John Wilkins of Evolving Thoughts is currently in Venice, Italy. He has just visited the Basilica San Marco (St. Mark's Basilica) according to What I am doing on my holidays….This visit is significant since the Spandrels of San Marco are famous in evolutionary biology. They are part of the attack on adaptationism launched in 1979 by Gould and Lewontin. This is a paper that every student of evolution should read. Here's an online version: The Spandrels of San Marco and the Panglossian Paradigm: A Critique of the Adaptationist Programme.
John has been struggling with adaptationism for almost fifteen years. When he first began studying evolutionary biology he, like many others, was unaware of the importance of random genetic drift and other anti-adaptationist perspectives. He certainly didn't know that random genetic drift is by far the dominant mechanism of evolution in terms of frequency of allele fixation. Over time John has developed an unusual perspective on adaptionism—one that I don't really understand.
Here's how he explains it in his latest posting ...
This is interesting, I think, in the context of Gould’s and Lewontin’s paper. It shows that claims of things being adaptive or not depend crucially on what one counts as the “task” of a structure. Since I think that everything is subjected to selection pressure at all times (sometimes not enough to overcome the noise of statistical properties), counting what is, and what isn’t, adaptive is a bit of a personal call, in the absence of access to the historical processes of particular traits. I am becoming more of an adaptationist these days.The idea that many alleles might be slightly beneficial or slightly detrimental isn't very controversial. But that's not what John is saying. As I understand him, he's saying there can be no such thing as a truly neutral allele. He seems to be saying that anyone who believes otherwise is making a "personal call." A personal call that he believes is wrong since he thinks (i.e. his personal call) that everything is subject to natural selection.
He's also making a somewhat trivial point that doesn't contribute to the debate, as far as I'm concerned. Many alleles that are slightly beneficial are lost due to random genetic drift and many alleles that are slightly deleterious are fixed by random genetic drift. To me, that says that adaptationism can't explain all of evolutionary biology. To call yourself an adaptationist while knowing that slightly deleterious alleles can be fixed by random genetic drift seems somewhat unsatisfying.
John has a paper in the latest issue of Biology and Philosophy, an issue devoted to Adaptationism. It's not a very enlightening issue, from my perspective. The main problem with adaptationism isn't that it can't explain adaptation and it isn't that some just-so stories are wrong. The main problem is that adaptationists don't even consider any other alternatives to fixation by natural selection. Everything, especially everything with a visible phenotype, is automatically assumed to be adaptive and the arguments proceed from there.
One of the papers I liked was Seven Types of Adaptationism by Tim Lewens (Lewens, 2009). The seven types are:A Empirical adaptationismsThere are problems with all seven forms of adaptationism but the nice thing about Lewens' paper is that he effectively refutes #4, #5, and #7. In the case of methodological adaptationism it's just not true that the default assumption has to be adaptation. Evolutionary biology will be just as productive in the long run if drift is the default assumption and adaptation has to be proven.
1. Pan-selectionism–natural selection is the most significant of the evolutionary forces that act on populations.
2. Good-designism–evolutionary processes tend to result in organisms with suites of well-designed traits. Most lineages are highly evolvable.
3. Gradualism–adaptation is always the result of selection acting on gradual
variation.
B Methodological Adaptationisms
4. Weak heuristic adaptationism–those traits that are adaptations are likely to be correctly recognised as such only if we begin by assuming that all traits are adaptations.
5. Strong heuristic adaptationism–only by beginning to think of traits as adaptations can we uncover their true status, whether they are adaptations or not.
C Disciplinary Adaptationism
6. Explanatory adaptationism–an evolutionary biologist’s proper business is the study of adaptations.
D Epistemological Adaptationism
7. Epistemological optimism–investigators have access to the data that reliably discriminate between conflicting evolutionary hypotheses.
In explanatory adaptationism, the assumption is that all of the interesting parts of evolution are adaptations and fixation of alleles by random genetic drift is so boring that it might as well not even be evolution. This is the stance taken by many adaptationists, like Richard Dawkins. As you might imagine, it doesn't take much effort to refute that kind of argument. One's personal opinion about what's interesting and what's not interesting should not play a role in determining how everyone else should go about studying evolution.
Gould, S.J. and Lewontin, R.C. (1979) The Spandrels of San Marco and the Panglossian paradigm: a critique of the adaptationist programme. Proc. R. Soc. Lond. B 205:581-598.
Lewens, T. (2009) Seven types of adaptionism. Biol. Philos. 24:161–182. [doi: 10.1007/s10539-008-9145-7]
The problem with your position vs adaptation"ism" is that drift is normally the null hypothesis, and selection has to be supported by evidence.
ReplyDeleteAnonymous said...
