To the antipode of Asia

By Razib Khan | July 18, 2011 1:57 am

Markers show populations sampled by HUGO Pan-Asian SNP Consortium

The Pith: Southeast Asia was settled by a series of distinct peoples. The pattern of settlement can be discerned in part by examination of patterns of genetic variation. It seems likely that Austro-Asiatic populations were dominant across the western half of Indonesia before the arrival of Austronesians.

About a year and a half ago I reviewed a paper in Science which did a first pass through some of the findings suggested by the HUGO Pan-Asian SNP Consortium data set, which pooled a wide range of Asian populations. You can see the locations on the map above (alas, the labels are too small to read the codes). The important issue in relation to this data set is that it has a thick coverage of Southeast Asia, which is not well represented in the HGDP. Unfortunately there are only ~50,000 markers, which is not optimal for really fine-grained intra-regional analysis in my opinion. But better than nothing, and definitely sufficient for coarser scale analysis.

A few things have changed since I first reviewed this paper. First, I pulled down a copy of the Pan-Asian SNP data set. I’m going to play with it myself soon. Second, after reading Strange Parallels, volume 1 and 2, I know a lot more about Southeast Asian history. Finally, the possibility of archaic admixture amongst Near Oceanians makes the genetics of the regions which were once Sundaland and Sahul of particular interest.

Before we hit the genetics, let’s review a little of the ethnography of Southeast Asia, as this may allow us to tease apart the meaning of some of the results. The largest ethno-linguistic group in Southeast Asia is that of Austronesians. An interesting point in relation to Austronesians is that they aren’t limited to Southeast Asia. As you can see the Austronesians range from off the coast of South America (Easter Island) to southeast Africa (Madagascar). Though there’s debate about this issue it seems to me that the most likely current point of departure of the Austronesian migration is Taiwan. Though today Taiwan is predominantly Han Chinese, that is an artifact of relatively recent migration. The indigenous population is clearly Austronesian.

A second language family which is somewhat expansive, though Southeast Asia focused, is Austro-Asiatic. There is a great deal of internal structure to this ethno-linguistic group, in that there is a well known coherent Mon-Khmer cluster, which includes some ethnic minorities in Burma and Thailand, as well as Cambodians. Additionally you have Vietnamese in the east and some tribal groups in northeast India. There has long been debate about whether these Indian tribes, the Munda, are the original Indians, to be supplanted later by Dravidian and Indo-Aryan speakers, or intrusive to the subcontinent. I believe that the most recent genetic data points to intrusion from the east into South Asia. Austro-Asiatic was likely less fragmented in mainland Southeast Asia before the historical period. Both the dominant ethnic groups in Burma and Thailand are intrusive and absorbed Mon-Khmer populations, the latter dynamic being historically attested.

Finally there are the ethno-linguistic clusters of Burma and Thailand (and Laos). The former nation is dominated by the Bamar, a Sino-Tibetan population with origins in South China ~1,500 years ago. In Burma the Mon substrate persists, while the Shan people of Thai affinity reign supreme across the northeastern fringe of the nation. In Thailand and Laos the Mon-Khmer substrate has been marginalized to isolated residual groups. But it is notable that in both these polities the Mon-Khmer populations set the tone for the civilizational orientation of the conquering ethnicities. The Thai abandoned Chinese influenced Mahayana Buddhism for the Indian influenced Theravada Buddhism of the conquered populace. Despite the notional ethnic chasm between the Thai and the Khmer of Cambodia, the broad cultural similarities due to the common roots in the society of the Khmer Empire is clear.

With the ethnographic context in place, let’s look at the two primary figures which we get from the paper. The first figure shows a phylogenetic tree of the relationships of the populations in their database, color-coded by ethnolinguistic group. Next to that tree there’s a STRUCTURE plot at K = 14, which means 14 ancestral populations. They’ve colored the bar components to match the ethno-linguistic classes (e.g., red = Austro-Asiatic, an Austro-Asiatic modal component). The second figure shows two PCA panels. PC 1 is the largest component of genetic variance in the data set, and PC 2 the second largest. I’ve added a label for the Papuan populations.

