Gene Flow III

This is a table of the distribution of edar370a, the allele that plays a big role in making Northeast Asians look the way they do.  I’ve talked about it before.

Note that the allele frequency is generally zero in African populations, and is over 90% in the Han. It’s also pretty common in pure-blooded Amerindians, which means it’s  fairly old, since they came over from Asia 15-20 k years ago. It’s been favored by selection, in some unclear way, which can easily drive an allele from one copy to the high frequencies we see today in China, given maybe 30,000 years.

Most alleles don’t vary enormously in frequency between different human populations, because those populations have a common origin  – maybe 70,000 years ago for people outside of sub-Saharan Africa, maybe 150-250 k years ago for the various groups in sub-Saharan Africa.  But this one does, and there are others, like DARC ( near 100% in Africa, unknown in China).

You can find populations all the way into Europe ( east and north) and India that have low, non-zero levels of EDAR370a. So there was some gene flow between such groups over the past few tens of thousands of  years, but not a lot.  How much gene flow was there between more distant groups ( over the past thirty thousand years or so) , say between China and the Khoisan? Or between China and Spain, China and England, China and Italy?




This entry was posted in Uncategorized. Bookmark the permalink.

77 Responses to Gene Flow III

  1. Steven E. Sailer says:

    Were still waiting for a theory of how some Andaman Islander-like genes got to the Amazon.

  2. JamesH says:

    Very interesting article!
    So, EDAR is really old I think about 40000 years or so per I recent paper I read (can’t remember which one atm), I suppose if gene flow was high, it would have time to reach Africa or Southern Europe.
    And gene flow was most likely low, but I am still wondering about the ancient DNA results that find East Asian admixture for example in some ancient Europeans (ANE and EHG). Do you know how valid these are?
    Furthermore, you have mentioned before that Fst values indicate a degree of isolation but is it not problematic to infer gene flow from Fst? (

    • gcochran9 says:

      Not that problematic. If Fst is , for example, 0.15, gene flow has been very very low for a long time. Certainly not high enough to interfere with local adaptation.

      • JamesH says:

        Gene flow from Europe to Africa was extremely low to zero, yes but East Asia to Europe is a different story I think. Fst is about 0.1 CEU-CHB.
        Lastly, could there not be gene flow from Europe to Africa? I think there was quite a lot of it in East African populations.

        • gcochran9 says:

          From the Middle East to sub-Saharan Africa, yes. Mainly East Africa.

          • JamesH says:

            Well, was there not some into the Yoruba too, per the most recent HGDP 929 genomes paper? . I think it showed them as 5% West Eurasian or something. But I am not sure if it has any impact on heredity issues.
            And funnily enough, while there was some (low/moderate?) gene flow between East Asians and Europeans, they seem to have similar IQ distributions (also GWASes are more easily transferable between them IIRC).

    • Zero Gene Flow? says:

      Zero gene flow is a weird claim for Greg to want to make:

      1) Petr et al 2019 ( suggests that there is a low and constant gene flow between Eurasians and Africans in order to explain the apparent illusion of negative selection against Neanderthal. (Other methods agree that West Africans have recent geneflow from Eurasia –

      2) As you refer, Fu et al 2016 seems reasonably certain of gene flow between either Holocene West Eurasians and East Asians (relative to Upper Paleolithic Europeans), then at least a third population that forms a clade with East Asians.

      I assume Greg is also well aware of the results finding Levant Neolithic farmer ancestry in the Khoi San, to boot. (Hence restricting himself to China and the Khoisan… but if Near Easterners and East Asians have geneflow, as Fu et al 2016 suggest). I’d actually assume Greg should be aware of all the above results.

      • JamesH says:

        We know that gene flow was from Europe to Africa that Petr et al,2019 are talking about, and not too significant in any case.
        As to Fu et al,2016 yes that is the paper I am referring to as well, there is some gene flow between West Eurasians and East Asians but given an Fst of 0.1 but I am not sure it was substantial.I wish Greg would clarify this a bit.
        It can also be common gene flow from a ghost population.
        As for the Khoisan gene flow, it is about 2% and seems to have been a pulse one-way admixture to them.

      • gcochran9 says:

        I know all that. But I also know that in the West African & Bushman samples we have, N = 100-200 per sample, we don’t see a single copy of EDAR370A.Maybe it’s there at a lower, unobservable frequency, say 1 in 10,000 or 1 in 100,000. For many practical purposes ( not all) that’s close enough to zero for government work.