ReplyDeleteThe problem with your position vs adaptation"ism" is that drift is normally the null hypothesis, and selection has to be supported by evidence.
How is that a problem with Larry's position? That is his position. Adaptationism, in most of the forms Larry just talked about, would reverse that and seem to place selection (adaptations) as the null hypothesis instead of neutrality.
That aside Larry I wanted to add that adaptationism has been one of my biggest beefs with popular writers of evolution such as Dawkins. Adaptations and natural selection might be some of the easiest concepts to convey, and provide some of the more interesting and intuitive examples for lay readers, but it is only showing a small part of the picture. Working in molecular evolution I have been amazed at some of the truly bizarre innovations that various protists and bacteria have come up at the molecular level where all of the data seems to support fixation by drift under a neutral model as opposed to selection. And some of them are very complex.
I would really love for an accessible popular book about evolution that at least touches on drift in a non-trivial manner would be written. There are a ton of unique features (like RNA editing) where adaptationist thinking produced totally backwards hypotheses about evolution and it was only people coming from a neutralist background that put forward explanations that were truly satisfactory and logical. (Such as constructuve neutral evolution)
DG says,
ReplyDeleteI would really love for an accessible popular book about evolution that at least touches on drift in a non-trivial manner would be written.
I'm working on it but, like everything else, progress is slow.
The working title is Evolution by Accident. You can read about the central idea in my essay Evolution by Accident.
-DG,
ReplyDeleteCan you give some specific examples? I'm not arguing with you or anything; I would just love to hear of some examples!
Kele:
ReplyDeleteRNA editing is the one floating around in my head right now since I listened to Dr. Mike Gray give a talk about it just recently. (A Review cam be found here)
Basically RNA editing encompasses several mechanistically diverse processes where the coding sequence of an RNA is edited so that it differs from the genomic coding sequence. In the organisms where this was first discovered it was noticed because genomic sequences often looked like pseudogenes, with premature stop codons for instance. RNA editing machinery fixes these coding mistakes so that a functional protein is produced.
Most adaptationists at the time (I believe) proposed that RNA editing was an adaptation to fix broken genes. But this doesn't make sense. Instead it is far more plausible that RNA editing machinery arose from other molecular processes and their presence meant that formerly deleterious mutations were now neutral because they could be repaired at the RNA level.
the nice thing about Lewen's paper is that he effectively refutes #4, #5, and #7.
ReplyDeleteYeah... He spends exactly 112 words and 625 characters to "effectively refute #4". LOL! And that "refutal" basically accounts to nothing more than stating that it ain't so. Because if something is not verified than it is proven ("if these predictions are not met, we have evidence that
the trait in question confers some fitness advantage over competitors"). Jeez...
Lewens is to be credited for being able to very nicely articulate the problem but he earned nothing but laugh for his childish approach to discount off hand an alternative POW. Even funnier, the way he articulated his approach exposes his blatant disregard to the notion of falsifiability.
Your painting anyone who asks a question of "can and how can it be adaptive?" as evil adaptationist isn't helping your cause either. Lots of those horrible folks have absolutely nothing against viewing something as neutral. Why can't you be reasonable and accept that Lewens' #4 is a perfectly productive way of doing science?
How is that a problem with Larry's position?
ReplyDeleteLarry's position often seems like this: The scientific community is populated by mindless selection-bots, with just a few enlightened ones like himself holding back the howling hoards.
Pete Duunkelberg
Larry’s view of an adaptationist doing science: “Wow, that’s an interesting feature. Let’s assume it’s an adaptation. Done. It’s an adaptation!”
ReplyDeleteAn actual adaptationist doing science: “Wow, that’s an interesting feature. Let’s assume it’s an adaptation. Now if so, then you would expect X, Y, Z, etc. Let’s design experiments to test those predictions...”
Larry’s view of the correct way to do science: “Wow, that’s an interesting feature. Let’s assume it’s due to drift. If so, then it should show no signs of being an adaptation. If it was an adaptation, then you would expect X, Y, Z, etc. Let’s design experiments to test those predictions...”
Divalent,
ReplyDeleteI think you've missed Larry's point:
The main problem with adaptationism isn't that it can't explain adaptation and it isn't that some just-so stories are wrong. The main problem is that adaptationists don't even consider any other alternatives to fixation by natural selection.
In other words, his view of an adaptionist doing science is this:
"Wow, that’s an interesting feature. Let’s assume it’s an adaptation for A. Now if so, then you could expect X, Y, Z, etc. Let’s design experiments to test those predictions. Oops, no, those predictions didn't turn out. OK, then, perhaps it's an adaptation for B, in which case we should expect P, Q, R, etc. Let's design experiments to test those predictions. OK, that's not it either. Maybe it's an adaptation for C....”