Going back to the chronology above, we know that the Thai came last. The Sino-Tibetans came before then. The issue I wonder about is the relationship of the Austronesians and Austro-Asiatic groups. Interestingly the Austronesian proportions are high not only in island Southeast Asia, but also among many South Chinese groups. In contrast, among the Mon-Khmer hill tribes of Thailand, who are presumably representative of groups which were present before the Thai migrations, it is absent. And it is notable to me that not only does Austro-Asiatic exhibit fragmentation in relation to Thai and Sino-Tibetan, but it does so to some extent with relation to Austronesian! The indigenous folk of central Malaysia seem to speak a Austro-Asiatic language. Finally, the Austro-Asiatic component rises in frequency on the southern fringes of island Southeast Asia, in densely populated Java.

Because of the thicker textual record for mainland Southeast Asia we know that the Austro-Asiatic groups predate the Thai and Sino-Tibetan ones. I believe that the Austro-Asiatic element also predates Austronesian in Southeast Asia. That is, I believe that an Austro-Asiatic substrate existed before the arrival of Austronesians from the zone between the Philippines and Taiwan. The Negritos of inner Malaysia, who are genetically and physically distinctive, speak Austro-Asiatic languages. This should not be surprising, it seems that hunter-gatherer groups often switch to the language of resident agriculturalists. Because of their isolation some of these groups have persisted in speaking the languages of the “first farmers” of Malaysia, even after those pioneers were absorbed by newcomers.

The PCA shows clearly that the Austronesians are the genetically most varied of these Southeast Asian groups. Why? I believe it is because they are late arrivals who have admixed in sequence with whoever was resident in their target zones. In the east of island Southeast Asia the admixture occurred with a Melanesian population. Both the STRUCTURE plot and the PCA show evidence of this sort of two-way admixture. The STRUCTURE is straightforward, but note the linear distribution of the Austronesians in relation to outgroups in the first panel, and implicitly on the second.

Why is the Austro-Asiatic fraction higher in Java than to the zones in the north? Java is today the most densely populated region of Indonesia because of its fertility. I hypothesize that the spread of the Austronesians was facilitated by a more effective form of agriculture which could squeeze more productivity out of marginal land. Relative to Java the Malay peninsula, Borneo, and Sumatra, are agriculturally marginal. The densities of the Austro-Asiatics was greatest in Java, while they were very thin in the regions to the north. It seems likely that the Austronesians engaged in a series of “leap-frogs” to islands and maritime fringes which were not cultivated by the Austro-Asiatic populations. Some Indonesian groups, such as the Mentawai who live on the island of the same name off the western coast of Sumatra, cluster with the Taiwanese, as if they transplanted their society in totality.

One thing that needs to be mentioned when talking about the genetics and prehistory of Southeast Asia are the “Negritos.” As indicated by their name these are a small people with African-like features. As is clear from the charts above these people are not particularly genetically close to Africans. The Philippine Negritos seem to have some relationship to the Melanesians. Interestingly they speak an Austronesian language; again following the trend where marginalized indigenes seem to pick up the language of their farming neighbors. The Negritos of Malaysia are somewhat different, but note that one of the populations exhibits Austro-Asiatic, but not Austronesian, admixture. This comports with my supposition that the Austro-Asiatic populations were the first to marginalize these tribes before themselves being assimilated by the Austronesians.

Someone with a better ethnographic understanding of Southeast Asia than I could probably decode the results above with greater power. But at this point I think we’ve got a chronology like so:

1) First you have hunter-gatherer populations of broad Melanesian affinities in Southeast Asia.

2) Then Austro-Asiatic populations move south from the fringes of southern China. Some push west to India, while others leap-frog south to zones suitable for agriculture such as Java.

3) Then Austronesian populations sweep south along water routes, and marginalize the Austro-Asiatics in island Southeast Asia, though the not on the mainland.

4) The Bamar arrive from southern China over 1,000 years ago, and marginalize the Austro-Asiatics in Burma.

5) The Thai arrive from southern China less than 1,000 years ago, take over the central zone of mainland Southeast Asia, and make inroads to the west in Burma.

