Wednesday, February 08, 2012

Michael Lynch on Adaptationism

I've been studying Michael Lynch's book The origins of Genome Architecture. It's a marvelous book, I wish everyone interested in evolution could read it and understand it.

The last chapter is very interesting. Lynch talks about the importance of understanding modern population genetics.
... I will comment on the current state of affairs in evolutionary biology, particularly the perception of softness in the field that has been encouraged by the propagation of evolutionary ideas by those with few intentions of being confined by the constraints of prior knowledge.
He also talks about adaptationism/panselectionism and about evolutionary-developmental biology. I'll get to evo-devo in another post [Michael Lynch on Evo-Devo] but here are some choice words about adapationism.
Despite the tremendous theoretical and physical resources now available, the field of evolutionary biology continues to be widely perceived as a soft science. Here I am referring not to the problems associated with those pushing the view that life was created by an intelligent designer, but to a more significant internal issue: a subset of academics who consider themselves strong advocates of evolution but who see no compelling reason to probe the substantial knowledge base of the field. Although this is a heavy charge, it is easy to document. For example, in his 2001 presidential address to the Society for the Study of Evolution, Nick Barton presented a survey that demonstrated that about half of the recent literature devoted to evolutionary issues is far removed from mainstream evolutionary biology.

With the possible exception of behavior, evolutionary biology is treated unlike any other science. Philosophers, sociologists, and ethicists expound on the central role of evolutionary theory in understanding our place in the world. Physicists excited about biocomplexity and computer scientists enamored with genetic algorithms promise a bold new understanding of evolution, and similar claims are made in the emerging field of evolutionary psychology (and its derivatives in political science, economics, and even the humanities). Numerous popularizers of evolution, some with careers focused on defending the teaching of evolution in public schools, are entirely satisfied that a blind adherence to the Darwinian concept of natural selection is a license for such activities. A commonality among all these groups is the near-absence of an appreciation of the most fundamental principles of evolution. Unfortunately, this list extends deep within the life sciences.


... the uncritical acceptance of natural selection as an explanatory force for all aspects of biodiversity (without any direct evidence) is not much different than invoking an intelligent designer (without any direct evidence). True, we have actually seen natural selection in action in a number of well-documented cases of phenotypic evolution (Endler 1986; Kingsolver et al. 2001), but it is a leap to assume that selection accounts for all evolutionary change, particularly at the molecular and cellular levels. The blind worship of natural selection is not evolutionary biology. It is arguably not even science. Natural selection is just one of several evolutionary mechanisms, and the failure to realize this is probably the most significant impediment to a fruitful integration of evolutionary theory with molecular, cellular, and developmental biology.

It should be emphasized here that the sins of panselectionism are by no means restricted to developmental biology, but simply follow the tradition embraced by many areas of evolutionary biology itself, including paleontology and evolutionary ecology (as cogently articulated by Gould and Lewontin in 1979). The vast majority of evolutionary biologists studying morphological, physiological, and or behavioral traits almost always interpret the results in terms of adaptive mechanisms, and they are so convinced of the validity of this approach that virtually no attention is given to the null hypothesis of neutral evolution, despite the availability of methods to do so (Lande 1976; Lynch and Hill 1986; Lynch 1994). For example, in a substantial series of books addressed to the general public, Dawkins (e,g., 1976, 1986, 1996, 2004) has deftly explained a bewildering array of observations in terms of hypothetical selection scenarios. Dawkins's effort to spread the gospel of the awesome power of natural selection has been quite successful, but it has come at the expense of reference to any other mechanisms, and because more people have probably read Dawkins than Darwin, his words have in some ways been profoundly misleading. To his credit, Gould, who is also widely read by the general public, frequently railed against adaptive storytelling, but it can be difficult to understand what alternative mechanisms of evolution Gould had in mind.
I agree with everything Lynch writes except that I have a pretty good idea what alternative mechanisms Gould proposed.


  1. That second-to-last paragraph is ripe for quote-mining. Expect to see a "Evolutionist admits ID is just as well-supported as NS" post at UD in 3...2...1...

    (Not criticizing Lynch here -- just wanted to get my prediction in early ;-) ).

    1. You're really good at thinking like a creationist. Is that because you're an engineer? :-)

    2. Joke noted, but seriously: it's 15 years of that permanently warped my thinking. I now have this little friend I call my Inner Creationist who can be relied upon to respond on cue.

    3. That second-to-last paragraph is ripe for quote-mining. Expect to see a "Evolutionist admits ID is just as well-supported as NS" post at UD in 3...2...1...

