Gene Expression

Selection, but for what???

In my post below about a possible locus to look at to explain the normal variation in hair form we see around us a reader asked:

I was once suckered into giving a course on animal ecophysiology (I was told it was basic ecology until after it was too late to back out) which was a traumatic experience as I’ve only taken one university-level animal physiology course in my life. One of the students asked what the advantage is of kinky hair. I wondered if it might be a better insulator against the heat of the sun but that’s just guessing. I also said that perhaps the question should be the other way round – why do people have straight hair. Perhaps to shed rain more readily? I was hoping this post would answer the question. Are there any suggestions?

Well, I didn’t have one. I’ve read the insulation-from-climate thesis before, and I’m skeptical. I think that these are pretty much the sorts of adaptationist tales which stimulated the counter-reaction in the 1970s. There are many ways to detect natural selection in the genome, or at least identify regions which are possible candidates. There are also more classic techniques derived from ecological genetics. But that doesn’t mean we know exactly why a region of the genome was subject to natural selection, or the exact ultimate cause of an observed fitness differential with a natural population. The lactase persistance story is probably one of the best ones we have. LCT is the region of the genome you want to look at if you want to make sure your test for selection actually works in humans (well, at least if you have sample of European humans). Not only that, we have a good idea about the phenotypic consequences of the genetic variation, we can also relate that phenotype to adaptive value, and hook up the function to cultural and historical correlates. Everything is just lined up there for you, it’s too easy.

Unfortunately, not everything is like that. Consider the likely selection around HERC2-OCA2, we know that the variation on this locus is a pretty good indicator of whether you have blue eyes or not. Well, obviously that means blue eyes are selected for, right? (the blue eye causing mutant has been increasing in frequency in Europe a great deal in the last 10,000 years) Perhaps. You could, for example, posit a sexual selection model. Unfortunately, sexual selection models can be introduced to explain almost anything that is physical, and often emerge when there isn’t anything else on the table plausible. Additionally, as I’ve noted, since blue eyes are a recessive trait it seems likely that it would have been at rather low frequencies for a long period (because most copies wouldn’t be expressed) which should have left particular stamp on the genome which I don’t see (in short, I think the haplotype is too long for selection on a recessive trait). This is why I have suggested that lighter skin color is the real adaptation that is driving selection on HER2-OCA2, on this trait the genetic effect does not exhibit recessivity, and so is amenable to the power of selection from the get-go. But there are more details which need to be filled in here. For example, I have been having a difficult time getting specific reflectance values from the researchers who reported the complexion trend; they simply binned their observations into 3 categories (fair, medium and olive). We need to know how large of an effect this particular locus has on skin color, as after all there are around half-a-dozen other major candidate regions, and some of them likely have a far stronger effect. Could a small effect have resulted in the third longest non-rare haplotype in the European genome? Additionally, why was the selection recent? With LCT we know that cattle showed up right before this locus was subject to selection, there’s a pretty obvious causal relation here. But if light skin allows one to synthesize Vitamin D at higher latitudes, why was it not selected for until relatively recently? I have some ideas, but the honest truth is that we aren’t totally sure, and there’s a lot of confusion right now. More questions than answers. Offering a tight adaptive explanation for selection on the region around HER2-OCA2 is contingent upon many parameters which haven’t been locked down yet.

I wanted to focus on blue eyes and HER2-OCA2 because though I don’t think anyone can be totally confident, we have some parts of the picture. We can start the process of hypothesis generation informed by a few plausibilities. On the other hand, for a lot of the genome we don’t have any clue as to why it might show up on these tests which imply selection. And there are many traits, like kinky hair, which don’t have an obvious adaptive explanation right now. Unfortunately there are many gaps in the superstructure into which we would like to slot our neat explanatory frameworks. I suspect that hypotheses will come more easily once we know a lot more. All the biochemical pathways, their dependencies, more history and archaeology, ethnography which might give us a grip of contemporary fitness differentials, and so forth. Until then, we do the best we can. And try not to be too embarrassed to say we have no clue when that’s all we can say.

Comments

  1. #1 Hayden
    February 29, 2008

    Interesting post. The simplest explanation I can think of is that the straight hair phenotype could just be a by-product of normal variation/drift mixed with the genetic bottleneck that happened as we ventured out of Africa. This bottleneck has been used to explain a bunch of different things, like the high frequency of deleterious mutations in Europeans, so surely we can just throw this one into the mix as well.

  2. #2 razib
    February 29, 2008

    many out of africa groups have very curly hair.

  3. #3 Lassi Hippeläinen
    February 29, 2008

    “I have been having a difficult time getting specific reflectance values from the researchers who reported the complexion trend; they simply binned their observations into 3 categories (fair, medium and olive).”