Not being an evolutionary biologist, I won't offer an opinion whether he's right, but his criticism is not so trivial as you suggest.
The main problem with adaptationism IMO is that it is a science killer, and not so much for the what-is-your-null-hypothesis? reason (although this is very important and relevant) but for the fact that it removes any quantitative assessment of what's going on from the discussion.
ReplyDeleteEffectively we're going back to the Middle Ages methods of reasoning with it. Once you start talking about actual allele frequencies, population sizes, selective pressures, fixation rate and probabilities, and all that cool stuff, adaptationism is not longer a viable option.
Ultimately it's a matter of education. The vast majority of students only ever hear about selection and they get stuck at that stage of their evolutionary biology education. It's like teaching kids how to add and subtract and letting them think they know mathematics.
However, I can't see that changing easily it is already difficult enough to get students to grasp natural selection and the very fact of evolution, that it few people in a position to change things would consider it a good idea to jump to drift and all that complicated math stuff.
P.S. I always find it deeply ironic that mathematics is the most heavily infested with creationists serious field of study, yet the most math-intensive and quantitatively rigorous area of biology is also the one that gives the bleakest view of evolution from a creationist perspective
Can you give some specific examples? I'm not arguing with you or anything; I would just love to hear of some examples!
ReplyDeleteMost features of the genome are only explainable by the neutral theory of molecular evolution. From the size of the genome and junk DNA (there is a reason why completely asexual amoebas have the biggest genome of all organisms while most bacteria have very compact ones) through splicing, the presence or absence of operon organization, RNA editing, to the excessive complexity of transcriptional regulatory networks, and many many others.
A quite thorough examination of the subject can be found here:
http://www.amazon.com/Origins-Genome-Architecture-Michael-Lynch/dp/0878934847
Not being an evolutionary biologist, I won't offer an opinion whether he's right
ReplyDeleteBeing an evolutionary biologist, I will complete that for you :-). Yes, I have encountered many evolutionary and ecological studies where adaptive solutions where being proposed, without giving any thought to neutral scenarios, such as genetic drift. The problem is even more severe outside of biology. For example, on this blog, Larry has offered many examples from evolutionary psychology that very clearly suffered from this problem. My personal feeling is that, yes, Larry is making a valid point.
Of course I posted that comment especially for Larry. I note that he didn't comment on my argument in the paper he cited of mine from B&P: that drift and adaptation are not contrary cases...
ReplyDeleteBeing en route and on leave, I will not get into this in more detail until I've managed to get home and sleep. In about five weeks...
Georgi Marinov says,
ReplyDeleteUltimately it's a matter of education. The vast majority of students only ever hear about selection and they get stuck at that stage of their evolutionary biology education. It's like teaching kids how to add and subtract and letting them think they know mathematics.
However, I can't see that changing easily it is already difficult enough to get students to grasp natural selection and the very fact of evolution, that it few people in a position to change things would consider it a good idea to jump to drift and all that complicated math stuff.
Unfortunately it's true that we do a very poor job of educating students about evolution. Here at the University of Toronto, the first year biology course (Organisms in Their Environment) is taught by adaptationists. I took a survey last month of 43 student who had completed this course and 77% of them could only identify natural selection as a mechanism of evolution.
When asked to define evolution, 60% used a definition that restricted evolution to adaptation. Only 10% of the students came up with either of the standard textbook definitions: "Change in the frequency or alleles in a population" or "Descent with modification."
I'm about to launch a campaign to change the course but I expect it to be about as useful as tilting at windmills. (One of the advantages of tenure is that I can afford to devote some time to this project without jeopardizing my job.)
The vast majority of my colleagues don't see a problem with the course as it currently exists. The instructor, Spencer Barrett, has won major teaching awards and holds the highest professorial rank in the university (University Professor).
Peter Dunkelberg says,
ReplyDeleteLarry's position often seems like this: The scientific community is populated by mindless selection-bots, with just a few enlightened ones like himself holding back the howling hoards.
By Jove, I think you've got it!
The only difference between your characterization and my actual position is that I'm not so certain that the howling hoards are being held back. :-)
DK says,
ReplyDeleteYeah... He spends exactly 112 words and 625 characters to "effectively refute #4". LOL! And that "refutal" basically accounts to nothing more than stating that it ain't so. Because if something is not verified than it is proven ("if these predictions are not met, we have evidence that the trait in question confers some fitness advantage over competitors"). Jeez...
< br />Jeez, yourself.
Adaptationism #4 is the proposition that, "those traits that are adaptations are likely to be correctly recognised as such only if we begin by assuming that all traits are adaptations."
The refutation is ....
