Genetic Convection

Advantageous gene variants certainly spread by people marrying someone from the next village, but they can also be spread by population movements – settling virgin lands, farmers pushing aside foragers, and of course conquest, our old friend. If a Fisher wave is diffusion, genetic transport because of those population movements could be compared to convection, as our commenter RCB has suggested.

People are sometimes interested in estimating the point of origin of a sweeping allele: this is probably effectively impossible even if diffusion were the only spread mechanism, since the selective advantage might well vary in both time and space. But that’s ok, since population movements – genetic convection – are real and very important. This means that the difficulties in estimating the origin of a Fisher wave are totally insignificant, compared to the difficulties of estimating the effects of past colonizations, conquests and Völkerwanderungs. So when Yuval Itan and Mark Thomas estimated that 13,910 T LCT allele originated in central Europe, in the early Neolithic, they didn’t just go wrong because of failing to notice that the same allele is fairly common in northern India: no, their whole notion was unsound in the first place. We’re talking turbulence on steroids. Hari Seldon couldn’t figure this one out from the existing geographic distribution.

Same thing with the EDAR allele: Kamberov et al estimate that it originated in central China, but there’s no reason to think that they’re right. Not least because we don’t even understand what the advantage is.

It looks as if the EDAR mutation was fairly common among the original Amerindians. If so, how rapidly did it spread in the Americas? Just as fast the the place was colonized : deep into South America in a thousand years.

On the other hand, the mere fact that an allele has spread very far is a strong hint that it was carried by population movements, not just a Fisher wave. As we gradually figure out ancient population movements, to a large extent using ancient DNA, we will have a better idea of the origin and trajectory of such mutations. Seeing LCT 13910 in both Europe (especially northern Europe) and India suggests that was it was carried along by the Indo-European expansion. The delf508 Cystic fibrosis mutation (also found in India) probably was as well, and perhaps the common 35delG deafness mutation. On the other hand, SLC24A5 and Factor V Leiden probably came with the Middle Eastern farmers.

Many of these allele look like defenses against infectious disease, or some kind of adjustment to an agricultural diet: you’d expect that they originated among either the EEF farmers or the Indo-Europeans, not the old Mesolithic foragers of Europe.

This entry was posted in Uncategorized. Bookmark the permalink.

46 Responses to Genetic Convection

  1. RCB says:

    A tangent:

    Whenever I try to imagine colonization of the Americas, I can’t help but think of selection on physiological and behavioral traits during the great Cane Toad conquest of Australia (some facts reviewed in this theory paper: Apart from the DRD4 novelty-seeking migration story (which Greg and Henry disputed, I think), I don’t know of much work of this sort in humans. Fun to imagine both cultural and genetic adaptations.

    At a theoretical level, I suspect one could derive something like spatial versions of fundamental selection theorems. e.g., in a spatially expanding population, the rate of change of allele frequencies at the population wave front is predicted by the additive genetic variance in mean dispersal distance (instead of fitness). Or something like that.

  2. MawBTS says:

    Genetic “wind” is going to become much more exaggerated in a world of fast, cheap transport. Any new mutation will find itself in hurricane-level conditions.

  3. Tidings says:

    Are there any genes that have seen a large spread in places like the U.S. where there are people from the entire world? Like a mutation that originally was confined to some part of the world but now has spread throughout different populations in the U.S.?

    • gcochran9 says:

      No. Insufficient time.

      • dearieme says:

        And insufficient intermarriage?

        • gcochran9 says:

          No, just not enough time. Every new allele begins with a single copy, and even when there’s an advantage, we’re almost always talking about an advantage of a few per cent. Suppose that one of the 20,000 Puritans emigrating to the Massachusetts Bay Colony carried an unusually advantageous allele, one that increased fitness by 10%. The effectvie population is, let’s say, 10,000, so the initial gene frequency is 1 in 20,000, 5 x 1–5. By now, assuming that it wasn’t lost by chance, the gene frequency would have increased by a factor of three or four: nothing to write home about. And that’s just among the Puritan genes, which since then been diluted by lots of other immigrations.

          On the other hand, if there was some allele that was already fairly common among the Puritans, but not overwhelmingly so – say a 10% gene frequency – and that for some odd reason had a big advantage in North America, say 10%, it could have increased substantially, say by a factor of two or more over the last 380 years. Of course there’s still massive dilution, with immigration from many sources.

    • setstamov says:

      Er…..DRD4? (In our genes,
      Henry Harpending*† and Gregory Cochran‡, – would have been favoured in both Americas in a low population density environment.

  4. The fourth doorman of the apocalypse says:

    The delf508 Cystic fibrosis mutation (also found in India) probably was as well

    But whatever it was for other defenses might have arisen.

    How does the frequency of delf508 compare with sickle cell?

    • The fourth doorman of the apocalypse says:

      OK, according to some sources on the internet (which surely cannot be doubted) the frequency of Sickle Cell is about 1 in 12 among Africans and the frequency of Cystic Fibrosis is about 1 in 40.

      Do those numbers sound reasonable?