    I can understand that. The skin colour isn’t stable, especially here in Northern Europe, where the amount of sunlight varies so much over the year. Even darkest Africans get paler during the winter.

    Hair colour isn’t stable either. I was born a blonde, but by my teens my hair was medium brown, and now it is slowly turning to transparent. (Or maybe it is falling off – how does baldness relate to these things?)

    The same goes for eyes. My sister is a blue-eyed blonde. Her two sons were born blue-eyed, but after a few months they had the brown eyes of their father.

    Obviously there are many things going on. Maybe the genes only define a tendency of a colour, and there are some other genes that define how quickly the tendency gets activated.

  4. #4 toto
    February 29, 2008

    many out of africa groups have very curly hair.

    Which ones? Except for some “Arabs” (in a wide sense), pretty much everybody I can think of has flat hair. This includes “black” people living in hot climates such as Southern Indians. Aboriginal Australians seem to range from straight to mild curl. Ironically, the only exception I can think of are Melanesians – supposedly the most “derived” group on Earth… Who did I miss?

  5. #5 Sven DiMilo
    February 29, 2008

    When in doubt, invoke sexual selection!
    Problem solved…just so!!

  6. #6 agnostic
    February 29, 2008

    I don’t think recessive vs. dominance leave different signatures in haplotype length. The prediction is that most of these sweeping alleles are dominant, as you point out, but no one knows whether they are or not. And since many labs have worked on this problem, they’d all have to be overlooking the signature you’re talking about.

  7. #7 Hayden
    February 29, 2008

    Aboriginal Australians seem to range from straight to mild curl

    Probably due to their being the result of their being one of the first waves of migration out of Africa (back when all humans had curly hair) and then subsequent mixing with straight-haired european colonials a few thousand years later.

  8. #8 razib
    February 29, 2008

    I can understand that. The skin colour isn’t stable, especially here in Northern Europe, where the amount of sunlight varies so much over the year. Even darkest Africans get paler during the winter.

    they use unexposed regions.

    Which ones?

    yes, melanesians and negritos.

    I don’t think recessive vs. dominance leave different signatures in haplotype length.

    Directional Positive Selection on an Allele of Arbitrary Dominance:

    “As can be seen, both reach 0 faster for smaller h. For example, for these parameters, the means of these statistics 18 kb from the selected site are ∼0 when h = 0.1, but they are still negative 40 kb away for h = 0.9. This finding suggests that, all else being equal, it will be more difficult to detect a selective sweep if the beneficial allele was recessive.”

  9. #9 razib
    February 29, 2008

    Probably due to their being the result of their being one of the first waves of migration out of Africa (back when all humans had curly hair) and then subsequent mixing with straight-haired european colonials a few thousand years later.

    no. there are studies which check abo genetic (not many, but enough). the unmixed ones are unmixed.

  10. #10 agnostic
    February 29, 2008

    I’m talking about finding a signature in the data people are using, where anything with minor allele freq greater than 5% gets looked at, and where the selection coefficients are mostly unknown.

    That paper assumes the allele has fixed. Even then, when selection is weak, degree of dominance doesn’t leave a different signature.

    Maybe there are ways of getting around these things, but it seems like the different labs would’ve already thought of and done something about this, since the results would be very interesting.

  11. #11 razib
    February 29, 2008

    That paper assumes the allele has fixed. Even then, when selection is weak, degree of dominance doesn’t leave a different signature.

    1) selection on OCA2-HERC2 isn’t weak, is it? so that’s an irrelevant point.

    2) the derived OCA2 frequently is pretty high in the european hapmap sample, 0.8 – 0.9 range.

    3) so you don’t think we can make any assessments of the conditional probabilities of such a long haplotype being selected for due to a trait with a low h?

  12. #12 agnostic
    February 29, 2008

    Yeah, I’m probably being too nitpicky about the general case, since you were only talking about a particular allele or two. Partly a reflex after reading about detecting selection every week, and partly to clarify to readers.

  13. #13 razib
    February 29, 2008

    i’ll be looking into this case further, but it seems to me that the expectation is strong enough that at this boundary condition we can make inferences based on conditionals. but i could be wrong. perhaps one of the authors of the cited article will stumble onto this post and comment if i am misconstruing their point….

  14. #14 p-ter
    February 29, 2008

    consider two alleles with a given s and a frequency f. one is recessive, the other dominant. the recessive one has been in the population much longer than the dominant one (on average). thus more time for recombination to break up the haplotype and less signal.

  15. #15 Neziha
    March 1, 2008

    Head lice can only specialize in holding on to one type of hair cross-section (oval for Africans, round otherwise) For instance, in the states African American children generally don’t get head lice. In Africa Europeans don’t.

The site is currently under maintenance and will be back shortly. New comments have been disabled during this time, please check back soon.