black man wrote:Btw, as far as I can judge after a short glimpse, the bars (fig. 2c) are notoriously misleading as for the recent paper. (E.g., the Japanese bar is very much orange as if the Japanese had very significant level of "Austronesian" admixture. Also, if I'm not mistaken, there is a lack of Han regional samples.) AFAIK, the bars in the paper by the pan-Asian consortium are more in accordance with linguistic and non-autosomal genetic data than this one.
I think that they are no more misleading than the bar graphs produced in any other populationwise ADMIXTURE analysis. I have seen this same pattern in many other studies in which a similar group of samples has been analyzed; if Oroqen had been included, they should have appeared with an even higher yellow ("Northeast Asian") / orange ("Austronesian") ratio than the Mongol, Daur, and Hezhen samples. Japanese regularly appear as approximately 75% "Northeast Asian" and approximately 25% "Southeast Asian" (in this case, "Austronesian" rather than "Austroasiatic") in ADMIXTURE analyses in which East/Southeast Asians are limited to those within East and Southeast Asia proper (i.e.
excluding North Asians). When North Asians are included, Japanese usually appear as something closer to 25% "North Asian" and 75% "East/Southeast Asian." The results do seem to vary significantly depending on what number of which samples of which ethnic groups are included in the analysis.
They have included a Minnan sample in addition to a "Han" sample, by the way. The Minnan sample appears to be roughly intermediate between their "Han" sample and their "Hmong" and "Hmong_Miao" samples, with less "(North)east Asian" (and, surprisingly, about the same or slightly less "Austroasiatic") than the "Han" sample, with the difference being taken up by "Austronesian," whereas the Hmong/Miao samples have somewhat more "Austroasiatic" than the Han or the Minnan but approximately the same amount of "Austronesian" as the Minnan.
black man wrote:However, the interpretation of fig. 2b seems to be easy: if the turquoise triangles represent Kusunda samples, Kusundas might simply be dropouts from three (or more) different source populations who chose to speak a language adapted to premodern life in a forest. (One of the videos on Kusunda actually refers to their language as something like that.) In that case, there would have been a "Kusunda lifestyle" but no "Kusunda ethnic group" (when "ethnic group" means something primordial based on ancestry first of all).
Interestingly, some of the "Kusunda" samples are close to the "Indian TB" cluster (which probably consists of Naga, Garo and Nyishi according to supporting table 11), another "Kusunda" sample even being inside of the latter and very close to several "SEA-AA" samples. So I'd say that there are possibly significant traces of AA ancestry in the genomes of certain Nepalese people.
As I mentioned in my original post in this thread, the Kusunda individuals in Figure 2b are somewhat spread out between East Asians and South Asians; the qualifier "somewhat" indicating that the Kusunda are not so spread out as the Spiti Tibetans, who range from fully "Chinese" to fully "Pakistani," but the Kusunda are more spread out than the Uttarakhand Tharu, who appear to be quite homogeneous/thoroughly homogenized (except one outlier who deviates strongly toward mainstream Indians). Only one of the Kusunda individuals appears to have a strong probability of "southern" (Indian AA?) admixture; the others all fall into a narrow band between mainstream North Indians ("Indian IE") and Chinese ("Altaic", "EA_TB," and "Hmong-Mien"), though one other Kusunda individual does deviate a bit "northward" from the majority of the Kusunda group, and falls squarely among the Spiti Tibetans as a result, and there is that one individual you have pointed out who deviates slightly toward SEA-AA and Tai-Kadai and falls among the Indian_TB.
black man wrote:Concerning Kusunda, one could suppose that the language was originally spoken by some meanwhile no longer identifiable people, people whose phenotypes were very different from present-day Nepalese phenotypes. Then the ancestors of people related to the Nyishi etc and others might have been pushed into the forests by Tibetans. (Historically, Tibetans expanded from a landscape of sparse vegetation and notoriously looked down on people who lived in forest regions.) The first group of these people possibly bullied by Tibetans might have been small in number and could have adopted proto-Kusunda language in order to learn from natives how to survive in the forests. But later on, multiple groups of Tibeto-Burman migrants into the forests might have gradually outnumbered those proto-Kusunda-speakers. So, in short, the original speakers of proto-Kusunda might have died out some time ago although a derivative of their language was still in use as a lingua franca until recently. (Strangely, those scattered triangles do seem to point to such a lingua franca in their case. It would be different if the triangles were close to each other.)
However, the Chepang, who are precisely one of those non-Tibetan Tibeto-Burman peoples who may have fled into the forest to avoid Tibetans or the like, supposedly dreaded the Kusunda as extremely primitive savages (like the Formosan headhunters were despised and dreaded by Han settlers, etc.), so it seems very unlikely that they would have adopted the Kusunda language for any reason.
I see a paradox here. The sampled Kusunda are clearly recently admixed because they have a fairly broad spread (mostly) along a vector between North Indians on one side and various populations of China on the other side despite the very small census population size and small geographic spread of the Kusunda ethnic group (i.e.
it is implausible that they could have sustained such great genetic diversity over time within their small population). However, the Kusunda appear to have been viewed as extremely primitive and noxious folk by their closest neighbors in recent times. Why would anyone admix into the Kusunda ethnic group and bother to learn their language? I suppose what we might be seeing in the PCA plot and the ADMIXTURE analysis is very recent admixture that has affected the genetic composition of the Kusunda people subsequent to or concomitant with the decline of the Kusunda language into its current moribund state. In other words, the oddly low homogeneity of the Kusunda sample is probably directly related to the decline of the Kusunda as a distinct ethnolinguistic group, so the genetic spread probably has been caused by admixture from whichever population(s) has/have been assimilating the Kusunda people. According to Wikipedia, "Nepali is now their language of everyday communication," and they seem to have been in contact mostly with Chepang and mainstream Nepali people, so the most likely sources of recent admixture in the Kusundas should be Nepali Aryans (probably similar to North Indians but with some Sino-Tibetan admixture) and early (sub-)Himalayan Tibeto-Burman peoples (Chepang, Magar, Limbu, etc.). Unfortunately, the position that should have been occupied on that PCA plot by the "original" Kusundas cannot be deduced without quantifying the amount of admixture from every population that has influenced the Kusundas recently (probably lower class members of the mainstream Nepali population, and maybe also some nearby tribal peoples, such as Chepangs, though it may be difficult to explain why descendants of Kusunda-Chepang intermarriage would choose to identify as Kusunda barring some sort of social rule regarding such cases, i.e.
a requirement that one identify with one's mother's ethnic group or a requirement that one identify with one's father's ethnic group).