Provine is an atheist, philosopher, and critic of intelligent design. He has engaged in prominent debates with theist philosophers and scientists about the existence of God and the viability of intelligent design. He has debated the founder of the intelligent design movement Phillip E. Johnson and the two have a friendly relationship. Provine has stated that he starts his course on evolutionary biology by having his students read Johnson's book "Darwin on Trial."When someone likes that publishes a book with the title, The 'Random Genetic Drift' Fallacy, I pay attention.
Provine is a determinist in biology, but not a determinist in physics or chemistry, thus rejecting the idea of free will in humans. Provine believes that there is no evidence for God, there is no life after death, there is no absolute foundation for right and wrong, there is no ultimate meaning for life, and that humans don't have free will.
Let's begin by reminding ourselves that random genetic drift is one of the mechanisms causing changes in allele frequencies in a population [Random Genetic Drift]. It's covered in all the genetics and evolution textbooks. The one I like to quote is by David Suzuki et al. from 1989.
If a population is finite in size (as all populations are) and if a given pair of parents have only a small number of offspring, then even in the absence of all selective forces, the frequency of a gene will not be exactly reproduced in the next generation because of sampling error. If in a population of 1000 individuals the frequency of "a" is 0.5 in one generation, then it may by chance be 0.493 or 0.505 in the next generation because of the chance production of a few more or less progeny of each genotype. In the second generation, there is another sampling error based on the new gene frequency, so the frequency of "a" may go from 0.505 to 0.501 or back to 0.498. This process of random fluctuation continues generation after generation, with no force pushing the frequency back to its initial state because the population has no "genetic memory" of its state many generations ago. Each generation is an independent event. The final result of this random change in allele frequency is that the population eventually drifts to p=1 or p=0. After this point, no further change is possible; the population has become homozygous. A different population, isolated from the first, also undergoes this random genetic drift, but it may become homozygous for allele "A", whereas the first population has become homozygous for allele "a". As time goes on, isolated populations diverge from each other, each losing heterozygosity. The variation originally present within populations now appears as variation between populations.
Suzuki, D.T., Griffiths, A.J.F., Miller, J.H. and Lewontin, R.C.
in An Introduction to Genetic Analysis 4th ed. W.H. Freeman (1989 p.704)
Provine thinks there's also some confusion between the segregation of single loci and whole chromosomes. Both of these misconceptions invalidate random genetic drift according to Provine. Here's part of what he says on page 46.
Fisher and Haldane agreed that "random genetic drift," on a single locus on a chromosome, was a measure of inbreeding in Mendelian populations, Both used the same models and did not understand meiosis, and made a biological mistake. They did not realize their mistake when meiosis, with no "random genetic drift," was understood by 1940. Population geneticists are taught now by other population geneticists, so they still today do not understand the problems with "random genetic drift."I've read these chapters several times and I still don't understand Provine's problem with random genetic drift. It's possibly just historical quibbling but I'm not sure.
I'm hoping that some Sandwalk readers will help me out.
With the development of Neutral Theory in the late 1960s, you'd think that random genetic drift was firmly entrenched in population genetics. After all, how else do allele frequencies change if alleles are neutral with respect to natural selection? How else do you explain all the differences between chimpanzee and human genomes in junk DNA?
The neutral theory and nearly neutral theory missed the supposed biological origin of the "random genetic drift." Neither Kimura nor Ohta saw where "random genetic drift" originated in population genetics. They inherited the theories of Fisher, Haldane, and Wright. Kimura often used the phrase "random sampling of gametes" as the source of "random genetic drift." This process produced maximum genetic variation fro recombination in meiosis in large populations, and minimum in small populations, and no "random genetic drift." Both thought of evolution over speciation; 10,000,000 generations are not much to either theory.Michael Lynch isn't going to like that!
Having no "random genetic drift" in evolution harms the neutral theories, No matter how we approach the neutral theories, "random genetic drift" is the crucial variable, and does not exist. I can see no way to preserve the neutral theory in population genetics."
Is Provine an adaptationist? Is that why he dismisses Neutral Theory and random genetic drift? I don't know. I can't decipher his argument in this book. I think he's arguing that fixation of apparently neutral alleles is always due to hitchhiking on selected loci but I get part of that idea from reading other things he's said over the years.
Here's a passage that causes me to think that there's more to his rejection of Neutral Theory than meets the eye. It's from pages 158-159.
In his last years Kimura had gained the highest honors of an evolutionist. His neutral theory of evolution seemed to have huge support, and he spent most time pointing how neutral evolution could lead to adaptive evolution, even in the 10% of DNA that seemed to code for something. Kimura was at the top of his profession until his death in 1994.Anyone guilty of this kind of fuzzy thinking in one area is probably guilty of the same kind of thinking in another.
Kimura believed that 90% of the DNA was neutral, but concerning percentages, beliefs with time have changed. Most biologists now believe that 70% of DNA codes for RNA, of so many kinds. For sexually breeding populations, mRNA is edited by an RNA editor that can make many different changes into ten or hundreds of different kinds of proteins. Many of those who study RNA think it existed prior to DNA and possibly evolved into using DNA for evolutionary history. The amount of neutral DNA has not decreased by a huge amount; perhaps 20% of neutral DNA is left, probably less. More than 80% of DNA is occupied by coding at various times in life. Kimura's thesis, neutral DNA, is getting smaller than he thought possible.
I can't wrap my head around this either -- what the hell does he mean by saying that random genetic drift does not exist?
ReplyDeleteThe random sapling of gametes is not really an assumption -- it's only a metaphysical assumption to the extent that we care about metaphysics, but otherwise it just follows from how meiosis and fertilization work. Of course, there are exceptions, but they're just that -- exceptions. Many of the key equations behind the neutral theory follow in a purely mathematical way just based on that.plus some variable sets of assumptions about the structure of genomes and populations. And then that drift will be a major factor under certain conditions is an inevitable conclusion. The open questions concern the relative influence of the different evolutionary forces in real life, but that's really a question of measuring parameters, not so much one of underlying theoretical framework (that is not to say that the theoretical framework is fixed and finished, but still, the basics are well understood).
For sexually breeding populations, mRNA is edited by an RNA editor that can make many different changes into ten or hundreds of different kinds of proteins. Many of those who study RNA think it existed prior to DNA and possibly evolved into using DNA for evolutionary history. The amount of neutral DNA has not decreased by a huge amount; perhaps 20% of neutral DNA is left, probably less. More than 80% of DNA is occupied by coding at various times in life. Kimura's thesis, neutral DNA, is getting smaller than he thought possible.
This isn't just "fuzzy thinking", this is full of non-sequiturs and also reveals gross ignorance of the subject..
The random sapling of gametes ...
DeletePresumably, when older, this has something to do with the evolutionary tree.
If there was an edit option, you would not have had a chance to make that comment
DeleteWhile I have not read this, I have it on good authority that Provine had a very difficult time getting this published, eventually resorting to Amazon's version of self-publishing. That doesn't mean his thesis is wrong, of course, but apparently he is having a hard time convincing people that he is right.
ReplyDeleteThe editing of the book is terrible indeed.
DeleteYep. That's painfully evident from just the quotes that Larry provided.
DeleteSame here - this all seems a bit confusing to me. I have seen strange controversies in biogeography and systematics where sometimes it turns out that somebody has completely redefined certain key terms, or has an extremely strange perspective on what something should be about (e.g. a systematist who defines macroevolution as the transition from one supraspecific taxon to another). Maybe that is what is going on here.
ReplyDeleteEven the little things, as in this Wikipedia quote: "not a determinist in physics or chemistry, thus rejecting the idea of free will in humans". Usually it is the other way around: one rejects (libertarian/supernatural) free will because of determinism.
