Monday, April 23, 2012

How Do Phenotypes Get To Be How They Get To Be? (or) Is Natural Selection Biology's Phlogiston?

Some creatures (and some plants) enjoy more reproductive success than others. This variability in reproductive success determines the allotments of genes that the next generation inherits. Specifically, the next generation inherits more of the genes of the more successful reproducers, and it inherits fewer of the genes of the less successful reproducers. These varying inheritances express themselves as various distributions of phenotypic traits, and that explains how phenotypes get to be how they get to be. (Read for "genes" shorthand for DNA, epigenetic markers and whatever else constitutes the machinery of inheritance.)

So far so good for the theory of natural selection. But the truisms raise a question: Why do some creatures (and some plants) enjoy more reproductive success than others?

The theory of natural selection assumes and asserts that reproductive success is a function of heritable phenotypic traits. According to the theory, variability among the heritable phenotypic traits in a local population causes the members of the population to exhibit variable reproductive success. Heritable phenotypic traits affect reproductive success by interacting with the environments in which their bearers live. A longer neck reaches higher fruits; a sharper eye detects more hidden prey, and so on. This is how traits lead to reproductive success, according to the theory.

But this explanation begs more questions. Which attributes of a creature constitute phenotypic "traits"? And why should the heritable ones be credited with determining reproductive success?

We coin a term, "trait", and identify, say, nose length, as one. We coin a term, "adaptation," and declare that a long (or short) nose is one. Its degree of adaptation relates the trait to "fitness," which determines reproductive success, which is a measure of the given creature's progeny: their number and viability and fertility.

If we measure the noses of a generation of offspring, we might find an over-representation of nose-types associated with certain members of the parental generation. Natural selection theorists would regard this outcome as testifying to the reproductive success of those members of the parental generation, which would testify to their fitness, which would testify to their being adapted, which would testify to their possessing nose lengths within some range. To repeat: Reproductive success is a function of fitness, which is a function of adaptation, which is a function of heritable traits, goes the theory.

The elaboration begs more questions: Which heritable traits are the adaptive ones? We can't say, offhand, because we can't distinguish between adaptive traits and other traits until we measure reproductive success. Once we do that, then we can credit whatever traits are overrepresented in the offspring generation with being adaptive and thereby conferring the fitness that led to the reproductive success of their earlier bearers. But we can credit traits with being adaptive only after reproductive success is measured, if we are to evaluate the theory of natural selection by its own terms.


One can argue that traits contribute to reproductive success without being determinative, but then what becomes of the formula that says reproductive success is a function of fitness, which is a function of adaptation, which is a function heritable phenotypic traits? If that formula fails--if reproductive success is due to something else--then the theory of natural selection goes out the window, and we are left without a theory of how phenotypes get to be how they get to be.

Variability of reproductive success necessarily reflects the interplay of countless variables, heritable phenotypic traits (nature) among them. Nonheritable phenotypic traits (nurture), along with countless environmental contingencies, also will affect reproductive success, positively or negatively. Dumb luck, being in the right place at the right time (or bad luck, being wrongly situated), might have more to do with reproductive success in many cases than having certain of one's phenotypic traits be more or less pronounced that the corresponding traits of stronger, faster, smarter--seemingly more fit--rivals.

For example, if the rivals are north of the river and you're south of the river when the blaze consumes half the forest, or the pack of predators converges upwind of you but downwind of your rivals, or the rivals are sterile for having been caught in the plague as infants, then you might emerge as the more fit, based on your reproductive success, but not due to heritable phenotypic traits. Factors responsible for your reproductive success need have no connection to heritability.

Nonetheless, in these scenarios your genes will be overrepresented in the next generation--all of your genes, whether we can concoct stories about any particular ones contributing to fitness, by playing adaptive roles, or not.

The same thing applies to traits acquired through diet and/or exercise. In the case of acquired traits, no matter how important the traits are in determining reproductive success, they are not heritable. Only their genetic potential is heritable. Identical twins, even in the same environment, will not necessarily enjoy identical reproductive success. Nonheritable effects might rule the day.

The natural selection theorist can counter that the local habitat's variability, typically small and random, will tend to cancel itself out over time, leaving genetic effects to determine reproductive outcomes. But the formula can be read either way: The local population's phenotypic variability, typically small and random, will tend to cancel itself out over time, leaving environmental (nonheritable) effects to determine reproductive outcomes.