I will hazard to guess that the Malagasy of Madagascar are Austronesians who have very little of the Austro-Asiatic element in their ancestry. I believe this is so because they were part of the leap-frog dynamic where societies were transplanted from suitable point to point by water (the Malagasy language seems to be a branch of dialects of southern Borneo!).

So far I’ve been talking about the north to south movement. And yet the paper observes a south or north gradient in genetic diversity, which implies to the authors migration from south to north (the northern East Asian groups being a subset of the southern). But the past may have been more complex than we give it credit for. It is entirely possible that modern humans arrived in northeast Asia via a southern route, retreated south during the glaciation, and expanded north, with some groups pushing back south again. As it is, looking at how distantly the Melanesians relate to East Eurasians I think the most plausible model is that there wasn’t a relatively recent expansion from Southeast Asia. Rather, the ancestors of most East Eurasians survived in refugia in China, and a sequence of agriculturally driven expansions have reshaped Southeast Asia more recently. These populations admixed with the indigenous substrate, more or less. This would have resulted in an uptake of genetic diversity. Finally, the massive expansion of Han from the Yellow river basin may have caused the extinction of many lineages across China within the past ~3,000 years.

Citation: ., Abdulla, M., Ahmed, I., Assawamakin, A., Bhak, J., Brahmachari, S., Calacal, G., Chaurasia, A., Chen, C., Chen, J., Chen, Y., Chu, J., Cutiongco-de la Paz, E., De Ungria, M., Delfin, F., Edo, J., Fuchareon, S., Ghang, H., Gojobori, T., Han, J., Ho, S., Hoh, B., Huang, W., Inoko, H., Jha, P., Jinam, T., Jin, L., Jung, J., Kangwanpong, D., Kampuansai, J., Kennedy, G., Khurana, P., Kim, H., Kim, K., Kim, S., Kim, W., Kimm, K., Kimura, R., Koike, T., Kulawonganunchai, S., Kumar, V., Lai, P., Lee, J., Lee, S., Liu, E., Majumder, P., Mandapati, K., Marzuki, S., Mitchell, W., Mukerji, M., Naritomi, K., Ngamphiw, C., Niikawa, N., Nishida, N., Oh, B., Oh, S., Ohashi, J., Oka, A., Ong, R., Padilla, C., Palittapongarnpim, P., Perdigon, H., Phipps, M., Png, E., Sakaki, Y., Salvador, J., Sandraling, Y., Scaria, V., Seielstad, M., Sidek, M., Sinha, A., Srikummool, M., Sudoyo, H., Sugano, S., Suryadi, H., Suzuki, Y., Tabbada, K., Tan, A., Tokunaga, K., Tongsima, S., Villamor, L., Wang, E., Wang, Y., Wang, H., Wu, J., Xiao, H., Xu, S., Yang, J., Shugart, Y., Yoo, H., Yuan, W., Zhao, G., & Zilfalil, B. (2009). Mapping Human Genetic Diversity in Asia Science, 326 (5959), 1541-1545 DOI: 10.1126/science.1177074

  • Maju

    I broadly agree with what you say, however I have also some differences.

    The Philippine Negritos do not appear to be more akin to Melanesians than to mainland SE Asians. Guess that a better experiment can be devised but in the first PC graph the Philippine Negrito closest to Papuans has the same visual distance to these as the Philippine Negrito closest to the Tai-Kadai or Hmong-Mien clusters. This is not true for the Malaysia Negritos however who, rather counter-intuitively do appear much closer to Papuans (and Wallacea Austronesians).

    This gets me to the fact that Papuans are and have been for many millennia agriculturalists (sago, tubers, pig). The first agriculturalist wave in SE Asia may have been Austroasiatic speaking (??) but not necessarily Austroasiatic in genetics, at least not everywhere. This is almost for sure the case east of Wallace Line (Timor, New Guinea…) but may have also been the case at some localities West of it (see this for an in depth archaeological discussion).

    This may be compared tentatively to the Austroasiatic penetration in South Asia, where it clearly absorbed the substrate while at the same time retaining distinctiveness, specially in the Y-DNA. But I’d say that Austroasiatic expansion already related to cerealistic (rice) Neolithic, whose roots are in South China (Hunan specially) and it may have spread on a previous primitive agriculturalist wave of sago and pig only, which does not appear related to any particular genetic pool (cultural diffusion).