I suppose it should be read like this: Provine (while not a determinist in physics or chemistry) is a determinist in biology, thus rejecting the idea of free will in humans.
DeleteBut even so, how can non-determinism in chemistry produce determinism in (molecular) biology?
Aaah... now I see. Then I guess the idea could be that whereas quantum in non-deterministic, it all collapses into deterministic behaviour at larger scales, at the scales that matter for biology. Not saying that is either right or wrong, but it would then have a logic to it.
DeleteWell, all arguments of this type use the law of large numbers. And often enough they use it without checking whether the conditions for a LLN are actually given. For starters all applications of laws of large numbers are about the convergence of the mean of a set of random variables to the expectation of that mean. That already restricts its applicability quite a bit. There are also conditions on how strongly these variables can be correlated (and a big chunk of the 2008 financial crisis comes down to banks repackaging mortgages in such a way that the correlations appeared to be lower than they actually were, leading to inflated ratings based on applications of the LLN).
DeleteIn biology we get some quantum mechanical effects to be magnified to macroscopic scales, because macroscopic structures are built from genomes. A single indel can have notable effects and that requires changes in two covalent bonds. Now, a covalent bond is something we can describe with QM and 2 is not a particualrly large number. And since somatic mutations can produce cancer and since cancer has an effect on the number of offspring an organism has, I don't think there is a reasonable case for no drift.
Life acts as a microscope, which makes the stochasticity at the molecular level apparent on macroscopic scales, because Genomes do not average out. That lead Darwin and Wallace to propose an explicitly stochastic theory, which in turn inspired Boltzmann to try out one in physics. And then Wien took that along with the Maxwell equations to figure out what the radiation profile of the sun should look like and got a wrong result. Planck fixed that, by taking a variable h Wien had introduced which he had go to 0 and simply making it go to a small constant. That's not a coincidence. Darwin and Wallace were laying the foundations for 20th century physics.
The whole "deterministic therefore no free will" is silly to begin with, because the alternative is randomness and that would not lead to free will either (free will is a self-contradictory concept no matter how it is cut).
DeleteLibertarian or supernatural free will yes; however, in my native language 'freiwillig' is simply the word for voluntary, and that is how I am happy to treat the concept even in English. I do something out of my own free will if 1. I want to do it (will) and 2. nobody has manipulated or forced me into it (free). That kind of compatibilist free will works perfectly well with determinism.
DeleteIt seems to me he conflates/confuses drift with (nearly) neutral evolution (I'm not sure, though, because his reasoning is hard to follow). I'm also puzzled by statements like "the 10% of DNA that seemed to code for something" (etc.). 10% of whose DNA? Is he another thinker for whom DNA = the human genome?
ReplyDeleteDNA codes for RNA
DeleteSo that's why they call it "noncoding DNA", right?
Thanks for the quotes. My hypothesis on what happened:
ReplyDeleteProvine is out of touch. He heard about ENCODE via the mass media or some popularization, but didn't read any of the blogs or the rebuttal papers. He's probably never heard of genome-size variation and the C-value paradox, or he's forgotten about it.
So, now he's got the naive view that most of the genome is functional, and he's trying to make population genetics work with that, by going back to some early-Modern-Synthesis views where everything-is-functional was the default assumption. But both the data and the theory actually lead pretty linearly to "most of the DNA isn't functional" and "there is a lot of drift happening", so what he's trying to do ends up just being muddled.
I dunno, that's my best guess.
most of the genome is functional -> most of the human genome is functional
DeleteWhen someone likes that publishes a book with the title, The 'Random Genetic Drift' Fallacy, I pay attention.
ReplyDeleteWell, for a very small definition of "publish", anyway. Looking it up on Amazon I see that it is self-published by what used to be called a "vanity press". Not that professional publishers are a perfect filter by any means, but it is some sort of filter anyway.
This is especially weird because Provine has published extensively on the work of Sewall Wright. You would think he would be knowledgable about neutral evolution.
ReplyDeleteProvine is the major biographer of Sewall Wright. He was in charge of disposition and preservation of Wright's scientific archives after Wright's death.
Delete(My involvement in all this was that, when Will and I were graduate students at the University of Chicago, he in history of science, I in zoology, I was one of the first people to suggest he go talk to Wright. At that point he was surprised to hear that Wright was alive. He was indeed alive, and continued to be for about another 23 years.)
Will used Wright as a major source for his first book The Origins of Theoretical Population Genetics, then later he more extensively covered Wright's life, with Wright's cooperation, in his major work Sewall Wright and Evolutionary Biology.
I think Will has always felt that even then, he had a more correct understanding of genetic drift than Wright did. Now you had to get up awfully early in the morning to outthink Wright on that. I suppose Will is either brave or foolhardy.
Will sent me a copy of his book for comments, but I was busy and have not been through it yet. Until I read it and understand Will's argument, I'll stick with Sewall Wright for now.
By the way, I don't think that anyone need resolve the issue of determinism before asking whether random genetic drift exists. The issue is not whether chromosome segregations and births and deaths are really really truly random, but whether they can be adequately modeled as random. If you are willing to model chromosome segregation as random, that should inform your view of genetic drift as well.
The way I think about randomness in that context is with respect to the "information content" of chromosomes and the future.
DeleteUltimately it comes down to movement of DNA molecules in space. That movement can be physically deterministic or non-deterministic, but the important thing is that it happens with neither the content of these molecules nor their future phenotypic effects having an influence on it. When you have two alleles A and a, the chromosome segregation machinery has no knowledge which chromosome has A, and which has an a, and when mating and especially fertilization happen, the gametes have no knowledge about their respective genotypes. There are exceptions, of course, and sometimes they're important, but this is still the key distinction to be kept in mind.
"How to Think about Evolution" was the topic of the Sandwalk blog for Dec 19th 2014. In the following commentary I suggested that 'you might add to your reading list a leading US biohistorian's take: "The 'Random Genetic Drift' Fallacy," which traces the story from the early work of John T. Gulick in the 1880s (William B. Provine, 2014).'
ReplyDeleteIt is good to see that Sandwalk is now bringing forward Provine's new work. But to understand Provine it might be helpful to emulate his deep study of history, perhaps beginning with Gulick .
Donald Forsdyke has done valuable service in bringing figures like Gulick, Bateson, and Romanes to our attention. Reading the links to Gulick, I am not sure I can see genetic drift among the mechanisms he suggests. He does talk about local variations in natural selection, but a force equivalent to genetic drift is not obvious. Don, where do you find it in Gulick.
DeletePS about 8 years ago we were on Kaua'i, and in the town of Waimea I saw the old parsonage, now a museum, about a block from us. But we were in a hurry so I never got to see whether there were any materials of Gulick's, Probably not, since he spend his later years on Oahu.
Sewell Wright cited Gulick and Romanes with approval as primary sources. From this can be derived a view of evolutionary biology that, I suspect, Provine now finds himself much in sympathy. For my understanding of this see my text Evolutionary Bioinformatics .
DeleteI am assuming that Gulick and Romanes discussed "chance variations" without putting into them the substantive specifics of changes of gene frequency -- because genes would not be appreciated until the rediscovery of Mendel's work in 1900.
DeleteFor the Gulick-Romanes discussions please see my 2001 speciation text The Origin of Species, Revisited . For anticipations of Mendel, please see our 2008 biography of William Bateson Treasure Your Exceptions .
DeleteI think it's pretty clear what William Provine is saying. William Provine is saying that William Provine is way smarter than Kimura, Ohta, et al. Way.
ReplyDeleteHere's a storified version of Joel McGlothin's review of it from Twitter last year. He found the book frustrating and confusing in equal measures, and scientifically unsatisfying:
ReplyDeleteJoel's review
I xab;r help but note that s evolutionist, who teaches, saying read a creationist book is someone more to be wary of. it shows a inner sincere confidence about letting his students get the best case of his opponents. He is not censoring in deed or spirit.