The foregoing suggests that natural selection theory be formulated as a problem of signal-to-noise ratio. That is, the burden on the theory is to show that the variability of heritable phenotypic traits within a species in a local population, limited as it is by developmental constraints, nonetheless is significant enough to account for the variability of reproductive success among the members of a generation. Can the variability of the heritable traits in a given generation, the signal, rise above the day-in day-out contingencies of the environment and the intrinsic developmental constraints that limit the variability of phenotypes in a given generation, the noise, to override these factors and determine reproductive outcomes generation after generation?

What magnitude of variability is necessary to tip the scales? A nose-length variability of one cell? A hundred cells? A nanometer? An inch?

In a given habitat, selection pressures will operate above a given magnitude of variability for any given trait. But for any given trait, how does nature determine that threshold of significance? If the determination depends on someone first measuring reproductive success, then we've made no progress beyond the truisms of the first paragraph of this post.

Even computers programmed to simulate evolution use criteria to decide which simulated creatures enjoy which levels of reproductive success. This must be so, if evolution is to occur among the simulated creatures. But nature has no criteria by which to select winners and losers. Fitness? By which criteria is that to be assigned? Adaptation? Ditto. The buck stops at varying reproductive success itself.

As a guiding narrative, natural selection is becoming shopworn. We notice the scent of biological phlogiston. Phenotypes get to be how they get to be by some other means.

6 comments:

  1. Your view of natural selection is entirely nonsensical.

    A couple of points:

    1) "Begging the question" is a logical fallacy. You mean "raise the question".

    2) Reproductive success is fitness. There are various ways to measure this in natural populations (see Natural Selection in the Wild, by John Endler, 1986) - the example you give of noses is not one; "an over-representation of nose-types associated with certain members of the parental generation" does not constitute what biologists would regard as evidence for natural selection.

    3) "In a given habitat, selection pressures will operate above a given magnitude of variability for any given trait. But for any given trait, how does nature determine that threshold of significance?" Here you display your lack of knowledge of evolutionary theory. It is quite well worked out what the magnitude of variability (aka selection coefficients) must be for selection to play a role. To a first approximation, the threshold is one divided by the population size. If the selection coefficient is greater than that, then the population can respond to the selection pressure brought on by the environment. Otherwise the frequency of that trait in the population will drift (another term that you don't seem to know - at least it always comes up in a discussion of the efficacy of natal selection).

    4) "But nature has no criteria by which to select winners and losers." How on Earth can you make that statement? Of course there are criteria! It works like this: If a (heritable) mutation confers a larger nose, and the environment affords individuals with larger noses a significant* reproductive advantage (for example by giving them a better sense of smell), then individuals with this mutation have a higher probability than those without of leaving more offspring in the next generations. That is natural selection.

    * Signifant as described in the third point.

    ReplyDelete
  2. 1) "Begging the question" is a logical fallacy. You mean "raise the question".

    Yes. Thanks. I'll fix that.

    2) Reproductive success is fitness.

    Always? Reproductive success can be due to many things that we probably don't want to ascribe to fitness. For example, if Creature1 gets caught in the tar pit as an infant and dies, and Creature2 luckily takes another path, then C2 ends up with more reproductive success than C1, even if C1's genes would have predisposed it toward being faster, stronger, smarter, or otherwise "more fit." C2 emerges with greater reproductive success, but not necessarily due to any adaptation that enhances fitness. BTW, where does adaptation fit in?

    (2 cont'd) There are various ways to measure this in natural populations (see Natural Selection in the Wild, by John Endler, 1986)

    It's not clear what "this" refers to: Reproductive success or fitness (same thing?) or natural selection?

    (2 cont'd)- the example you give of noses is not one; "an over-representation of nose-types associated with certain members of the parental generation" does not constitute what biologists would regard as evidence for natural selection.

    Even if, as in your point #4, it improves the sense of smell? What distinguishes my nose example (not natural selection) from yours (natural selection)? See my comment on your point #4.


    3) "In a given habitat, selection pressures will operate above a given magnitude of variability for any given trait. But for any given trait, how does nature determine that threshold of significance?" Here you display your lack of knowledge of evolutionary theory. It is quite well worked out what the magnitude of variability (aka selection coefficients) must be for selection to play a role. To a first approximation, the threshold is one divided by the population size. If the selection coefficient is greater than that, then the population can respond to the selection pressure brought on by the environment. Otherwise the frequency of that trait in the population will drift (another term that you don't seem to know - at least it always comes up in a discussion of the efficacy of natal selection).

