What intra- & inter- population genetic variance tells us

By Razib Khan | November 8, 2010 2:30 pm

uyafrThe figure to the left is a composite merged from two different papers. One analyzes the patterns of genetic variation within African Americans, and the other the patterns within the East Turkic ethnic group, the Uyghurs. The bar plots show the ancestral element which is similar to two parent populations which resemble Europeans and Africans or East Asians. Looking at total aggregate ancestral quanta we infer that African Americans are on the order of 15-25% European in ancestry, and 75-85% African. Uyghurs seem to be a composite in even measure of a European-like group, and an East Asian-like group. This makes total sense phenotypically; most African Americans look more African, while Uyghurs seem to exhibit a phenotype on average which spans the middle-range between West and East Eurasians.

Central_Asian_Buddhist_MonkBut we’re clearly missing something when we focus purely on a population level statistic. Each “slice” of the bar plot actually represents an individual. Note the contrast between African Americans and Uyghurs. There is relatively little intra-individual variation among Uyghurs, while there is a great deal of such variation among African Americans. Why? Population geneticists have looked at linkage disequilibrium in both African Americans and Uyghurs, and inferred that the former went through an admixture phase much more recently than the latter. Though you don’t really have to be a population geneticist to have known that about African Americans. The ethnogenesis of the group African Americans as a cultural entity occurred in the period between 1650 and 1850. Genetically they are a compound of African, European, to some extent Native American, ancestry. For the Uyghurs we have thinner textual evidence, but the visual and genetic data point to a “western” Indo-European speaking population in the Tarim basin before the arrival of the Turks sometime in the second half of the first millenium A.D. The assumption is that after the initial admixture event and the absorption of the pre-Turkic substrate there was no population substructure. Over time the two components distributed themselves evenly across the population over a period of 1,000-1,500 years.

From this we can infer that patterns of individual variation within populations, as well as between closely related populations, can tell us a great deal. Today the Dodecad Ancestry Project posted a file with the population ancestries broken down by individuals. Looking at this sort of fine-grained data patterns can jump out based on what you already know. Below is a slide show I created which highlights some patterns of interest.

[zenphotopress album=213 sort=sort_order number=4]

The first slide is confirmation of what we already know, or should expect. The Burusho are a linguistic isolate in the mountains of northern Pakistan. Their lack of inter-individual variation within the population is suggestive of long term isolation, as is common in mountainous regions. The very fact that they speak a linguistic isolate should lead us to expect this, as the flow of culture and genes often correlate. The Sindhi are the dominant Indo-Aryan speaking ethnic group of the lower Indus watershed. Because of their geographic position they have been conquered many times, being under Persian, Arab, and Turkic rule. Genetically they’re very similar to the Burusho, but observe that there are two individuals with substantial West African ancestry. The presence of black Africans in the armies of the Muslims who conquered the subcontinent is well known, and the origin of the Indian Siddi community. Some of the Sindhis also have appreciable ancestral components which are probably derived from Muslims from West Asia, the “Southern European” and “Southwest Asian” ancestral element which the Burusho lack.

418px-Kim_Kardashian_6Next you see a comparison between Assyrians and Armenians. These two groups seem very similar, and both have deep textually attested roots in the Middle East. The Armenians date to the Persian Empire, at least, while the Assyrians are clearly the descendants of the indigenous Semitic population of Mesopotamia before the arrival of the Arabs. In the Muslim period many of them retreated to mountainous areas of northern Iraq, before emigrating to the cities of modern Iraq with the relaxation of their status as marginalized dhimmis. Today the Assyrian community is scattered across the world. The portion which adheres to the Church of the East was nearly totally extirpated from Iraq early in the 20th century, while that which is in union with the Roman Catholic Church, the Chaldeas, is currently leaving Iraq en masse.