ReplyDeleteHis confidence he can handle creationists, however perhaps, is mistaken like his apposition to drift.
yEC doesn't mind drift as long as its within kind and likely very trivial.
Here's a case where it seems to me genetic drift is real and does not immediately result from inbreeding. Imagine a population of 1000 individuals, 500 females and 500 males, mated in 500 pairs. Each pair has exactly two offspring. If we have a gene with gene frequency 0.5 in the parent population, the offspring generation will not necessarily have a gene frequency of 0.5. Rather the frequency will vary owing to random Mendelian segregation.
ReplyDeleteThose differences between parent and offspring generation are genetic drift, owing to Mendelian segregation at the locus. (The other possible sources of genetic drift, random births and random deaths, have been ruled out in this simplified case).
Can somebody persuade us that these random changes of gene frequency "are really inbreeding"?
If the population were able to expand indefinitely, without regard to resources, wouldn't the ratio remain constant?
DeleteJust asking.
No, it wouldn't remain constant.
DeleteIf we take my example and have each pair have 4 offspring instead, there are still changes between each parent generation and their offspring generation. However, as the population size doubles, then doubles again and again, the amount of change gets less and less. ultimately stalling out, in the limit.
To close approximation, the total amount of genetic drift in a case of repeated doubling of the population size is roughly equivalent to two generations of change in the original constant-size model.
Technical notes:
1. I am implicitly imagining that the offspring in each generation are paired up at random, with each pair of parents having 4 offspring, two males and two females, and the N males and N females are then paired at random.
2. The algebra can get messy but if we look at the gene copy level, each copy has 4 chances to get into an offspring, each of which is a 50:50 toss of a coin, with Mendelian segregation doing the tossing.
(After some algebra) the variance of gene frequncy change in each generation is proportional to 1/N(t), where N(t) is the population size in that generation, which is generation t. For the case of 4 offspring, the sum of reciprocals of N is:
1/N + 1/(2N) + 1/(4N) + 1/(8N) + 1/(16N) + ...
which is 1/N times 1 + 1/2 + 1/4 + 1/8 + 1/16 + ... which sums to 2. Thus the equivalence to two generations of change at the original population size of N when N is constant. This is an approximation for large N.
Joe,
DeleteAt what population size in sexually reproducing organisms would you estimate that the meiotic/mendelian Genetic Drift (mGD) becomes irrelevant as an evolutionary factor for alleles present at 0.5 frequency, in the absence of random births and deaths (bdGD) and selective forces?
@Joe Felsenstein
DeleteYou responded to the question above on another thread, so to continue the conversation, I'll post it here:
JF: Technically, never, if by "irrelevant" you mean completely irrelevant. The larger the population size the less relevant, but genetic drift caused by Mendelian segregation is never totally absent.
CB: "Thanks, but lets be practical and reasonable in predicting the evolutionary impact of mGD on alleles' frequency in context of other factors such as bdGD and selection. I would say that in populations of a few thousand individuals, the mGD is basically irrelevant from a biological (not statistical) perspective. Do you agree?
JF: No, I do not agree. In a normal Wright-Fisher model of genetic drift, it can be shown that the part of it due to the randomness of meiosis is half of the genetic drift. So the mGD and the bdGD (to use your terminology) are equal in impact.
Ok, lets consider a population of ten thousand individuals. In how many generation would you expect that mGD (under conditions you described above) would change the frequency of an allele from 0.5 to 0.6, or from 0.5 to 0.4, for example?
@Claudiu:
DeleteIn a population of few thousand individuals, the average allele drifts to fixation or loss in about 4N generations, or (if N = 3,000) about 12,000 generations. Lots of species have generation time about 1 year.
If there was only meiotic genetic drift, that accounts for half of the genetic drift, so it would talk about twice as long.
As for changes from 0.5 to 0.6, the first passage time is a bit of a nuisance to compute, but we can roughly say that once the gene frequency changes enough by genetic drift that one standard deviation of its change is 0.1, that is plenty long enough. That is when the variance of gene frequency (around the starting gene frequency will be 0.01. In one generation at N = 10,000, with only meiotic genetic drift, the variance accumulated would be 0.5*0.5/(4*10,000) = 1/160,000. So to accumulate that to 1/100 one needs 1600 generations.
With both kinds of genetic drift, about twice as fast as that.
Happy to oblige. So now you conclude ... (what?)
If we think of mGD in terms of a coin toss, as you suggested, intuitively we would expect that in a population of a few thousand individuals, an allele present at 0.5 will remain at the same frequency. Obviously, this intuition is wrong, so if we keep tossing a coin, eventually, we will only be getting heads or tales; correct? -:)
DeleteThe probability of hitting precisely 5000 heads in 10000 coin tosses is a mere 0.7978646%. It is thus far more probable that the frequency will change. And in the next generation there is actually a bias, because now there are more copies of one allele. In combination this eventually leads to the loss of an allele.
DeleteThanks Simon. So, the alleles that are present at less than 0.5 frequency can't be fixed by mGD. Correct?
DeleteOf course they can. For any allele the probability of ultimate fixation in the neutral case is equal to its frequency.
DeleteClaudiu,
DeleteNo, alleles at less that 0.5 frequency can be fixed, it is just less likely that they will be. It would make genetic drift irrelevant if things were deterministic, as you are thinking. After all, considering a point mutation, it occurs all by itself, and in a diploid organism, on only one copy of two chromosomes.
"we will only be getting heads or tales; correct?"
'tales' just a Freudian slip, there?
Kidding aside, though, you are still thinking in terms of an idealistic Wright-Fisher model, but considering only the outcome of a single mutation in one individual in a population.
Consider a point mutation happening in an individual in a population. By the Wright-Fisher model, this has a probability of fixing in the population of 1/2N, where N is the population size of diploid organisms. In a large population this is indeed a low probability. However, mutations per haploid genome occur at some frequency in all individuals in this population; let's call this mutation rate M. The total rate of mutations in this population is 2NM. Multiplying the population level mutations by the probability that any 1 mutation will fixate, we get (1/2N)(2NM) = M. Therefore the fixation rate of any given mutation in a population is the mutation rate.
You see, adding more individuals to a population decreases the chance that one particular mutation fixates, but those additional individuals also produce mutations, so that in the end, fixation rate = mutation rate.
Ok, let's take population with an effective size of 3000 individuals as described above by Joe Felsenstein. In this population, according to Joe, "the average allele drifts to fixation or loss in about 4N generations, or (if N = 3,000) about 12,000 generations". And, if we only consider mGD, it will take twice as long.
DeleteLet consider the same a population (i.e. 3000 individuals) in which a previously fixed allele A1 undergoes in one individual a neutral mutation to A2 (in one chromosome). Under the technical parameters described by Joe, how many generations it would take A2 to be fixed in the population and how many generations it would take A1 to be lost?
I think Joe made an error there, because the mean time to fixation in this case is about 12,000 generations, while the mean time to loss of is 17.4 and for no variation in offspring numbers between individuals these numbers would double. Note that the fixation of A2 is the loss of A1 and vice versa. So the numbers given above for fixation and loss of A2 can simply be reversed for A1.
DeleteLet's focus, for a moment, on concepts/principles rather that exact numbers. Simon, you mentioned above that "...in the next generation there is actually a bias, because now there are more copies of one allele. In combination this eventually leads to the loss of an allele."
DeleteSo. if there is a bias in the favor of the allele with most copies, then the probability for a neutral allele at relatively low frequency to become fixed by mGD in a relatively large population (e.g. 3000) would be infantisimal. True?
