More Recent Comments

Tuesday, November 26, 2013

Natural selection may not lead to evolution?

I recently discovered a new book called The Princeton Guide to Evolution. It looks pretty authoritative so I ordered a copy.

There are excerpts online. The first chapter is "What Is Evolution?" by Jonathan Losos. I'm not very impressed with his answer but I was shocked to read the following passage.
The logic behind natural selection is unassailable. If some trait variant is causally related to greater reproductive success, then more members of the population will have that variant in the next generation; continued over many generations, such selection can greatly change the constitution of a population.

But there is a catch. Natural selection can occur without leading to evolution if differences among individuals are not genetically based. For natural selection to cause evolutionary change, trait variants must be transmitted from parent to offspring; if that is the case, then offspring will resemble their parents and the trait variants possessed by the parents that produce the most offspring will increase in frequency in the next generation.

However, offspring do not always resemble their parents. In some cases, individuals vary phenotypically not because they are different genetically, but because they experienced different environments during growth (this is the “nurture” part of the nature versus nurture debate; see chapters III.10 and VII.1). If, in fact, variation in a population is not genetically based, then selection will have no evolutionary consequence; individuals surviving and producing many offspring will not differ genetically from those that fail to prosper, and as a result, the gene pool of the population will not change. Nonetheless, much of the phenotypic variation within a population is, in fact, genetically based; consequently, natural selection often does lead to evolutionary change.
I never heard to this idea before (that natural selection may not lead to evolution). I thought that natural selection was DEFINED as a change in the frequency of alleles in a population due to selection. Doesn't it have to have a genetic component?

Does this mean that natural selection may not lead to adaptation? Or, does it mean that adaptation isn't necessarily evolution?

The chapters are written by an impressive group of authors (Jonathan Losos is the editor-in-chief). It must represent the current consensus among evolutionary biologists. I'm surprised that I never heard of this definition of natural selection.


Diogenes said...

Larry: "I never heard to this idea before (that natural selection may not lead to evolution)."

I never heard any such ridiculous thing-- natural selection refers only to traits that are heritable. If I lift weights and get bigger muscles, and you don't, would these gentlemen call that "natural selection that doesn't lead to evolution"? Ridiculous.

Marcoli said...

That is strange, but I often often find errors in text books and in other technical books about basic concepts. Maybe the basic things do not get fact-checked as well. One that I have seen more than once is where an author describes viruses as 'organisms'. Another one that we are going to see more and more are statements like 'scientists used to think non-coding DNA was junk'.

Piotr Gąsiorowski said...

I thought that natural selection was DEFINED as a change in the frequency of alleles in a population due to selection.

Natural selection due to selection? Wouldn't it be better to say "due to the differentiating effect of heritable traits on reproductive success"?

Nullifidian said...

As a creationist once reliably informed me: “Adaptation still isn't evolution. At least not the evolution that Darwin was talking about.”

rich lawler said...

It's axiomatic in some evolutionary biology circles that this is absolutely the case. For example, one of the most highly cited papers in evolutionary biology (2900+ citations on google scholar) starts out with the following sentence: "Natural selection acts on phenotypes regardless of their genetic basis, and produces immediate phenotypic effects within a generation that can be measured without recourse to principles of heredity or evolution." In this sentence, which is not incorrect, the authors are taking a quantitative genetics perspective on evolution and breaking up the process of "evolution by natural selection" into two parts: selection on phenotypes within a generation and the response to selection across generations. Hence, to a quantitative geneticist one can measure natural selection occurring within a population independent of whether this produces a response to the selection, such as when a bunch of sparrows die in an ice storm due to larger wing size (see Lande R, Arnold SJ. 1983. The measurement of selection on correlated characters. Evolution 37: 1210-1226). This is standard practice in quantitative genetic theory.

Palo Palinger said...

How can something which is defined as a change _lead_ to something? It is like Newton would have only defined gravitation as a change in the position and that's all. How to test it? That's not explanation of anything.

How is evolutionary biology supposed to explain Darwin's development of his theory, when it only looks on his heritable traits and his reproductive success, so according to his own theory he is not different from other writers like Oscar Wilde? Was he a miraculous organism with miraculous metabolism?

AllanMiller said...

Futuyma in Evolutionary Biology p349 (3rd Ed.) notes that evolutionary biologists differ on whether the definition should include a genetic component or not. It seems fair enough to separate the response to selection from the selection itself to me. Selection does act on phenotypes after all. Heritability is unlikely to be 100%, and so selection would still be accepted to occur at lower levels. Yet how low? To insist it must be something other than zero seems as arbitrary as would setting the limit anywhere else on the continuum.

BradFoley said...

I think the whole epigenetics "revolution" has led to a redefinition of terms (like natural selection). This leads to caveats like the one you just quoted.

