search this blog

Monday, June 20, 2016

Ancient genomes from the Himalayas


Open access at PNAS:

Abstract: The high-altitude transverse valleys [>3,000 m above sea level (masl)] of the Himalayan arc from Arunachal Pradesh to Ladahk were among the last habitable places permanently colonized by prehistoric humans due to the challenges of resource scarcity, cold stress, and hypoxia. The modern populations of these valleys, who share cultural and linguistic affinities with peoples found today on the Tibetan plateau, are commonly assumed to be the descendants of the earliest inhabitants of the Himalayan arc. However, this assumption has been challenged by archaeological and osteological evidence suggesting that these valleys may have been originally populated from areas other than the Tibetan plateau, including those at low elevation. To investigate the peopling and early population history of this dynamic high-altitude contact zone, we sequenced the genomes (0.04×–7.25×, mean 2.16×) and mitochondrial genomes (20.8×–1,311.0×, mean 482.1×) of eight individuals dating to three periods with distinct material culture in the Annapurna Conservation Area (ACA) of Nepal, spanning 3,150–1,250 y before present (yBP). We demonstrate that the region is characterized by long-term stability of the population genetic make-up despite marked changes in material culture. The ancient genomes, uniparental haplotypes, and high-altitude adaptive alleles suggest a high-altitude East Asian origin for prehistoric Himalayan populations.


Jeong et al., Long-term genetic stability and a high-altitude East Asian origin for the peoples of the high valleys of the Himalayan arc, PNAS June 20, 2016, doi: 10.1073/pnas.1520844113

23 comments:

andrew said...

For an area that has been inhabited since pre-LGM the time frame of available genomes isn't necessarily all that informative.

Rob said...

Treu. But surely a ~ 1000 BC sample from south Asia will be useful ?

Chad Rohlfsen said...

If 1000 BCE is the same as today, at least we know Mahayana was a cultural exchange only. It's better than nothing.

Hector said...

The main point of this paper is that despite the "archaeological and osteological evidences" the genetics of this region has been stable over the last 3000 years, NOT that it has been so since 10000 years ago or something even though it may be suggestive of such.

ryukendo kendow said...

The ADMIXTURE analysis is clearer evidence than ever that Sino-Tibetans are distinguished by a similarity to Tibetan highlanders and the North Chinese preserve elements of a movement from the Tibetan Plateau, accounting for the Sinitic language family. The original agriculturalists on the N China plain were probably much like the She or the non-Jomon half of Japanese, having no such Tibetan affinity. This is very obvious in the PCAs too, where the Chinese are bracketed by the Japanese and SE Asians in the NE-SE Asian cline but are distinguished from all other members of that cline by a shift to Tibet.

Interestingly this shift also exists in Altaic populations, which I would account for by placing the origin of their 'East Asian Agriculturalist' ancestry, appearing from the Iron Age onwards from the Russian aDNA evidence, in NW China, at the Gansu corridor and surrounds, among the belt of formerly agriculturalist cultures such as the Zhukaige, Tianshanbeilu, Qijia, etc. later lost to the Steppe- infused 'Northern Zone' during the Bronze Age. Also it looks like the Sinitic movement was not from Sichuan into the Central Plains, but from Qinghai down to the Central Plains.

Grey said...

Makes sense i think.

If population A has a tech edge over pop B then they A will usually push into B's territory unless they can't for some reason - in which case a border forms and diffusion occurs instead.

From the paper

"It is interesting that the high-altitude barrier to migration seems to be more permeable from the northern, as opposed to the southern, side of the Himalayan arc."

This may have been true for the denisovans etc as well and may have effected other Himalaya related adaptions if there were any.

Grey said...

mountain mommas

ak2014b said...

@Chad Rohlfsen

"If 1000 BCE is the same as today, at least we know Mahayana was a cultural exchange only. It's better than nothing."

Buddhism is dated to around 500 BCE.

But Mahayana is dated on average to somewhere in the 1st century CE, evolving from precursor Mahasamghika.

On the other hand, Buddhism in Tibet entered in the 7th century and took root there from the mid 8th century onward.

@Hector

"NOT that it has been so since 10000 years ago or something even though it may be suggestive of such."

