observer_two theory on the selection and spread of EA alcohol flush

[okarinaofsteiner]

observer_two from Anthrogenica did a great deal of research in their free time on research papers on the East Asian alcohol flush to better understand how the mutation spread historically and why it has the distribution it does today. They wrote a cumulative summary on their findings and kindly shared it with me: Alcohol-Processing Genes

in general, ALDH2*2 and part of 47his look like they are selected for in specific conditions relating to high population density life in agricultural EAs, while another part of 47his looks like its correlated to populations living on the ice age far coastal seaboard.
its not "east hunter-gatherers" though— it actually seems to be something more "austronesian"

Overall:
1) Both ADH1B 48His (H7) and ALDH2*2 seem to be selected for, but not associated with the start of farming- instead, they seem to be selected for in relation to more recent, possibly large-scale/mass farming, or aspects of "civilized" life altogether in East Asia. It is difficult to comment on why the Middle East specific ADH1B 48his haplotype would have been selected for owing to not enough data.
2) ADH1B 48His (H6) has an older distribution that seems to confirm the ice age era East Asian shoreline experienced shared demographic processes, without being impacted by recent selection.
3) ADH1B 48His (H7) is more consistently selected for, with similarly high frequencies in most "core" East Asians. On the other hand, while ALDH2*2 only shows selection in places that show selection for H7, it has a narrower distribution and generally lower frequency. For example, Hmongs and non-southern Chinese show strong selection for H7, but only a portion of southern Chinese show strong selection for ALDH2*2. Okinawans, Koreans, and northern Japanese have similar H7 frequencies as central Japanese, but central Japanese have more ALDH2*2. As such, it seems that conditions leading to H7 selection are a prerequisite for ALDH2*2 selection, but not the other way around.
4) ALDH2*2 is mostly strongly tied to specific rice farming communities, but its selection effect seems to vary across different rice farming communities, and it also seems to have impacted certain less rice farming communities as well. As such, ALDH2*2/alcohol flush is correlated with rice farming, but not fully- its selection is likely something far more specific that is also only correlated with rice farming communities.

I will post more snippets of observer_two's deep dives on various research papers in this thread. For now, here is a list of some of the research papers they cited as references for the cumulative summary:

https://pmc.ncbi.nlm.nih.gov/articles/PMC1821113/ [Han et al 2007]
Evidence of Positive Selection on a Class I ADH Locus
The alcohol dehydrogenase (ADH) family of enzymes catalyzes the reversible oxidation of alcohol to acetaldehyde. Seven ADH genes exist in a segment of ∼370 kb on 4q21. Products of the three class I ADH genes that share 95% sequence identity are …

https://pmc.ncbi.nlm.nih.gov/articles/PMC2268739/ [Li et al 2008]
Ethnic Related Selection for an ADH Class I Variant within East Asia
The alcohol dehydrogenases (ADH) are widely studied enzymes and the evolution of the mammalian gene cluster encoding these enzymes is also well studied. Previous studies have shown that the ADH1B*47His allele at one of the seven genes in humans is …

https://pmc.ncbi.nlm.nih.gov/articles/PMC2823730/
The ADH1B Arg47His polymorphism in East Asian populations and expansion of rice domestication in history [Peng et al 2010]
The emergence of agriculture about 10,000 years ago marks a dramatic change in human evolutionary history. The diet shift in agriculture societies might have a great impact on the genetic makeup of Neolithic human populations. The regionally …

https://www.sciencedirect.com/science/ar...9707630614
Geographically Separate Increases in the Frequency of the Derived ADH1B*47His Allele in Eastern and Western Asia [Li et al 2007]
MENA vs East Asia alcohol flush reaction

https://www.nature.com/articles/s41467-018-03274-0
Deep whole-genome sequencing reveals recent selection signatures linked to evolution and disease risk of Japanese [Okada et al 2018]
Nature Communications - Recent natural selection left signals in human genomes. Here, Okada et al. generate high-depth whole-genome sequence (WGS) data (25.9×) from 2,234 Japanese people of…