Claudiu Bandae
DeleteAfter reading your posts for a few months now I have no doubt that you are a really insightful fellow. I hope you don't mind if I ask you to state your position on evolution? You seem to be questioning the majority of the so-called neo-Darwinian evolution, but at the same time you often refer to evolution as a real process, at least in my mind. Would you feel comfortable stating your position on the matter in public? If not, you can email me with your response at tomashull987@gmail.com
Thank you for you great contribution on this blog.
So. if there is a bias in the favor of the allele with most copies, then the probability for a neutral allele at relatively low frequency to become fixed by mGD in a relatively large population (e.g. 3000) would be infantisimal. True?
DeleteHe already told you this. Read more closely. The probability is 1/(2N), or, if we count multiple instances of the same mutation, µ. But because there are so many mutations happening in the population, some of them do become fixed.
So. if there is a bias in the favor of the allele with most copies, then the probability for a neutral allele at relatively low frequency to become fixed by mGD in a relatively large population (e.g. 3000) would be infantisimal. True?
DeleteIt is small but not something you can dismiss. For a population of 3000 a novel allele has a frequency of 1/6000 and if it is neutral then it has a 1/6000 chance of eventually getting fixed. But there are a lot of novel (near) neutral mutations in that population each generation and as a result fixation of neutral mutations happens regularly. It's worth noting that this bias in favor of the wild type doesn't stop if there is selection. A novel allele in a population of 3000 would need s=0.35 to get to a fixation probability of 0.5. And that would mean that homozygotic carriers would need to produce about twice as many offspring as homozygotic non-carriers. That'd be an outlandlishly rare occurence. Given the distribution of values for S we find in the literature, about 1 in 10 million novel mutations reaches such a level (though estimates on these types of tails are hard and this estimate could easily be a few OOMs to the high side). An allele for which we would detect selection has a probability of ~1/9500 to get fixed.
Joe and Simon are making the point that, technically, we cannot *completely* dismiss the role of genetic drift in populations with effective population size of a few thousand individuals. The point I was trying to make is that, although this is true, mGD is negligible as an evolutionary force and that, even in relatively small population, the vast majority of alleles responsible for phenotypic traits are fixed primarily through adaptive evolution driven by natural selection, just as stated by Motoo Kimura. Is Kimura’s view wrong?
DeleteSimon, on a more specific point, you wrote above that “a novel allele has a frequency of 1/6000 and if it is neutral then it has a 1/6000 chance of eventually getting fixed”, and that “An allele for which we would detect selection has a probability of ~1/9500 to get fixed.” Are these numbers correct?
We don't know how many alleles affecting phenotype have been fixed in the human lineage and we certainly don't know what percentage of those were adaptive.
DeleteHowever, of all the alleles affecting phenotype that are currently segregating in our species, most seem to be non-adaptive. In the absence of evidence to the contrary, it's safe to assume that this was true in the past.
True, but what is the evidence that most of the current alleles affecting the phenotype are non-adaptive?
Delete"True, but what is the evidence that most of the current alleles affecting the phenotype are non-adaptive?"
DeleteBecause across the great variation of human phenotype, most of it does not seem to interfere with reproduction and rearing offspring successfully to reproduce themselves. Do you consider this a controversial statement?
I somehow mixed up the numbers. For the allele under significant selection it's 1/2594.
DeleteChris, it is not a controversial statement; however, phenotypic variation in human populations is primarily the result of natural selection.
DeleteI will be interested to discover how you know that.
DeleteI've usually disagreed with Claudiu in his arguments here. On this one I'm not so sure. In a population of (effective) size N, selection coefficients down to about 1/(4N) have effects on the gene frequencies. If human populations have had effective sizes as small as 10,000 about 100,000 years ago, that means selection coefficients as small as +/- 0.000025.
DeleteJohn, Larry, and Chris B, how do you know that phenotypic variations large enough for you to see don't differ in fitness by this amount?
Kimura's statement, which Claudiu is citing, is probably made with population sizes of nonhuman organisms in mind, so more like N = 1,000,000 and |s| > 0.00000025. Just because we don't notice any effect on fitness offhand isn't a string argument.
I meant to type "... a strong argument."
DeleteI am not invoking string theory.
@Joe
DeleteI can't rule out the possibility that all phenotypic variation in humans is due to underlying alleles that are being selected, negatively and positively, due to tiny selection coefficients.
But it doesn't make a lot of sense. Do you really believe that all distinguishing features of Nigerians are maladaptive in the Japanese environment? How about Nigerians living in Alabama compared to the native Americans who used to live there?
What's so special about alleles that are responsible for "phenotypic change" that makes them so different from all other alleles in the functional perts of the genome? Doesn't the burden of proof lie with people like you and Kimura who are making the extraordinary claim that almost all phenotypic change is adaptive?
Can we clarify what we mean with "phenotypic" changes here? Because I would count functional differences, even if they aren't leading to huge morphological differences. So I'd count things like lactose tolerance, although that's not something that produces a visible change.
DeleteThis in turn would make Kimuras view pretty tautological - by definition functional loci are under significant selection. If we restrict "phenotypic chance" to morphological change then it's not very clear - we don't really have an inventory of loci that are relevant here and therefore don't have the option of looking what the distribution of selection coefficients looks like.
What is clear is that stating that the alleles responsible for phenotypic change are a subset of the functional genome implies they are under significant selection, Larry.
Do you really believe that all distinguishing features of Nigerians are maladaptive in the Japanese environment?
DeleteThe current Japanese environment, probably not. But that's not where they evolved. It is perfectly plausible that visible differences were adaptive when they arose, in the region and population in which they arose. The formulae show how much more powerful selection is even at levels way below anything realistically detectable when trying to make a causal link between a trait and its survival value, and although they don't prove anything, they do argue for caution in dismissing non-drift causes.
It does stretch credulity somewhat that the differences in eye shape (say) result from adaptation to local conditions, but there are other non-drift causes to consider - pleiotropy, sexual selection for example. And part of the difference may also be attributable to contingency - the differential mutations arising in the separate populations, which again, is not drift per se.
Typically "phenotypic change" refers to traits that could possibly be under selection so, as Simon says, it's a tautology. When Dawkins uses the phrase, he means anything that's visible.
DeleteI think that things like blood types and amino acid substitutions are phenotypes so that makes the discussion complicated.
I suppose we should ask those who make the the distinction to explain their meaning. What alleles in the E. coli genome are responsible for phenotypes and which ones aren't?
Allan Miller says,
DeleteThe current Japanese environment, probably not. But that's not where they evolved.
So, do you agree with me that alleles for many obvious phenotypes are currently neutral?
And part of the difference may also be attributable to contingency - the differential mutations arising in the separate populations, which again, is not drift per se.
How do those alleles become fixed in the local population?
Professor Moran writes:
DeleteTypically "phenotypic change" refers to traits that could possibly be under selection so, as Simon says, it's a tautology. When Dawkins uses the phrase, he means anything that's visible.
I think that things like blood types and amino acid substitutions are phenotypes so that makes the discussion complicated.
I doubt Dawkins takes such a simplistic view. Not to speak for him but I suspect he agrees with you that any phenotypic trait that was beneficial in the current niche would be visible (that is subject) to selection.
It's hard to pin down exactly what Dawkins means even when talking to him directly. I think his views had modified over the past few decades but I'm not certain.
DeleteRichard Dawkins' View of Random Genetic Drift
Richard Dawkins on Visible Changes and Adaptationism
Strangely, because your post on Will Provine's book prompted me to try and understand the dole of drift in adaptive evolution, I dug out <1>The Greatest Show on Earth a couple of days ago to see what Dawkins' most recent view was and checked the index and found there is no specific reference.just as you did.
DeleteIs there a good source to better understand the role of drift for the layman?
So, do you agree with me that alleles for many obvious phenotypes are currently neutral?