        Some idiots are trying to argue that there’s oodles of gene flow between all populations, which is why those populations can’t become different. Except that they are different, which ought to put a stopper on their capers, but idiots never say die. It is true that ongoing high levels of gene flow could prevent adaptive differences from emerging – but that takes ~1% gene flow per generation. We know that the total amount of gene flow from East Asia to Africa or Western Europe never reached 1% in thirty thousand years ( around 1000 generations): so we’re talking levels of gene flow thousands to tens of thousands of times lower than the amount required to ensure sameness.

        There are even more sensitive tests of gene flow. Even a tiny flow of an adaptive allele could eventually result in a high frequency, since the frequency of an adaptive allele grows exponentially with time. So, for example, even a few imported copies of the sickle-cell mutation could eventually lead to a high frequency in a pop exposed to lots of malaria. And HbS has spread – originating in West Africa, it’s spread over a lot f Africa, and you can find some in places like Sicily or Greece. India even: but it never went any farther. There is plenty of malaria in Southeast Asia and New Guinea. and there are defensive mutations – but they’re all different from the ones in Africa and the Middle East.

        Next: let’s suppose that (fairly recently) a few copies of EDAR370A waltzed into the Middle East, say with Turks. Not many: and there’s a further vast dilution if we’re talking transmission into sub-Saharan Africa.

        • JamesH says:

          Do you think that the HbS allele in India could be due to the Siddi ? ( -though you probably have heard of them already)
          As for East Asian flow, it is most likely 0 in most of Africa and either really low or close to 0 in some places in the Eastern part.

          But, Fu,2016 has found evidence of it in Europe though which is a bit confusing given that you would expect Europeans to be more similar to East Asians if admixture was that widespread.

        • Zero Gene Flow says:

          Sure. If you state “zero” though, expect your precision oriented audience that knows ll this stuff to respond as if you had said “zero” rather than “Actually closer to 0.01% than 1% (1 migrant in 10000 per generation, not 1 migrant in 100), and punctuated, and thus not inhibiting local adaptation.”.

          Rather than inferring that you are arguing against some non-specified idiots who imagine implausibly high levels of gene flow (and who presumably appear at genetics conferences and pronounce that all the work mapping local variants in diverse populations is actually an enormous waste of time).

          • JamesH says:

            In some cases it has been 0 or extremely close to that. What is your point?

            • Zero Gene Flow says:

              Accuracy? To a degree that is not abnormal among intelligent people who discuss this topic seriously, and that Greg may save some time if he puts what he already knows up front rather than simplifications which play to the dumber people in his audience. I’m not sure what’s up with your oddly hostile response here.

              • JamesH says:

                To be entirely honest, I misread your reply. It’s just that I have heard a lot of really odd statements about ancient gene flow between populations like East Asians and Africans (in the geographic sense) hence my less-than polite reply.
                Apologies for that, I agree with you.

    • Think Fst, Malkovich! says:

      Regarding Fst, Native Mexican groups show Fst as follows:
      “Seri and Lacandon show the highest level of population differentiation as measured with Wright’s fixation index FST (0.136, Fig. 1B and table S4), higher than the FST between Europeans and Chinese populations in HapMap3 (0.11)” (“The genetics of Mexico recapitulates Native American substructure and affects biomedical traits”)

      FST doesn’t tell you straightforwardly much immediate about time. These Native Mexican populations are unlikely to be differentiated too deep in time.

      • JamesH says:

        That result could be due to a variety of factors, I would not take it too seriously.

        • Think Fst. says:

          Well, it clearly results from very high degrees of genetic drift in a short space of time. But the main point is that because this can happen, Fst on its own does not tell you very much about the time depth for which two groups are evolving separately. Obviously.

  3. You’re making an assumption there — you’re assuming that EDAR wasn’t selected against in European populations.

    I recall reading a study years ago about some hunter-gatherers from Sweden who had significant levels of the EDAR gene. It struck me as quite odd. As I see it there are two possibilities: 1. This particular band of hunters was racially mixed (had some east Asian ancestry), but died out completely without leaving any descendants. 2. There were descendants of these hunter gatherers in Sweden, but none of the descendants had the EDAR gene. In both of these possibilities it seems like EDAR is selected against in European populations. (I’ll have to read that study again… I’ll see if I can find it).

    Perhaps there’s some gene in European and African populations, but not in east Asian and American Indian populations which interferes with EDAR in some way. I believe this is called “outbreeding depression”.

    • JamesH says:

      These hunter-gatherers are exactly the ones that I am asking about, I think they did have some East Asian ancestry via Ancient North Eurasians.
      But I do not think there is any evidence of selection against the EDAR gene in Europeans , also given that we do find it in some Europeans today. I think Scandinavians can be modeled about 2-3% East Asian (Finns are a different story).

    • Greying Wanderer says:

      might Europeans have a gene which duplicates something the EDAR gene does but in a way somehow better suited to the region?