In some cases, the quibble is boring (not trivial, but not interesting). Steve Arnold did a great job parsing selection on P versus selection on G a long time ago.

But, in fact, natural selection on epigenetic variation (not classical genetic variation) can in some instances lead to heritable phenotypic change. Whether any of this heritable change is robustly persistent enough to "matter" evolutionarily is an open question, still. For instance, do epi-alleles stick around for more than one or two generations? do culturally transmitted changes persist for more than 5 generations? do heritable hormonal changes caused by, e.g., maternal grooming or maternal provisioning persist?

Some people are pretty evangelical about the idea that epigentics has changed everything we know, and think most of evolution can be reconstrued as due to epigenetics. I think this is bollocks. But adaptive intergenerational plasticity is a real thing, so I guess we need to be careful in our wording. Even if it makes it really to read (your quote is a good example).

Robert Byers said...

As a creationist it says here again there is problems with concepts and reasoning in evolutionary biology. Creationists always find close attention reveals chaos in methodology and even language.
The top people should not be confused about there subject of expertise.
the truth is evolutionary biology is a speculative subject or rather mere lines of reasoning from trivial data.
As the writer says "the logic is unassailable" . THE logic or lines of reasoning is simplistic.
Its not science. Its just a hunch. Its impossible. Its not logical to see variation equals threshold complexity revolution.

Carl said...

Losos is referring to selection in terms of the correlation between a trait and reproductive success. But the change in allele frequencies--ie, evolution--will depend on heritability. Basically, Losos is talking about the breeder's equation:

Palo Palinger said...

In other words, because we are ignorant of all those factors that have been affecting developmental processes in evolution, but we can culturally construct (culturally because !Kung people won't do it) abstract fiction from the correlation called selection, and then we can self-deceive ourselves that this abstract "selection" leads to evolution. No prediction and falsification, just inference to our loveliest explanation. Atheistic platonism.

Once upon a time, there was a response to selection pressure to study the nature of reality in Greece...

Larry Moran said...

Sorry, I didn't even try to specify an exact definition. I just meant to specify the genetic component. Apparently there are a lot of biologists who are comfortable talking about natural selection that has nothing to do with evolution. I find this very frustrating. It means that when we're talking about mechanisms or evolution we have to say "random genetic drift and some forms of natural selection."

I don't think I've ever heard anyone say that but it seems absolutely necessary in order to avoid confusion.

William Bell said...

The thing is that given that a non-genetically linked trait will appear without respect to the genetics of the individuals (which kind of goes with the of "non-genetically linked" part) so on average it should affect the fitness of each genotype about equally, so really it has no effect on evolution. At least that is what this immediately brought to mind upon reading it. Unless the epigenetic trait gave a certain genetic trait a spectacular boost or bust in which case it would interfere.
This is certainly making me think although I'm not taking this to mean we should throwout natural selection just yet. ;)

Corneel said...

Larry said
we have to say "random genetic drift and some forms of natural selection."
Absolutely. Think of balancing selection (such as overdominance). There will be natural selection and genetic variation, but no evolution, because none of the genetic variation is heritable. Allele frequencies will be driven towards a stable equilibrium.

troy said...

To add to Corneel's example: selection-mutation balance, selection-migration balance.

According to the breeder's equation R = h^2 S, where S is the (within generation) selection differential, R is the (between generation) response to selection and h^2 is the trait's heritability. If the latter is zero, there will be no between generation evolution even for nonzero S. This is standard stuff.

AllanMiller said...

It does seem reasonable to say that selection may not lead to evolution. All the processes of evolution are perpetually in train, even if they are currently 'freewheeling' due to having nothing to get their mixed-metaphorical teeth into. But if variant phenotypes are being sorted by an environment, it would seem that only in certain very restricted conditions (eg small population/limited genetic variability) would no opportunity exist for a genotype to catch a ride on the effect, and compound its returns.

Joe Felsenstein said...

We can carry out artificial selection on a population, whether or not it has any genetic variability. There was important work by the Danish biologist Johannsen, published in 1910. He took some lines of the bean Phaseolus vulgaris and inbred them until no genetic variation was left. He then showed that artificial selection was ineffective, not changing any of these inbred lines. While this seems trivial to us now, it was dramatic then, as Mendel's work was just recently rediscovered, and natural selection was competing with many other schemes, including Lamarckism and orthogenesis. (I would have called Johannsen a geneticist, but actually he was the one who introduced the word gene).

If you are willing to call Johannsen's action artificial selection, you should be willing to call Jonathon Losos's natural selection. Neither will result in any genetic change.

middle.professor said...

Just to echo Lawler's and Felsenstein's response: this is standard stuff representing a quantitative genetic view of natural selection. I think it led to a definitional debate between Lande/Arnold and John Endler, but I may be getting my history confused. And it's not just inbred lines. In wild and lab populations, there can be (and are) directions in multivariate phenotypic space with effectively zero additive genetic variance. Were (natural) selection in these directions there would be no (evolutionary) response.