      Seems like there are few ways to deal with malaria …

      • The fourth doorman of the apocalypse says:

        Of course, it could be that Africans (and others) have had to deal with Malaria longer than we have had to deal with whatever it is that delf508 protects against.

        • gcochran9 says:

          The frequency of something like sickle-cell, or the common CF mutation, is limited by its lethal effects in double copy. If the advantage in carriers, s, isn’t too big, the equilibrium gene frequency is approximately s. So with CF, the mutation gives you a 2% fitness advantage and has a frequency of 2%. The frequency rises to 2% and sticks there forever. In a highly malarial environment, sickle-cell (HbS) can be worth more, up to an s of 20% and a correspondingly high equilibrium gene frequency. Not really a matter of time, except for the initial rise in frequency.

  5. a very knowing American says:

    Would you expect “convection” to be unique, or uniquely strong in our species? Human beings often organize themselves at the level of whole tribes (or larger) which may be relatively endogamous, and act (somewhat) like whole reproductively isolated species, when moving into new territory. It seems that in a species without that kind of social packaging, successful migrants would merge with the natives and you’d be back to Fisher and diffusion.

  6. chrisdavies09 says:

    “Seeing LCT 13910 in both Europe (especially northern Europe) and India suggests that was it was carried along by the Indo-European expansion.”

    Maybe so.
    But I don’t think that the Indo-European expansion also reached the Fulani of Mali, where LCT 13910 is found at 37%.

    Or the Cameroon Fulani, where it is found at 39%.

    Click to access Ingram_HG07_LCT-africa.pdf

    Maybe LCT 13910 spread as a result of migrations from West Asia into Europe, South Asia, and Africa.

    • gcochran9 says:

      Probably not. If there had been some in the Middle East, you would never had a different lactose-tolerance mutation spread among the Bedouin. in the same way, we can be sure that very few, probably zero, copies of the European LCT mutation reached south of the Sahara until fairly recently, because the Nilotics and Cushitics developed several of their own.

      Suppose that even 10 copies of the European allele (which is older) had reached the Sudan a few thousand years ago. With that head start, it would likely dominate.

      The Fulani originated relatively recently, a mixture of people from western north Africa (Berbers) and sub-Saharan Africans. The European LCT allele must have come from the north (the Tuareg also have it), but it’s hard to say how it arrived there. I’ve guessed the Vandals, but who can really say?

      Scanning the surrounding haplotype ought to tell you something.

      • Michael says:

        The Fulani have a significant amount of R1 Y-haplogroups. Sounds Indo-European to me.

        • chrisdavies09 says:

          Wrong clade, they are mostly R-V88.

          • Michael says:

            I guess I don’t understand. Are you saying that R1b originated in North-East Africa? Wouldn’t it make more sense that the LCT allele migrated together with the R1b haplogroups into Africa from their origin in Eurasia, but before the other major expansions of the Indo-Europeans?

      • chrisdavies09 says:

        “Suppose that even 10 copies of the European allele (which is older)..”

        According to this paper the Fulani one is slightly older:- [Table 5]
        Age of allele (g ± SD)
        Fulani T−13910 245 (±28)
        Utahans T−13910 223 (±25)
        Finns, East T−13910 217 (±34)
        Basques T−13910 208 (±23)
        Finns, West T−13910 208 (±23)
        Saharaawi T−13910 149 (±74)
        Moroccans T−13910 82 (±41)
        Arabs T−13910 39 (±28)

        Also: “The network analysis would indicate that the European T−13910 and the East African G−13907 LP alleles share a common ancestral allelic background (here labeled as H84) whereas the Arab G−13915 allele has a completely different background allele (labeled here as H107)”

        Perhaps it has an origin in North-East Africa, reaching the Finns via the Caucasus/Urals, and the Basque/British Isles via the Balkans, who knows.

        • gcochran9 says:

          Since all of the -13910 alleles have a common origin, they are exactly the same age. All are older than the other lactose-tolerance alleles found in the Bedouins, Nilotics, and Cushitics.

          • chrisdavies09 says:

            So if the Fulani are believed to have harbored T-13910 for an estimated 245 generations (approx) according to the data, while other T-13910-bearing populations are believed to have harbored it for between 39 and 223 generations (approx.), then this would imply that it probably originated in a population other than Berbers or Europeans that was ancestral to the Fulani.

            • gcochran9 says:

              Look, if you looked at the antiquity of R1b Y chromosomes in Ohio, by the usual method of estimating the time to the most recent ancestor, they’d look just as old as they do in England. That means that R1b could have originated in Columbus, right? In fact, since there are plenty of other R1b lineages in Columbus, from Germans and other western Europeans, the TMRCA will likely be older in Columbus than it is in England.

              If an allele is brought into a population by, say, a single individual, the TMRCA is the time that that individual joined the population. If quite a few guys with that allele join, practically all the variation is preserved, and the TMRCA will be far older than the date of admixture.

              For example, the Fulani didn’t even exist (with their current genetic mix) 100 generations ago.