    • JRM says:

      Smaller breasts were definitely selected against.

    • Woodpecker says:

      Makes sense. Comments in the previous EDAR thread suggested the advantage was due to lipids in breast milk. Presumably there are also costs. If cow milk consumption negates the advantages of EDAR we would see it deselected in Europe.

      • gcochran9 says:

        I don’t think anyone knows what the advantage is. Complicated by the fact that every time we check, we find something new that it does. It strikes me as very odd that a mutation that changes so many things could be adaptive.

      • Bob says:

        The problem with this explanation is that breast milk and cow milk aren’t substitutes. Breast milk is for babies in general, and cow milk is consumed after weaning and into and throughout adulthood by lactose tolerant populations.

  4. Josh says:

    How far was can we say there was zero gene flow from China over 30k years?

    • JamesH says:

      I doubt there was any south of Europe and west of the Himalayas (with the exception of the Mongolian invasions).
      And it also seems there was more gene flow from East Asia than to East Asia (or Europe)

    • Josh says:

      That should have said “how far west”.

  5. Alex says:

    We do have Tatars and other Asian looking people in Romania. The Khanates were nearby. I also have a friend with the slanted eyes look whose Chinese great grandfather jumped ship in the Danubian principalities from a Russian ship.

    Nothing really systemic though.

  6. Jim says:

    African-Americans have a few percent, presumably admixture from Amerindians. European-Americans are zero.

  7. catte says:

    How much gene flow was there between more distant groups ( over the past thirty thousand years or so) , say between China and the Khoisan? Or between China and Spain, China and England, China and Italy? Zero.

    China and Africa, sure I can believe that. But surely Italy would have gotten a little bit of second-hand gene flow due to steppe invasions into Europe?

    • gcochran9 says:

      If any, well under 1%. There are other alleles that are fairly common to very common in East Asia that you don’t see that far West, like the ALDH2*504Lys allele.

    • JamesH says:

      It depends on which steppe invasions you are talking about. Huns and related peoples did not really leave a genetic mark, and they were largely West Eurasian too.
      The Steppe Yamnaya people I do not think they were too East Asian, perhaps 4% or so?

  8. Eponymous says:

    Question: in Greg’s old post on edar (linked) he quotes a study claiming it is about 30k years old, had a 7% selective advantage, and was at 80% in the Han. But based on my calculations, that level of advantage should have gone to fixation long ago. So what gives? Surely it didn’t confer a 7% advantage that whole time! Or am I doing the math wrong?

    • David Chamberlin says:

      Cochran says in this post it is at 90% in the Han. Selection advantage moves quickly from 20 to 80 percent but slows down progressively after that. But even so I think you are right, it cannot have remained a 7% advantage continuously or it would be far closer to fixation.

      I am guessing such a powerful advantage that has not yet been discovered is increased resistance to some common disease in the rice growing area.

  9. Cpluskx says:

    On another note, Edar is almost non-existent in Turkey (2018 samples) meaning that invading Central Asian Turkics were heavily Caucasoid.

    • Jim says:

      The genetic impact of invading Turks in Anatolia was minor.

      • J says:

        And those Turks, like the Huns, were strongly Caucasian. Even the Easternmost steppe populations, like the Uyghurs, are fifty/fifty.

        • JamesH says:

          I think some of the Uyghurs are more than 50% European/West Eurasian, but I could be wrong. In any case, yes many were more Caucasian than East Asian.

        • Difference Maker says:

          The modern Uighurs are created from a mix of Tang era Uighur Turks and the Indo European Silk Road population – we even have murals from that period – so yes, in the Tang era they were originally East Asian.

          In their sojourn over Iran and the Mideast of course they mixed with the local populations – but keep in mind this is sometimes only a century after the Tang, and so many early sources describe them as East Asian

          • Difference Maker says:

            Besides, it’s not so shameful. Steppe military cannot be discounted, and Tang had a great empire for a reason. I do love my pet topics; humor me and know that even at Talas it seems the Chinese had the upper hand until Turkic intervention.

            It seems the Muslim sources themselves suggest the Tang had the advantage in both the shootout and the melee. I must smugly take this as evidence of the usefulness of those old style crossbows as well as what the Muslims attributed as superior Tang armor.

            Now, it could be that having prevailed over both the heavy Roman infantry and the heavy Sassanian cataphracts in recent memory the Muslims decided they didn’t need all that much armor, and though there are tradeoffs against chasing the Turks vs resisting their arrows

          • JamesH says:

            During the Tang period yes, they were probably already heavily East Asian, given that East Asian ancestry in the region is already evident since Tocharian times. There was a lot of back and forth there it seems, and even Mongolians have a lot of European-like ancestry and carry Ydna R.

