Indoerupean (may be wanderworts, coincidences of sound or even remotely shared etymologies).

      [lehen] "likely cognate to isnin/ithnen/eka (first)"

      After such blatant idiocy, I don't want to see your shadow again in this blog. It'd be a brutal waste of time.

      I know the saying "ignorance is arrogant" but never thought I'd be faced with it such an insulting way.

      "nekaz = azkena?"

      FYI no. Nekaz or rather nekez is with effort. Form the verbal root neka(-tu): to get tired.

      You have not the slightest idea but you dare to make such arrogant claims... get lost!

  5. planets "wander", English-Dutch "week" ~ wander/vika/weave back & forth/via/vector/vaya.

    "week" has nothing to do with "wander" - instead, it is related to the Germanic word for "change" - still surviving in the German word "Wechsel" with the same meaning. It refers to the periodic changeover of planet-gods that watched over specific days of the week. See also the Latin word "vices" with the same meaning.

    1. Yes it's all crazy! I'm not deleting those comments because he's a one-time commenter but anyone posting that kind of junk repeatedly would for sure get me very annoyed.

  6. I can only say that I did not expect emotional reaction, just objective discussion. I'm not claiming Basque and Ainu are closely related, nor am I claiming the opposite.

    Nekaz = (low) furrow (of week); Igandea = high (peak of week)

    montaine Basque wednesday midweek bottom (sunday high)
    lowland Dutch & English wednesday midweek "hump"

    English "last" means both la(te)st and track/furrow (probably related to the last of a shoe in some way).

    Long before there were names for planet gods, there was weaving (all great apes weave their bowl nests, probably for 10+ million years), very likely earliest Hs were weavers with associated words.

    1. In order to deserve objective discussion your proposals would need to have some merit. You're just like that imaginary illiterate character who imagined a "tree" in the letter T and "forest" where there were Ts, etc.: no absolute connection with reality.

      All you say is mere rambling.

      I don't want you wasting my time or that of my readers anymore so it's your last warning: any future posts will be systematically deleted.

    2. Again, I can only say that I did not expect emotional reaction, just objective discussion.

  7. Where do you get this stuff from? Wednesday is derived from Wodan's day (the planet-god), and that by itself is most likely a translation from Latin (dies Mercurii = Mercury's day). By the way, German, Polish and Russian don't share this, but instead use literally "middle of the week."

    This is how the planet-gods were thought to permutate watch over the hours of each day during the week (each was responsible for the first hour, the others for the subsequent hours, in order; the 25th hour is then the first hour of the next day associated with that respective planet-god; the line segments below are simply shortcuts so one does not have to count to 25, and form a heptagram):



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 (sorry, too many trolls).