      ...and here it is - !

    4. Michael Behe has been jumping all over the work of Michael Lynch and Joe Thornton, as well as the multicellular yeast work from Ratcliff and Travisano. He argues that because all of this work demonstrates that "Darwinists" can't show a trait that can be created by natural selection; instead it has been gene duplications and loss-of-function mutations. It's interesting how such work can be twisted in support of intelligent design (although Behe never states that this is his goal in these blog posts).

  2. The irony of Lynch's statement is that it is similar to John Gillespie's statement about folks uncritically accepting the neutral theory without ever entertaining other models. He talks about most geneticists being contented to demonstrate the effects of drift "on the back of an envelope" (the whole k = u, neutral theory equation).

    In fact, Gillespie has proposed some studies that explain nucleotide and amino acid evolution from a selectionist point of view, but since the math is hard and the neutral theory is seductively easy to demonstrate, it leaves the some geneticists to be contented that they "know" the truth: that most molecular evolution is neutral, even though equally plausible selectionist models are out there.

    1. Are you (or Gillespie) willing to defend the proposition that most base substitutions in the human lineage have been fixed by natural selection?

      It requires proof that most of our genome isn't junk. Do you have any evidence to support such a silly idea?

      And what about the approximate molecular clock that we see in all molecular phylogenies? Do you have a selectionist explanation for that?

    2. Proof positive of what I'm talking about: If you already "know" that alternative models are silly, why bother looking into them? But if you want to move past your certitude, you can read all about Gillespie's models in his 1991 book, including fluctuating selection models and an episodic clock. And please note that Lynch isn't talking humans, nor is Gillespie, nor am I.

      But natural selection has certainly played a role in establishing junk DNA (including in humans) by selecting for mobile genetic elements to nefariously intersperse w/in our genome; then they can be rendered functionless by additional mutations resulting in Junk DNA. This doesn't sound like drift to me. Are you reading Lynch's book in reverse? He discusses this issue in Chapter 7 (and page 57 in Lynch shows that mobile genetic elements constitute about 44% of the human genome). There are also adaptive models of gene conversion that render one of the duplicates to be junk DNA.

      It's possible we're talking past each other but I view selection as playing a role in shaping non-functional aspects of genomes, including the human genome.

  3. I wonder what advice Lynch would give Dawkins to get up to speed. Read Futuyma's text?, start reading Trends in Ecology/Evolution cover to cover?

    1. Dawkins is very familiar with the criticism. He claims that fixation of nearly neutral alleles isn't very interesting and hardly even counts as evolution. He also claims that (almost) every visible phenotype is subject to selection.

      What that means is that practically everything he writes about is due to selection, according to his view of life. I think he's wrong. To me, life does not have the overwhelming appearance of design that he sees.

  4. Unfortunately Lynch is just repeating the vague and hackneyed accusations of 'panselectionism' that Gould and Lewontin initiated decades ago. Granted, it is not always clear what people like Dawkins are implying so far as the ubiquity of selection goes, but in this regard it is helpful to keep in view a clear set of distinctions made by the philosopher of biology Peter Godfrey-Smith. He has outlined three forms -- more precisely continua -- of adaptationism: 'empirical adaptationism', 'methodological adaptationism', and finally 'explanatory adaptationism'.

    'Empirical adaptationism' roughly holds that all traits are adaptations (though since all three distinctions run on a continuum, it is important to bear in mind that the extent to which selection has directly sculpted the traits of organisms is ultimately a matter of degree). Rather than assume that empirical adaptationism is true or false -- or more accurately guess the point along the continuum which is on average the most accurate depiction for life in general (or any given species) -- the merit of empirical adaptationism is to be assessed (unsurprisingly) empirically.

    'Methodological adaptationism' is the view that evolutionary biologists should proceed under the useful assumption that all traits are adaptations. This of course is not intended to be a conclusion granted on a priori grounds; rather it's to be seen as a useful methodological heuristic for guiding investigation. By treating all traits as if they were adaptations, it will assist in accomplishing at least two chief aims: 1) allowing researchers to systematically test all adaptationist hypotheses before concluding that a non-adaptationist hypothesis is the more likely explanation; and 2) allows researchers to discover adaptations that they would otherwise have missed had they not been operating under such a heuristic guide -- that is, actively searching for adaptations. Indeed, this latter point is vital by dint of the fact that nature often crafts designs whose details deviate in significant and counterintuitive ways from the manner in which the prototypical human engineer devises and implements their artifactual analogues. (Incidentally this last point is related to Leslie Orgel's reminder -- his 'second rule' -- that "evolution is cleverer than you are.")

