PDA

View Full Version : East Asian Ancient PCA and K9 nMonte



K33
06-13-2019, 09:32 PM
I put this together to try and parse out ancient E Asian population history. We have lots of gaps obviously, but, at least there's enough to form some rough outlines. Ryukendo made this post (https://anthrogenica.com/showthread.php?7094-Ancient-DNA-from-Asia&p=155724&viewfull=1#post155724) a while back, before we had all this ancient DNA. I think a lot of these predictions are starting to come to fruition, with maybe some slight modifications.

Below is a PCA I made with some ancient E Asian samples. (VUT2900BP = Vanuatu proto-Polynesians; NPL= Nepal). I used Corded_Ware to sop up West Eurasian ancestry, which I figure typically appeared in the form of steppe-related ancestry (though for some South Asians it could be mostly Iran_N-related). The Czech CWC sample also plots closest to Sintashta on PCA.

https://i.imgur.com/ea6XgMX.jpg

Here is a spreadsheet with how moderns match to these components in nMonte:

http://i.imgur.com/QCBanMa.jpg

K33
06-13-2019, 10:05 PM
My thoughts:

- It is not surprising the Jarawa and Onge overwhelmingly gravitate towards Hoabhinians between all these references. What is surprising is how tight the % fits are. So contrary to some initial speculation (ie, that they have been isolated for tens of thousands of years, they represent an archaic OOA wave, etc) these guys are quite literally the 1:1 descendants of the Hoabhinians, and their isolation only dates to sometime from the late Neolithic to the Iron Age, when farmers finally ousted them from Mainland SE Asia

- The amount of Vanuatu_2900BP picked up by Austro-Asiatic speakers and Chinese, Koreans, and Japanese (as opposed to Jomon or Hoabhinian for these groups) is pretty surprising. You can see that Vanuatu basically looks a Jomon drifted out into "left field"; a group sharing much of this drift must have been common across much of SE Asia, not just on Taiwan. I think the idea has always been of an Austro-Asiatic substratum in China prior to the arrival of Sino-Tibetan speakers. Well, do these results lean towards a genetic relationship between the Austro-Asiatic and Austronesian language families? (ie, the proposed Austric macrofamily (https://en.wikipedia.org/wiki/Austric_languages)?)

- Chokhopani is obviously "too late" to be a proto-Sino-Tibetan speaker, but from a principal components perspective, it looks to be a decent proxy for whoever spread this ancestry thru East/SE Asia

- Some of the worst fits in these runs involve South Asian tribals. There is obviously additional substructure in AASI that cannot be accounted for by Hoabhinian alone (adding HotuCave to sop up West Eurasian ancestry doesn't help)

Kulin
06-13-2019, 11:23 PM
Mind putting me on it if possible? I'm not exactly east Asian, but I have around 12% East Asian-like admixture.

Michalis Moriopoulos
06-14-2019, 06:45 AM
- Some of the worst fits in these runs involve South Asian tribals. There is obviously additional substructure in AASI that cannot be accounted for by Hoabhinian alone (adding HotuCave to sop up West Eurasian ancestry doesn't help)

Interesting. I was hoping an Iran source would take care of that problem. Maybe try Iran Neolithic instead of Hotu?

The fits on the Negritos look pretty good. But I'm surprised the Aeta fit is as good as the Jehai fit. I thought the Aeta's Australasian ancestry was more like Papuans than Onge, whereas Jehai should indeed be more Hoabinhian than Papuan. A Papuan reference might be worth trying out to see how it affects some of the other groups, too.

Tsakhur
06-14-2019, 03:28 PM
I put this together to try and parse out ancient E Asian population history. We have lots of gaps obviously, but, at least there's enough to form some rough outlines. Ryukendo made this post (https://anthrogenica.com/showthread.php?7094-Ancient-DNA-from-Asia&p=155724&viewfull=1#post155724) a while back, before we had all this ancient DNA. I think a lot of these predictions are starting to come to fruition, with maybe some slight modifications.

Below is a PCA I made with some ancient E Asian samples. (VUT2900BP = Vanuatu proto-Polynesians; NPL= Nepal). I used Corded_Ware to sop up West Eurasian ancestry, which I figure typically appeared in the form of steppe-related ancestry (though for some South Asians it could be mostly Iran_N-related). The Czech CWC sample also plots closest to Sintashta on PCA.



You should also add Malaysia_LN, Lao_LN_BA/Lao_BA, Thailand_IA and VNM_BA_all/VNM_BA_DongSonCulture.

okarinaofsteiner
08-02-2019, 11:23 PM
My thoughts:

- It is not surprising the Jarawa and Onge overwhelmingly gravitate towards Hoabhinians between all these references. What is surprising is how tight the % fits are. So contrary to some initial speculation (ie, that they have been isolated for tens of thousands of years, they represent an archaic OOA wave, etc) these guys are quite literally the 1:1 descendants of the Hoabhinians, and their isolation only dates to sometime from the late Neolithic to the Iron Age, when farmers finally ousted them from Mainland SE Asia

- The amount of Vanuatu_2900BP picked up by Austro-Asiatic speakers and Chinese, Koreans, and Japanese (as opposed to Jomon or Hoabhinian for these groups) is pretty surprising. You can see that Vanuatu basically looks a Jomon drifted out into "left field"; a group sharing much of this drift must have been common across much of SE Asia, not just on Taiwan. I think the idea has always been of an Austro-Asiatic substratum in China prior to the arrival of Sino-Tibetan speakers. Well, do these results lean towards a genetic relationship between the Austro-Asiatic and Austronesian language families? (ie, the proposed Austric macrofamily (https://en.wikipedia.org/wiki/Austric_languages)?)

- Chokhopani is obviously "too late" to be a proto-Sino-Tibetan speaker, but from a principal components perspective, it looks to be a decent proxy for whoever spread this ancestry thru East/SE Asia

- Some of the worst fits in these runs involve South Asian tribals. There is obviously additional substructure in AASI that cannot be accounted for by Hoabhinian alone (adding HotuCave to sop up West Eurasian ancestry doesn't help)

I think VUT_2900BP is just a proxy for "South Mongoloid"-like ancestry and/or shared genetic drift.

I made an Excel graph that shows how your modern East Asian populations fall on the PCA- this only includes populations with a relatively good fit (under 10). Based on how I score on various GEDmatch Oracle calculators and old nMonte calculators, I definitely fall somewhere in the triangle between Miao, Korean, and Japanese.

https://i.imgur.com/UyvCpU2.png

Kanenas
08-03-2019, 03:29 AM
What will happen if you remove CW and add WHG and Natufian? Can you do it?