DeleteHard to say, isnt it? The maths of the situation caution against being dogmatic either way. A selection coefficient of 1 extra birth in 1000 produces a significant selective effect, but you'd be hard pressed to notice that level of discrepancy.
But to take a particular example, I don't think that skin colour is currently neutral. There is a disproportion of Vitamin D deficiency among darker-skinned people in high latitudes, likewise more skin damage on white skin in equatorial regions. That said, people are maybe not the best example species, due to society, clothing, etc.
Me: And part of the difference may also be attributable to contingency - the differential mutations arising in the separate populations, which again, is not drift per se.
Larry: How do those alleles become fixed in the local population?
Again, hard to say. We don't know what the competing alleles were, nor the selection coefficients. But either way, selection and drift can only promote what turns up.
The most sensible position to take, as far as I can tell, is that we don't know what proportion of phenotypic features are under selection in any population of any species. (Though I'm willing to be disabused if anyone can cite a clear counterexample, with evidence.) The great thing about molecular evolution is that there are clues to neutrality in comparing sequences among individuals and species. But that doesn't seem to work as well when considering phenotypes. Conceivably, genetic load arguments might help. But so far all anyone has presented here are claims, backed up by, at most, quotes from famous people.
DeleteThere isn't. Drift is extremely hard to explain, but it is not that hard to explain why it is so hard to explain.
DeleteLet's start with the basics: You have a population of organisms and the number of offspring the ith organism has is a random variable Xi. We note that we can sum the X for organisms that have AA at some locus, for those that have Aa and those that have aa (or more, if there are more alleles for that locus) to get random variables that describe the total number of offspring for each of these groups. And we can easily use them to get random variables for the number of copies of A and a.
This then gives us a distribution for the allele frequency of A in the next generation.
Now, if you do this you end up with a population genetic model of population resampling, which is the same as drift and selection combined.
Let's say our allele frequency at time t=0 is p and at t=1 it is the random variable p1. We can then define the change in allele frequencies as d=p1-p.
Now, we usually make some simplifying assumptions along the way, which makes d easier to understand. For instance in the diffusion approximation d is a normal distribution.
Now, for all random variables X that have an expected value E(X), we can write X=E(X)+(X-E(X)) (we simply add and subtract E(X)). So for our change in allele frequencies d this works out as
d=E(d)+(d-E(d))
This decomposition is sometimes useful (d-E(d) is centered, which is a prerequisite for using the central limit theorem. Which makes it a normal distribution and that's how we get to the diffusion approximation in the first place).
Now here is the big issue: We call E(d) selection and d-E(d) drift (it's worth noting that there is a similar decomposition in physics, but it calls the part we call selection drift and the part we call drift diffusion).
Now the E(d) part is also relatively easy to describe. But drift is the difference between the full model and the selection bit. That is tricky. The analogy I always use is a silhouette of a wine glass with the silhouette of a air freshener tree cut out. It's a bugger to describe in a simple way, unless you say that it's the silhouette of a wine glass with the silhouette of a air freshener tree cut out...
It's generally not just something that does require doing some math, it requires doing some ugly and unnecessary math, because in between you have worked out the combined effect of selection and drift, which is nearly always a far more elegant description than your decomposition and because it describes both selection and drift gives you more "bang for your buck" as well.
Don't know whether this random find is of any interest:
Deletehttp://www.molecularecologist.com/2013/10/random-drift-and-phenotypic-evolution/
John Harshman says,
DeleteThe most sensible position to take, as far as I can tell, is that we don't know what proportion of phenotypic features are under selection in any population of any species.
Yes, that's the "sensible" scientific position. Let's all remember that whenever anyone claims that all phenotypic features are adaptive and starts telling just-so stories.
However, there can still be some serious discussion about whether it's "reasonable" to assume that all phenotypic variation is adaptive. And it's reasonable to debate whether the null hypothesis should be adaptation or random genetic drift.
Surely the null hypothesis must depend on the test hypothesis, not on some abstract and universal principle. If we're testing for selection, neutrality is a great null hypothesis, and we suppose selection only if we can reject it. If we're testing a hypothesis of neutrality, it becomes more difficult to frame a null hypothesis. All we can say would be that selection is below a level detectable by that test.
Delete@Larry
DeleteLet me add a citation from another founder of the field of population genetics, Ronald Fisher who wrote:
“Evolution is progressive adaptation, and consists of nothing else”
In this thread, I quoted 4 renowned scientists, Fisher, Wright, Kimura and Felsenstein, who have founded or developed much of the thinking in population genetics, and all of them support the view that natural selection is the primary driving force in evolution.
Are all these scientists who have spent most of their life establishing the theoretical foundations of the field of population genetics, including the concepts of genetic drift and neutral evolution, wrong?
John,
DeleteIf I remember well, in previous posts you have argued that all functional genome sequences are products of selection. Have you had a change of heart?
Why all these arguments from authority? ("Are all these scientists who have spent most of their life establishing the theoretical foundations of the field of population genetics, including the concepts of genetic drift and neutral evolution, wrong?")
DeleteWhat you've presented above ("“Evolution is progressive adaptation, and consists of nothing else”") is a definition of evolution, one that seems excessively limited to me and doubtless to many others, no matter who used this definition.
Drift seems very common. It seems like it must be common, given the numerous differences among taxa that seem roughly proportional to the time they have been diverging. Drift usually doesn't help organisms be more adapted to their environments. However, it plays roles. Alleles that increase in frequency due to chance alone may turn out to be beneficial in a different environment or what a new allele shows up at a different locus, for example. And then we'll call the further change in frequencies of that allele the result of selection.
It's not really an either/or question. The forms of resampling we call drift and those we call selection both happen. Selection produces adaptation. Drift is far more common, but usually has little effect on adaptation. Once in a while, though, drift plays a critical role in the evolution of new processes. It results in new features selection will act on. Both drift and selection contribute to change in populations over time (evolution).
If I remember well, in previous posts you have argued that all functional genome sequences are products of selection. Have you had a change of heart?
DeleteNo, you just remember poorly. All functional genome sequences are in fact products of mutation. Their fixation is very likely to be a product of selection, though.
@bwilson295
DeleteI hear you, but the arguments and the views presented by Fisher, Wright, Kimura and Felsenstein are based on deep understanding of population genetics, “random genetic drift” and “neutral evolution” concepts. How do we know that? Well, they are the scientists who put them on the map, so obviously they knew what they are talking about. How can you argue with that? (Apparently, you can't, because even 'knowledgeable scientists' like Larry, who have tried to question these authorities are running out of valid arguments).
After originating and developing the theoretical foundation of population genetics and the concepts of “random genetic drift” and “neutral evolution”, these great scientists concluded that natural selection is the primary force in evolution, or to put it in the insightful and metaphoric words of Fisher: “Evolution is progressive adaptation, and consists of nothing else”
The scientific integrity of these scholars is just extraordinary as they minimized the significance of their own concepts and work, and recognized that Darwin’s work remains supreme. Again, this is just remarkable.
Claudiu Bandea says,
Delete... these great scientists concluded that natural selection is the primary force in evolution, or to put it in the insightful and metaphoric words of Fisher: “Evolution is progressive adaptation, and consists of nothing else”
Well if that's what Fisher said then he's just wrong. As for Wright, Kimura, and Felsenstein, if they mean that more alleles are affected by positive and negative selection than by random genetic drift then they are probably wrong as well. However, this depends on how much junk is in a genome and how you define "evolution."
Again, this is just remarkable.
What's remarkable is your steadfast refusal to move into the 20th century, let alone the 21st.
Claudiu's continuing reliance on the argument from authority is not encouraging. Of course one of those authorities is conveniently present and able to speak for himself, which is what Claudiu ought to be doing.
DeleteJoe Felsenstein did speak clear and 'loud' (see above) but, apparently, you and Larry didn't hear him.