Show Content

Spoiler

https://www.mdpi.com/2073-4425/9/9/452
Recent Selection on a Class I ADH Locus Distinguishes Southwest Asian Populations Including Ashkenazi Jews [Gu et al 2018]
The derived human alcohol dehydrogenase (ADH)1B*48His allele of the ADH1B Arg48His polymorphism (rs1229984) has been identified as one component of an East Asian specific core haplotype that underwent recent positive selection. Our study has been extended to Southwest Asia and additional markers in East Asia. Fst values (Sewall Wright’s fixation…

[okarinaofsteiner]

going back to this article-
https://pmc.ncbi.nlm.nih.gov/articles/PMC2846302/
Refined Geographic Distribution of the Oriental ALDH2*504Lys (nee 487Lys) Variant

1) I find it interesting because of its rather comprehensive review of ALDH2_2 frequency literature, including many underrepresented groups. A problem though is that the selection process used for which frequencies made into their map is a bit unclear imo (eg. it would be helpful if they noted on the table why a number was included or not)
2) the number included in Table S1 and Figure 1 is the allele frequency of ALDH2_2, and they calculated the total homozygote + heterozygote proportion used for Figure 2 by doing q = allele freq, p = 1-q (eg. 2pq + q^2 = 2(1-0.4)(0.4) + (0.4)^2) = ~0.64; see scale where Figure 1 max is 0.4, Figure 2 max is 0.65). To compare to the commonly posted map, we have to use:
(heterozygote) 2pq = light green
(homozygote) q^2 = dark green
(heterozygote + homozygote) 2pq + q^2 = total non-yellow

I will present all the listed frequencies in Table S1 (rather than just the ones selected by the authors) with comparison calculations https://upload.wikimedia.org/wikipedia/c...quency.png

---

Han Chinese subgroups + Taiwanese aborigines + Manchuria ethnic minorities

Taiwanese Aborigines:


Manchurian Ethnic Minorities:

Wu 2007 is apparently a poster presentation at an American Society for Human Genetics annual conference, they give really janky results for these tungusics...
I can believe Daur having more than average and obvs Manchu also, but Wu 2007 is pretty contradictory vs others
the interesting thing is that Manchu have so much more than Xibe
...Wu 2007 gives pretty reasonable results for the other Chinese groups...
so these tungusic results might potentially be accurate...? they also have the highest sample size for tungusics (ALFRED database has ~5-10 for many)
there might be somehow different scoring tungusic populations in china

the authors of this article decided to include Yan 2003 (the 0.4 allele freq shandong) into their map, which is why there is a peak in Shandong
Northern Chinese (sans "Beijing):


Southwest and Inland Central Chinese:
(note Hunan...)

and another one (Yanting Hakkas)


Jiangzhe/Eastern Chinese:


Fujian/Guangdong:

something that stands out to me is that Hakka might score more than average...?
sample might be rather small though

[okarinaofsteiner]

going back to this article-
https://pmc.ncbi.nlm.nih.gov/articles/PMC2846302/
Refined Geographic Distribution of the Oriental ALDH2*504Lys (nee 487Lys) Variant

1) I find it interesting because of its rather comprehensive review of ALDH2_2 frequency literature, including many underrepresented groups. A problem though is that the selection process used for which frequencies made into their map is a bit unclear imo (eg. it would be helpful if they noted on the table why a number was included or not)
2) the number included in Table S1 and Figure 1 is the allele frequency of ALDH2_2, and they calculated the total homozygote + heterozygote proportion used for Figure 2 by doing q = allele freq, p = 1-q (eg. 2pq + q^2 = 2(1-0.4)(0.4) + (0.4)^2) = ~0.64; see scale where Figure 1 max is 0.4, Figure 2 max is 0.65). To compare to the commonly posted map, we have to use:
(heterozygote) 2pq = light green
(homozygote) q^2 = dark green
(heterozygote + homozygote) 2pq + q^2 = total non-yellow

I will present all the listed frequencies in Table S1 (rather than just the ones selected by the authors) with comparison calculations
https://upload.wikimedia.org/wikipedia/c...quency.png

Japanese

I will comment on others later, but for now Japan:
(from Figure 2)

• there is lowered alcohol flush in Tohoku/Hokkaido, which is about similar to Korea in its range
  
• Kanto, Chubu, and Kansai have relatively high alcohol flush (around Guangdong levels except for a couple low points).
  
