evolgen

Whither Adaptation?

In a post at the Panda’s Thumb, Ian Musgrave cites this paper by Bakewell et al claiming that 154 genes out of 13,888 surveyed show evidence for adaptive evolution in humans since the divergence with chimps (this is the “chimps more evolved than humans” paper). Ian brings this up in a discussion of Haldane’s dilemma — which is only a dilemma to creationists (biologists are more interested in his sieve and his rule) — but he cites the paper as an authority on the amount of adaptive evolution in humans. A discussion of adaptive evolution in human genes would be incomplete without mentioning other estimates, such as 35 genes out of 13,731 evolved adaptively and 304 out of 11,624. So I present these estimates to round out Ian’s post.

But the real reason I’m bringing this up is in response to this post from Larry Moran. Larry takes a hard-line anti-adaptationist stance, summing up his position thusly:

There seems to be little doubt that most of the fixed alleles at the molecular level are probably neutral in their effect. Thus, they have been fixed by random genetic drift. This includes many amino acid substitutions in proteins. Even though these substitutions change the structure of a protein by a small amount, it does not seem reasonable to assume that they have all been selected.

Depending on the genome, many, if not most, nucleotide changes were fixed independently of natural selection. Additionally, some amino acid changes have also been fixed by drift. “Some” can range from the majority of amino acid changes (as seen in humans, as shown above) to very few (as seen in Drosophila, where 95% of amino acid changes appear to have been fixed by selection between two species). In his attempt to criticize the adaptationist position, Larry creates a caricature of the anti-adaptationist who demands evidence for natural selection without knowing what that evidence would look like.

Larry’s post critiques the adaptationist who finds evidence of selection in everything, from a single amino acid change in a protein to the horns of a rhinoceros. But, as p-ter points out in the comments to Larry’s post, that’s not how population geneticists detect selection. Conveniently, I have previously gone through some of the more common techniques we use to detect natural selection using molecular data (this post is the last in the series and contains links to the previous posts). In a nutshell, we employ many different statistical tests (which one we chose depends on the structure of our data and what assumptions we can make) to detect selection.

The modern selectionist does not invoke adatationist explanations for every evolutionary change. But he does not, by default, say “drift did it”, either. Instead, he requires evidence for whichever conclusion he reaches. Maybe he’s not really a selectionist; maybe he’s just a population geneticist who understands how to detect selection. It’s quite possible that there are still people who point to every evolutionary change as an adaptation, but I don’t know them.

At this point, you should be saying, “Neutrality is the null hypothesis! How on earth can we hope to prove the null?” It’s okay if you weren’t saying it . . . I’ll let you off easy this time. Yes, tests for detecting natural selection revolve around rejecting neutrality, but support for neutrality can be found if we rule out all selective and demographic explanations. And what about those demographic explanations? It’s quite common to invoke changes in population size, population structure, or other complex demographic scenarios when explaining molecular data. But, without evidence, those explanations are just as frail as adaptationist stories without an appropriate test.

Where are we now? Well, Larry does have a point that adaptationist explanations can get a bit silly. But the modern canon of evolutionary theory allows us to test for evidence of natural selection. Larry’s anti-adaptationist stance requires one to believe that detecting natural selection comes down to creating just-so stories (dare I say, straw man?). Instead, there are rigorous tests for selection that any adaptationist must employ to be taken seriously in the scientific community. Perhaps the two horned and one horned rhinoceroses differentiated by purely chance processes, but without a detailed look at the evidence, we cannot conclude either neutrality or selection were predominantly responsible for the differences in horns. Neither drift nor selection are appropriate a priori assumptions; even though neutrality is the null hypothesis for statistical tests, other scenarios must be ruled out before reaching an entirely neutral conclusion.


Bakewell MA, Shi P, and Zhang J. 2007. More genes underwent positive selection in chimpanzee evolution than in human evolution. Proc Natl Acad Sci USA 104:7489-7494 doi:10.1073/pnas.0701705104

Bustamante CD, Fledel-Alon A, Williamson S, Nielsen R, Hubisz MT, et al. 2005. Natural selection on protein-coding genes in the human genome. Nature 437:1153-1157 doi:10.1038/nature04240

Nielsen R, Bustamante C, Clark AG, Glanowski S, Sackton TB, et al. 2005. A scan for positively selected genes in the genomes of humans and chimpanzees. PLoS Biol 3:e170 doi:10.1371/journal.pbio.0030170

Sawyer SA, Parsch J, Zhang Z, Hartl DL. 2007. Prevalence of positive selection among nearly neutral amino acid replacements in Drosophila. Proc Natl Acad Sci USA 104:6504-6510 doi:10.1073/pnas.0701572104

Comments

  1. #1 CCP
    July 3, 2007

    Agreed.
    Unfortunately, evidence for (or against) adaptation is exptremely difficult for the phenotypic characters of interest to those of us who study organisms instead of molecules.