  7. chrisdavies09 says:

    @ Michael –

    The purported origin of Indo-European languages is either in Anatolia &/or Pontic-Caspian steppe or proximal regions.

    The Fulani language belongs to the Atlantic branch of the Niger-Congo language family, not Indo-European.

    In terms of R1b Y-DNA, while there actually is some R-M269 found in African populations [a clade which is typical of various Indo-European language speaking groups], by far the most common R1b clade found in populations like the Fulani is R-V88 of which there is little or none found in Indo-European language speakers.

    • Michael says:

      I’m sorry. I was using the term “Indo-European” to mean the population in Eurasia where the R1 haplogroups and the LCT alleles first came together (if they didn’t actually arise within the same population). I didn’t mean to imply that these people spoke an Indo-European language at that time, or that every person who is a descendant of any one of them also speaks an Indo-European language today.

  8. chrisdavies09 says:

    @ Greg –

    Thanks for you reply, I take your comments on board.

    When T-13910 is referred to as ‘European’, clearly it is European, given its prevalence and frequency across Europe.

    However in terms of its origin – while I now agree that it is unlikely to be West Asian, I still am not entirely convinced that it is definitely of European origin.

    You point out that the Fulani are of somewhat recent origin, being a combination of Berber and Sub-Saharan African. You suggest that T-13910 in the Fulani must be of European origin, and would only have required a small number of European-derived individuals having an input in order to reach the frequencies seen today.

    If European-derived individuals brought in T-13910 in to the ancestors of the Fulani, why have the present-day European populations included in the data above harbored this allele for so many fewer generations than the mystery European population it allegedly originated from? Are we simply lacking sufficient data from other European population such as Italy and France? How can e be so certain that it definitely originated in Europe as opposed to, say, the Chad Basin thousands of years ago?

    • Michael says:

      A European light skin allele (the A111T mutation in SLC24A5) is also found at a range of 15-30% in tribes of R1b-V88 cattle herders like the Fulani and the Hausa. Isn’t it most reasonable to suppose that this allele was already present among all R1b people before the split between V88 and P297? And isn’t it just as reasonable to assume that the lactase persistence alleles were aready also present at that time?

      Clearly, the R1b people of Africa are descendants of the same group in Eurasia who also gave rise to the Indo-European speaking, R1 possessing, lactose tolerant, light skinned, nomadic pastoralists who spread these traits and lifestyle to huge chunks of Europe, the Middle-East and India.

  9. Matt says:

    Not sure how informative this site is –

    However, it suggests that the main European LP SNP we talk about, 13910T (rs4988235 –, is, in Europe (or West Europe at least), found in a European modal haplotype with another SNP, 22018A (rs182549 – (SNPedia agrees with this assessment of the European haplotypes).

    22018A is an independent predicator of LP in East Asian populations, while 13910T has no presence there and so does not predict LP (i.e. it underpredicts phenotype if you use 13910T as a predictor) – and

    13910T by itself probably does not explain as much about LP as we think, in Europeans – a lot of its power is “borrowed glory” that it is generally also on a haplotype with 22018A.

    Seems like this haplotype should not date to the Indo-Europeans, as Kazakhs who likely have Indo-European ancestry from the steppe don’t seem to have that.

    Also raises the question of whether 13910T by itself does actually explain LP in Indian populations as well as we think. IRC there’s actually a big gap between LP in North India and 13910T.

  10. RCB says:

    More tangent:

    As for the effects of selective migration, consider Mark Twain’s thoughts about Californians in Roughin’ It:
    “It was a splendid population—for all the slow, sleepy, sluggish-brained sloths staid at home—you never find that sort of people among pioneers—you cannot build pioneers out of that sort of material. It was that population that gave to California a name for getting up astounding enterprises and rushing them through with a magnificent dash and daring and a recklessness of cost or consequences, which she bears unto this day—and when she projects a new surprise, the grave world smiles as usual, and says ‘Well, that is California all over.'”

  11. chrisdavies09 says:

    “[…]The T-13910 and A-22018 variants occurred on the same haplotype background, consistent with previous studies of European populations (Figures 2, 3, and 4A).8, 20 and 22 Furthermore, some African populations with the T-13910 and A-22018 variants, such as the Bulala from Chad and the Fulani from Cameroon, shared the same haplotype background with Europeans. However, the Fulani also shared a distinct T-13910/A-22018 microsatellite haplotype with the Arabic Baggara from Cameroon (Figures 4A–4C). Further, we noted that the Mozabite from Algeria had the T-13910 and A-22018 variants on the same haplotype background as did the Palestinians from the Middle East (Figures 2, 3, and 4A–4C).[…]

  12. chrisdavies09 says:

    Interestingly, there is another LP allele, 13913-C, which is found in Africa, Europe, and West Asia [Italy, a Danish person of German origin, Jordan, Saudi Arabia Bedouin, Cameroon, Sudan, and Ethiopia].

  13. n/a says:


    Please clarify. Did you think you were waltzing into a debate about group selection, or are you genuinely confused about a basic population genetics concept?

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