    This is confusing. You say that for selection to "play a role," the magnitude of variability (aka selection coefficient) must be greater than a threshold, which is 1/ pop size (as a first approximation). How is the value of the selection coefficient determined in the first place, so that it can be compared to the "threshold"?

    4) "But nature has no criteria by which to select winners and losers." How on Earth can you make that statement? Of course there are criteria! It works like this: If a (heritable) mutation confers a larger nose, and the environment affords individuals with larger noses a significant* reproductive advantage (for example by giving them a better sense of smell),

    So now the significance (as described in your point #3) of reproductive advantage is something afforded by the environment? You just said it was a function of the selection coefficient relative to a "threshold".

    4 cont'd) then individuals with this mutation have a higher probability than those without of leaving more offspring in the next generations. That is natural selection.

    Should I understand you to say that if the offspring generation on average has bigger noses than the parent generation, then the designation of drift vs reproductive advantage is based on some statistical convention? Or on someone's opinion as to whether the environment afforded an advantage? Or on some formula that factors in both? Or what?

    ReplyDelete
  3. There is indeed a stochastic component to evolution, but that does not mean fitness and NS are muddled concepts. Yes, the individual that would otherwise have produced more offspring might die from a meteor, but that does not mean we can't talk about the effects of fitness and NS. And adaptation is that which increases fitness in the current environment: a trait that has evolved by means of NS.

    John Endler's book is about measuring NS. And again, yes, reproductive success is the ability to pass on genes to the next generation(s), and this is also called fitness.

    About the nose example: "an over-representation of nose-types associated with certain members of the parental generation" does not constitute what biologists would regard as evidence for natural selection, because it also needs to be shown that this over-representation is not due to stochastic processes (drift) or linkage with some other adaptive trait.

    The selection coefficient is defined as s=w'/w-1, where w is fitness before the mutation, and w' is the fitness with the mutation.

    Yes, whether a mutation has a large or small s depends on the environment. A mutation for a larger nose may be beneficial in an environment where the prey with a distinctive smell, but make no difference in another environment. In the first environment, w' is larger than w, so s>0, but in the second s=0. Whether a non-zero s can be "detected" by natural selection depends then on 1/N.

    I don't really understand your last question. Of course no person's opinion matters these questions. Whether a mutation will drift or be affected by NS depends on how much it affects fitness (again, with s>1/N, there is a chance that it will go to fixation by NS; for s<1/N it can still go to fixation, but in that case it would be by drift).

    ReplyDelete
  4. You refer to a formula that is supposed to distinguish traits due to genetic drift from traits due to natural selection.

    According to the formula, the distinction rests on the value of the selection coefficient (s - the magnitude of variability of the trait in a local population) relative to a threshold (T). The threshold is given as 1/population size (as a first approximation).

    s is defined as w'/w-1, where w is fitness before the mutation (so now we're talking about mutations, and not just normal variation of a trait in the population?), and w' is fitness with the mutation.

    If s > T then the population can respond to selection pressures relevant to a given trait (i.e., the trait confers a significant advantage), and the frequency of the trait will increase.

    If s < T then the trait frequency will drift.

    So far so good, except that the variable w is not defined, either in terms of empirical measurements or of other variables. So, what determines the values of w and w' in a particular case?

    You said fitness = reproductive success. So, w and w' must be measures of reproductive success. But how is that success quantified? By the number of offspring? That can't be right, because it doesn't take into account things like r/K selection theory. Let me know how w is quantified. That seems to be the last piece of the puzzle.

    ReplyDelete
  5. Fitness can be quantified in many ways. Growth rate and number of offspring are common measures (proxies, really). And this is right despite r/K selection, because fitness is measured within populations in which organisms will be in the same range in terms of r/K. Fitness (and selection) is not measured between populations or between species.

    Hope that helps.

    ReplyDelete
  6. Bjørn, thanks for the input. As I considered your comments, I could see that a fundamental issue was being swept aside.

    You say that we need to distinguish selection from drift and that statistics can separate the two. But that actually does beg the question, because it assumes in the first place that selection is available as an explanatory mechanism.

    My point is that I don't think that's the case. While statistical definitions might help operationalize variables, they can't establish, or distinguish among, mechanisms.

    In any case, I've posted a follow-up. If you're interested, see the next
    blog post
    .

    ReplyDelete