    Finally, 'explanatory adaptationism' is particularly what writers like Dawkins are most interested in. This distinction points to the complex, intricate, functionally co-ordinated, co-evolved components, processes, and design features which have been selected for specific purposes (e.g., the eye). These are the phenomena that explanatory adaptationists are especially keen to highlight. These are the biological phenomena that are as cosmically improbable to arise by chance as is a Boeing 747 being formed by the proverbial tornado sweeping through a junkyard. Note also that explanatory adaptationism is logically independent of empirical adaptationism; that is, making salient the former does not entail anything about the latter. In other words, it remains possible that not all phenotypic traits may be adaptations, even given the special emphasis on explanatory adaptationism. Explanatory adaptationism gains its uniqueness by its ontological significance. It is for these phenomena that natural selection is the only available explanation (even if phenomena such as self-organization, inter alia, figure in various important senses as the 'order' with which selection co-opts 'for free' in forging its designs, to borrow Stuart Kauffman's nice phrase).

    Unfortunately, many people glaze over these above details and distinctions, which Lynch has done, and which is unhelpful.

    1. "Empirical adaptationism" is stupid and conflicts with known facts.

      "Methodological adaptationism" is illogical as a null hypothesis. You simply cannot rule out every single adaptationist just-so story so a scientist following this methodology will never consider non-adaptive explanations. That's exactly the point that Gould, Lewontin, Lynch, and many other very smart people, are saying. Drift is the null hypothesis. If it's consistent with the data, then the onus is on adapationists to demonstrate that an additional step (natural selection) provides a better explanation.

      "Explanatory adaptationism" is useless. If something really is an adaptation (e.g. the eye) then the adaptive explanation is true and doesn't need an excuse. If some phenotypic traits are not adaptations then it makes no sense to rule out non-adaptive explanations for those traits and "explanatory adaptationism" doesn't apply.

      You say that there's a class of observations where "natural selection is the only available explanation" but that's exactly where the problem arises. Adaptationists may believe that there's no other explanation but, as Michale Lynch points out, their lack of knowledge and lack of imagination is not his problem. Lynch provides many examples of how the proper application of population genetics can show that the adaptationist assumption is wrong.

    2. Firstly, I find it troubling that you use the disparaging term “just-so story.” There's no need for such things in scientific discourse. (A big reason why Gould and Lewontin did not convince many practicing biologists was precisely because of their rhetorical tact.) Secondly, non-adaptationist explanations are not the default explanation -- they require evidence no differently than adaptationist hypotheses. Remember that I said views like 'empirical adaptationism' run on continua; and neither does it entail the view that all traits are optimal, mind you. Calling empirical adaptationism “stupid” misses the point; the answer lies somewhere on the continuum, so setting up a false dichotomy doesn't help matters.

      'Explanatory adaptationism' is about emphasis: It claims that the most interesting biological phenomena are adaptations. I'd invite you to offer a purely non-selectionist explanation for, say, the bat's wings. Lynch's work is more interesting at the genomic level, and even there there are issues that need to be settled.

    3. And I would invite you, Nietzsche, to offer a purely adaptive explanation of gRNA-based RNA editing (

      Is it just me, or does the pan-adaptionist crowd tend to be biased towards focusing on the handful of obvious, macro-scale traits in animal/plants? Biology has many more interesting things than wings up its sleeve.

  5. To explain evolutionary events by natural selection just make sexier Stories than e.g. genetic drift that wouldn't render much of an exciting narrative. In addition, direct links between alleged causes and effects are just easier to depict and they are so appealing to the human brain that we just want to beleave them. If two researchers would send papers based on the same results to Nature, Science or Cell, one stating that natural selection caused fixation and the other presenting genetic drift as the cause, which story do you think they would the publish?

    1. Which explanation do you think is sexier?

      1. Almost all of the DNA in our genome is under selective pressure and has a function of some sort even though we have no idea what that could be.

      2. Based on the best available evidence, most of our genome is junk.

  6. "it is a leap to assume that selection accounts for all evolutionary change, particularly at the molecular and cellular levels."

    How about a little fire, scarecrow?
    Does Lynch offer even a single example of an actual biologist who makes this leap?
    It's fucking tiresome.