If we understand adaptations as those traits selected for their function, then the existence of such adaptations can be uncovered, in principle at least, from the opening assumption that all traits are selectively neutral. Such an assumption leads to predictions for how traits should fluctuate in a population. If these predictions are not met, we have evidence that the trait in question confers some fitness advantage over competitors. In other words, the opening assumption that all traits are selectively neutral can lead to the identification of adaptations.
There are people who argue that adaptation should be the default position because otherwise we would never recognize an adaptive trait. I think Lewens has an effective counter-argument. He points out that assuming random genetic drift as the default hypothesis does not prevent you from looking for evidence of selection.
Do you really want to defend #4?
John Wilkins says,
ReplyDeleteI note that he didn't comment on my argument in the paper he cited of mine from B&P: that drift and adaptation are not contrary cases...
He's right. I didn't comment on that argument.
And I still think that "no comment" is the most polite way of dealing with the argument. :-)
qetzal: you've missed Larry's point...
ReplyDeleteActually, qetzal, I suspect you missed mine. *Of course* that is the progression.
Let me revise my (tongue-in-cheek) characterization of Larry’s view of the correct way to do science: "Wow, that’s an interesting feature. Let’s assume it’s drift. If it's drift, then it should show no signs of being an adaptation. If it was an adaptation for A, then you could expect X, Y, Z, etc. Let’s design experiments to test those predictions. Oops, no, those predictions didn't turn out. OK, then, perhaps it's an adaptation for B, in which case we should expect P, Q, R, etc. Let's design experiments to test those predictions. OK, that's not it either. Maybe it's an adaptation for C....”
IOW, for phenotypically visible characteristics, regardless of what notion you might have about whether it came about by drift or as an adaptation, your attention must immediately turn to possible adaptive functions, and you don't stop until you reach the limits of your ability to generate plausible adaptive hypotheses.
Taking the stance that it is probably due exclusively to drift unless you can prove its an adaptation might be warranted for point mutations in junk DNA, but for phenotypically visible characteristics, the ratio of features that have clear adaptive value relative to those strongly suspected of being due to drift is exceedingly large. So, from a methodological standpoint, there usually is no value in momentarily postulating that it’s due to drift, since you immediately jump to a consideration of possible adaptive functions.
That’s not to say that their aren’t some instances where drift seems the most plausible at the outset: the loss of the ability to synthesize vitamin C in primates is a very likely candidate. (I.e., the general class consisting of the loss or diminution of functions/structures that are no longer needed, or only weakly beneficial, owing to changes in environment or habit.) But others, such as hairlessness in humans, IMO, is not (but that’s another discussion for another thread).
OTOH, I’m more and more coming to the conclusion that this drift vs adaptation debate is a just a pedagogical diversion. It really is a false dichotomy, because even alleles that provide strong adaptive value “drift” in the sense that the population frequency changes that occur after an allele arises are subject to random fluctuations due to events unrelated to the value of the allele. Thus, it is absolutely correct to say that every allele fixation is due to drift, even ones that provide an very strong selective advantage. What the debate really boils down to in particular instances is the question of what is the selective advantage, if any.
Using the analogy of a drunk randomly staggering on a planer surface, the strength of the fitness advantage (or disadvantage) is represented by the slope of the plane. A positive fitness (negative slope) is no guarantee that the first step won’t be in the uphill direction (causing the drunk to fall off the edge of the plane) nor a negative fitness (positive slope) a guarantee that he won’t make it to the other side; the slope just determines the probability that these events occur.
For a particular population size, a given neutral mutation might have a 1% chance of becoming fixed through pure neutral drift. Suppose a slight fitness advantage increases the probability to 1.5%. If such an allele gets fixed, what do we say? Was it due to drift or to the selective advantage? It seems to me that debating this question is rather pointless. Framing it as “drift vs adaptation” doesn’t really provide any insight.
Divalent wrote:
ReplyDeleteIOW, for phenotypically visible characteristics, regardless of what notion you might have about whether it came about by drift or as an adaptation, your attention must immediately turn to possible adaptive functions, and you don't stop until you reach the limits of your ability to generate plausible adaptive hypotheses.
Really? The only way to show that a character arose by neutral fixation is to rule out all plausible adaptive hypotheses? If so, then your characterization may well be the only experimental approach. (I hope that’s not the case though.)
That still doesn’t address Larry’s overall claim that adaptionists (as a rule) wrongly assume that almost everything is an adaption. IOW, Larry is saying that most adaptionists never reach a point where they agree that all plausible adaptive hypotheses have been excluded, and we should conclude neutral fixation. As a consequence, many things that are properly attributable to neutral fixation are instead wrongly attributed to adaptation.
I think Larry might also dispute this statement from your previous comment:
[F]or phenotypically visible characteristics, the ratio of features that have clear adaptive value relative to those strongly suspected of being due to drift is exceedingly large.
Even if that’s technically true, it’s irrelevant if the possibility of drift is often improperly ignored or rejected.
qetzal: Really?