Rich Lawler said...

And by the way, Joe Felsenstein wins the prize for having one of the most highly cited evolutionary biology papers: 23000+ citations for his bootstrap/CI on phylogenies paper. Yes, that's 23000+ citations...just one paper!

S Johnson said...

Try substituting "differential reproduction" for "natural selection."

The first paragraph then ends in saying differential reproduction can change "the constitution of a population." This is somewhat ambiguous, referring either to the allele frequencies of the population or the morphology of the population, but overall the statement is correct.

The second paragraph then says that differential reproduction can only lead to an increase in allele frequency when the trait selected is heritable. Again, this is overall correct.

The third paragraph then takes up the case where traits are not heritable. It says not all phenotypes are heritable (correctly. ) It also says that differential reproduction of nonheritable traits has no effect on evolution (also correctly.) The last paragraph refutes Lamarckism, which is actually good, as well as correct. But it uses "selection" to refer to changes in phenotype frequency in a population in a context where the topic of discussion is allele frequency. I don't think this is particularly confusing though.

Thirty years ago, when I was trying to explain natural selection, I said that natural selection had three component processes or conditions: differential reproduction, heritable variation and reproductive isolation. This reading selection equates differential reproduction and natural selection.
It's telling me I got it wrong, but it won't be the first time I've been accused of that.

Rolf Aalberg said...

The problem is the difficulty of observing natural evolution of animal life in realtime. Let's cordon off a suitable piece of Madagascar or something and observe what happens over some 10.000 years. A Faraday screen may be required to keep invisible designers out.

troy said...

To put that in perspective: that's more citations for that one paper than the total number of citations of all of William D. Hamilton's papers combined.

Reed A. Cartwright said...

This is pretty standard in population genetics and goes back to Darwin's original argument I believe.

I was just teaching last week that natural selection is a force that results from the variation of reproductive success between different phenotype. And this force will only result in evolution (i.e. affect the distribution of allele or genotype frequencies) if there is a genetic component behind the phenotype variation.

Larry Moran said...

One of my population genetics textbooks is by Wen-Hsiung and Dan Graur (you may have heard of them). They DEFINE natural selection as ...

Natural election is defined as the differential reproduction of genetically distinct individuals or genotypes within a population. Differential reproduction is caused by differences among individuals in such factors as mortality, fertility, fecundity, mating success, and the viability of offspring. Natural selection is predicated on the availability of genetic variation among individuals in characters related to reproduction. When a population consists of individuals that do not differ from one another in such traits, it is not subject to natural selection. Selection leads to changes in allele frequencies over time.

Li and Graur are going to be angry at you for not using their view of "standard population genetics." I'm telling.

I also have a textbook of population genetics by Daniel Hartl and Andrew Clark. They describe natural selection as ...

Modern formulations of natural selection are ... compacted into a form resembling a logical syllogism.

In all species, more offspring are produced than can possibly survive and reproduce. Organisms differ in their ability to survive and reproduce—in part because of differences in genotype. In every generation, genotypes that promote survival in the current environment are present in excess at the reproductive age and thus contribute disproportionally to the offspring of the next generation.

Through natural selection, therefore, alleles that enhance survival and reproduction increase in frequency from generation to generation, snd the population becomes progressively better able to survive and reproduce in the environment.

This description also restricts the definition of natural selection to an effect on genotypes.

Could you give me the definition of natural selection from the population genetics textbook you use in your class? Almost all the books I have on my shelf talk about natural selection only in terms of selection on genotype.

Anonymous said...


I don't mean to be disrespectful, as you know I have a degree of respect for you....But how much did you pay for this piece of shitty confusion...?

It seems to me, the best minds in the world in the biz, like Joker Frankenstein, Cat-Jerry the Coy--ny and you Larry the are in town....Do you have any reason left????

WHY CAN'T YOU AGREE ON ANYTHING??? WHY??? If evolution is a fuck, then so is my previous post....

Paul C. Anagnostopoulos said...

Here is Marvin Minsky's Theorem of Evolution, which I often use when trying to explain evolution to newbies. Note that it cleverly avoids the term "natural selection." It's a bit like Hartl and Clark.

The Process of Evolution is the following abstract idea:

There is a population of things that reproduce, at different rates in different environments. Those rates depend, statistically, on a collection of inheritable traits. Those traits are subject to occasional mutations, some of which are then inherited.

Then one can deduce, from logic alone, without any need for evidence, that:

THEOREM: Each population will tend to increase the proportion of traits that have higher reproduction rates in its current environment.

Tim Tyler said...

Re: Marvin Minsky's Theorem of Evolution

Um, that's nonsense. What about if the mutation rate exceeds the rate at which deleterious mutations are eliminated by selection? Then you get a meltdown - not progress.