DeleteIt just happen that on this issue I agree with him and with Fisher, Wright, and Kimura, at least partially. So, I'm speaking primarily for myself.
Nevertheless John, I'm really happy to second you on your statement above:
"All functional genome sequences are in fact products of mutation. Their fixation is very likely to be a product of selection, though."
Let's see if Larry agrees with you; do you Larry? Does anybody else agree with that?
"All functional genome sequences are in fact products of mutation. Their fixation is very likely to be a product of selection, though."
DeleteI think I don't agree. There are functional regions of the genome (spacers) whose sequence is irrelevant. If you simply say, "All functional regions are subject to selection even if the actual sequence is unimportant," then I can agree with that statement.
BTW, I "heard" Joe speak and he didn't say what you think he said.
@Larry and John
DeleteSo, if the functional genomic DNA sequences are fixed by selection, then genetic drift fixes non-functional DNA sequences?
More correctly, it fixes DNA sequences whose differences do not matter to any functions they may have. Or more strictly, it fixes sequences whose differences in fitness are small enough that selection is not significant, "not significant" depending on population size.
DeleteI feel you're trying to lead me into some kind of trap. Why not just spring it now and see what I think?
If natural selection acts on all functional genome sequences, and the genetic drift acts on sequences whose differences do not matter to the functions they may have, then it appears to me that natural selection is the primary force of evolution and genetic drift represents the stochastic dimension of natural selection.
DeleteThat all depends on what you mean by "primary force". I would agree, though Larry may not, that natural selection is the primary force leading to adaptation. But if we count by number of fixations, then fixations by drift far outnumber fixations by selection, simply because most of your genome is junk, and even within functional sequences, many changes are neutral or nearly so.
Delete@Claudiu Bandea
DeleteYou seem to have trouble following an argument and reading for comprehension. You should try to improve your skills because it makes you look bad.
Nobody in this thread ever said that natural selection was the only mechanism of evolution acting on functional sequences. Nobody but you would ever believe such a ridiculous idea.
@Larry
DeleteAs a host of this blog, perhaps you should delete your own comment because it does not contribute the discussion. Apparently, though, as stated by John, you might be among the few 'contributors' on this blog who doesn't accept the scientific fact that that natural selection is the primary force leading to adaptation.
This is what John wrote: "I would agree, though Larry may not, that natural selection is the primary force leading to adaptation."
@Claudiu Bandea,
DeleteLarry's comment contributed pointedly to this over-long discussion. Your issues have been explained pretty thoroughly to you but you have been talking in circles. Early on in the thread, you wanted to focus on numbers and probabilities. Maybe you thought that fixation by genetic drift was a highly improbable event and that people here didn't know what they were talking about. In any case, popgen 101 shows the emergent property that fixation rate = mutation rate at the population level. Then you didn't want to focus on the numbers so much and talk about 'concepts'. I don't know how many different ways it can be explained, but genomic evidence shows pretty clearly that fixation of neutral alleles by genetic drift is a significant 'force' of genomic change. How much of that is adaptive is like trying to guess how many horses in the world are running. It's a real number, but trying to pin it down is logistically more difficult. As Joe Felsenstein pointed out earlier in the thread, even very small selection coefficients can be significant in producing change in populations, and small selection coefficients would be hard to disentangle from drift in real world populations. And as Larry also pointed out, a great deal of morphological variation within populations and differences between populations is most likely due to fixation of more or less neutral variants.
Apparently, though, as stated by John, you might be among the few 'contributors' on this blog who doesn't accept the scientific fact that that natural selection is the primary force leading to adaptation.
DeleteI have no idea where John Harshman ever got such a bizarre idea. For the record, I believe that natural selection is the primary, but not the exclusive, cause of adaptation.
Larry: “I have no idea where John Harshman ever got such a bizarre idea”
DeletePerhaps from being active reader and contributor to this blog, or you insinuate that he also has “trouble following an argument and reading for comprehension”.
However, I’m pleased to know that you believe that natural selection is the primary cause of adaptation.
DeleteFor the record, I didn't get that bizarre idea. I was just being cautious in attributing notions to you. Why is Claudiu trying to start a fight?
DeleteDon't be silly John, this is not about 'traps', 'fights', or whatever, but about science. So now, we are all clear: Natural Selection is the primary force leading to adaptive evolution, period. (Perhaps Larry should write a post with that title).
DeleteThere is, however, one more thing to clarify about the forces of evolution. Two years ago, we had the following discussion, which was not resolved:
Claudiu Bandea, Tuesday, June 04, 2013 5:41:00 PM:
In his upcoming book, Masatoshi Nei outlines Mutation-Driven Evolution as follows:
“…many evolutionists including Motoo Kimura and Jack King believed that phenotypic evolution (in contrast to molecular evolution) is caused primarily by natural selection. By contrast, Nei (1975, 1987, 2007) proposed that since phenotypic evolution is ultimately controlled by DNA and RNA molecules, both molecular and phenotypic evolution must be primarily caused by mutation. Nei’s view has been based on the new findings in the study of molecular evolution and developmental biology. He considered all kinds of DNA changes (nucleotide substitution, gene duplication, polyploidization, epigenetics, etc.) as mutations and tried to explain all phenotypic evolution by mutation. Previously I called this view neomutationism (Nei 1983, 1984), neoclassic theory (Nei 1987), and the new mutation theory (Nei 2007), but in this book I have decided to call it the theory of mutation-driven evolution or neomutationism depending on convenience” (first parenthesis mine)."
John Harshman, Wednesday, June 05, 2013 4:36:00 PM:
"Larry, this seems to conflict seriously with your (and my) preferred definition of evolution. Nei seems to say here that mutation is evolution, and that the spread of those mutant features within a population is unimportant. I'm not sure what new findings support that view, but I guess we'll see."
Do you and Larry care to clarify your position?
What is unclear about my position? Evolution happens to populations (or higher-level units). Mutation happens to individuals.
DeleteSo, can you explain your comment:
Delete"Larry, this seems to conflict seriously with your (and my) preferred definition of evolution. Nei seems to say here that mutation is evolution, and that the spread of those mutant features within a population is unimportant.
Is Nei's view about evolution flawed?
Is Nei's view about evolution flawed?
DeleteI think so. Of course I haven't read the book. But as described by Larry it sounds strange to me. What about you?
John,
DeleteLike you, I haven’t read Nei’s book “Mutation-Driven Evolution.” However, two years ago in the post I mentioned above (http://sandwalk.blogspot.com/2013/06/mutation-driven-evolution.html), Larry wrote: “I can't wait to get my hands on a copy of this book. Look for a review in a few months”, so I presume he read Nei’s book, although I don’t think he wrote a review on, but I might be wrong.
I any case, according to his post, Larry is a student of Nei’s work, so it would be great if he let’s us know his view on Nei’s theory. Is this theory flawed?
@Claudiu Bandea
DeleteYou don't listen to, or believe, anything I say so why in the world would you want me to tell you about mutationism when you can go directly to the source? Read Nei's book yourself. If reading a book is too challenging, you can read this blog post and follow the links within it.
Arlin Stoltzfus explains evolutionary theory
Don't expect me to spoon feed you if you don't bother to do your homework. You've already demonstrated that you can post comments on this blog for years without ever bothering to understand random genetic drift.
Here we go again.
ReplyDeleteEvolutionists can’t agree on how evolution happened but they agree on one sure thing that evolution did happened.
How? It is a matter of personal speculation and interpretation called science.
Genetic drift can play some evolutionary role in small populations and probably accounts for some non-adaptive features of DNA.
As purely random process, genetic drift can’t cause the evolution of adaptations. It could never build a wing or an eye. That takes nonrandom natural selection. What drift can do is cause the evolution of features that are neither useful nor harmful to the organism.
And so on.