ReplyDeleteFor all practical purposes, Really.
qetzal: That still doesn’t address Larry’s overall claim that adaptionists (as a rule) wrongly assume that almost everything is an adaption. IOW, Larry is saying that most adaptionists never reach a point where they agree that all plausible adaptive hypotheses have been excluded, and we should conclude neutral fixation. As a consequence, many things that are properly attributable to neutral fixation are instead wrongly attributed to adaptation.
I don't know. It seems to me that most good evolutionary biologists are well aware of drift. It's really kind of a silly argument what one "assumes" at the beginning of an investigation, because that assumption is not proof. BTW, within the set of phenotypically visible characteristics or traits, what are examples of the “many things that are properly attributable to neutral fixation”? (I know Larry occasionally tosses out an example that he thinks is a good candidate for being due to drift, but those are just drifter "just so stories".)
Larry "cheats" a bit in inflating drift’s important by considering any change in the genome (e.g., a base change in junk DNA) as equivalent to any other change as a measure of evolution (and by that metric, Dawkins agrees with Larry that neutral drift is the most common mechanism). But most field biologists are more concerned with changes that have some phenotypic effect. On a numbers basis, we know that vastly more of those things are adaptations rather than the likely the result of drift.
On a numbers basis, we know that vastly more of those things are adaptations rather than the likely the result of drift.
ReplyDeleteWhat are the numbers? Just curious...
Divalent claims,
ReplyDeleteBut most field biologists are more concerned with changes that have some phenotypic effect. On a numbers basis, we know that vastly more of those things are adaptations rather than the likely the result of drift.
We know no such thing. That's adaptationist talk.
Let's look at something simple like the phenotypic variation among humans. We'll restrict ourselves to the variation that's due to underlying genes and therefore subject to evolution.
Walking down the streets of Toronto, I see different shapes of noses, teeth, mouths, eyes, and ears. I see people of differing heights, different skin color, different eye color, different body shapes different hair color, some with less hair than others, and different head shapes.
In fact, people are so different from each other that we have no trouble identifying individuals from their appearance alone. We can even distinguish between members of the same family.
So, Divalent, the question before us is how many of those genetic variations are adaptive? Do you really want to defend the idea that most of them are?
The refutation is ....
ReplyDeleteThe refutation it isn't.
All he says is that the neutral null can, "in principle at least", lead to the identification of adaptations. This claim is neither supported by any reasoning, nor shown to be advantageous over its alternative. That's a far cry from anything remotely resembling refutation.
Do you really want to defend #4?
Sure. First, on a most far-out philosophical level, "the method" is unimportant (see Feyerabend). Second, for as long as one decides to stick to a particular method, the falsifiability principle is among the most important rules. As it stands today, due to the complexity of the subject the neutrality hypothesis is usually less falsifiable than adaptation. Thus, we stand to learn more from trying to come up with crazy adaptationist explanations instead of defaulting to the "no real reason" stance. Third, I invoke emotional argument. Neutrality is boring while adaptations are thought-provoking.
Of course, none of that implies that anyone should not have a neutral drift option firmly in mind.
Larry: “We know no such thing. That's adaptationist talk.”
ReplyDeleteWell, I concede that for most of the phenotypically visible traits that we *could* catalog, we have no idea one way or the other. But there is a *huge* number of things we know are adaptations (eyes ears muscles bones etc etc etc). Give me an hour and I can probably fill pages of them (there are probably scores of them in our eyes alone). OTOH, we don't really know one phenotypically visible trait in humans that was *fixed* by purely neutral drift, although there are several plausible candidates. (Can you name any that we know for reasonably for sure? I keep asking every time this topic comes up, and you never answer this) So, among the fairly substantial pool of things we know one way or the other, it is balanced vastly in favor of adaptations. So maybe it is "adaptationist talk", but it’s also a fact.
Larry: So, Divalent, the question before us is how many of those genetic variations are adaptive?
Good grief! You can’t seriously believe that these features (and the variations you see) are each due to particular alleles, each of which has that function and no other? And that body height, frame type, particular combinations of facial features, eye color etc are not now, and never were, subject to selection of any type? And that none of these differences makes a wit of difference in how that individual could function in the environment they are in now, nor in any of the many different environments that every one of their ancestors found themselves in the past? What makes you think the variation in features you listed are any different from, say, the variation in beak size of finches on the Galapagos Islands?
Pointing to variation in particular features, particularly in a cosmopolitan population, is no evidence at all that these things are maintained, or became fixed, by drift. Of course, that doesn't establish that anything you mentioned absolutely is an adaptation, but there is certainly no evidence *at all* that they merely represent, now and in the past, neutral alleles waxing and waning in the population.
