The shadow of the Emishi

By Razib Khan | March 24, 2011 3:57 pm

Randy McDonald just pointed me to a 2008 paper in AJHG, Japanese Population Structure, Based on SNP Genotypes from 7003 Individuals Compared to Other Ethnic Groups: Effects on Population-Based Association Studies. It speaks to an issue I brought up earlier in my post, Sons of the farmers, the story of Japan, which describes the ethnogenesis of the Japanese modern people from the Yayoi culture. The Yayoi presumably brought rice from the Asian mainland, probably from what is today southern Korea. But the Japanese islands were not uninhabited before this period. Japan was home to the Jomon culture, which has a rather storied history in the annals of archaeology. The Jomon seem to have been a predominantly hunter-gatherer population which was also sedentary, and engaged in the production of objects such as pottery which are normally associated with more advanced farming societies. I have a difficult time crediting the ~13,000 year period of continuous development which is attributed to the Jomon, but, it does seem likely that the period between 2,000 and 2,500 years before the present did mark a sharp cultural discontinuity in the Japanese islands, as Jomon gave way to Yayoi.

A related issue to this indisputable cultural shift is the question of whether it was accompanied by a demographic transition. This particular debate is fraught with politics, but we have enough genetic information that we can hazard a tentative guess. It does look like the Jomon-Yayoi cultural shift was accompanied by a significant demographic transition. In particular, the Ainu of the north and the inhabitants of the Ryukyu islands in the south seem distinctive from the majority of Japanese who inhabit the core islands. The hypothesis that these peoples are more related to the Jomon, or directly descended from them. One must distinguish these two groups though; the Ainu remained culturally distinctive from the Japanese, in lifestyle and language before their de facto absorption into the Japanese of late. In contrast, the people of the Ryukyus today seem to be clearly related to the southern Japanese in both language and lifestyle. If the Ryukyu islanders preserve more of the Jomon ancestral heritage, it may simply be due to the dilution of the signal of the original Yayoi pioneers as they moved south.

But there is another piece of the puzzle which has always been a point of curiosity for me: what happened to the non-Japanese populations of northern Honshu? Termed Emishi, these people retained a distinctive identity in northern Honshu until ~1,000 years ago. Fragmentary references in the historical texts make it clear that these people did not speak Japanese natively, and were physically different in appearance, being a “hairy” and “bearded” people. This is how the Ainu were also described, and because of the Emishi’s geographical proximity to Hokkaido it is presumed there may have been a cultural continuity. It turns out that the 2008 paper hints at the genetic imprint of the Emishi.

First, some preliminaries. The authors drilled down to between 100 and 150 thousand SNPs. While 10,000 random markers is sufficient for inter-continental distinctions, a floor of 100,000 is probably optimal for more fine-grained examinations. They had the HapMap populations, which included Africans, Europeans, Chinese, and, some Japanese. But their big data set were nearly 7,000 Japanese from all over the islands. I assume this is large enough that one can down-weight the probability of problems with representativeness due to small sample sizes. Below is a table of Fst values. This basically measures between population genetic distance. There are two things to focus on. First, the Okinawan sample, from the Ryukyu islands, is clearly more distant from all the main island samples. This is what we’d expect. But second, notice that the highest value of genetic distance is between Tokai-Hokuriku and Tohoku.

Tokai-Hokuriku and Tohoku are both on Honshu, the former in the center of the island, between Tokyo and Osaka, and the latter to the far north. Tohoku then is coterminous with the former Emishi region. Some of the patterns here are made clear by the recent history of migration within the Japanese islands. Tokyo is in Kanto-Koshinstsu, and is naturally a magnet for individuals from all over the country. Hokkaido was settled by Japanese only within the last 200 years, and not through gradual expansion from Tohoku to the south.