    "The vast majority of evolutionary biologists studying morphological, physiological, and or behavioral traits almost always interpret the results in terms of adaptive mechanisms, and they are so convinced of the validity of this approach that virtually no attention is given to the null hypothesis of neutral evolution, despite the availability of methods to do so (Lande 1976; Lynch and Hill 1986; Lynch 1994)."

    a) note that molecular and cellular traits are notably missing from this claim.
    b) the last part is simply bullshit. Every statistical test for selection in natural populations has as the statistical null hypothesis the absence of a relationship between phenotypic variation and (some estimate of) fitness. All of them.

    When are you going to review this book?

  7. Larry, since your recent post on the top 5 scientists I have followed a few of your links and ended up reading Evolution by Accident. I'm an undergraduate; not in Biology but Anthropological Science(and Geology) with a heavy lean on BioAnth, from that confession I can assume you know where I'm soon about to head.

    I think I'm a little confused on your exact stance, so hopefully I can try and understand it without attacking a straw man.

    When it comes to the evolution of humans, is your opinion that the major morphological (bipedalism, encephalization, etc) changes are due to natural selection but made possible because of random mutations? I can easily see how chance plays a large role in evolution, one doesn't need to look further than mass extinctions, but does it also depend on the time scale of the evolution that you are interpreting? For example: without the last mass extinction I think it's reasonable to say primates wouldn't have found the niche they did and evolved as well as they have but that the evolution of their morphological features were mainly selected upon by natural selection but owed this particular selection due to their underlying genes that largely came about as a result of chance due to the mass extinction, other genetic drift and such?

    Of course, that is an over-generalization of the 65my of primate evolution with much genetic drift taking place in that time but when it comes down to studying the period over which we evolved obligate bipedalism is natural selection not the more important mechanism to study? One would naturally acknowledge the random mutations that led to the pre-bipedal ancestral species but down play it compared to natural selection?

    I take it from reading your 'What Causes Speciation" paragraph that I have kind of reflected your thoughts but with a human analogy; that the environmental changes that led to the necessity for selected adaptations of our ancestors came about as a result of chance but the adaptations themselves were mainly due to natural selection. "The path of evolution is influenced by the frequency and randomness of mutations", so even when there are environmental selective pressures, the actual mutations that end up being beneficial for an individual come about as a result of chance and the probability that they end up becoming fixed in a population depends more on their frequency than selective pressure? Confusion aside, I find it difficult to side with Dawkins; assuming I've interpreted your essay correctly.

    By the way, would you describe evolutionary stasis as a lack of random mutations that have high enough probabilities of being fixed in a population?

    Sorry about this ramble, just trying to piece things together in my head. Guess it's about time I take some biology courses.

  8. While I can appreciate many of Lynch's criticisms of the field, this comment -- "the sins of panselectionism are by no means restricted to developmental biology" -- is baffling. Panselectionism is not particularly rife in db; he mentioned West-Eberhard in your evo-devo post, and if you've read anything by her, you'd know that selection is the least of the mechanisms she discusses.

    1. I believe that adaptationism is rife in the field of developmental biology. For example, one of the dominant themes is evolvability and that's not a reference to the fact that nearly neutral alleles can be fixed by drift.

      I'm not familiar with West-Eberhard so I looked up a few of her papers. Here's one: West-Eberhard, M.J. (2005) Phenotypic accommodation: adaptive innovation due to developmental plasticity. J Exp Zool B Mol Dev Evol. 304:610-8.

      Phenotypic accommodation is adaptive adjustment, without genetic change, of variable aspects of the phenotype following a novel input during development. Phenotypic accommodation can facilitate the evolution of novel morphology by alleviating the negative effects of change, and by giving a head start to adaptive evolution in a new direction. Whether induced by a mutation or a novel environmental factor, innovative morphological form comes from ancestral developmental responses, not from the novel inducing factor itself. Phenotypic accommodation is the result of adaptive developmental responses, so the novel morphologies that result are not "random" variants, but to some degree reflect past functionality. Phenotypic accommodation is the first step in a process of Darwinian adaptive evolution, or evolution by natural selection, where fitness differences among genetically variable developmental variants cause phenotype-frequency change due to gene-frequency change.

      This seems to be consistent with her emphasis on the evolution of developmental plasticity. It it atypical?

  9. West-Eberhard believes in the almost wondrous powers of developmental plasticity to get almost the best animal for any particular situation: not pan-selectionism but pan-adaptionism.

    Phenotypic accommodation is adaptive adjustment, without genetic change, of variable aspects of the phenotype following a novel input during development.
    I can't understand how this would be possible, at least without previous selection, and then the input would not be 'novel'.
    Phenotypic plasticity (what might be the same as developmental plasticity, depending upon author) is perfectly able to evolve, however.