Larry: Do you really want to defend the idea that most of them are [adaptations]?
Sorry, but the burden is on you to defend the idea that most of them are due to drift of neutral alleles, and with something more than an argument from personal incredulity. Again, for features known one way or the other, the score card is: # of adaptations >> # fixed by drift of neutral alleles. So although we don't know now (at least I don't), based on the past, I would bet a lot that these features were, and are, shaped by selection.
Well, I concede that for most of the phenotypically visible traits that we *could* catalog, we have no idea one way or the other. But there is a *huge* number of things we know are adaptations (eyes ears muscles bones etc etc etc). Give me an hour and I can probably fill pages of them (there are probably scores of them in our eyes alone).
ReplyDeleteI don't think anyone who criticizes adaptationism is doing the mistake of going to the other extreme and rejecting selection as a major factor behind evolution. If you do that, the creationists would actually have a point that complexity does not randomly arises out of nowhere.
What people are rightfully worried about is the simplistic assumption that everything we see as trait was selected, which is a very wrong way of approaching these questions. The right way is to consider both drift and selection and then try to come up with a quantitative way of figuring it out, if possible.
If, however, for example, you go out and say that something that has appeared only very recently, and after modern humans stopped being a single population, and is almost clearly a cultural trait, was the result of selection, as has happened many time, is a sign that you have no idea what you're talking about
Divalent says,
ReplyDelete... but for phenotypically visible characteristics, the ratio of features that have clear adaptive value relative to those strongly suspected of being due to drift is exceedingly large.
This is a typical adaptationist statement. Real scientists like to see evidence.
Do you have any?
You still don't get it, do you? The problem with the adaptationist program is that adaptation is often assumed without any evidence of selection or even of adaptive value. In the absence of evidence the favorite ploy of adaptationists is to make up stories that might be plausible.
Your statement above illustrates the worst excesses of the adaptationist program. You have no evidence at all that the majority of phenotypically visible mutations must be subject to selection.
Nevertheless, you and your adaptationist colleagues continue to chant the mantra that visible traits must be adaptive. Why do you do this when time after time your statement is challenged and you can never provide evidence to back it up?
Is that how to do science?
The reason why drift is the null hypothesis is that, if we assume that evolution is a stochastic process (which is a whole other problem for people who have only had a non-technical introduction to evolutionary biology), there are many outcomes for any given set of initial conditions. For instance, if you draw ten beads (with replacement) from an urn containing five red beads and five blue beads, there are 11 different possibilities for the number of red beads drawn (i.e., 0 to 10 red beads). Additionally, even though the probability of drawing a given number of red beads is not uniformly distributed, no single outcome has the majority probability. For instance, in the bead example above, the most probable outcome, when each bead has a equal probability of being drawn, is 5 red beads, with a probability of .2461 (approximately 1 out of 4 times). The second most probable outcome is either 4 red beads or 6 blue beads, both of which happen with a probability of .2051 (approximately 1 out of 5 times). Finally, the third most probable outcome is either 3 red beads or 7 red beads, both of which happen with a probability .1172 (approximately 1 out of 8 times).
ReplyDeleteThe thing to note here is that approximately 3 out of 4 times a sample is drawn the proportion of red beads (and blue beads) to total beads changes. In other words, it is far more likely that a change in frequency is observed, even though there is no bias toward any individual or type of individuals.
Within the context if evolution, this means that you cannot assume a priori that an observed change in frequency has occurred through a bias toward an individual or type of individuals (i.e., through natural selection) because such a change can (and, more importantly does,) occur through the process of sampling a population without bias (i.e., genetic drift). Moreover, drift cannot be eliminated, because it is caused by taking a finite sample, as is always the case in evolution. Drift, therefore, must the null hypothesis when a change in frequency is observed.
Divalent wrote:
ReplyDeleteIOW, for phenotypically visible characteristics, regardless of what notion you might have about whether it came about by drift or as an adaptation, your attention must immediately turn to possible adaptive functions, and you don't stop until you reach the limits of your ability to generate plausible adaptive hypotheses.
A last remark, before the thread stops: this approach only is necessary for fixed differences. Whenever the trait is still polymorph, you have the possiblity of measuring fitness differences between heritable phenotypes. In many instances, it thus becomes possible to identify selection, while being oblivious of the adaptive mechanism. In these situations, it is only natural to pose neutrality as the null hypothesis, and selection as its alternative, and distinguishing between them becomes a statistical issue.
Divalent says,
ReplyDeleteSorry, but the burden is on you to defend the idea that most of them are due to drift of neutral alleles, and with something more than an argument from personal incredulity.
That's ridiculous. I make no such claim.
I'm merely reacting to your specific claim as stated in an earlier comment you made.