Oh, no criticism? I must be right then and now. ;-)
DeleteBut Larry is working on a book. I'm pretty sure he is not going to preach that the drift is the main evolutionary mechanism because his buddy Coyne just published his book on facts vs faith and he promotes natural selection and he totally disregards the drift as having almost no influence on evolution at all when evolution needs to build new things, such as wings out of almost nothing.
DeleteIt is very interesting how the world of evolutionary bullies works.
Larry and Coyne have just had lunch in Vancouver Canada few weeks ago. One inquiring mind would ask if they ever talked about their humongous differences when it comes to the mechanism of evolution. None has ever been officially mentioned.
Well, what's new?
Is there a case, as Provine attempts, for suggesting that "random genetic drift" is synonymous with inbreeding. Does drift only become significant in small populations?
ReplyDeleteI have skimmed Provine's book, and I can say a few things about it. I would not imagine for a moment that Provine is some kind of fool meddling in something he doesn't understand. He is making a conceptual argument.
ReplyDeleteThe background for considering this *kind* of argument is that generations of scientists can be wrong or confused about something conceptual. For instance, the idea of a "rate-controlling" or "rate-limiting" step in a biochemical pathway is invalid: the rate of flux through a pathway can be affected by changing the rate of any step, not just one key step (metabolic control theory provides a theory for this). Yet, the idea persists. Likewise, sometimes we rely on concepts that are pure fantasy but aren't questioned because they don't seem to be causing any problem. For instance, look at a reaction energy diagram and ask yourself what is the meaning of the x axis, the "reaction coordinate". It has no physical meaning. It does not represent time, for instance.
Concepts like "drift" or "entropy" are intangible. They don't have a color or a shape. We make up equations using these concepts and they may be useful, but this doesn't mean that we understand the concepts. It doesn't mean that they correspond to anything real in the world. They could be just useful fictions.
Provine is clearly trying to argue that the concept of drift is in some way invalid or confused. In his account, the historic shift from references to "inbreeding" to references to "drift" is the sign of this confusion. Inbreeding is something tangible, and drift is this intangible thing that is treated as a force.
But that is all I can say. Like Larry, I tried to read this and just could not understand what Provine is really trying to say. Eventually I put down the book and stopped trying. I wish he could state the argument in terms of a critical test that the concept of drift clearly fails.
Provine’s book is a fascinating historical presentation of the “random genetic drift” concept, which according to the author has been central to the field of population genetics. Provine has spent a half century or so studying the history of population genetics and he has interacted for ten years with Sewall Wright, one of the founders of population genetics. So, it would be an understatement to say that Provine is knowledgeable about the historical developments in the field of population genetics.
DeleteHowever, apparently, few people seem to understand his thesis that the concept of "random genetic drift" is flawed. This is unexpected considering that, as an historian, he should be highly skilled in presenting facts and concepts as clear as possible. Indeed, even Arlin who is a thoughtful and tenacious scholar has given up the effort to make sense of Provine’s view.
I just read Provine’s book and I have similar concerns. However, it is likely that we are missing something here. Perhaps, in Provine’s mind his thesis is so obvious, that he neglected to present it in a clear mode.
I think the key in understanding Provine’s thesis is to start with the way he defines "random genetic drift". And he does that, as you would expect, at the beginning of his book:
"Fisher's first paper on population genetics (Fisher 1922) showed his belief that inbreeding (loss of chromosomes) could be modeled by "fortuitous extinction of genes" (I will call this term "random genetic drift" in this book) using a genic F he attached to a chromosome. Fisher's intent was to make "random genetic drift" measure the inbreeding effects in a population, but inbreeding is tied to whole chromosomes and Mendelian heredity, and "random genetic drift" is tied to the locus F. Fisher invented locus F in 1922 to represent any variable in the population including inbreeding and out breeding, selection, survival of individual genes, neutral genes, Hagedoorns's argument, and many other issues in one short paper. Fisher's genic F become a fundamental assumption of population genetics then and to the present."
In brief, Provine makes the case that Fisher, Wright and the other scientists working in the field of population genetics mistakenly conflated “inbreeding” with "random genetic drift" (as defined by him and the founders of population genetics: Fisher, Wright, Haldane and later on by Kimura and Ohta).
If that is true, as supported by the historical facts presented by Provine, then, the "random genetic drift" is indeed a fallacy.
"inbreeding (loss of chromosomes)" -- That can't be right. Is the error Provine's or yours?
DeleteBelow is a link to an article about inbreeding. Whether all of the statements in it are correct I don't know, but I thought that some of you might find it interesting:
Deletehttp://io9.com/5863666/why-inbreeding-really-isnt-as-bad-as-you-think-it-is
@Barbara, the text "inbreeding (loss of chromosomes)" is Provine's, and I don't think it's an error.
DeleteClaudiu If that is true, as supported by the historical facts presented by Provine, then, the "random genetic drift" is indeed a fallacy.
DeleteDon't really get this (though I confess I haven't the inclination to read the book). In any random-mating population, you must be mating with relatives at some remove or other, in order for the union to be fertile. That's not what is typically meant by inbreeding, but nonetheless it has an unavoidable element of drift (sample error). Meanwhile, the apparent inbreeding displayed by any finite population has an upward tendency, simply because variation is steadily eliminated without new mutation.
On a historical note, Wright did regard drift and inbreeding as broadly the same - see this nice article by James Crow.
Alan, indeed a nice article, which refers several times to Provine's work on historical developments in population genetics.
DeleteIt would be great if you, Joe, Simon, Larry, Arlin and other interested readers would further evaluate Provine's thesis about the "random genetic drift" in context of our discussion here, so we can make some progress.
Stimulated by this conversation, I read Wright's famous 1932 paper "Evolution in Mendelian Populations" (http://www.esp.org/foundations/genetics/classical/holdings/w/sw-31.pdf), and I want to bring forward a quote that I think remains contemporary after all this time:
"The conclusion seems warranted that the enormous recent additions to knowledge of heredity have merely strengthen the general conception of the evolutionary process by Darwin in his exhaustive analysis of the data available 70 years ago."
@Claudiu Bandea
DeleteIt's time to come clean with your own view of random genetic drift instead of hiding behind quotations from the past. Human and chimp genomes differ at milllions of loci. By what evolutionary processes did these alleles become fixed in their respective modern populations? Give your view on the approximate contributions of each process that you identify.
Different human populations differ considerably in phenotypes under genetic control (think Japanese vs. Nigerians). Of all the phehotypic variation in human populations, what percentage of alleles do you think are adaptive and/or detrimental? What percentage do you think are segregating as if they were neutral?
What process is responsible for the changes in allele frequency of neutral alleles? What process is responsible for the loss of the majority of new advantageous alleles in a population?
@Larry
DeleteIt's time to come clean with your own view of random genetic drift instead of hiding behind quotations from the past.
I can't wait to hear Caldiu to answer your questions even if they are indirect.
Larry,
DeleteAlthough I have occasionally commented on "random genetic drift" on your blog, I have yet to fully present my view on this concept here or elsewhere. However, it just happen that I'm in process of writing on paper addressing the "random genetic drift" and "neutral evolution" concepts. As a matter of fact, I'm contemplating now the idea that maybe I should experiment with a 'real-time' approach of writing the article right here at Sandwalk in context of this post on Provine's book, The "Random Genetic Drift" Fallacy. One way or another, though, I'll be answering your questions.
As you mentioned, I posted some "quotations from the past" on the driving forces in evolution; however, I did that not to "hide behind them", but to emphasize the fundamental ideas promoted by the most prominent thinkers in the field population genetics, including Wright and Kimura.
