Tuesday, March 9, 2021

23 and Me Korea-Japan Glitch Fixed

When the home genetic testing firm 23andMe first provided reports for my wife, her parents, her sister, my children, and my nice and nephew who are children of her sister, the reports showed a significant share of Japanese ancestry. 

This would be a scandalous fact, if true, as it would indicate, in all likelihood, a child born from a Japanese father during Japanese occupation (from 1910 to 1945), or from a Korean man collaborating with the Japanese in that era who married a Japanese woman in connection with that event, something that came be to seen as dishonorable. 

The bite was particularly cutting in this case because my wife's mother's family, in particular, was part of the big business economic elite of the pre-Korean War Korean nation, and would have had lots of economic and social ties to members of the Japanese occupation elites as a result.

But while it wasn't absolute impossible that there was cryptic Japanese ancestry in the family, her parents family ancestry was know to them to be exclusively Korean in their parents and grandparents generations, at a minimum, so the percentages shown were impossible. Also, pretty much all other Koreans we knew who had done consumer genetic testing, were seeing the same thing.

The reality was a methodological artifact rooted in a historical reality that blind cluster analysis performed by computers couldn't grok. 

The modern Japanese people are the admixed product of the indigenous Japanese Jomon people (mostly with a sedentary fishing village based economic foundation) whose ancestors arrived on these islands on the order of 8,000 to 16,000 years ago, and the Yayoi soldier-rice farmers who arrived around 2,300 years ago by sea from Korea and conquered the Jomon and their land.

Modern Japanese people are 9-13% Jomon in origin (measured by autosomal genetics), although modern Japanese men are about 1/3rd Jomon in Y-DNA and a significantly smaller share of modern Japanese people are Jomon in mtDNA, which is inherited from mother to child. 

This mix of uniparental markers is quite surprising. Usually more maternal than paternal uniparental DNA survive by introgression into a conquering people from the people who conquerer them, as male soldiers or pioneers or explorers take local wives. The fact that there is virtually no Jomon linguistic substrate in the Japanese language also limits to the scope of the possible narratives that could cause this unual uniparental DNA marker mix to arise. The cultural process by which this happened in Japan starting around 2,300 years ago is still not well understood. 

A best guess would be that Yayoi soldiers recruited and then thoroughly assimilated local Jomon boys into their army who would leave few linguistic substrates due to their superior youthful language learning abilities. Then, the Yayoi brought wives for all of its soldiers and veterans, including Jomon boy recruits, from Korea, which was not far away, rather than taking local wives. Given the overall percentage of Jomon automsomal ancestry in the Japanese, most Japanese Jomon autosomal DNA probably derives from Jomon origin fathers rather than Jomon origin mothers. But, this is little more than an educated guess. 

Anyway, the bottom line is that modern Japanese people are predominantly Korean in remote genetic origin. Admittedly, the genetic makeup of Korea 2300 years ago was somewhat different than it is today. Also, there was significant later Chinese genetic introgression into Japan via wives imported from China to Japan for elite Japanese men for many centuries after the initial ethnogenesis of the Japanese people. This happened because Chinese wives were considered prestigious because China had the most sophisticated civilization in East Asia at the time. Strong ties to China also did have significant linguistic impact on the Japanese language.

But because the 23andMe database has more Japanese samples than Korean samples, and because there is heavy overlap between the two populations, a computerized blind cluster analysis by 23andMe assumed that many Koreans had Japanese admixture. 

In reality, however, unless someone has a significant minority of Jomon origin genes in their genome as a whole, someone with genes found in both Japanese people and Korean people almost certainly has no Japanese admixture, even if they have many genes that are common in Japanese people.

Eventually 23andMe figured this out, and in a December 2020 update, it now reports that my wife's parents and their descendants all have 100% Korean ancestry (even though my wife's father's Y-DNA haplogroup, O-CTS723 which is a subclade of O-M268, is one that was common among the Yayoi).

Now, the handling of their Korean origins still isn't perfect. 23andMe says that everyone in the family has localized Korean origins that are most likely in Seoul, South Korea. And, this is true as of the post-Korean War era in the 1950s. My wife's parents emigrated there during the Korean War (i.e. June 25, 1950-July 27, 1953) in their tweens or teens about fifteen years before moving to the United States. But, her parents lived the first half of their Korean lives in North Korea, where they were born, as did at least a couple of generations of their ancestors. Yet, 23andMe states that they have 0% North Korean origins. 

Still, two cheers for progress.


Tom Bridgeland said...

Nice to clear that up. Anything that indicates close relations between Koreans and Japanese is going to be controversial on both sides, and no need to roil the family. I know my Japanese wife would be extremely unhappy to discover Korean (or Chinese) ancestry in her family, for all that she married a gaijin.

Guy said...


Hi Andrew, that higher Y-DNA than mtDMA really stands out as an anomaly to be explained. Almost like the Jomon overcame the Yayoi, at least the 1st wave. Compare to the Basque?


Ryan said...

The Y-DNA/mtDNA mismatch is similar to the situation with hunter-gatherers in Europe in the Middle Neolithic though.

andrew said...

@Ryan @Guy Good ideas.