Molecular Evolution Activities

The Early Reception of the Neutral Theory
What was your initial reaction to the Kimura (1968) and/or King and Jukes (1969) papers? Why? Please click on the "Add a comment" link.

Click below for relevant online documents:

John Beatty

Response to Non-Darwinian Evolution
The initial response to the non-Darwinian evolution or the neutral theory was caused by the lack of understandings among researchers of different fields. Evolution and population genetics people were accustomed to think in terms of neo-Darwinian model. To them, any mutations were either advantageous or deleterious, and no intermediate class existed. On the contrary, biochemistry people did not know population genetics and did not care how natural selection worked. They used the term"natural selection" vaguely. So most oppositions to the neutral theory were from evolution and population genetics people.

-- Tomoko Ohta, October 31, 2001

The neutral theory
I was most interested in the proposal of the neutral theory when Kimura's paper came out in 1968. It seemed to me at the time to be a very bold theory, in that over many years neutrality had often been claimed for specific alleles (eg in the ABO system), only to be shown later that selection prevailed for those alleles. Indeed Dobzhansky had reglarly made derisive comements on those who claimed neutrality for the alleles at this or that locus, over the years.

However Kimura was clearly too clever to fall into this trap, and his theory had to be taken seriously.

As a statistician/mathematician, perhaps my main interest was in the theory leading to the neutrality claim, together with the obvious thought: how can we test this theory, given gene frequency data?

On the theoretical support for the theory, I felt then, and feel just as strongly now, that the "substitutional load" argument for the theory was total nonsense, and derived from extremely poor modeling on Kimura's part. I have give the reasons for this view, in detail, in my 1979 book "Mathematical Population Genetics" and elsewhere, and there is no point in repeating the details here. I am happy however to discuss them with anyone who wants to.

The "truth or otherwise" (whatever that can mean) of the neutral theory, however, does not depend on load arguments, so in my view it is a pity that these were ever advanced.

On the point of testing for neutrality, I started to think hard about this two or three years after Kimura's paper came out, and this resulted in my 1972 paper "The sampling theory of selectively neutral alleles". This gave, in a very simplified evolutionary model, the null hypothesis (i.e. neutrality) distribution of a sample of genes, under the infinitely-many alleles model. This could then be used as a basis for a test of neutrality.

After a year or two I came to think that testing for neutrality would be very difficult, because of the very low power that any test one could think about at the time would have. I still think that this is largely true - one has to make many(and to me probably unrealistic) assumptions to get a test with much power.

On the other hand, the sampling theory in part led Kingman to the idea of the coalescent, and that is something that has proved very useful in data analysis. So I don't feel that my efforts were wasted.

-- Warren Ewens, November 1, 2001

Structural Reduction
I well remember the appearance of the King and Jukes paper and the papers by Motoo Kimura. I had already published a prediction concerning the consequences of the mose likely mutation based on the same appraisal of the molecular control of heredity. That was entitled "Structural reduction in evolution" and it appeared in the American Naturalist in 1963 (vo. 97, pp. 39-49). That had provoked a reaction by Sewall Wright himself the next year ("Pleiotropy in the evolution of dominance," vol. 98, pp. 65-69). I had suggested that his term "mutation pressure" would better be called the "probable mutation effect" and would generally interfere with the development of the phenotypica manifestation which it controlled. If selection were suspended, then the structure would reduce over time. What had gotten me started on that gambit was reading Christian B. Anfinsen's work The Molecular Basis of Evolution (1959). When the King and Jukes paper appeared I regarded it as support for what I had proposed. I did not then and do not now regard it as opposed to a Darwinian interpretation of evolution but just an extension of it. Darwin had observed the reductions that follow disuse, and it seemed to me that this was the mechanism at the molecular level that could account for the phenomenon. As a student of human evolution, I felt that it cold account for a series of reductions that had occurred by which "modern" form emerged from the more robust Pleistocene hominids who were the only people present from a million years ago until about 50,000 years ago. In regard to the "neutralist" approach, my take has been very much that of Francisco Ayala et al. in the PNAS in July of 1974 (91:687-6794). "The neutralist theory may . . . be considered a modification or extension of the synthetic theory of evolution" (1974:6793). My own attempt to harness a portion of it to explain particular aspects of human evolution has met with thunderous silence and outright opposition in the realm of anthropology. With two co-authors I did get an application of it published in Evolution in 1987 ("Gradual change in human tooth size in the laate Pleistocene and post-Pleistocene" 41[4]:705-720), but my attempt to expand on the approach was rejected twice by anthropological journals over the last year. Recently, however, molecular genetics has emerged as a major segment of biological anthropology, and I suspect that there may yet be a change in views as the molecular perspective becomes part of the general outlook. At the moment, however, the majority of the business is still reacting the way the biological world did to Kind and Jukes when they first published.