Robert Byers said...

Byers Theorem.
The origin of the complexity within and diversity between biological entities is professed to be from reproductive success of diverging traits within a reproducing population leading to segregated reproducing populations.
Since reproduction is the source for the success of a new population its impossible for divergence to be great enough for any important change . Any mutation leading to a trait would always be restrained within the reproducing population.
So thresholds could never be crossed and thus impossible for biology to have evolved beyond the genus level. This should be self evident.

Tim Tyler said...

It's unorthodox usage of the term. By long convention, selection refers to non-random change. To use the term for something else, one should start by arguing that the conventional usage is misleading and worthy of reform.

Schenck said...

1) How is the genetic variation not heritable in balancing selection?

2) The breeder's eqn, including the description linked to be Carl Zimmer below, talks about "S" being the selection differential 'as related to phenotype', so that seems to be begging the question at hand here, namely: "Doesn't natural selection mean selection on heritable traits/selection between genotypes"? That source also notes the heritability is gotten at by looking at the similarity between parents and offspring (which is how Lewontin defined it too), without mention of genotypes or mechanisms of and influences on inheritance.

Palo Palinger said...

Natural selection is apparently a buzzword, which people use when they are ignorant of history and developmental processes and create western progressive ontology, Darwinian Platonism, instead. Ontology with no testable and refutable theory, but with many "perspectives":

"...first, individuals might be trading-off offspring quantity vs. quality to maximize Darwinian fitness and that a smaller family is optimal in the modern environment (behavioural ecology perspective); second, reduced fertility might be a maladaptive response to an environment changing radically (evolutionary psychology perspective); and third, the influence of kin on individual reproductive decisions may have decreased with the widening of social networks, limiting the incentives
for having big families (cultural transmission hypothesis)."

I refuse to regard Darwin as an organism with the brain which mirrored "nature out there" after all those millions years. That is ridiculous. Look at the chimpanzee's brain, whether it is able to mirror.

Read Richard Rorty's Philosophy and the Mirror of Nature as a warning that brains of Spencer, Darwin, Haeckel, Kropotkin, Fisher, Mayr, Lewontin, Gould, Dennett, Dawkins, E.O. Wilson, Nowak, mine and yours can mirror what is going on in evolution.

Corneel said...

@Robert Schenk
1) Because genetic differences in reproductive succes do not get passed on to the offspring. Consider for example a case of balanced lethals, where both homozygotes (say aa and bb) are not viable, but heterozygotes (ab) are. The allele frequencies will immediately converge to p(a) = p(b) = 0.5 so there is genetic variation. However, half of the offspring from heterozygotes will be homozygotes again, resulting in a mean offspring fitness that equals the original parental population fitness. Hence, that genetic variation is not heritable
2)True, but by that defintition balancing selection is not a form of natural selection, which I think woud be terribly confusing.

TheOtherJim said...

Johannsen, published in 1910

Yet another paper Dr. J. Shapiro has missed.

AllanMiller said...

Li and Graur clearly offer your preferred definition, but Hartl and Clark don't. They do say that where there is genetically-based variation in reproduction rate and nonzero heritability, NS acts to promote adaptation, but they don't appear to me to be saying that NS doesn't happen unless it affects genotype frequencies.

Tim Tyler said...

Wen-Hsiung and Dan Graur's definition of NS in terms of reproduction risks missing much. If you have a population in which the tallest half die, most would be OK with referring to that as "natural selection". Confining natural selection to cases where differential reproduction is involved eliminates lots of cases where reproduction rates are fixed, but there is differential mortality - e.g. the differential mortality of stellar photons hitting dust. You would then need to have another category of selection to cover adaptive phenomena in those cases.

One category for differential reproduction and another one for differential mortality is a reasonable idea. However, *normally* the term "natural selection" refers to *both* cases - and such usage is good.

Larry Moran said...

@Allan Miller

I agree that we can nitpick the Hartl and Clark definition but their use of words like "genotype" and "allele" makes it pretty clear to me that they are only considering heritable changes.

Allan, I don't think I"ve ever seen you write that one of the mechanisms of evolution can be SOME FORMS of natural selection. Is that what you believe?

Larry Moran said...


How do you explain random genetic drift using the Marvin Minsky Theory of EVOLUTION?

The whole truth said...

Is evolution always equivalent to progress? When considering evolution, or just natural election, on a scale that encompasses everything that has ever lived on Earth, is it correct to say that progress is always or ever attained? Doesn't evolution/natural selection often result in extirpation or extinction? And even when other organisms fill the niches left by extirpated or extinct organisms, is that necessarily progress?

The whole truth said...

Larry (and anyone else who would like to respond), I'm not playing down the importance of genetic drift but I'm wondering how you would answer this question:

Is it accurate to say that the results of genetic drift are or subject to natural selection at some point?