Another way to look at the genetic variation is a PCA plot. The y axis represents eigenvector 1, the largest dimensions of variation in the data set, and the x axis eigenvector 2, the second largest dimension. On all the plots you see the HapMap Chinese from Beijing on the top left of the plots. You can see here the difference in sample sizes. The cluster to the bottom and center represents Okinawans. The primary central cluster represents individuals from the main islands of Japan. Each panel highlights a different geographical region. So you see, for example, that individuals from Kanto-Koshinetsu are relatively well distributed across the whole “Japanese” area of the plot. Okinawans are primarily in their own cluster, as expected. The regions closer to mainland Asia are shifted in that direction on eigenvector 2. Finally, notice that the Tohoku population represents a vertex of a triangle where the two other positions are held by the Okinawans and Chinese. The authors here argue that skewed distribution of the Tohoku sample is evidence of disproportionate assimilation of a native pre-Yayoi element. Of course it could just be isolation by distance, as Tohoku is furthest from the Asian mainland aside from Hokkaido.

I don’t think we have much more to go on right now. But as we proceed into the future and data sets become more widely available, and analytic techniques more powerful, I wonder if we can reconstruct the Emishi from the chromosomal segments of the people of Tohoku which seem to “jump out” of the Japanese genetic background. And we can always hold out hope for DNA extraction from Jomon burials!

CATEGORIZED UNDER: Genetics, History

Comments (12)

  1. Diogenes

    Well, isn’t Okinawan distinctiveness more likely due to influence from Taiwanese aboriginal “Neolithic” movements? The “Malayo-Polynesian-Madagascar” wave?

    Also Tohoku was colonized only around 1000AD after a long period of Japanese-Eimishi contact. Rice apparently was less of an advantage in the colder climate. Japan is also likely a very good territory for hunter-gatherers surviving mostly on fishing, and likely allowed for higher densities.
    What really surprises me is the apparently largely insignificant effect of the Eimishi in the current gene pool. These were unusually advanced high density settled hunter-gatherers as you say. Yet if significant admixture occured we would expect a significant shift away from Chinese with every sampled region further out, but Kyushu and Kanto are still basically fully overlapping. Japanese distinctiveness from Chinese in PCA (small but observable) and shift away in more Northern regions are probably largely also due to drift (founding effects) and Japanese being derived from a subset of Northeastern Yellow Riverers likely subject to other subsequent waves inside China and Korea.
    Would be interesting to see where both Aboriginal Taiwanese and “unadmixed” Ainu would fall on that PCA plot.

  2. Diogenes

    Also the population I refered earlier as “Taiwanese aborigines” are presumably also derived from mainland China and ultimately to a large extent derived from the Yellow-River/Yangtze River Neolithic Core Area. They would be a “Southwestern subset” population, also less affected by subsequent “higher drag” waves (at least until recently…). Taiwan’s stark archaeological Paleolithic-Neolithic discontinuity strongly suggests this.

  3. .Well, isn’t Okinawan distinctiveness more likely due to influence from Taiwanese aboriginal “Neolithic” movements? The “Malayo-Polynesian-Madagascar” wave?

    if we had the CHD sample, which is more cantonese, that would help.

  4. “And we can always hold out hope for DNA extraction from Jomon burials!”

    FYI: in the case of mtDNA, we have a sample from the Funadomari Jomon site (ca. 3,800-3,500 BP),. and one from the Nakazuma site in eastern Honshu (2500 BC) The Nakazuma one showed a version of hg B, while the Funadomari one didn’t.

  5. The relatively minor contribution of the Emishi to the Tohoku gene pool isn’t surprising. While the Emishi were incorporated into Heian Japan, not exterminated wholesale, this incorporation did follow centuries of war–massacres of Emishi prisoners have been documented. After Tohoku’s subjugation, agriculturalists with superior technology overwhelmed the subjugated Emishi on terms that probably weren’t too favourable

    Leaping across the Pacific Ocean to the scene of other population replacements of hunter-gatherers by farmers, after a century and a half of American rule less than a sixth of Alaska’s population is of indigenous hunter-gatherer background, and less than 5% of British Columbia’s. And those proportions are high, following relatively painless incorporations of indigenous populations into the Canadian and American states–there were no Indian Wars in northwestern North America. A low proportion in Tohoku sounds quite plausible to me.

  6. Actually now that I think of it, I wonder what the stats would show for traditionally Burakumin villages in the Tohoku region. I think that might be interesting, given the possible origins of Burakumin villages to the south.

  7. i thought the eta emerged only during the tokugawa period?