  2. #2 RM
    July 3, 2007

    Yes,
    but luckily we can still either attempt to measure correlates of fitness acting on natural variation in our trait of interest, or perform some nifty phenotypic engineering, modify our trait of interest and ask how that changes fitness. This won’t neccesarily tell us that a trait is an adaptation per se but it will tell us if selective significantly affects it.

  3. #3 Torbjörn Larsson, OM
    July 3, 2007

    Thank you, this was incredibly useful! You answer the question I posed at Larry’s blog, if it hasn’t been answered there. I didn’t realize yesterday that we were conflating the distinction between null hypotheses and a priori assumptions, but it is obvious know. (Um, a priori seems here to be the usual “don’t know”, in the specific case – though I guess one can play a tenuous probability game as well, with drift being likely perhaps).

    And I would probably benefit from digging into your reference posts at some leisure time – the posts on polymorphisms/divergence synonymous/non-synonymous site tests was readable even for this layman.

  4. #4 p-ter
    July 3, 2007

    dunno what the general consensus is on that bakewell paper, but it certainly didn’t show what it claimed to. claims about the absolute number of genes under selection in a lineage are dubious, at best (though perhaps necessary). Naive applications of a single test to compare the absolute numbers of genes under selection in different lineages should be a criminal offense.

    in the supplementary material she even admits they have more power to detect selection in chimps, which utterly discredits her main conclusion (i.e. the title). you were at smbe–she get any questions about it?

  5. #5 RPM
    July 3, 2007

    I saw the talk, and she did get questions, but I can’t really remember any of them. I wasn’t all that impressed by the paper either, but I’m too apathetic to pay close attention to the “how many genes under selection in humans” debate. She didn’t win the Fitch prize, for what that’s worth. I do know that I’m dubious of any claims of selection based solely on dn/ds.

  6. #6 Larry Moran
    July 4, 2007

    RPM says,

    Larry’s post critiques the adaptationist who finds evidence of selection in everything, from a single amino acid change in a protein to the horns of a rhinoceros. But, as p-ter points out in the comments to Larry’s post, that’s not how population geneticists detect selection.

    Duh! What I’m objecting to is the assumption that most visible phenotypic changes must be adaptations. There are plenty of people over on Sandwalk who seem to share this silly point of view. I’m well aware of how population geneticists try to detect natural selection and I’ve been posting about it since 1990.

    Notice that I said “try to detect.” This qualifier was meant as a caution. Just because the adaptationists say they’ve detected natural selection does not make it true. Some of the so-called evidence is not worth the paper it’s printed on. I’m astonished that some of you will accept everything in the scientific literature as long as it confirms your bias. What ever happened to skepticism?

    The modern selectionist does not invoke adatationist explanations for every evolutionary change. But he does not, by default, say “drift did it”, either. Instead, he requires evidence for whichever conclusion he reaches. Maybe he’s not really a selectionist; maybe he’s just a population geneticist who understands how to detect selection. It’s quite possible that there are still people who point to every evolutionary change as an adaptation, but I don’t know them.

    Most evolutionary biologists have conceded that the majority of evolution at the molecular level is due to random genetic drift. However, there are still many who attribute the majority of visible phenotypic changes to natural selection. Richard Dawkins is the most obvious example.

    Where are we now? Well, Larry does have a point that adaptationist explanations can get a bit silly. But the modern canon of evolutionary theory allows us to test for evidence of natural selection. Larry’s anti-adaptationist stance requires one to believe that detecting natural selection comes down to creating just-so stories (dare I say, straw man?).

    That is a silly strawman version of what I’m saying. I’ve posted dozens of articles about evidence for natural selection. I’m well aware of the evidence that’s required to demonstrate adaptation. I’m also well aware of the lack of evidence for many just-so stories.

    What I’m saying is that visible phenotypic changes can also become fixed by random genetic drift. Thus, even at the morphological level it may be true that most evolution is due to random genetic drift and not natural selection. This does not mean that natural selection is unimportant and it does not mean that there’s no solid evidence for selection.

    This is a debate about the relative importance of random genetic drift and natural selection in evolution. In particular, it’s a debate about their relative importance at the level of visible phenotypes since the pluralists have already won the battle at the molecular level. Please try and keep up.

  7. #7 p-ter
    July 5, 2007

    This is a debate about the relative importance of random genetic drift and natural selection in evolution. In particular, it’s a debate about their relative importance at the level of visible phenotypes since the pluralists have already won the battle at the molecular level. Please try and keep up.

    ahahahaha. Larry Moran, everyone. he’ll be here all week.

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