• Nagano/Japanese alps has a relatively lower alcohol flush, but there's only one report for it. It could possibly be lower?
  
• Shikoku (with only one datapoint though) and western Kyushu are similar to Kanto, Chubu, Kansai, but Chugoku and eastrn Kyushu (Miyazaki, Kumamoto, Kagoshima) have possibly slightly lower scores (more at Jiangzhe level than Guangdong)
  
• Ryukyuans only have one datapoint but it scored rather low, lower than all the northern Japanese numbers
  

[okarinaofsteiner]

Han Chinese subgroups + Taiwanese aborigines + Manchuria ethnic minorities

-snip-

(https://i.imgur.com/SC3kPAp.png)
added yunnan and jilin in


(https://i.imgur.com/3fshTwI.png)
has elevation and allele freq; too lazy to add in the separate shapefile for taiwan (the elevation data for china comes with taiwan)
not sure how useful this map is though

maybe the communities that are contiguous with the north china plain but not in it, eg. hubei/wuhan and to a smaller extent nanjing, also have relatively lower ALDH2*2? and then the highest values belonging to the mountainous areas in fujian/GD (they may have slightly higher values than surroundings/coastal areas)?
but on the other hand hmongs have extremely low ALDH2*2 (I'll post hmongs and TKs later but hmongs are around TW Abo/north Tungusic level)

(https://i.imgur.com/4eDlHBt.png)
as the paper's map suggests the local maxima of ALDH2*2 should be about here
all that's significant about this place is that its mountainous and has hakkas?

the "true native" people of inland southern china, prob hmongs and TKs, don't have particularly high ALDH2*2 values...

so either
1) the specific "yue" groups living in this particular area must have had exorbitantly high ALDH2*2 initially, or
2) like the paper argues, the allele's original peak may have been somewhere in central china but ended up proliferating more south than its place of origin
then, current central chinese would have lower ALDH2*2 than they did before, due to mixture from further north
but hubei people for example, have basically similar ALDH2*2 to north china plains

so it probably occurred further south than hubei- hunan and jiangxi still seem to have current "highs" not that much lower than the inland fujian/north GD high
much of the settler populations of Guangdong, Fujian may have come from hunan and jiangxi originally; I believe that southern settlement as it occurred on land was stepwise with people only really migrating from close places to the next (rather than eg. one stroke from henan to guangdong)
https://upload.wikimedia.org/wikipedia/c...ansion.png

Hakka people are in general rather closely related to Jiangxi (linguistically as well)

so I postulate that prior to the formation of "China" and associated large-scale migrations, the peak ALDH2*2/alcohol flush carriers may have been a local community in Hunan/Jiangxi, with areas further south of it then lower than nowadays. This was likely separate from the ancestors of modern Hmongs and local TKs, who have lower alcohol flush.
when "China" happened, people from probably Hubei migrated to now Hunan and Jiangxi, while people from Hunan and Jiangxi probably migrated to now Fujian and Guangdong
it is unlikely that the allele got selected for after "China" happened, as if the allele was selected for, the estimated time of selection was some at least 7k+ years ago
not a single TK or hmong-mien group scores high in ALDH2*2, which for some may be due to austroasiatic mixture, but others still living in the vicinity don't have much of it either
there is then reason to believe that going westwards in hunan, the historical dropoff for ALDH2*2 may have been fairly high
going southwards, some dropoff, eastwards, some dropoff, and northwards, a lot of dropoff