Wikipedia

"Some archaeological data suggests archaic humans may have passed through Tibet at the time India was first inhabited, half a million years ago.[1] Modern humans first inhabited the Tibetan Plateau at least twenty-one thousand years ago.[2] This population was largely replaced around 3,000 BC by Neolithic immigrants from northern China. However, there is a "partial genetic continuity between the Paleolithic inhabitants and the contemporary Tibetan populations.[2]" referring to Zhao et al 2009

However, such conclusions still require confirmation from Paleolithic and later aDNA from the Tibetan Plateau. The current Himalayan aDNA study doesn't touch on that time period though.

ak2014b said...

@Chad Rohlfsen

So I think checking either side of 7th century CE and 8th century CE would give a better idea of whether the earliest introduction of Buddhism in Tibet involved population movement too. Regional history indicates that there was some in the early 8th century.

"The king builds temples in his capital, Lhasa, to house his wives' sacred treasures. This is the first visible foothold of Buddhism in Tibet. Early in the next century the Indian religion receives a further boost when Buddhists from central Asia flee to this remote region to escape the advance of the Muslims. But it is not until the second half of the 8th century that Tibetan kings actively promote Buddhism as their state cult."

Tobus said...

@ak2014b: If there were a significant population change at 700-800CE then 1000BCE would *not* be the same as today.

Chad Rohlfsen said...

ak,

I'm very familiar with Tibetan Buddhism, Padmasambhava, and the like.

Nirjhar007 said...

Indian culture has great resiliency . That is established through culture,archaeology,anthropology etc . This is just a peripheral glimpse.

Matt said...

Interesting finding, though these high altitude Tibetans are people would expect to have the stronger probability of surviving as a population, due to the adaptation advantage (which looked old). At least over 3k years.

Also, in the context of Jeong's earlier paper (which I feel they should have commented on explicitly) -
http://www.nature.com/ncomms/2014/140210/ncomms4281/full/ncomms4281.html

"The date of this East Asian admixture into the Sherpa was estimated to be 23.4 generations ago, based on the decay of linkage disequilibrium (LD) (see Methods; Supplementary Figs 4 and 5, and Supplementary Table 2).

This date is in close agreement with the historical record of a Sherpa migration out of their ancestral homeland in Eastern Tibet 400–600 years ago to their current place, Solu-Khumbu.

In subsequent analyses, we considered the subset of 21 individuals with 100% high-altitude component (referred to as HA-proxy sample; ‘HA’ for high altitude) to be the descendants of an ancient high-altitude population with a broad geographic distribution across the plateau.

Taken together, these results strongly suggest that Tibetans are admixed descendants of two populations currently represented by the HA-proxy and low-altitude East Asians (such as Han Chinese), while the Sherpa more recently experienced admixture with nearby East Asian populations, most likely Tibetans (Supplementary Fig. 5)."


So how much does that stand in relation to this? These samples cluster with Tibetans, not the High Altitude Adapted group.

Presumably that still stands... but the lowland East Asian admixture into Tibet had already happened by the 1st millennium BC (Chokhopani (3,150–2,400 yBP) ). The admixture time between lowland and highland components couldn't be dated in Jeong's previous.

However, at the same time, these are the earliest human settlements. So where were the HAA until that time? Presumably hunter gatherers?

Lathdrinor said...

The paper's earliest samples are from as late as 2,400 ya. It cannot be used to argue about the Neolithic and/or Paleolithic migrations that shaped the plateau. The fact that Y-DNA D is so plentiful among both Tibetans and Sherpa must be explained, as it is a generally rare haplogroup in low-land populations as well as populations to the north that might've migrated into the plateau. Consequently it is, in my opinion, the best signal for the original highland hunter-gatherers that evolved the unique plateau adaptations we see in Tibetans and Sherpa today, since it is this haplogroup that best divides Tibetans/Sherpa from low-land East Asian populations. In that case, Tibetans being a mixture of this high-land proxy with incoming low-land migrants is still the theory to beat, in my opinion. The Sherpa are unlikely to be a relic population of those original hunter-gatherers, despite possible interpretations as such in ADMIXTURE. This is because they only migrated away from the plateau 400-600 years ago and are also a low-population group with likely drift & isolation, making them more likely to show up as vectors in ADMIXTURE, even if their Y-DNA shows that they are quite diverse themselves.

Much earlier samples are needed for a proper assessment of the Tibetan plateau's pre-history. We still don't have anything equivalent to, say, Yamnaya or the recent Near East paper.

ryukendo kendow said...