But the Armenians are a far different case in terms of their interactions with the rest of the world. They have been present as “middlemen minorities” as far east as Southeast Asia, and north into the Russian Empire, and south into the Muslim world. The most parsimonious explanation for the individuals with Northern European ancestry is that like Kim Kardashian they are products of mixed-marriages, but I wonder if the centuries of the Armenian Diaspora has resulted in a change in the gene frequencies in the Armenian homeland in part because of back-migration. With larger data sets this will be testable, as well as the hypothesis that Diaspora communities are admixed while the Armenians in Armenia proper are not.

The third slide compares Scandinavians, Finns, and Lithuanians. Scandinavia refers to the Germanic speaking lands of Norden. Lithuania has historically been just outside the arc of Nordic influence (in contrast to Estonia and Latvia), so it can serve as a Northern European control. I believe some of the Finnish samples in Dodecad are related, so one shouldn’t make too much of them. But, contrast the relatively constant level of Southern European in the Scandinavian samples, and their variance in the Finnish ones. Inversely, the Finns show relative constancy of the “Northeast Asian” proportion, while the Scandinavians vary, with some lacking it. This is likely evidence of recent population exchange, and cultural switching. Finland was under Swedish rule for most of the past 1,000 years, and there still remains a large ethnic Swedish population in Finland, and an ethnic Finnish population in Sweden. Some families in Finland likely switched from Finnish to Swedish to Finnish within the last 500 years. The Southern European and West Asian elements more prominent in the Scandinavians tend to increase as one goes south in Europe, with the former modal in Sardinia (in fact, Sardinians are nearly fixed for the Southern European component), and the latter more prominent among southeast European groups. Geography may then explain why the Lithuanians have similar amounts of the West Asian, but less of the Southern European.

UygurFinally we compare Turks, Greeks, and Cypriots. The historical ethnography strongly implies that the major component of Anatolian Turkish ancestry is Greek and Armenian. A broad similarity to the Greeks here is rather clear (with an elevated West Asian component probably from the Armenian ancestry). But notice the differences. There is a consistent East and Northeast Asian component of ancestry among Turks which is lacking in the Greeks. Since the origin of the Turks is in what would today be termed Greater Mongolia, this makes sense. What surprised me though is the presence of a South Asian component among the Turks. This is where looking at individual level results yields results; I’d assumed that like the Romanians the South Asian element was due to a few assimilated Roma. That seems unlikely now, it’s too evenly distributed. So what then? I think here looking at the Uyghur plot illuminates this for us. I don’t know what to make of the South Asian component which you can find in the Uyghur, and even to a trace extent, but again consistent, among the Chuvash, who inhabit the South Urals. Some readers have long claimed that some of the West Eurasian Uyghur ancestry was somehow connected to South Asia, and to be honest I’ve kind of seen that in other HGDP bar plots, but ignored it as of secondary importance. The Turkic group to the north and east of the Uyghurs, the Yakut, totally lack it. From what little we know it seems that the Turks pushed west to Europe, the Middle East, and South Asia, via what is today Xinjiang and Kazakhstan. The existence of this South Asian element in the Turks of Anatolia may be because of their sojourn in this region. There were Iranian speaking Indo-Europeans in Xinjiang, and certainly in Central Asia. Additionally we know historically that northwest India was connected to Xinjiang culturally, as some Indians arrived in China after a period of residence in Xinjiang. But instead of an “Out of South Asia” event I think what we may be looking at is part of the old “Ancient North Indian” genetic variation which pushed into South Asia from the north, and was eventually overlain in Central Asia with other components. I had assumed that the South Asian component among the Finns was noise or Kale, but perhaps it could be that.