(https://i.imgur.com/vWsnRMF.png)
jiangnan and the coast of fujian/GD probably had local alcohol flush themselves, jiangnan being a bit higher than now, fujian/GD being a bit lower than now (if my thory about their settlement and alc flush increase by hunanese/jiangxi people is correct)
therefore, the range of increased alcohol flush (ie. above the usual ~15-20% found in most east asians) in pre-recorded history/"China" creation southern china, probably was around the purple circle I drew here, but with more concentration going to the end of the vector
but why it peaks specifically in that place I am not sure

(https://i.imgur.com/BfF12am.png)
on the korean peninsula, I drew where I suspect the local peak there historically may have been; its hard to discern from korean stats sadly due to homogeneity but Japanese phenotypes of central areas with higher ALDH2*2 look more like koreans from there
the interesting thing though, is that the purported vector of increased historical ALDH2*2, is literally facing in the opposite direction from the Korean peninsula in basically a straight line, and for some weird reason, the flush maximizing at the ends of this line on each side, rather than in the middle
a curious parallel may be drawn to the range of the O1b2 yhg which has its sole uniquely non-KRJP clade in Hunan of all places, again forming a dumbell-like structure across the yellow sea and east china sea

Me: this all confirms my own observations and hypotheses- the selection for flush reaction is related to an ancestry component (or selection process) that is specific to Southern Han and not found among modern-day Tai-Kradai, Hmong-Mien, Austroasiatic, or Austronesian speaking groups

if it was primarily a selection process, the papers suggesting that put its time of selection at 7-10kya, so long before southern han were "han",
so either way this shows that there were clear barriers between various populations in now southern china, despite being all lumped as "yue" they were probably discrete from each other
assuming that the historical alcohol flush for eastern Hunanese was higher than now, and if proper Hmong-Mien and TKs populated western hunan, the ALDH2*2 discrepancy must have been insane

[okarinaofsteiner]

going back to this article-
https://pmc.ncbi.nlm.nih.gov/articles/PMC2846302/
Refined Geographic Distribution of the Oriental ALDH2*504Lys (nee 487Lys) Variant

1) I find it interesting because of its rather comprehensive review of ALDH2_2 frequency literature, including many underrepresented groups. A problem though is that the selection process used for which frequencies made into their map is a bit unclear imo (eg. it would be helpful if they noted on the table why a number was included or not)
2) the number included in Table S1 and Figure 1 is the allele frequency of ALDH2_2, and they calculated the total homozygote + heterozygote proportion used for Figure 2 by doing q = allele freq, p = 1-q (eg. 2pq + q^2 = 2(1-0.4)(0.4) + (0.4)^2) = ~0.64; see scale where Figure 1 max is 0.4, Figure 2 max is 0.65). To compare to the commonly posted map, we have to use:
(heterozygote) 2pq = light green
(homozygote) q^2 = dark green
(heterozygote + homozygote) 2pq + q^2 = total non-yellow

I will present all the listed frequencies in Table S1 (rather than just the ones selected by the authors) with comparison calculations https://upload.wikimedia.org/wikipedia/c...quency.png

---

(https://i.imgur.com/uDBMWkn.png)

Austroasiatics
the only group with substantial flush are Viets who are arguably Tai-Kradai, Chinese Kinh reach very high levels of flush but the number decreases going southwards in Vietnam
Cambodians have slightly higher flush than Laotian Austroasiatics
in fact about the same as central-southern Viets

correction to the "China Kinh/Viet"- these were Viets from Seattle
(https://i.imgur.com/Hmn05xW.png)

not all Viets in Seattle in the 90s should be Hoa though... so this number is perplexing
I don't know what to do with it; it is blatantly inconsistent with all other reports, so I will disregard it

Me: This tracks with my anecdotal observations of flush reaction among VietAms tho- lower than South China Han but much higher than all other SEA groups

that number would be higher than the highest number for Zhuang (also by Chen 1994) and higher than all but one of the Guangdong numbers
I would be fine with something like 0.3 flat but imo that is way too high