The whole truth said...

The word 'or' shouldn't be in my question above. That's what I get for typing when I'm sleepy.

Schenck said...

"genetic drift['s results] are...subject to natural selection at some point?"
Drift is a result of not having an infinitely large population, with less than ideal population sizes there can be a mismatch between the alleles in the parental and offspring generation, even with random survival, reproduction, and mating of parents.
So any selection is always going to be acting on 'stuff' that drift has already acted on.

Larry Moran said...

@The Whole Truth

It is not correct to say that the results of genetic drift are subject to natural selection at some point. Think, for example, of all those mutations in junk DNA that have been fixed by random genetic drift.

Schenck said...

Isn't heritability /rarely/ 100%/perfect? Clones would approach it (and even they'll have errors), and in sexual species heritability is quite below 100%. If heritability was absolutely perfect, we could have natural selection, but no long term, cumulative adaptation.
So we'd have red and blue individuals, cloning, perfect inheritance, and blue selected for, so that's fine the population becomes blue. But if purple was adaptive, there's no way to get to it.

Tim Tyler said...

RE: "The logic behind natural selection is unassailable. If some trait variant is causally related to greater reproductive success, then more members of the population will have that variant in the next generation"

Ironically, that logic is wrong. The *proportion* will probably increase, but the absolute numbers might easily decline - if the entire population is shrinking quickly.

It seems like a pretty embarrassing logical mistake to make - after starting with the claim that the logic is "unassailable".

Larry Moran said...

Robert Schenk says,

Drift is a result of not having an infinitely large population, with less than ideal population sizes there can be a mismatch between the alleles in the parental and offspring generation, even with random survival, reproduction, and mating of parents.

Just to be clear to readers, the fixation of alleles by random genetic drift is not restricted to small populations and it is not restricted to species with sexual reproduction. It occurs just as frequently in large population and in asexually reproducing species. These are just two of the many misconceptions that we are teaching to students. [see Random Genetic Drift and Population Size]

Rich Lawler said...

It's not surprising that Li/Grauer and Hartl/Clark use a different definition than the one given by Losos. These guys either study molecular evolution or "traditional" population genetics, where the emphasis is on alleles. And since one is considering the very things that get transmitted between parents and offspring, their definitions incorporate heritability. But quantitative genetics comes from a different tradition, focusing on phenotypes and artificial selection experiments. It's completely fine to state that natural selection acts on phenotypes within a generation...because that's what it does! If you're a farmer who has been selecting leg number, or if you are "nature" selecting finger number in humans, if there is no heritability in these traits, then there is no response to selection (since leg number and finger number variation is often driven by teratogens, not alleles). But selection is still going on. It's a semantic point coming from two different traditions in evolutionary biology. Understanding selection within a generation is insteresting in its own right, independent of the heritable consequences. But to other evolutionary biologists the two components (selection within and the response across a generation) are wrapped up into a single definition. But separating the two isn't problematic, unless you're a pedant. It's sort of like insisting that fitness involves reproduction and survival, whereas a demographer might define fitness as survival only.

Schenck said...

"genetic differences in reproductive succes do not get passed on to the offspring."
Maybe I'm being a bit thick here but the heterozygotes are able to reproduce, their offspring are all heterozygotes, and their reproductive success is inherited from their parents.
Perhaps did you mean that the homozygotes are viable (grow from zygotes to living individuals), but sterile? So then the frequencies of the alleles in a population includes these sterile homozygotes but of course they didn't inherit those genes from homozygous parents? But even then their sterility is a result of inheritance of the alleles from their heterozygous parents. Forgive my misunderstanding here.

Unknown said...

If you lift weights, and if your new strength means you live longer and leave more descendants, then natural selection is favoring you. That seems to me a reasonable use of the term. I think it could be very useful to be able to disconnect the actual selection of a class of individuals from the genetic consequences of such selection
Lou Jost.

The whole truth said...

Thanks for your response, Larry. So that I can more fully understand your train of thought, do you think that there are ever any results of genetic drift that are subject to natural selection?

Fukuda said...

Wen-Hsiung and Dan Graur's definition of NS in terms of reproduction risks missing much. If you have a population in which the tallest half die, most would be OK with referring to that as "natural selection". Confining natural selection to cases where differential reproduction is involved eliminates lots of cases where reproduction rates are fixed, but there is differential mortality - e.g. the differential mortality of stellar photons hitting dust. You would then need to have another category of selection to cover adaptive phenomena in those cases.

The reason why mortality is relevant to natural selection is because it may curtail reproductive success, i.e. dead organisms don't reproduce (or live less, but see later), which lowers the proportion of their alleles in the following generation. How is the allele supposed to spread by natural selection if it doesn't affect reproduction rates?