Then there is Cyprus. Today the island is ethnically divided between Greek Cypriots and Turkish Cypriots. But in the Bronze Age Cyprus seems to have had a civilization with a close connection with the Near East, in particular Egypt. Sometime between the Bronze Age and the Classical Era it became an outpost of Greece. But notice the near total absence of Northern European among the Cypriots. Like the people of Sardinia, but unlike Sicily, Cyprus is relatively far from the Eurasian mainland. So how did Cyprus become Greek? If the Greeks always had a noticeable Northern European component, or at least during the Bronze Age, that would indicate that the Cypriots are a case of cultural diffusion and emulation of a small Greek elite which arrived during the migrations of the Sea Peoples. Or, the Northern European element could be due to admixture with the Slavic peoples who arrived in Greece after the collapse of East Roman frontier in the 6th century. Or it could be a combination of both. In any case, the Cypriots look most like the Syrians genetically, though the Syrians seem to have a lot more trace exogenous components.

There’s a lot more one could say. I invite readers to download the RAR file with the bar plots. I will leave you with one last comparison, without comment:


Image Credit: Tocharian Buddhist monk of European appearance, and Kim Kardashian, by Luke Ford

CATEGORIZED UNDER: Genetics, Genomics, History

Comments (19)

  1. pconroy

    Fascinating stuff!

    I notice that in the Assyrians vs Armenians, the 3 Armenians with the most North European component, also are the only ones to have a sliver of Northeast Asian, and the Armenian with almost 50% North European, also has no Southwest Asian (Semitic) so it may indicate admixture with a population like Russians??

  2. onur

    I notice that in the Assyrians vs Armenians, the 3 Armenians with the most North European component, also are the only ones to have a sliver of Northeast Asian, and the Armenian with almost 50% North European, also has no Southwest Asian (Semitic) so it may indicate admixture with a population like Russians??

    I strongly suspect so, as the same pattern also exists in a few Georgians. Besides, those Armenian and Georgian individuals with the North European component are almost certainly the ones who appear closer to Russians than the bulk of Armenians and Georgians on the PCA plot of Behar et al., thus further confirming your conclusion:


  3. good call. that seems plausible. the correlation of admixture elements is almost as interesting at the specific ones themselves. in part, we shouldn’t assume that these elements represent real distinct ancestral populations.

  4. toto

    Ok, the difference between Basque and non-Basque French is pretty impressive.

    The only origin I can suggest for the SW Asian component is a remnant of the original Neolithic wave from the Levant. It has a clear E-W, N-S gradient over all European populations (except the Basques), while being evenly mixed within each population (ruling out a very recent origin). The Saracen invasions would have left an African component, and would not have affected Northern France (much). The Jewish input presumably wouldn’t suffice (…?) I can’t see a reason why the Celts would carry a strong West Asian component, much less create a South-to-North gradient over Europe.

    That might leave a possible Roman origin, but Latinization of Europe was supposedly cultural rather than demographic. Also, you would again expect a difference between North and South. If anything, the SW Asian component is inversely correlated with “South European” among the French.

  5. VG

    A Google search tells me the Fulani are a west African people. How did they end up with so much European blood?

  6. VG

    It’s very interesting to see how informative this data can be. The main difference between the Burusho and Sindhis seem to be in the presence of northeast and east Asian ancestry among the Burusho, which is absent or miniscule among the Sindhi. Also notice that some of the Sindhi samples have bits of east African ancestry, which supports your theory of the presence of black Africans in the armies of the Muslims who conquered the subcontinent.

  7. A Google search tells me the Fulani are a west African people. How did they end up with so much European blood?

    some of the tuareg probably mixed with the fulani. historically there were also periods when the fringes of the sahel and southern spain were under the same muslim polity. so the connections are pretty plausible. the fulani have the eurasian lactase persistence allele too.

  8. toto, i think there may have been multiple agricultural waves out of the east. i think the basques got wave 1, but not wave 2. the west asian component is very widespread, so i think it hitchhiked with some cultural leapfrog. the finns don’t have much of it. the wave 1 (‘southern european’) may have brought r1b to western europe. though i don’t know really

  9. Matt B.

    There was a pseudoscience book out decades ago called “The Sirius Mystery” by Robert K.G. Temple. He claimed that some ancient Egyptians had migrated west to end up as part of the modern Dogon tribe in Mauritania or Mali. It would be interesting to test that.