@ Lathdrinor

The Sherpa are unlikely to be a relic population of those original hunter-gatherers, despite possible interpretations as such in ADMIXTURE. This is because they only migrated away from the plateau 400-600 years ago and are also a low-population group with likely drift & isolation, making them more likely to show up as vectors in ADMIXTURE, even if their Y-DNA shows that they are quite diverse themselves.

This is not likely. ADMIXTURE typically does not produce components centering on drift in a single population, e.g. 'Palestinian' or 'Kalash' components, until it reaches very high K; in all papers on Sherpa and Tibetans, the Highland Asian component appears at very low K. This is relevant as ADMIXTURE will not pick a component to appear at low K unless it explains a great deal of variation in the dataset across many of the populations. Also, 'Palestinian' and 'Kalash' components drop off extremely rapidly outside of the endogamous groups in which they center, no other populations scores more than, say 40% typically, this is not how the 'Sherpa' component behaves.

Also, drift or variation in single populations is extremely unlikely to dominate low PCs in PCA, but the Highland-Lowland cline dominates one of the lowest 2 dimensions in PCA of Asian populations in all papers with Sherpa and Tibetans, with Sherpa being highly differentiated in that dimension. Its not very logical to claim that Sherpa drift is somehow especially correlated with the Highland-lowland direction, away from Lowland East Asians, as opposed to Sherpa being really distinct from them in terms of deep ancestry.

This paper:
http://www.genetics.org/content/early/2015/10/20/genetics.115.178673.supplemental
in the supplements show that, other than Ainu, Sherpa is the most different from lowland east asians in terms of D and F stats and Treemix, which are not sensitive to recent drift like ADMIXTURE is.

ryukendo kendow said...

I'm inclined to think that Sherpa really are the best representatives of Tibetan Hunter Gatherers. We have no idea how divergent that population really was, it may be the case that they were extremely divergent, even more so than Ainu, but they contribute less than 10% of ancestry of Sherpa. Or they were much less divergent than Ainu and contribute up to 100% of ancestry of Sherpa. That will take aDNA. But that Sherpa-She form a cline of coancestry across East Asia, with Sherpa on one end, cannot really be denied at this point.

Lathdrinor said...

@ryukendo kendow Sherpa are rather diverse in single-sex markers. Their distribution is quite similar to that of Tibetans, with similar percentages of D and O, 45% and 30%, respectively. But importantly, they have up to 17% of R1a, J, and F. Their mtDNA is largely shared with lowland East Asians, but also contain a few distant clades such as M5c2, M21d, and U. This is not how a relic population looks like, and is further supported by the historical fact that they came from eastern Tibet/Kham within the last few centuries.

While I do believe you are correct to say that they are the most distinguished from East Asian lowlanders, and are probably the best representative of a "highland East Asian population" today, I do not think you can argue that this makes them an excellent proxy for Neolithic/Paleolithic highland hunter-gatherers. Tibetans can be modeled as a mixture because Tibetan samples usually include both plateau Tibetans and lowland Tibetans, who aren't necessarily all that different in Y-DNA and mtDNA, but are probably different autosomally, since survival at high altitudes inducted a selective process, as has been shown in many papers; bi-directional gene flow, however, helped to keep major Tibetan populations closer to each other than to the Sherpa, who were more isolated from this flow.

Thus, while "Tibetans" can be modeled as a mixture of Sherpa and Dai, that doesn't represent a simple highland hunter-gatherer + lowland farmer model for Tibetan origins, since ultimately it fails to account for the fact that historical highland populations such as the Sherpa were likely derived from a mixture of lowland hunter-gatherers/farmers and highland hunter-gatherers during the Neolithic themselves. Since we do not have a proper proxy for this highland hunter-gatherer population today, ADMIXTURE and PCA cannot distinguish it at low K. Consequently the Sherpa are mistaken for a proxy on the basis of being a more isolated plateau population, who could be used as a stand-in for "highland genetics" as it exists today, but *not* as it existed during the Neolithic and Paleolithic.

Lathdrinor said...


I suppose another way of describing this distinction is to take issue with your theory of a "movement from the Tibetan Plateau" becoming a major factor in distinguishing different lowland East Asian populations. I cannot see any evidence of this either archaeologically or genetically. Lowland Sino-Tibetan groups such as the Chinese and the Bai do not have the highland adaptations that Sherpa and plateau Tibetans possess. Further, D and especially D1, which diverged from D3 around 30,000 years ago and which is found in abundance within Sherpa and Tibetan populations, is very rare among lowlanders - virtually 0% of Chinese. This cannot be explained under a model of "movement from the Tibetan Plateau" unless we assume that D arrived after the movement, which also cannot be supported in light of 1) the lack of a neighboring source for D and 2) this study.