---

(https://i.imgur.com/JsY2gvE.png)

Chinese and Laotian Hmong/Miao have practically the same level of flush, both being slightly higher than Austroasiatics/TW Aborigines but lower than mainstream EAs

(https://i.imgur.com/VpWGAsz.png)

Tai-Kradais
there is actually some interesting variance between these
Zhuang have the highest alcohol flush, reaching but slightly lower than Guangdong Han values (numbers taken as a group)
despite this, Hlai have almost 0
Kam, Maonan, Mulam who live in Guizhou area have lower flush than Zhuang

Ethnic Lao have middling-low flush, and then TK ethnic minorities in Laos have rather low flush
northern Thais (black, Khon Kaen) have low flush, but then southern Thai (Songkhla) have slightly increased flush

so the value for southern Thai is interesting; it could be due to Chinese Thais being more likely to be in the south?

in main China TKs, there's in general an increase in flush going towards Zhuang (Kam are the northernmost in SW Hunan, Guizhou, then Maonan across Guizhou, and then Mulam mostly in Guangxi)
Lao and Thai (other than what probably is Chinese admixture) have low flush also
but near Zhuang, ethnic Kinh in China have similar levels of flush to Zhuang, which rapidly decreases going into Vietnam proper
what I'm at a bit of a loss on explaining rn is why Cambodian levels are higher than southern and central Viets, and also most Laotian/Thai groups
Hlai people might originally be from Leizhou in southern Guangdong, which is right next to the Zhuang peak, yet their flush is so low

I'm not going to post values for Austronesians (sans TW Abos), but none of them incl Malays have have flush

[okarinaofsteiner]

Show Content

Spoiler

(https://i.imgur.com/uDBMWkn.png)

Austroasiatics
the only group with substantial flush are Viets who are arguably Tai-Kradai, Chinese Kinh reach very high levels of flush but the number decreases going southwards in Vietnam
Cambodians have slightly higher flush than Laotian Austroasiatics
in fact about the same as central-southern Viets

correction to the "China Kinh/Viet"- these were Viets from Seattle
(https://i.imgur.com/Hmn05xW.png)

not all Viets in Seattle in the 90s should be Hoa though... so this number is perplexing
I don't know what to do with it; it is blatantly inconsistent with all other reports, so I will disregard it

Me: This tracks with my anecdotal observations of flush reaction among VietAms tho- lower than South China Han but much higher than all other SEA groups

that number would be higher than the highest number for Zhuang (also by Chen 1994) and higher than all but one of the Guangdong numbers
I would be fine with something like 0.3 flat but imo that is way too high

(https://i.imgur.com/JsY2gvE.png)

Chinese and Laotian Hmong/Miao have practically the same level of flush, both being slightly higher than Austroasiatics/TW Aborigines but lower than mainstream EAs

(https://i.imgur.com/VpWGAsz.png)

Tai-Kradais
there is actually some interesting variance between these
Zhuang have the highest alcohol flush, reaching but slightly lower than Guangdong Han values (numbers taken as a group)
despite this, Hlai have almost 0
Kam, Maonan, Mulam who live in Guizhou area have lower flush than Zhuang

Ethnic Lao have middling-low flush, and then TK ethnic minorities in Laos have rather low flush
northern Thais (black, Khon Kaen) have low flush, but then southern Thai (Songkhla) have slightly increased flush

so the value for southern Thai is interesting; it could be due to Chinese Thais being more likely to be in the south?

in main China TKs, there's in general an increase in flush going towards Zhuang (Kam are the northernmost in SW Hunan, Guizhou, then Maonan across Guizhou, and then Mulam mostly in Guangxi)
Lao and Thai (other than what probably is Chinese admixture) have low flush also
but near Zhuang, ethnic Kinh in China have similar levels of flush to Zhuang, which rapidly decreases going into Vietnam proper
what I'm at a bit of a loss on explaining rn is why Cambodian levels are higher than southern and central Viets, and also most Laotian/Thai groups
Hlai people might originally be from Leizhou in southern Guangdong, which is right next to the Zhuang peak, yet their flush is so low