  10. Ali

    Fulani with European blood. Perhaps it is the “Moorish” connection. The Islamic presence in Hispanic lands lasted nigh on a millenia. Many Visigoths converted to Islam, some even tend to the majority. In time the Islamic communities heavily intermarried to an extent that the remaining Christian Spaniards called their ethnic kin “Moors” too (see the meaning of blue-blood).
    After the Reconquista & Inquisitions, many Moors fled to North Africa, still wistfully cradling the keys of their homes. To this day, there exist people who claim descent from these Muslim Spaniards.
    Moroccan society is heavily influenced by Sufis. The Tijaniyya Sufi order is known to have spread towards West Africa.

  11. just to be clear: the ‘european’ part of the fulani might not be european. could just be standard north african berber. in the world wide genetic context that’s the same as european at this level of K.

  12. Antonio

    As someone from a recently mixed country like Brazil I wonder how much can be learn about population and political history using such data. Sadly I don’t think that has been done yet or even whether the Brazilian population have been properly sampled.

  13. There are two outlier populations in Northern Cameroon, one Fulani who speak a Niger-Congo language, the other are several nomadic Chadic language speaking groups (including the Hide, Kotoko, Mafa and Masa), that have atypical Y-DNA haplogroups.

    In the Fulani it is a rather heavily admixed share of Y-DNA haplogroup T. Y-DNA haplogroup T is also found in Upper Egypt, Sudan, the Danubian basin, Mesopotamia and parts of Dravidian language speaking India with concentrations highest among the linguistically hypothesized Proto-Dravidian homeland in India.

    About 8% of Fulani mtDNA has Eurasian affinities (I don’t know if this is concentrated in North Cameroon or not).

    In the other it is a less heavily admixed share of the Y-DNA haplogroup R1b1c (aka M335). Y-DNA haplogroup R1b1c is most closely related to a Y-DNA haplogroup are found in the Coptic Sudanese and another Chadic language speaking population (P-25). A relatively distantly related variant of Y-DNA haplogroup R1b is the Atlantic modal haplogroup, and the African variation is not found in North Africa. The Chadic-speaking groups of northern Cameroon share more similarities in mtDNA with the populations of the Upper and Middle Nile Valley and East Africa than with populations from Central Africa.

    There is not continuity of either group anywhere close in North Africa. Both groups are just over the hill in the Lake Chad basin from the source of the White Nile River. A plausible hypothesis is that these groups are descendants of two separate waves of post-Neolithic era migration down the White Nile until they hit the end of the road in Cameroon and called it home. Given that the vast majority of members of the ethnicities and language groups to which each group of Cameroonian outliers belong are typically Subsaharan African from a population genetics perspective, it is safe to assume in the case of the Fulani that this is a case of an introgressing population adopting membership in a local tribe and taking on its culture and language. The presence of strong Y-DNA and mtDNA lineages in the Chadic speaking groups, and their Afro-Asiatic affiliated language makes their case less clear — they could be the source group for Chadic culture rather than the recipient of it.

  14. Antonio

    Thanks Razib! Your blog has been a major source of ideas and new knowledge and interests for me! I just think that many very very interesting things are coming from genetics but they are still very much under appreciated in other fields. I think you once mentioned that in a post on the Monkey Cage blog – and you’re totally right. Part of it is due to academic specialization, but a great deal of it come from the fact that many people in “social sciences” don’t actually have much of a scientific training or approach – nor they are interested in developing it – so that the conversation across those fields became very difficult. And one of the nicest aspects of your blog to me is the combination of science and more traditional social science knowledge.

  15. no worries. as long as readers don’t behave like assholes i’m very happy to answer questions when i can. though i do apologize to people who i don’t get to via comments, email, etc.

  16. Williams
  17. ss

    A similar study of genetic variation among the various indian caste populations may give an idea about intercaste mobility in India…..


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 http://www.razib.com


See More


RSS Razib’s Pinboard

Edifying books

Collapse bottom bar