This is why I continue to believe that the original population of D bearers is the proper proxy for highland hunter-gatherers on the Tibetan plateau, while O bearers were later migrants dating to the Neolithic. The lack of O diversity within Tibetan and Sherpa populations is obvious, and can be used to rule out M175, M122, or even M134 being native, since we don't find basal instances of the above haplogroups on the plateau or linkage clades between separate branches. At the same time, evidence of modern human habitation on the Tibetan plateau dates back >21,000 years ago, yet archaeological evidence indicates that the main Neolithic complexes resembled those in nearby lowland *agricultural* sites such as Qijia and Majiayao, which had earlier lowland precedents. The evidence therefore is much more supportive of a movement from the lowlands onto the plateau, after which the migrants then underwent high-altitude selection, with only those who sufficiently mixed with the native hunter-gatherers, who had already developed high-altitude adaptations, surviving, which then explains the very high % of D in Tibetans and Sherpa today.

Thus I stand by my statement that the population which evolved highland adaptations on the plateau should be older than the arrival of O, and should correspond to a population bearing, especially, D1 and D3.

ryukendo kendow said...

@ Lathdrinor

Lathdrinor, for your first comment, I don't think we are disagreeing really. I never said that Sherpa are a proxy for Tibet during the Paleolithic, as the relevant HG ancestry in Sherpa, even though greater than in other regions, may be as low as ~10% only, which we can't know unless we get aDNA. Whatever the case, it is a fact that Sherpa forms an outgroup to all E Asians in D and F stats and Treemix.

Since we do not have a proper proxy for this highland hunter-gatherer population today, ADMIXTURE and PCA cannot distinguish it at low K.

This is likely false, ADMIXTURE runs where the most ASI population is Punjabi still produces a S Asian component without problems, we just have to remember that even though Punjabi end up 100% S Asian this does not correspond to Punjabi being 100% ASI, just that they are at the end of the ASI cline for that particular run.

ADMIXTURE cannot determine the direction of gene flow, but for PCA, if Han, Naxi, Yi, Bai and so on are pulled away from their non SinoTibetan neighbours like Dai, Lahu, Japanese or She, towards the Highland Asian dimension dominated by Sherpa at one end at very low PCs, furthermore the distinctiveness of Sherpa in the East Asian tree is clear in D and F stats, then that Sherpa-She dimension of correlated variation is very strong and testifies to Han, Naxi etc. really being closer to Sherpa, and also to this not being mediated by Han-->Sherpa or Han<>Sherpa, but really mediated by Sherpa-->Han, in comparison to their Japanese and SE Asian neighbours, such that Han/Naxi/Yi is intermediate between Sherpa and Dai/She/Japanese in a very important dimension of variation, instead of any other scenario.

Did you know the Denisovan high-altitude EPAS1 allele is found in Han chinese at a couple of percents, despite it not being found in any other low-altitude population? Its absent in Melanesians, making me think its actively selected against at low altitude, implying a late date of introduction of this to the Han Chinese gene pool. I expect a mixed Highland-Lowland population would have influenced Sino-Tibetans in Asia.

Lathdrinor said...

@ryukendo kendow I think we agree on two facts: 1) Treemix shows that Sherpa are an out group to East Asians 2) PCA shows that for some low PC, Sherpa-Han/Naxi/Yi-Japanese/Dai/She form a dimension of variation. The issue is interpretation.

This is likely false, ADMIXTURE runs where the most ASI population is Punjabi still produces a S Asian component without problems, we just have to remember that even though Punjabi end up 100% S Asian this does not correspond to Punjabi being 100% ASI, just that they are at the end of the ASI cline for that particular run.

My point is that ADMIXTURE cannot necessarily distinguish between recently and more distantly admixed populations. The single-sex results clearly indicates that Sherpa are an mixed population, and probably even recently so. You might want to consult: http://www.nature.com/articles/srep16249, which states: "Analysis showed that Sherpas share most of their paternal and maternal lineages with indigenous Tibetans, representing a recently derived sub-lineage... These findings reject the previous theory that Sherpa and Han Chinese served as dual ancestral populations of Tibetans, and conversely suggest that Tibetans are the ancestral populations of the Sherpas, whose adaptive traits for high altitude were recently inherited from their ancestors in Tibet... The detailed genetic diversity pattern revealed in the Y-STR network and the mtDNA phylogenetic trees indicate that Sherpas represent a recently derived lineage from Tibetans dated to less than 1,500 years ago."