  • Razib Khan

    thanks. makes sense. i might explore some issues when i get the pan-asian set started.

  • ohwilleke

    In broad outlines, I think that you have it right, although it is hard to match the colors in the phylogenetic tree to those in the K=14 analysis. They don’t seem to be a match as they seem, e.g., to implausibly suggest as large Tai-Kadai ancestry in the Japanese and Han Chinese, and there are simply more colors in the K=14 chart than in the phylogenetic tree.

    The PCA charts suggest both broad, gradual population genetic shifts with pre-maritime geography, and yet, with a couple of exceptions fair neat genetic discriminants between linguistic groups.

    * The great distinctiveness of Philippine Negrito and Malaysian Negrito populations on the PC chart is surprising. At K=14 there is almost no overlap of common components between the two populations. The Malaysian Negrito population shows an affinity to the Papuan population, perhaps with varying degrees of Austronesian and Austro-Asiatic admixture.

    But, the Philippine Negrito population is closer to the population of India than to the Papuan population and it is no closer to the Papuans than it is to the Uygars. They are not in the range of the broad Papua to Northeast Asian continuity in genotype, showing a much strong West Eurasian affinity.

    The PC chart doesn’t specifically break out ANI and ASI components, but if the North-South; West-East tendencies in the PC chart are applicable within India, then Philippine Negrito may have a suggestively small distance from where ASI would appear if there were a pure ASI individual. Put another way, Philippine Negritos appear quite close genetically to where the Andaman Onge should be.

    I’d be curious to see where the Ainu (indigeneous Japanese) fit. My intuition is that they would be fairly close to the Philippine Negrito (with Ainu also having a modest NE Asian admixture) despite not looking all that similar in physical appearance – perhaps belonging to a single broad island hopping wave of settlement.

    If the PC chart here is providing valid intuitions, one would also expect high Denisovian admixture levels in Malaysian Negritos but little Denisovian admixture in Philippine Negritos.

    * Also interesting is the almost complete absence of Austronesian elements in Japan, even in Okinawans which are geographically closest to Taiwain the putative Austronesian homeland (Okinawans have even less of an Austronesian component than the Han Chinese from China). Every now and then, you see suppositions of Austronesian influences on Japan (linguistic or cultural or genetic), but this data doesn’t bear it out. Indeed, Okinawa looks genetically like the most purely Altaic (and hence historically Yaoyi) of the the Japanese samples, suggesting an almost complete replacement of the local population (if any) by the Yaoyi, followed by much lower levels of migration from China than was experienced in the rest of Japan.

    Indeed, the absence of apparent substrate or admixture influence suggests that Ryukyukans might be a better choice of looking at ancestral Japanese lingustics than Japan, because there would have been little or no substrate influence and would have been less areal borrowings from Chinese.

  • Ortu Kan

    I’d be curious to see where the Ainu (indigenous Japanese) fit. My intuition is that they would be fairly close to the Philippine Negrito (with Ainu also having a modest NE Asian admixture) despite not looking all that similar in physical appearance – perhaps belonging to a single broad island hopping wave of settlement.

    Are the STRUCTURE and PCA plots shown above, in that case, compelling you to update your priors re: the significance of “Jomon” ancestry in non-Ainu Japanese? Is the negligibility of the Mamanwa-modal component amongst them, for instance, problematic?

    The comparatively greater length of the “Altaic” bar in Okinawans — and the natural conclusion that this is indicative of “almost complete replacement of the local population … by the Yayoi” — are interesting in light of evidence from physical anthropology* that Ainu-like ancestral components might follow a bimodal distribution with peaks both the extreme south and extreme north of the island chain (i.e., in Ryukyuans and Hokkaido Ainu, though more pronouncedly in the latter than in the former).

    * — Cranial and dental data, earwax type, etc. In mainland stereotypes, it’s interesting to note, Okinawans are distinguished physically by their hirsuteness and waviness of hair, shorter stature, lower incidence of epicanthic folds, and dark skin (cf. perceptions of the Ainu and emishi).