I'm not going to post values for Austronesians (sans TW Abos), but none of them incl Malays have have flush

---

btw on Hlai people- https://academic.oup.com/mbe/article/39/...10/6731089

"HNL show closer genetic connections with ancient southern East Asian ancestry and Austronesian-related ancestry, which may also be preserved by the early migration to Hainan Island. Collectively, we propose that the ancient Bai-Yue population lived in mainland South China before ∼4,000 ya, and a part of the ancient Bai-Yue population, that is, the proto-HNL, started migrating from the mainland to Hainan Island and became the main settlers ∼4,000–3,000 ya. The isolated circumstance of Hainan Island well preserved the ancient Bai-Yue ancestry in the HNL and prevented admixture with other populations, thus restricting the increase of Ne growth of the HNL. In turn, the mainland Bai-Yue populations were admixed in various degrees with ancestries from other surrounding groups on the mainland, and this contributed to the increase of Ne since ∼4,000 ya."

Show Content

Spoiler


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

there are two clusters of Hlai people; one looks to be Han-admixed while the other is more "extreme" than the main Chinese TK grouping; it looks closer to Lachi people who live in some mountain area of NW Vietnam

Show Content

Spoiler



https://i.imgur.com/7JsszxY.png

Maonan are more extreme than Zhuang, but Kam/Dong people are less extreme
hence in terms of "Han closeness", Kam/Dong are closer to populations with high alcohol flush cumulatively, but have lower alcohol flush than Zhuang, and Maonan further, but also have lower alcohol flush than Zhuang
Hlai are mildly pulled towards the direction of Atayal

https://upload.wikimedia.org/wikipedia/c...people.png

Show Content

Spoiler

Both Zhuang and (Chinese) Kinh likely have experienced more introgression from Han groups than the Maonan and Hlai have (Chinese Kinh may also have affinity to Zhuang due to living near them?)

I can then make the following comments on SW China, FSC, and continental SEA alcohol flush distributions:

• Going off of my previous inference that the historical peak of alcohol flush was likely somewhere in Hunan and Jiangxi, this patch likely extended further towards the southwest (eg. into western Guangdong, Guangxi), than towards the west of Hunan proper (judging by lower values among Hmongs and northern TKs like Kam/Dong in Hunan/Guizhou).
  
• Hlai people have lower flush than the continental Guangxi TK groups, incl Mulam and Maonan which are not as "Han-pulled" as Zhuang, as well as northern Viets and Lao. Assuming the theory they are from Leizhou is correct, then historically, there might have been a separation of sorts between inland Guangxi TK groups, who may be more directly contiguous with north Viet/Lao, than to the continental ancestors of the Hlai. In other words, it is possible that the actual southern coast (as in people who actually lived on the coast) of historical FSC was originally not part of the increased alcohol flush region. Increased values among Han in this area now, may be due to migration from areas slightly further north, and even today these seem to have slightly lower values than say Hakkas.
  
• Alcohol flush decreases to almost nonexistent among tribal mountain Austroasiatics and non-mainstream continental SEA TKs, but Cambodians and Songkhla Thais have higher flush. This is easy to explain for the latter, as there is probably a non-insignificant SE Chinese component there. But I don't think this is the case for Cambodians as much, who have higher numbers than southern Viets even. This is nowhere as high as in southern China or Japan, but one can surmise that there might be a minor patch of increased flush in Cambodia compared to neighboring areas, that may be local rather than the result of migration from the southern China peak.
  

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

on historical southern Chinese communities:
We can probably infer now that the following likely had boundaries between each other (whether cultural or geographic) prior to known historical migrations based on their differing alcohol flush values:

• eastern Hunan/Jiangxi/possibly inland Guangdong
  
• western Hunan/Guizhou mountain (?)
  