This would go against a naive interpretation of the ADMIXTURE results, from which one might mistakenly read that Tibetans are a mixture of a Sherpa-like population and a Dai-like population.

Han, Naxi etc. really being closer to Sherpa, and also to this not being mediated by Han-->Sherpa or Han<>Sherpa, but really mediated by Sherpa-->Han

Or a third population that influenced both Sherpa and Han/Naxi, but less so the She and Dai. Say... The Neolithic inhabitants of Qijia and Majiayao, both of whom were lowlanders.

Did you know the Denisovan high-altitude EPAS1 allele is found in Han chinese at a couple of percents, despite it not being found in any other low-altitude population?

If you are talking about: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC4134395/#SD1, the paper indicates that the 5 SNP motif present in Tibetans in Han is 0-1%, which doesn't indicate any sort of significant gene flow, much less recent gene flow. To be sure, *some* populations similar to Tibetans must have influenced neighboring lowland populations. But it cannot justify the large-scale gene flow from highland->lowland that you argue. I simply cannot imagine highlanders sweeping across the lowlands - they lacked the population, the genetic adaptation, and the technological edge for lowland success. Perhaps the Tibetan Empire did have a small effect on surrounding populations, but it was short-lived.

ryukendo kendow said...

@ Lathdrinor

Lathdrinor, it is entirely possible that the uniparental data indicates a much shallower separation between a population and its neighbours than the autosomal data indicates, for example, the Orang Asli of malaysia are not that distinctive from the peoples around them in uniparentals despite preserving within themselves a ~30% genetic substrate from a population clading with Onge, basal to other East Asians. Ainu are also much more similar to Japanese and populations around the Okhotsk sea in uniparentals than their autosomes would indicate.

Either way, autosomal analysis indicates good enough symmetry between Tibetans, Sherpa, and lowland East Asians with West Eurasians, indicating that West Eurasian ancestry, even if it exists, is too low to create the phylogenetic distinctiveness in formal analysis that Sherpa display. Also, if Sherpa are only distinctive due to recent drift, the pattern of a Sherpa-She cline dominating PC2 in East Asia, or its behaviour in Treemix and Ds and Fs, cannot be explained.

Or a third population that influenced both Sherpa and Han/Naxi, but less so the She and Dai. Say... The Neolithic inhabitants of Qijia and Majiayao, both of whom were lowlanders.

Maybe, but note that this population has to influence the Sherpa the most, and influence Tibetans less, and Naxi even less, and Han the least, which makes a pure lowland East Asian population unlikely. Maybe it was the Majiayao, I can actually see this happening, since they in NW China are in the right position to contribute both to Lowland China and also via Qijia and the 'Northern Zone' to the Steppe, to explain the distribution of the affinities we see here, but in that case they would have to admix with the ultimate source population that explains this cline, i.e. some kind of population whose penetrance in modern times is greatest in marginal Tibetan populations, probably Tibetan HGs.

they lacked the population, the genetic adaptation, and the technological edge for lowland success

Another interesting point of disagreement; I think, if we look at history, technological superiority has counted for very little until the early modern periods. Just look at what the Thais did to the Khmers, or the Goths did to the Romans, or for that matter the Semitics to the pre-Semitics and IEs to the pre-IEs. What matters for cultural group survival is the capacity for internal mobilisation of a large proportion of people within the in-group to win intergroup conflicts, either via organic social relations, e.g. tribal social relations, or inorganic management strategies, e.g. conscription in large-scale societies. Whether small groups of pastoralists or highland hill tribes can descend and overrun huge masses of people, which they've proven they can--repeatedly--depends on whether their subsistence/economic system and their sense of imagined community can support the level of coordinated collective action needed to overcome another cultural group, and these factors are often more supportive of mass coordinated action in technologically primitive societies than in advanced ones, who generally lose such capacities due to internal status competition and sociopolitical decay.

Lathdrinor said...