In fact, mortality and reproduction are intertwined in a complex way. Because differential reproduction is what really matters for the spread of a given allele under natural selection, alleles favoring reproduction at the expense of lifespan are selected for. This is called antagonistic pleiotropy and actually is one of the main competing hypotheses explaining the evolution of aging.

Larry Moran said...

@The Whole Truth

Your question is so confused that I'm reluctant to give you an answer. One of the results of random genetic drift will be the fixation of an allele and the elimination of all other alleles. Is it possible that a beneficial mutation or a deleterious mutation will arise at the exact same nucleotide at some time in the future? I suppose it is, in which case the allele originally fixed by drift will be subject to natural selection.

Is this what you meant?

Larry Moran said...

@Rich Lawler

Does the quantitative genetics tradition distinguish between the natural selection that leads to evolution and the natural selection that leads to non-evolution? Or, do you have to figure it out by the context of the discussion?

Do quantitative geneticists ever refer to "the theory of change by natural selection that's not evolution"? Or, do they have another name for that concept?

When quantitative geneticists teach about changes in peppered moths or Darwin's finches are they always careful to tell students that these changes may be due to natural selection but they may not be examples of evolution?

These are new ideas to me. I don't think I've ever met anyone who says that you can have natural selection that may not be evolution. I know lots of people who talk about different meanings of "adaptation" but I always though that natural selection had to involve genetic traits.

rich lawler said...

Usually it's the context of the discussion, though often in quant-gen literature you'll see folks talk about the "evolutionary response to selection" They don't have a conventional phrase/term for "the theory of change by natural selection that's not evolution" but again, there is talk of either "a selection response" or "no response" to selection, or sometimes folks use "phenotypic selection" to denote they are concerning themselves with selection and not the response to selection (e.g., Kingsolver et al. 2001. The strength of phenotypic selection in natural populations. Am Nat 157: 245-261). Evolutionary quantitative geneticists are interested in both selection and the evolutionary response to selection. But cases in which there is selection acting on a population but no evolutionary response to selection are interesting in their own right. One example is the study of selection limits where selection has chewed up all the additive genetic variance and thus you can continue to select traits but not observe the expected response to this selection. And middle.professor gave another example where you can have selection acting on multiple traits (all with significant heritability) but given how each trait is genetically correlated with the other traits, there will be no response to selection. These examples actually illustrate the strength of talking about selection as separate process from the response to selection, as they demonstrate that selection can occur in a population and one of two things will happen: 1) there will be no evolutionary response to selection, or 2) there will be an evolutionary response to selection. More simply, if you read the link supplied by Carl Zimmer on the breeders equation, you'll see two terms on the right side of the equation: the heritability and the selection differential. These two terms mathematically capture the two processes quantitative geneticists are talking about (and that seem to be vexing you). If the heritability term is zero, there is no response to selection, even if the selection differential is non-zero (i.e., selection is occurring). Hence, one can talk about (and study) selection independently of its evolutionary consequences.

"When quantitative geneticists teach about changes in peppered moths or Darwin's finches are they always careful to tell students that these changes may be due to natural selection but they may not be examples of evolution?"

No, those traits are heritable and there was a response to selection across many generations, hence they are discussed in the context of the evolutionary consequences of this selection.

"These are new ideas to me. I don't think I've ever met anyone who says that you can have natural selection that may not be evolution."

You mentioned the Hartl/Clark book, their chapter 8 (depending on which edition you have) deals with quantitative genetics and I would be surprised if they didn't talk about the same distinction I'm talking about here. Or pick up a copy of "Evolutionary Quantitative Genetics" by Derek Roff.

Corneel said...

@Robert Schenck
Usually you measure fitness from the zygote stage onwards, so the population does not merely consist of heterozygotes, but yes, you can think of sterility instead of lethality if that works better for you.
Your mistake is in thinking that we are discussing the origin of the alleles, but we are asking how much of the PHENOTYPIC variation in fitness gets passed on. The higher fitness of the heterozygotes is not heritable (though it is genetic) because they cannot produce only heterozygotes. Homozygotes are newly formed every generation, preventing this population from evolving/ adapting.
Hope that makes things clear

Corneel said...

words like "genotype" and "allele" makes it pretty clear to me that they are only considering heritable changes.

This is a mistake. Genetic variation does not equate heritable variation. It is very well possible to have plenty of genetic variation, but zero heritable variation. Several examples were mentioned earlier in this thread (e.g. overdominance at equilibrium). I don't think the defintion Hartl and Clark use automatically implies evolution is involved.

AllanMiller said...


I agree that we can nitpick the Hartl and Clark definition but their use of words like "genotype" and "allele" makes it pretty clear to me that they are only considering heritable changes.

I agree it's nipicky, but they say that "Through natural selection, therefore, alleles that enhance survival and reproduction increase in frequency", which does not explicitly exclude non-heritable change from the definition of NS, but rather describes the impact of NS upon heritable change.