You said,
But most field biologists are more concerned with changes that have some phenotypic effect. On a numbers basis, we know that vastly more of those things are adaptations rather than the likely the result of drift.
You are making a positive claim. I am expressing skepticism. The onus is on you to back up your claim with evidence.
What I'm saying is that of all possible visible traits, the number that are known to be adaptive is a very small percentage. This directly contradicts your claim because you say that we "know" that most of them are adaptive.
If we "know" it then you should have no trouble providing the evidence. If you can't do that, then I conclude that it's just adaptationist talk. You're expressing your bias n the guise of scientific knowledge.
Larry: “Divalent said ’... the ratio of features that have clear adaptive value relative to those strongly suspected of being due to drift is exceedingly large.’ This is a typical adaptationist statement. Real scientists like to see evidence.” And later: “You have no evidence at all that the majority of phenotypically visible mutations must be subject to selection.”
ReplyDeleteNo no! I did try to clarify this in my follow up. The pool of features I’m referring to is not *all features* that we can catalog, it the *SUBSET* of phenotypically visible features that we know, to a reasonably confident level, are either adaptations or purely neutral drift. *That* list is overwhelmingly populated with adaptations. My point is merely that, in the absence of evidence either way, the odds it will turn out to be an adaptation. But I am NOT saying that merely because the odds are in its favor that therefore it IS an adaptation.
It seems every time this topic comes up, I ask you for examples of traits in humans that have been shown to likely have resulted from neutral drift. You have NEVER EVER stated one. Instead, you toss out things like epicanthal eye folds, baldness in men, human hairlessness, and now variations in weight, height, head shape, eye color, etc. and imply that these are likely due to neutral drift. But never any evidence to support your gut feeling. You might be right about these things, but they are nonetheless merely guesses without evidence on your part, and IMO you are suggesting these things in a face of reasonably plausible selective hypotheses. If you want to assert that they are due to drift, you have to do more than your gut feeling.
I absolutely agree with you that merely proposing a plausible adaptive hypothesis in no way establishes that some trait is an adaptation, (just as merely proposing that it is due to neutral drift in no way establishes *that* “fact”). Thus, I agree that it is an error to treat that hypothesis as an established conclusion for the purpose of making some other point without some evidence that makes it more likely than not to be true. And that is true of a drift hypothesis as well.
In the absence of evidence at least indicating something as more likely than chance to be correct, the proper position really should be agnosticism. But it *is* the case that the nature of the trait does make the different hypotheses more or less likely. For example, a point mutation in the middle of a broken ERV has a high probability of being due to neutral drift. A point mutation that affects, say, body size clearly will be subject to a variety of selective pressures (e.g., bigger size might mean better defense against predators and mating rivals, but with an ongoing energetic cost; net selection pressure may vary depending on the particular environment: e.g., current population density, relative food availability, etc). Sure, the net selection pressure might be neutral, and remain so over the many generations it will take to fix that allele, but in this case drift is clearly less likely than in the case of the ERV point mutation.
Generally, IMO if your intent is to study the question about a phenotypically visible feature, you have to start with the plausible adaptive explanations (and then test them). Starting with the drift explanations either takes you nowhere, or is just a momentary diversion before you then start considering the possible adaptive explanations.
...Your statement above illustrates the worst excesses of the adaptationist program. You have no evidence at all that the majority of phenotypically visible mutations must be subject to selection. Nevertheless, you and your adaptationist colleagues continue to chant the mantra that visible traits must be adaptive. Why do you do this when time after time your statement is challenged and you can never provide evidence to back it up?
Wow, you slander me! :)
I think it is important to point out again (or was that in the other adaptationism thread) that when we need to differentiate between fixed and polymorphic traits. It is with traits (alleles) that are fixed in a population that we are talking about whether the fixation occurred primarily due to drift ot selection. With polymorphic traits we can, theoretically, given enough population level knowledge directly assess fitness costs/benefits and therefore strength of selection.
ReplyDeleteIts just important to keep in mind that talking about fixed and polymorphic traits in a population are two different things.
Larry Moran said...
ReplyDelete"In fact, people are so different from each other [sic] that we have no trouble identifying individuals from their appearance alone. We can even distinguish between members of the same family."
I hope I'm not committing heresy by invoking group selection (sort of), but if the ability to be identified, ie. distinguished from other members of the population, is positively adaptive, then each/all of these features could have been adaptive in a diverse population.
Here, we might see "racial" features as being fixed by drift (in small founder populations) despite the fact that they were adaptive only along with alternative features in the larger, more diverse, population from which the founders were drawn. (Should the founder effect be completely distinguished from drift, or do most people regard them as fading into one another as I do?)
Consider the situation when any particular feature is positively selected when present in small but significant amounts (due to the adaptive effects of being identifiable), but negatively selected when present in such small amounts as to trigger xenophobia. How would the random activity of drift and/or founder effects be classified in this situation?