• coastal Guangdong/Hainan
  
• Guangxi/northern Vietnam
  

[okarinaofsteiner]

going back to this article-
https://pmc.ncbi.nlm.nih.gov/articles/PMC2846302/
Refined Geographic Distribution of the Oriental ALDH2*504Lys (nee 487Lys) Variant

1) I find it interesting because of its rather comprehensive review of ALDH2_2 frequency literature, including many underrepresented groups. A problem though is that the selection process used for which frequencies made into their map is a bit unclear imo (eg. it would be helpful if they noted on the table why a number was included or not)
2) the number included in Table S1 and Figure 1 is the allele frequency of ALDH2_2, and they calculated the total homozygote + heterozygote proportion used for Figure 2 by doing q = allele freq, p = 1-q (eg. 2pq + q^2 = 2(1-0.4)(0.4) + (0.4)^2) = ~0.64; see scale where Figure 1 max is 0.4, Figure 2 max is 0.65). To compare to the commonly posted map, we have to use:
(heterozygote) 2pq = light green
(homozygote) q^2 = dark green
(heterozygote + homozygote) 2pq + q^2 = total non-yellow

I will present all the listed frequencies in Table S1 (rather than just the ones selected by the authors) with comparison calculations https://upload.wikimedia.org/wikipedia/c...quency.png

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

on TBs (non-Sinitic Tibeto-Burmans)

lowland TBs seem to have similar ranges to northern Chinese, but the Tibetans have lower scores
Burmese are in range of Laotian Austroasiatics and hill Tais

of note is that the Lhasa Tibetan has slightly higher flush than Shigatse Tibetans- this indicates that Amdowa might be even higher, possibly close to the lowland TB/northern Han range
but I don't think any of these are Amdowa

I won't make a separate table for this but Hui and Tu/Mongour have effectively the same scores as lowland TBs/northern Han also
as such, it seems possible to comment that variance in alcohol flush is lower in northern China and resident, as well as branch-off groups than in southern China
additionally, one can infer that there is a barrier between Tibetans and lowland TBs, as well as between Burmese and Chinese TBs, but these are for obvious reasons

btw, if we use the usual autosomal estimates for the proportion of non-ST (indigenous plateau) peoples in Tibetans, even for southern Tibetans the allele frequency looks too low to be accounted for by admixture alone, if the source population that migrated to the Tibetan plateau was similar to eg. Qiang. Assuming the non-ST group had 0%, that would only get you to at lowest still in the mid-teens %.
That leaves a few options, including:
1) The pre-ST population that contributed to both Tibetans and lowland TBs/northern Han had lower alcohol flush than modern lowland TBs and northern Han, possibly only on the order of ~5-7% allele freq already. Admixture on the part of the latter with people from the Yangtze river (which seems to be have happened) may have increased their flush to modern levels.
2) Flush may have been reduced as a function of possibly linkage to an environmental adaptation (?)
3) Flush may have been reduced as a function of plateau Tibetan communities being possibly more prone to drift
4) The indigenous plateau component in southern Tibetans has been underestimated

the Burmese number is also slightly lower than what you'd expect exactly given their admixture proportions but it seems closer to variance (eg. if you had a Yi community more on the order of low 10% range as a 50-60% component in Burmese, and the rest close to 0 you could expect something like 6% for them)

somewhat related to option 1 here: what Koreans currently get on average is also a bit lower than what you'd expect from a population that's autosomally ~50% northern Chinese or lowland TB-like + ~50% Japanese, it would work better if the non-Japonic side had roughly 0.1 flat or lower
of interest is that Liaoning Shenyang does get around the low end (0.11), there aren't any samples for Shanxi/Shaanbei or northern Hebei but I wonder if those would also be lower than the zhongyuan-heavier samples provided

[ZZ131]

ADHD2 evolved to protect against schistosomiasis, related to rice farming, and perhaps related to some O1b lineages, Kra-dais