  • Hla Thein

    A few points about Burmese history that might be relevant:

    (1) Modern Burmese is part of the Lolo-Burmese subgroup of Tibeto-Burman. Tibeto-Burman may or may not be genetically (in terms of linguistics) related to Chinese.

    (2) Lolo-Burmese branched off into its main Lolo (or Yi) and Burmese branches less than 2,000 years ago. Lolo-Burmese was the predominant language in southwestern China until at least 1,000 AD and likely the language of various independent kingdoms that existed in the region (eg Yelang, Dian) since bronze-age times. There were strong links to the steppe cultures to the north and northwest.

    (3) Prior to the arrival of Lolo-Burmese speakers in the second half of the 1st millennium (in the wake of Nanzhao invasions of the 9th century), what is now central Burma was primarily inhabited by the Pyu or Tircul who spoke a related Tibeto-Burman language.

    (4) Pyu or Tircul civilization shows strong connections with much earlier bronze-age cultures going back more than 3,000 years.

    (5) The 18th century Burmese chronicles say the modern Burmese (or “Myanmar”) people are an amalgam of earlier pre-existing populations. The Mons (speaking an Austro-A language) existed primarily along the southeast coast of Burma, not in the centre. In addition to the Pyu, other Tibeto-B populations in the center may have included the Thet (or Sak) and the Kadu. Tai or Shan speakers may already in the late 1st millennium (or even much earlier) have been present further east, in present-day Shan State, Laos and northern Thailand.

    For more on Burmese history see “The River of Lost Footsteps”.

  • Paul Ó Duḃṫaiġ

    The highest language diversity in Austronesian family is in Taiwan. One classification has 10 branches of the family. 9 of which are based in Taiwan and the 10th consists of all member-languages outside of Taiwan.

    The reconstructed expansion seems to imply colonisation of the Philippines and then onward expansion. Interesting enough the languages in Madagascar are closest to those of Borneo, the standard position on this is that Madagascar was unoccupied until the first centuries of the Common Era after which time it was colonised from the sea by Austronesian speakers.

    Funnily enough the narrative often in the Philippines is that they are really “Malay” who were colonised by the Spanish. The linguistic evidence however points to austronesians been in the Philippines longer then the ethnogenesis of the Malays.

    For me an interesting feature is the languages are VSO (Verb subject Object) in order just like Irish (insular celtic languages are unique among IE languages in this regard). That and my son is half filipino.

  • Razib Khan

    Funnily enough the narrative often in the Philippines is that they are really “Malay” who were colonised by the Spanish.

    yes. i’ve heard this from filipino friends, and have repeated it myself. it seems that “malay” is more archetypical for some reason, probably because the filipinos have so many influences from spain. but it isn’t as if malays don’t have outside influences either….

  • Pavlova

    I find it weird how Japanese have more Altaic than Koreans. And Okinawans have more Altaic than Japanese and Koreans? It seems that the lack of “true Altaic” samples, such as the Manchurians, Turkic peoples and Siberians have made this Altaic cluster more of a Korean-Japanese cluster.

    It’s interesting to note that Okinawans show some Papuan admixture (more so than other East Asian groups). This could mean that the original Jomon were at least partly Papuan-related. The Okinawans could be a product of inter-breeding with the Yayoi population.

    The PCA is also quite strange. Indians are usually more related to Papuans than Southeast Asians or East Asians are (or am I mistaken?). Possibly the lack of actual Papuan samples could indicate that the “Papuan” cluster is actually Melanesian. Melanesians are a product of Papuans inter-breeding with Austronesian populations.

  • Razib Khan

    The PCA is also quite strange. Indians are usually more related to Papuans than Southeast Asians or East Asians are (or am I mistaken?).

    you’re kind of wrong. not in the first two dimensions of variation. but they are closer in lower dimensions.

  • Maju

    @8 Pavlova:

    Don’t think the yellow component as “Altaic” but as “Okinawan”, “Japanese” or “NE Asian”.

    Actually the classification of Japanese and Korean under the Altaic linguistic umbrella has been challenged a lot as of late and Wikipedia (reflecting the consensus) rather treats these two languages as isolates. “Micro-Altaic”, gathering Tungusic, Mongol and Turkic is in good shape however.