Allan, I don't think I"ve ever seen you write that one of the mechanisms of evolution can be SOME FORMS of natural selection. Is that what you believe?

Well, there's only one 'form' broadly speaking, and it acts through phenotype. The coupling to genotype is variable along a continuum from 0 to 100%. One could imagine pressing a 'heritability clutch'. So long as some torque is getting through to the genetic wheels, we would agree that this is Natural Selection. But I think it's still Natural Selection when the genetic clutch is fully disengaged and the engine is merely spinning. If the environment is favouring an increase in size, for instance, it favours both better-fed individuals and those whose greater size has a genetic component, and that genetic component can result from alleles at several different loci. Something (NS?) sorts 'em all, and adaptation emerges from any genetic component. If there is none of the latter, the same thing is still going on, sorting of phenotypes; the engine will spin until a heritable variant arises. Drift, meanwhile, I would regard as operating upon genotype alone - it is the random error from sampling a population's genes.

Anonymous said...

Natural selection is rarely defined as needing allele frequency change, as other commenters note that would potentially exclude some forms of balancing selection, and mutation-selection balance.

There is somewhat of a split in the field about whether we define natural selection as needing heritable variation. I think it is helpful to define natural selection as non-random reproductive success, survivorship, transmission etc based on phenotype. And that this can lead to evol by NS, if this phenotypic variation is heritable. If we are talking about selection from a phenotypic perspective it is certainly more helpful to not require the heritable component. This allows us to talk about selection gradients, i.e. covariance between phenotype and fitness.

I cover this briefly in my notes there's a slightly more up to date copy here

Anonymous said...

Sorry I forgot to sign my comment.

Graham Coop

Schenck said...

Wrt population size, as far as I understand it the only population that isn't effected by drift is infinitely large (and perfectly randomly mixed, random death, etc).

Schenck said...

Yes thanks! I was in fact thinking of the phenotypes being inherited b/c the alleles are inherited, thanks dor taking the time to clear this up for me.

The whole truth said...

Larry, no, that isn't what I meant although I suppose that what you mentioned could be a part of it. Please correct me if I'm wrong but it appears that you think that natural selection only occurs at the level of an allele, or nucleotide, or mutation. My view of natural selection is much more broad than that. My view is that natural selection can and does occur at pretty much any level of life and at pretty much any level of the ingredients/processes/events that make life possible, or not.

I certainly do respect your knowledge and the knowledge of many other scientists but one of the things I've noticed (and I'm not the only one) is that the opinions and definitions that scientists adhere to depend largely, if not completely, on the particular schooling/training that scientists have gotten and on what they specialize in. In many of the discussions here and on other sites there is often disagreement between scientists about definitions and some or even most of the particulars of evolutionary theory. Sometimes the disagreements are minor but sometimes they're major. Sometimes it's fairly easy to tell who is wrong but sometimes it's not, and depending on what evidence is found in the future, the scientists who appear to be right may turn out to be wrong about some things. I'm not saying that there's no such thing as wrong and right when it comes to evolutionary theory (some things are well substantiated) but I do think that scientists should communicate much more between their specialized fields of study and that they should work hard at finding agreement on at least definitions, such as the definition of natural selection, and evolution.

When I see such disagreements I try to discern who has the most substantiated argument, but as I'm sure that you are aware, some arguments won't be substantiated and settled unless and until more evidence is found, or unless and until there is more agreement between scientists, especially in regard to definitions and the explanations that depend on them.

I may be confused but if I am I'm certainly not the only member of that club. I ask questions because I want to see what others think. I like to learn and I'm always open to serious consideration of evidence based arguments and even to reasonable, speculative opinions. There are many things that I don't know but I do know that no one knows everything.

Anyway, I just wanted to see what you think because I value your knowledge and opinion. I may not always agree with you though. :)

Larry Moran said...

@The Whole Truth

You are correct that there are many genuine scientific controversies and you are correct when you say that scientists in different fields often have conflicting views on some fundamental issues.

In this case, I am encountering, for the first time, the notion that natural selection can occur without evolution. I thought that natural selection was DEFINED as a nonrandom change in alleles or heritable traits. Apparently there are a whole group of scientists who want to extend Darwin's theory of natural selection beyond evolution.

I knew that you could have ADAPTATION without evolution but I didn't know that all forms of biological adaptation were now called NATURAL SELECTION by certain groups of biologists. This is very confusing.

I agree that SELECTION can occur at levels higher than individuals in populations. That's why I listen to arguments about species selection (or, more correctly, "species sorting"). However, I haven't heard anyone refer to species selection as NATURAL selection.

The whole truth said...

Well now I'm more confused than I was before! ;)

I respect your view, Larry, and even though I'm nowhere near as educated in molecular and cellular biology as you are I too am frustrated by multiple definitions and the confusingly variable explanations that depend on the multiple definitions of natural selection and evolution.