  8. The legal codifications, family and village registers and cultural mores of it, yes. The roots however go back to at least the Heian period, if you look into accounts of the period. It’s all pretty fascinating and messy, and most Japanese would rather not look into it. My personal theory is that there isn’t a single locus out of which they emerged other than by groups that for whatever reasons were “unclean”. Residual cultural practices could certainly stand as markers for it. Japanese thought Ainu slaughtering and food preparation practices were culturally revolting. Maybe Emishi groups would have evinced the same feelings?

    It’s also interesting to note that even with pretty late contact and pretty heavy discrimination until the 1970s, nearly if not all Ainu alive today have at least some Japanese blood, the majority being 50% or more Japanese. I would imagine the Emishi got swamped out, but if patterns are the same as sorta happened in Hokkaido, some rural out of the way villages probably had/have a higher percentage of ancestry even after the last vestiges of the culture disappeared.

    Granted, this is all arm chair quarterbacking, but the precedent exists elsewhere.

  9. The historical insight on the Emishi is interesting as is the fact that “Hokkaido was settled by Japanese only within the last 200 years, and not through gradual expansion from Tohoku to the south,” neither of which I had known. I wonder if there are any linguistic data on the language spoken in Emishi? This was well within the historic era among a people that kept good histories at that point in time, and presumably, the Emishi language would have lingered on at least a century or two after it fell politically.

    Two other little footnotes from the world of anime.

    1. The Ryukyu accent is distinctive enough that there is actually a standard convention for translating this distinction in dubbed Japanese media. It is generally translated as English in the accent of the American South.

    2. I wonder if some of this history inspires at all the anime feature “The Place Promised in Our Early Days” (2004) which features an alternate reality in which Japan is under two regimes after the country’s loss in World War II, with Hokkaido in Japanese hands and the remainder in American hands.

    Finally, a question. Does anyone know if there is a lingering Emishi ethnic or national identity that people identify with today? Do the people of North Honshu have the kind of feelings towards their pre-Japanese days that, for exampe, the people of Scotland have towards the English? Is the region politically distinctive?

  10. Hokkaido was settled only in the last quarter of the 19th century, when the Meiji state wanted to solidify its northern frontier versus Russia and noted that Hokkaido was a frontier home to a bit more than a hundred thousand people. The settlement of Hokkaido was systematic, planned, and quite successful–Hokkaido’s identification as one of the Home Islands is proof of that.

    Little is known of the Emishi language, if anything. There’s no particular reason for it to be similar to Ainu, or for there to have been a single one–given the length of Jomon era Japan’s settlement, pre-Yayoi Japan might have been as linguistically diverse as New Guinea!

    Ryukyuan is frequently identified as the other “Japonic” language, and–who knows?–if Ryukyu had become an independent country the sociolinguistics of the islands might be different. As things stand, language shift to standard Japanese is ongoing.

    As for “The Place Promised in Our Early Days”, Stalin was interested in a Soviet zone of occupation in Japan–the northern coast if Japanese forces didn’t resist, all of Hokkaido if it did. That was a non-starter.

    I have no idea about the extent of Tohoku identification with the Emishi past, although I suspect it’s trivial. How many East Germans identify with the Wends? The rupture was near-complete.

  11. @ Randy McDonald

    Thanks for the input.

    East German identity with the Wends might be modest, but the divide between Protestant Reformed German and Catholic German on North-South lines and dialectal lines is quite pronounced. Similarly, the divide between the regions of modern day Spain which can be traced, in part, to which regions did and did not resist assimilation into the Roman Empire, are significant; and Macedonia managed to re-emerge into a political entity after many centuries of being subsumed in other political bodies. Similarly, the E.U.’s boundaries closely match those of the Holy Roman Empire.

    I ask because retention of regional identity that far back in a history minded population seems plausible.


Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!

Gene Expression

This blog is about evolution, genetics, genomics and their interstices. Please beware that comments are aggressively moderated. Uncivil or churlish comments will likely get you banned immediately, so make any contribution count!

About Razib Khan

I have degrees in biology and biochemistry, a passion for genetics, history, and philosophy, and shrimp is my favorite food. In relation to nationality I'm a American Northwesterner, in politics I'm a reactionary, and as for religion I have none (I'm an atheist). If you want to know more, see the links at


See More


RSS Razib’s Pinboard

Edifying books

Collapse bottom bar