    What I find most interesting about the yellow and light green components is that there is a clear barrier between Okinawa and Taiwan, so by this way there was no much migration in at least a long time. The flows seem to have gone through mainland China instead.

    “It’s interesting to note that Okinawans show some Papuan admixture”…

    Affinity can be admixture or just remainder from ancient common origin or even an statistical fluke. This last is specially likely when the bands are so narrow as the one you mention: it is probably just “noise”.

    “… the lack of actual Papuan samples”…

    You are right that they are insular Melanesian (of “Papuan” language but “Papuan” is a catch-all family that only indicates “non-Austronesian” in fact). However island Melanesian and “true Papuans” are closely related in all studies (not identical but the closest to each other on Earth.

    And of course I agree with Razib that they have no particular genetic relation with South Asians, being closest to East Asians instead.

  • Pavlova

    “In broad outlines, I think that you have it right, although it is hard to match the colors in the phylogenetic tree to those in the K=14 analysis. They don’t seem to be a match as they seem, e.g., to implausibly suggest as large Tai-Kadai ancestry in the Japanese and Han Chinese, and there are simply more colors in the K=14 chart than in the phylogenetic tree.”

    There are more colors in the K=14 chart than the legend indicates, however if you look at the PCA, the Malaysian Negrito circles are colored maroon. On the K=14 chart, the Malaysian Negrito cluster is also maroon. The same can be implied for the Phillipine Negritos, who are colored purple both on the PCA and their cluster on the K=14 chart. The authors should have just listed all the colors on the legend to avoid any confusion.

    Also I’m pretty sure that the “Tai-Kadai” legend on the K=14 chart is actually “Sino-Tibetan”. Could be a mistake in the legend or did the authors really mean it? The Mlabri of Thailand (an Austro-Asiatic population) have all the pink “Sino-Tibetan”, whereas the Han Chinese, Koreans and Japanese have no Sino-Tibetan and have significant “Tai-Kadai”, which seems a bit strange to me.

  • Maju

    The colors of the legend only refer to the colors in the phylogenetic tree, where the mean ONLY ethno-linguistic adscription and nothing else. The authors have tried to have the structure chart (at the right) to use similar colors where possible but you should not let you be mislead by them, there is no legend for the genetics, only for the linguistics.

  • Pavlova

    Mlabri are Austro-Asiatic, but their cluster is all Sino-Tibetan (pink). Obviously this means the authors made a mistake in labeling that cluster as “Sino-Tibetan”.

    I disagree that Papuans are more related to East Asians. Nearly all genetic studies I’ve seen show that Papuans are the closest to South Asians. Note that the Australoids originated from South Asia. Also, Siberians and Native Americans seem to be the furthest away from Papuans out of all non-African populations.

    Among Europeans, Northern Europeans are genetically more related to Papuans than Southern Europeans are to Papuans. The difference in FST between European-Papuan and Northeast Asian-Papuan is negligible. This probably means Papuans split off from the same Eurasian branch as the Europeans and East Eurasians.

  • ohwilleke

    @4 Ortu Kan

    Interesting observations that I take to heart. I am deeply puzzled by the apparent disconnect between the whole genome inferences and the inferences one would make from NRY and mtDNA about the proportions of Ainu v. Yayoi v. Chinese ancestry in the Japanese which seem very much at odds with each other, so I’m not stunned by your suggestions. The point about an apparent lack of Austronesian influence, however, still stands, as does the striking lack of similarity between Philippino Negritos and Malyasian Negritos (particularly in light of later comments to this post).

    Your suggestions seem to disfavor a Ainu-Philippino link intuition, however.

  • Razib Khan

    Nearly all genetic studies I’ve seen show that Papuans are the closest to South Asians.

    which genetic ones? i post a lot of genetic studies here, so i won’t repeat myself. i want you to dig the studies out so i can check this out.

  • Pavlova

    I’m trying to find studies to show that.