Here I want to quote an a contemporary giant in the field, who happens to be a frequent contributor to Sandwalk, including this particular post. So, here is what Joe Felsenstein wrote (http://evolution.gs.washington.edu/pgbook/pgbook.pdf):
Natural selection is the evolutionary force responsible for the progressive adaptational aspects of evolution - for the fact that organisms are as good as they are at surviving and reproducing.
Natural selection "is the primary force which causes evolution to be adaptive, the creative and progressive element in the evolutionary process".
Hi Claudiu
DeleteI'm struggling with the idea of genetic drift. I have a blog post up here and I wonder if you'd be interested in having a look.
It would be great if Professor Moran had a minute to look in, too!
DeleteHi Alan,
DeleteThanks for the link to your post and the associated comments, which are very interesting. It’s amazing that after almost a century of thinking and writing about “random genetic drift”, it is still a difficult concept to define; as pointed out by Arlin at the beginning of this thread that is true to many concepts. [Parenthetically, and to add to Arlin’s examples of scientific fallacies that persist despite being flagrantly wrong, you might want to take a look at this paper ( http://precedings.nature.com/documents/3886/version/1) in which I tried to bring attention to what might be one of the most enduring misconception in biology, if not the entire science; for a summary you take a look at my comments here: (http://sandwalk.blogspot.com/2015/05/ford-doolittle-talks-about-tree-of-life.html)].
So according to Provine and apparently the founders of population genetics, the “Random Genetic Drift” refers to "fortuitous extinction of genes" (see quote above). To keep the same approach, we now could define Random Genetic Drift (GD) as fortuitous extinction or fixation (and any fluctuation in between) of genes and alleles.
First, I want to point to the obvious, that even if GD never leads to fixation or extinction, nevertheless, it still might be a significant force in evolution. And, I think it would makes sense also to think of GD not only as changing the frequency of alleles (at a locus), but also of its potential role in the extinction or fixation of genes per se (think in terms of new genes coming into existence or disappearing, and the forces associated with that).
Second, I want to emphasize that, theoretically, GD acts on all alleles/genes, whether they are “neutral” (as usually implied) or under positive or negative selection, or whatever other forces one might think of. Technically, GD is always present, but its effects might or might not be relevant for evolution.
I’ll add more to this comment, but does it make sense what I wrote up to this point? Also, I’m thinking of posting some comments on your blog; can I re-post some of the comments I made here?
Claudiu writes:
Delete...does it make sense what I wrote up to this point?
Unfortunately I have RL to deal with but will follow your links later tonight. I'm a layman so don't rely on my opinion. However you would be very welcome to join the discussion. Not my blog, BTW. Blog-owner is neuroscientist Dr. Elizabeth Liddle. I
There must be something I am missing here. I find that the process so clearly described by David Suzuki above (I almost forgot he was a regular, working scientist) to be almost self-evident. I have no difficulty understanding it, unless that is an illusion created by my complete inability to understand it. However, the confused and turgid writing above from of doubters like Purvine and Claudiu Bandea make me suspect the fault is not mine. "Technically, GD is always present, but its effects might or might not be relevant for evolution," Claudiu? Seriously, what is that even supposed to mean?
Delete@Alan,
DeleteI have no doubt that you have the knowledge to understand GD, and it is very important that this concept is presented in such way that basically anyone with interest in the subject would understand it. Obviously, English is not my 'forte' -:), but there are other people, such as Allen Miller and Joe Felsenstein, who have commented both here and at your post, and are not only knowledgeable but excellent writers. In any case, I'll be interested to know what you think regarding the points I was trying to make about GD.
So, Claudiu, what you're saying is that most of the traits subject to selection are fixed by selection. Who can argue with that? Nevertheless, the overwhelming majority of fixations in the human genome have been neutral, and the number of neutral fixations doesn't depend on population size. Nearly neutral fixations do, though. Maybe you should look at that. I'm not sure what you intend to convey by your apparent rejection of modern population genetics.
ReplyDeleteJohn,
DeleteI said that "the vast majority of alleles responsible for phenotypic traits are fixed primarily through adaptive evolution driven by natural selection, just as stated by Motoo Kimura."
It appears that one of you, John and Claudiu, is either ignorant of or simply does not understand modern population genetics. I know where my bet is...
DeleteI fell flattered, latesuite, but John has a lot to offer, too, if you give him the chance -:)
DeleteJust heard the sad news that Will Provine passed away yesterday. For more information see:
ReplyDeletehttps://caseybergman.wordpress.com/2015/09/01/from-the-library-of-prof-william-b-provine/
Will was a good guy, and the preeminent historian of theoretical population genetics, particularly in his excellent books The Origins of Theoretical Population Genetics and Sewall Wright and Evolutionary Biology. They are well-thought-out and very readable.
ReplyDeleteI hope to post at Panda's Thumb soon some recollections of my interactions with him over the years. Others will probably write interesting obituaries.
Will had struggled for years with brain cancer, which he was very open and brave about. I think that it affected the clarity and thinking in his recent self-published book The Genetic Drift Fallacy. I did not want to say that earlier, for obvious reasons.
Will also, over the years, was convinced at various times that he had a more correct view of adaptive surfaces and of genetic drift than Sewall Wright did. Put simply, he didn't -- he hadn't corrected Wright's views.
I will miss Will, a brave and forthright person, if occasionally wrong about things technical.
Thanks, Joe. I will look forward to reading your thoughts on Panda's Thumb.
DeleteHe's saying something, albeit rather incoherently, about constructivist mathematics, which underlies the neutral theory. I suggest you read what is now a classic critique of this mathematics: A. Garciadiego, BERTRAND RUSSELL AND THE ORIGINS OF THE SET THEORETIC 'PARADOXES.'
ReplyDeleteProvine had decided that the population-genetic models of Fisher, Haldane and Wright were not merely mathematically wrong, but were conceptually wrong. This contradicts what he wrote in his biography of Wright where he wrote, "I would emphasize in conclusion that Wright's shifting balance theory of evolution in no way depends upon the usefulness of his fitness surfaces as heuristic devices. The shifting balance theory gains its place among the few really robust theories of evolutionary change…" Towards the end of his life Provine concluded that Wright's shifting balance theory was based on a fundamental confusion on Wright's part between genetic drift and the effects of inbreeding (followed by cross-breeding) that he (and Jones and East, and the Hagedoorns) had observed in the nineteen twenties. According to Provine, Wright knew (or should have known) that these inbreeding and cross-breeding effects were due to recombination during cross-over in meiosis and were quite different from the effects of genetic drift based on the Fisher's conception of a very large population associated with an unstructured gene pool. Provine concluded that Wright wanted to create a theory essentially the same as Fisher's but with a focus on the division of a large population in small subpopulations which would follow the patterns observed in breeding experiments. However, Wright ignored meiosis and attempted to model these effects on genetic drift, then further confused the matter by introducing his model of adaptive (or fitness) surfaces in the early nineteen thirties. Late in the process of writing his biography of Wright, Provine concluded that Wright's graphical presentations of adaptive surfaces were incoherent, and he claims that Wright agreed with him. I found Provine's critical analysis of adaptive surfaces in his biography of Wright to be clear and convincing.
ReplyDeleteThe" Random Genetic Drift" Fallacy is completely different in style from Provine's biography of Wright. It appears to be a first draft, written in a polemical style, that badly needs editing. A fully developed argument would have to be at least twice as long, and considering the state of Provine's health, he probably should have sought out a collaborator. I believe I understood most of what Provine wrote, and perhaps if I were an evolutionary biologist better acquainted with meiosis, I would understand what he meant when he repeated talked about "losing chromosomes" in meiosis. Provine appears to be angry with himself and exasperated that Wright stuck with what Provine concluded was a worthless and misleading analysis. Provine's description of how experiments were misinterpreted to support Wright's analysis and how most evolutionary biologists failed to point out obvious problems with Wright's analysis is an indictment of much of the work of twentieth century evolutionary biology.