@ryukendo kendow I'm not sure I'd agree with you that Ainu are much closer to Japanese in uniparental markers than they are in autosomes - they seem rather different in both. For example, up to 90% of Ainu paternal DNA belong to Y-haplogroup D, while only 50% of Japanese paternal DNA do. Ainu also have a significant - up to 20% - of mtDNA Y, which among Japanese reaches only 1-2%. This is quite a large difference and matches up well with the autosomal difference.

I also offer a simpler explanation than the case of the Orang Asli: Sherpa descended from Tibetan highlanders, who should be distinguished from Tibetan lowlanders, but aren't in many samples that include both under "Tibetans," which would explain why Sherpa, being strictly highlanders, would show up as a single component in ADMIXTURE while Tibetans would show up as two. I don't disagree with you that Sherpa are one of the best representatives of Himalayan highlanders today, and do think that the patterns you are seeing with respect to PCA and Treemix indicate an altitude cline running, from low to high, Dai->Han->Tibetan->Sherpa. But I'd say that this is the effects of selection and drift among highlanders, rather than them being a relic population of Neolithic/Paleolithic hunter-gatherers.

Maybe, but note that this population has to influence the Sherpa the most, and influence Tibetans less, and Naxi even less, and Han the least, which makes a pure lowland East Asian population unlikely.

They probably did have a larger effect on groups in northwest and west China than they did on groups in east China, considering the geography, but I actually don't see why it has to be so. Aren't you disregarding the potential impact of archaic Himalayan hunter-gatherers on the Tibetans and the Sherpa? The high amount of Haplogroup D1 and D3 has to come from somewhere. I'd say it came from them, while the lowlanders were primarily O-M117. We can then imagine a highland-lowland cline resulting from degree of admixture with the archaic Himalayan hunter-gatherers, with Sherpa=Tibetan highlanders > Tibetan lowlanders > Naxi > Han > Dai.

This would make more sense to me in terms of explaining the virtual absence of D1 and EPAS1 5 SNP markers in lowlander populations, and the archaeological record, which for the moment seems to favor a lowland->highland migration route. It seems to me your main opposition to this theory is the fact that Sherpa showed up as a component in ADMIXTURE and that Treemix indicates Sherpa are the most distinct from East Asian populations. Do you still feel that this is a definite contradiction?

ryukendo kendow said...

Lathdrinor, as you know I tend not to respect archaeological interpretations too much, as they are subject to the fashions and fads of the day, and often reveal an ignorance on the part of academics about the dynamics of small scale-societies, such that some archaeologists are still saying that there is no evidence for IE movement into Europe, or that the expanding neolithic does not reflect physical movement, or that there is no an IE movement into India, though I'm sure that a new generation of young academics armed with gene science will find the signs of such a movement the moment they start researching with different priors in mind.

But I'd say that this is the effects of selection and drift among highlanders, rather than them being a relic population of Neolithic/Paleolithic hunter-gatherers.
Formal modelling does not suffer from the effects of drift to anywhere near the extent that ADMIXTURE does. In fact, in theory, f4s (not f3s), Ds, treemix topology (not branch length) and ADMIXTUREGRAPH topology (not branch length) should not suffer from recent endogamy/drift effects at all. So drift in Sherpa should not cause Treemix to identify it as an outgroup to lowland East Asians; Kalash, Palestinian and Nganasan are not identified as outgroups to their regions in this way in formals.

Anyway the f3s, which indicate very strong signs of admixture in Tibetans as a mix of Sherpa and Han, indicate Sherpa-->Tibetan; and recent drift effects private to Sherpa and Tibetan should not be expected to pull Naxi, Yi and especially Han away from other lowland east asian populations at very low PCs, unless Sherpa-She differences are reflected also in differences between Han, Naxi, Yi vs She and Japanese as well, and in fact also in Mongolian vs Yakut, since Mongolian, Daur etc. also evince this pull.

About EPAS1, if you look at the paper there is a very strong and very smooth correlation between the allele freq of the population and its altitude, which tells us that selection effects both for and against the high-altitude allele due to environment is extremely strong, such that movements of people at the historical baseline rate does not erase this cline or result in interruptions of the cline's smoothness; this is anyway expected, as EPAS1 influences metabolism and oxygen use and is thus global in effects and strongly influences the phenotype. So I would expect the small allele freq in lowland Han vs. e.g. HGDP Dai as strong evidence of recent gene flow into Han, otherwise the allele, which is probably purged very quickly, c.f. the altitude-cline relationship, will have zero occurrence.