    • Difference Maker says:

      Are you trying to excuse conquest by Asiatics? 😉

      Arabs had difficulty distinguishing Turks and Tibetans. Now, Tibetans are not the most representative East Asian group, but they are East Asian

      The expansionist Turks did start in the far west of Mongolia around the Altai on contrast to their Siberian branches, and so were on the frontier of contact between east and west

      • Cpluskx says:

        Arabs also said Khazars were gingers and Cumans were blonde/blue eyed guys. Do you have a quantitative argument?

        • J says:

          I never met a Khazar but met lots of Cumans. They are called Kun in Hungary (their dogs are called Komondor) and they look like the rest of the Hungarians. There are many blonde/blue eyes among them. It is understandable that for a totally black haired people like the Arabs, the Cumans appeared blonde. In Peru, Italians are considered blonde. Everything is relative.

        • Difference Maker says:

          The Arabs also said the Turks were pan faced and small eyed. Do you know that the Wusun were a redhaired blue/green eyed people? Surely you know of the Tarim basin mummies. The steppe is a big highway. In fact, the Greeks said Scythians were blond and white skinned

          So for the Khazars and Cumans, sure they didn’t exterminate the native population. Why would they? In fact I have wondered about the seeming contrast with Mongols and other peoples

          Now, we can claim they were already Caucasoid, etc etc but surely you must know about the theory of Indo European origins on the Pontic steppe.

          Interestingly, the name Wusun seems to literally translate to “children of the crow”. I wonder if there is some connection to Nordic crow mythologies

        • Difference Maker says:

          So are you denying the Mongol conquest of Russia? The Arabs said the Turks were pan faced and small eyed. Do you know that the Wusun were a redhaired blue/green eyed people? Surely you know of the Tarim basin mummies. The steppe is a big highway. In fact, the Greeks said Scythians were blond and white skinned

          So for the Khazars and Cumans, sure they didn’t exterminate the native population. Why would they? In fact I have wondered about the seeming contrast with Mongols and other peoples, though even the Mongols stopped massacring after a while. Indeed, see the Tatars today

          Now, we can claim they were already Caucasoid, etc etc but surely you must know about the theory of Indo European origins on the Pontic steppe. They were the first horse people

          Interestingly, the name Wusun seems to literally translate to “children of the crow”. I wonder if there is some connection to Nordic crow mythologies

          • Cpluskx says:

            East Asians have 80% – 90% EDAR. With 0.5% EDAR Turks must be less than 1% East Asian, -IF- you get the avg of two samples (2015-2018), about 3%. Given that Turks have around 15% Central Asian admixture, invading Turks should be 80% Caucasoid – 20% East Asian.
            Earliest Turkic samples are mostly Caucasoid, Yamnaya-Corded Ware descendants becomes Indo-Aryans with farmer admix then experince language shift with minority North East Asian admix.

  10. TWS says:

    Is this in response to the knuckleheads on Unz?

  11. JamesH says:

    Interestingly, this new paper about EDAR just came out.

  12. gcochran9 says:

    It was low enough to allow significant Fst between population: that’s on the order of one guy per generation.

  13. Seth says:

    And yet plenty of Sub-Saharan Africans are smarter than Mr. Cochran.

  14. Dividualist says:

    I’d like to ask a question. When parents of a high genetic distance have kids, what happens? Beyond the obvious averaging of the genes? Something positive like hybrid vigor, or the opposite of that, something negative out of not well compatible genes?

    Like, Whites and East Asians both tend to be high-IQ and succesful. What happens with Half-Asian kids? Any particular strengths or weaknesses compared to both Whites and East Asians?

    In theory, is the an opposite kind of problem to the well known problem of too much inbreeding, a too much outbreeding kind of problem, even if both parents are high-IQ, healthy, strong, and their parents and grandparents too so no regression to the mean kinds of problems?

    • gcochran9 says:

      Mostly they seem to be somewhere in-between. In principle there could be either heterosis or outbreeding depression, but I don’t think there’s really much evidence for either. Someone that was half African would probably have weaker defenses against tropical diseases, which would be a disadvantage in Africa, and might be more prone to rickets in Ireland. But that’s just splitting the difference.

      There’s a European haplotype that seems to do nothing in Europeans but is a heart-disease risk on a generally African genetic background – or so the study claims. We don’t know of many things like that.

      There is a logical extension of these ideas that nobody seems to have thought of. Except me.

    • TB says:

      My kids seem to be smarter than either my wife or myself, but not exceptionally so. Probably just luck, or they caught her strengths and my strengths and missed our weaknesses. Physically I don’t notice any advantages.
      One study I saw suggested half-Asian/Whites have a risk for allergies, and slightly higher risk of miscarriage.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s