Divalent wrote:
ReplyDelete"The pool of features I’m referring to is not *all features* that we can catalog, it the *SUBSET* of phenotypically visible features that we know, to a reasonably confident level, are either adaptations or purely neutral drift. *That* list is overwhelmingly populated with adaptations."
I agree this is very likely true.
"My point is merely that, in the absence of evidence either way, the odds it will turn out to be an adaptation."
That doesn't necessarily follow. If people overwhelmingly look for adaptionist explanations and rarely if ever consider neutral ones, that would very likely bias the list. The fact that most 'proven' explanations are adaptionist might mean that most features are adaptive, or it might mean that we're not doing a good job recognizing the neutrally fixed features.
That said, I too would like to see some examples of fixed traits that are known to have resulted from neutral drift, in humans or any other multicellular organism.
How about it, Dr. Moran? Any good examples you can give us?
As said the problem is demonstrating genetic drift while excluding selection as a factor.
ReplyDeleteFirst example I know of was Aspinwall, 1974
http://www.jstor.org/pss/2407331
As said the problem is demonstrating genetic drift while excluding selection as a factor.
ReplyDeleteWe can argue about that (whether this is a problem or not)
While it would be nice if we were able to know which trait was selected for ans which wasn't, that's clearly impossible because the information that will allow us to do that simply isn't there.
But that's not what the argument is about and on its own it is something we can definitely live without. What is important is understanding the general principles of the process, in this case the appreciation of the different mechanisms driving evolution. To demonstrate that drift plays a major role we don't need to prove drift while disproving evolution, we just need to collect a number of examples of absolutely non-adaptive and even deleterious traits that nevertheless have not been eliminated by selection. And we have a lot of these, transposons being the boring but still the best example. After all, species can go extinct because of transposon accumulation in their genome, and most eukaryotes have developed elaborate small RNA pathways to silence them from the germline. Why would this be the case if selection was all-powerful? One would expect them to be eliminated from the genome. But they're not, and this should be enough to persuade anyone that there is only that much that selection can do and it can not and should not be the default explanation for anything
Georgi Marinov said...
ReplyDelete"[...] transposons being the boring but still the best example. After all, species can go extinct because of transposon accumulation in their genome, and most eukaryotes have developed elaborate small RNA pathways to silence them from the germline. Why would this be the case if selection was all-powerful? One would expect them to be eliminated from the genome."
Perhaps because they perform a valuable (and irreplaceable) function in the creation of new valuable mutations, such that lineages that eliminated transposons didn't evolve and were overrun by those that kept them along with the "elaborate small RNA pathways to silence them". It may also be that these pathways, once developed, performed other useful functions, so that the entire package created a better adapted organism than one without it.
Perhaps because they perform a valuable (and irreplaceable) function in the creation of new valuable mutations, such that lineages that eliminated transposons didn't evolve and were overrun by those that kept them along with the "elaborate small RNA pathways to silence them". It may also be that these pathways, once developed, performed other useful functions, so that the entire package created a better adapted organism than one without it.
ReplyDeleteThere are plenty of species that have eliminated transposons (or never accumulated them), you will have hard time finding mobile elements in bacteria and there are plenty of eukaryotes with very few of them. The common theme is the relation between effective population size and genome size (the expansion of the genome being primarily driven by mobile elements). Completely asexual amoebas have the biggest genomes, with an effective population size of 1.
These observations do not fit your hypothesis.
In my experience this argument, and the people debating, can be split primarily in to two distinct groups. The most ardent adaptationists (and here I don't want to imply that they are all bad, ignore drift, etc. Just the people who lean most heavily towards explaining features via adaptation) come from what I would call the more organismal biology end of evolutionary biology, particularly people who work with multicellular eukaryotes, Metazoa in particular.
ReplyDeleteThose who lean more towards a neutralist default stance tend to come from a more Molecular Evolution or Population Genetics background.
Just an observation. I suspect it is mostly do to training and just what people in those different fields think are interesting.
-DG said:
ReplyDeleteIn my experience this argument, and the people debating, can be split primarily in to two distinct groups. The most ardent adaptationists (and here I don't want to imply that they are all bad, ignore drift, etc. Just the people who lean most heavily towards explaining features via adaptation) come from what I would call the more organismal biology end of evolutionary biology, particularly people who work with multicellular eukaryotes, Metazoa in particular.
Those who lean more towards a neutralist default stance tend to come from a more Molecular Evolution or Population Genetics background.
I've even run into people who will claim that, on the molecular level, most evolution is neutral and therefore is a stochastic process, but that, in some not very well explain and almost magical way, on the phenotypic level, most evolution is selective and therefore a determinisitic process.