Unfortunately, the confusion makes it harder to learn about natural selection and evolution, and it gives the IDiots something to attack. Oh well, as long as scientists are free to continue studying natural selection/evolution I suppose that eventually more agreements will come about as more evidence is discovered, and in a thousand years or so (hopefully a lot sooner) the fact and the theory of evolution and its particulars may be agreed upon and accepted by all but the most stubborn people. In the meantime the search for more answers is interesting and rewarding.

Larry Moran said...


Balancing selection is quite rare but it's clearly affecting alleles (i.e. genes) the way you describe it. How would balancing selection work if it had no effect on genotypes? Could it be stable?

Larry Moran said...

@Allan Miller,

You used the word "selection" in your comment and not the term "natural selection." I think everyone agrees that you can have "selection" on phenotypes even if there's no genetic component. The question is whether this is "natural selection."

Larry Moran said...

I think the whole epigenetics "revolution" has led to a redefinition of terms (like natural selection).

That would make me very sad. I don't see much evidence of this, do you?

Larry Moran said...

I think we all know that he is probably referring to "selection" but it's when he equates all forms of selection with "natural selection" that we get into trouble.

Larry Moran said...

It does seem reasonable to say that selection may not lead to evolution.

I think everyone will agree with that statement. The question is whether "natural selection" may not lead to evolution.

Larry Moran said...

Joe, do you use "natural selection" in that way? When you are teaching population genetics are you careful to say that you are only referring to one particular form of natural selection? When you are describing mechanisms of evolution do you always say "random genetic drift and some kinds of natural selection"?

Larry Moran said...

@rich lawler,

I notice that in your lengthy response you use the word "selection" a great deal but you never say "natural selection."

Why is that? There's no dispute over the multiple meanings of the word "selection." Would you be just as comfortable with your response if you replaced "selection" with "natural selection" in every instance?

Anonymous said...

This may help.... ;)

"Epigenetics drives phenotype?

Researchers have identified a possible mechanism by which DNA regions that don't encode proteins can still determine phenotypic traits such as a person's height or susceptibility to a particular disease, researchers report online in Science today. Image: WikipediaThe scientists found that certain chromatin modifications often considered to be epigenetic -- meaning, regulated by factors other than genetic sequence -- are in fact determined by a person's DNA..."

rich lawler said...

Economy of phrasing. I was using "selection" as a stand-in for "natural selection" and I'd be comfortable with substituting one for the other. One could just as easily say "the response to natural selection" as "the selection response." Isn't that what we are talking about given your opening post and question? In this context, the word "natural" really doesn't matter unless we are contrasting it with "artificial." Make no mistake about it, not all evolutionary biologists use this two part distinction about "natural selection" (or selection) and the "response to natural selection," I'm just trying to clearly outline the quantitative genetics position on this issue. If you don't like this distinction, then don't worry about it. Usually it's a non-issue as most folks are interested in heritable traits anyways.

Unknown said...

Natural selection doesn't equal evolution.

Natural selection doesn't create new kinds or new structures. In fact, natural selection does the opposite, it loses genes. When an allele (trait) is selected because of a selection pressure, that allele may be passes onto the next generation but the other "rejected" alleles will be lost.

This means that the hundreds of genes needed for a kind to change (evolve) into another kind (not speciation) isn't being created but is being lost.

Scientist know and understand this problem so they have come back with mutations as their answer.

If you look into mutation you will understand that this doesn't solve the problem either. Mutation is the damage of genes.

Mutation would need to create hundreds of new base pairs in the correct order, giving the correct information in the correct amount with "on switches" before and "off switches" at the end. DNA is far too complex for mutation to accidentally create.

The DNA has many defensive systems against mutation. Mutation isn't wanted. Natural selection wouldn't select for something that isn't wanted.

Even when mutation "helped" it was because of the destruction of a gene not the creation of new base pairs let alone alleles.

Natural selection and mutation doesn't equal evolution.

rtmcdge said...

That sounds right. "The Origin of the Species" is supposed to explain the new species. And in this instance Darwin is actually referring to new kinds of organisms. Which seems to be made up of the different species members of each of the different kinds of organisms. An example of this is seen when there is a discovery of a "new" species. The announcement in the News Paper, or on Google News, even in the "Times Magazine", would always be something like, "A new species of shrimp, or bat, or bird or fish or some other kind of organism.

This shows that there are a variety of species of different kinds of organisms.

And realizing this completely shatters the foundation of the speculation of evolution. Because it tries to make the term species not only serve to represent the variety of species of each of the kinds of organisms, but they would twist this same term to represent the individual kinds of organisms.
Seeing this the error of evolution comes into focus. They are actually saying that "A new species of species" has been found.
And this is not what evolution can not do.