    However you did a run before on ADMIXTURE. It’s titled: Eurasia + Mozabites + Papuans
    In decreasing order of genetic affinity to the Papuan cluster:
    South Asian > Southeast Asian > Northeast Asian > European > West Asian > Kalash

    From the other analysis you did about Negritos not being a single population, your FST results showed that Northeast Asians are closer to Melanesians than the SE Asians are. This is strange since SE Asians usually have a higher genetic affinity to Papuans, due to intermixing of the natives of SE Asia (especially Indonesia, Cambodia among others) with Papuan-like populations. I have a suspicion that this Northeast Asian cluster, at a higher K, would split into 2 or more different clusters, with one “Okinawan” cluster and one “Northern Mongoloid” cluster. The Okinawan cluster would have closer affinities with the Papuan cluster, while the Northern Mongoloid cluster would have less affinities. This is because Okinawans are close to full descendants of the Jomon who are, as some say, Australoid.

  • Razib Khan

    I’m trying to find studies to show that.

    find them. i don’t care about the rest of your interpretations. you don’t seem to understand how best to read ADMIXTURE from what i can gather (for example, my negrito run was overloaded with southeast asians, so you shouldn’t be generalizing about non-southeast asian groups to the granularity you are) so i’m curious about your primary literature citations. i know some of it myself, but you made the claim so i’m asking you to substantiate it.

  • Pavlova

    I gave you the evidence already. From your own ADMIXTURE run:

    South Asian cluster is the closest to the Papuan cluster.

    FST values:
    South Asian + Papuan = 0.201
    Southeast Asian + Papuan = 0.214
    Northeast Asian + Papuan = 0.225

    These FST values make much more sense. It is logical that since Australoids originated in South Asia, they are most closely related. And due to gene flow from Papuans to neighboring SE Asians and vice versa, they would have developed affinities to eachother too.

    This Northeast Asian cluster should be different to the cluster in the negrito admixture run. However, this is due to the inclusion of Okinawans who most likely have genetic affinities to Melanesians compared to Koreans or Japanese.

    Even the Southeast Asian + Papuan FST is higher than the South Asian + Papuan FST. So the Northeast Asian cluster in your negrito admixture run does indeed have some element that has higher genetic affinities to Melanesians, which I suspect comes from Okinawans.

    Tishkoff et al. 2009 supplementary data:
    Figure S7: Neighbor-joining tree from pairwise (δμ)
    2 microsatellite genetic distances between populations

    The global phylogenetic tree makes it clear that Papuans are closer to South Asians than to East Asians. In fact, Papuans are closer to Europeans than to East Asians. Amerindians are a branch from East Asians, which is also true with respect to most modern genetic studies. Figure S14 shows an unrooted neighbor-joining tree with a similar interpretation.

    I’m not saying all studies have to have consistent results. I’m just saying your FST values in the negrito admixture run seem illogical and don’t make any sense.

  • Razib Khan

    I gave you the evidence already. From your own ADMIXTURE run:

    i can read my own run better than you. i said that south asians are not closer to melanesians because the “south asian” cluster in that run is not the totality of south asian ancestry. it’s something i labelled as south asian because it was distinctive to south asia, but in the north and west of south asia it can go to less than 50%. do you understand what i’m saying? the fact that you rambled on at length indicates that you’re not getting that i wanted a concise citation so i could check what you are trying to get at. you don’t know as much as me quite obviously, but you can say valuable things, but be spare about it. i’m annoyed that you took this long to point me to a paper and kept rambling on your own interpretations, which don’t add much value (i set up the parameters for these runs so i’m more aware then you how sensitive they can be to population sizes, marker density, and geographical mixes).

    don’t respond to the above, i’ll ban you if you do. further conversation will be on the literature you cited, not on interpretations of my ADMIXTURE runs.

    i will check the tishkoff link now. though please note that microsatellites have been known to differ from SNPs

  • Razib Khan

    Pavlova, also, to be clear i’m not sure you know this (though from the model you outline you implicitly at least do), but south asians are a compound. one component of their ancestry is west eurasian. the other component is south/east eurasian. this is the “australoid” element you’re talking about. this is helping generate the “south asian” component (as a stabilized admix). this element may have distant affinities to the melanesians. but that doesn’t mean that south asians as an aggregate do, because you pool the ancestry together.

    the main issue with the tree you provided is that it’s obviously distorted by low effec. population size and geographic isolation, which means that oceanians and new worlders are always further separated from core eurasians than they should be. i’m going to check up some autosomal runs, and perhaps run ADMIXTURE with the HGDP set myself and visualize that. stay tuned….

  • Razib Khan

    here’s a similar tree btw:

    using snps instead of microsatellites. the lengths are a little different, very close to the boundary. i wonder if denisovan might be pushing the oceanians out so far, and not just isolation/bottlenecks.

    you may be right depending on some of the results. i might do some analyses myself and try to visualize things more clearly re: melanesians vs. everyone else. the HGDP has some good papuan sets.

  • Pavlova

    You’re right about South Asians not being of purely “South Asian cluster”. Also, Melanesians have a non-Papuan source to their genome, most notably East Eurasian. I’m not sure if there are any Papuan samples available who don’t have any significant East Eurasian admixture.

    Anyway, does this mean that Northeast Asians are closer to Papuans than Southeast Asians are?

    I’ve seen some trees (from new and old studies) with East Eurasians being closer to Papuans. But in some other studies (notably newer studies), West Eurasians appear to be closer to Papuans. Could the relationship be like that of Africans? Maybe East and West Eurasians are quite equidistant to Papuans, just like Europeans and Asians are equidistant to Africans?

    One study in 1994 showed that the closest to Australian Aboriginies, genetically, appeared to be the Mongol-Tungus population, then the Japanese, then the Koreans. What is similar about these populations is that they are Northern Mongoloids. I’m not sure if this is even remotely accurate and what type of genetic markers were used.

    Here’s the table:

    Actually a lot of the values don’t make sense and don’t reflect FSTs from modern genetic studies. However the African to non-African FSTs look quite accurate.

  • Razib Khan

    I’m not sure if there are any Papuan samples available who don’t have any significant East Eurasian admixture.

    the HGDP has a papuan sample, where only a few individuals have austronesian ancestry. i will use that in the near future.

    I’ve seen some trees (from new and old studies) with East Eurasians being closer to Papuans. But in some other studies (notably newer studies), West Eurasians appear to be closer to Papuans. Could the relationship be like that of Africans? Maybe East and West Eurasians are quite equidistant to Papuans, just like Europeans and Asians are equidistant to Africans? Or am I missing something here…

    i think this is the problem. different marker sets will give different results because of this problem. as you note we do need unadmixed sources, as austronesian mixture in many melanesians will bias the trees.

    Actually a lot of the values don’t make sense and don’t reflect FSTs from modern genetic studies. However the African to non-African FSTs look quite accurate.

    the power of the old classical marker studies is low to detect a lot of fine-grained differences. they’re good with african vs. non-african cuz that’s the cleanest distinction in the data set. fyi, for robust human phylogenetics i wouldn’t look at anything before 2005. the old stuff isn’t wrong always, but for fine scale stuff the SNP-chip studies are the best.

    i will do a follow up a post on the issue of papuans you raised. i’m not sure things are quite cut & dried in any direction to be honest, though part of the issue is probably denisovan admixture, which is pulling the melanesians “away” from continental eurasians in general.

    thanks for taking the time to look at the literature. i’m kind of aware of some of these confusions re: melanesians, and i wanted to you to see that too before move on.


Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!

Gene Expression

This blog is about evolution, genetics, genomics and their interstices. Please beware that comments are aggressively moderated. Uncivil or churlish comments will likely get you banned immediately, so make any contribution count!

About Razib Khan

I have degrees in biology and biochemistry, a passion for genetics, history, and philosophy, and shrimp is my favorite food. In relation to nationality I'm a American Northwesterner, in politics I'm a reactionary, and as for religion I have none (I'm an atheist). If you want to know more, see the links at


See More


RSS Razib’s Pinboard

Edifying books

Collapse bottom bar

Login to your Account

E-mail address:
Remember me
Forgot your password?
No problem. Click here to have it e-mailed to you.

Not Registered Yet?

Register now for FREE. Registration only takes a few minutes to complete. Register now »