C. Loring Brace, Museum of Anthropology, University of Michigan, Ann Arbor, Mi.

-- C. Loring Brace, November 5, 2001

Dismay and disbelief
As quoted on this site, James Crow stated that the initial reaction was one of dismay and disbelief. I wonder if the contributors so far, who seem by contrast to have been interested, feel that their interest was unusual at the time?

-- Arne Hessenbruch, November 24, 2001

The neutral theory - a further comment
I would like to respond to the request that those who have contributed so far respond to Jim Crow's comment that the neutral theory was greeted with dismay.

There certainly was an element of that - after all, we were all convinced Darwinians at the time, meaning in this context that it was felt that the vast bulk of the variation observed in natural populations was selection- based.

On the other hand, for those (such as myself) who were interested in the mathematical aspects, the neutral theory opened up many lines of research, simply because the math for the theory was much simpler than that for selectively- based theory. What I find myself constantly fighting against is biasing my view of the neutral theory as a scientific theory just because of the enjoyment of the math that it led to.

I want to emphasize again that it is meaningless to me to say that the neutral theory is true or not true. It comes down to a question of proportions, of asking whether one is asking if the alleles at a locus are selectively neutral now or have been for millions of years, of whether one argues at the DNA or protein level, and so on.

-- Warren Ewens, November 26, 2001

Response to Biology vs. statistics
Contrary to Ewens statement, the intention of both Kimura and King/Jukes was to show that drift was the main force of molecular evolution. Mathematical developent of the neutral theory was of course very important, but the main interest had been on the evolutionary aspect of the neutral theory. Kimura and King/Jukes did not make any statements on the relationship between the evolutionary change of molecules and that at the phenotypic level. Their theory was solely on molecular evolution. Many evolution people misunderstood this point, and thought that the neutral theory was applicable to both molecules and phenotypes. So, many criticisms were based on the misunderstanding. To clarify the mechanisms of molecular evolution is a most important purpose of evolutionary studies at the molecular level. The Ewens sampling theory and other statistics are very useful for the purpose.

-- Tomoko Ohta, November 28, 2001

From the biochemical side...
I was very surprised to get an e-mail suggesting that I was "one of the first botanists to comment on the neutral theory"! I cited the papers by King and Jukes, Nei (1969) and Kimura because they provided estimates of mutation rates. I was trying to explain why the %GC of moss and fern DNAs fell in a rather narrow distribution with about the same mean as angiosperms. Using a rough average of these estimates and others, I calculated that there hadn't been enough time in the 400 million years since divergence from a common ancestor for the %GC to have changed very much.(BBA 254:402-406, 1971; BBA 277: 29-34, 1972).

Dr. Ohta is quite correct in what he says about biochemists. I did my Ph.D. in Biochemistry at the Univ. of Washington (1965), and the idea that most mutations had no effect seemed quite sensible to us. It still does. Take a look at the molecular models of the amino acids in Stryer or any other biochem text. It is obvious that some of them are going to "pack" in the same spaces. By that time, there were enough sequences of hemoglobins so it was clear that oxygen was carried quite efficiently by molecules with a variety of substitutions, especially on the surface of the protein. If you are being chased by an annoyed elephant, it will be pretty clear that your hemoglobin sequence and his are not going to be the deciding factor in the race!

We were taught that selection acts on the whole organism, generally by affecting the relative reproductive success of individuals in a population. Unless a mutation is quite deleterious, there will be other factors, e.g. behavioural, that will swamp out any small effect due to a small change in a protein molecule. Also, by 1965, Jacob and Monod's paper on the regulation of the lac operon had appeared, and there were a number of examples of negative feedback regulation in metabolic pathways. So, even if a given enzyme doesn't operate quite as efficiently, the steady state level of end-product will be the same, because it depends on feedback regulation.

My work on families of chlorophyll-binding proteins leads me to the view that there is quite a large degree of adaptability in protein structure (see the attached Commentary on "Molecular Opportunism" in PNAS 98:2119-2121,Feb27, 2001. One of my students once commented that "evolution is the survival of the adequate". He had a point.

-- Beverley R. Green, November 28, 2001
Attachment: PNAScomment.pdf

Response to More again
I would like to add one word to Tomoko Ohta'a comment.

Tomoko implies that there is some contradiction, in my previous comments, to what Kimura and others were doing, by pointing out that Kimura's main focus and emphasis was biological.

I have never questioned this, or said anything to the contrary. What I said was the MY OWN INTEREST was in the statistics relevant to this, and in how one could test the theory by statistical methods. I agre that this is a narow focus, but it was nevertheless my focus at the time.

-- Warren Ewens, November 29, 2001

Kimura and Jukes-King papers
This is response to the question of my initial reaction to the Kimura and King and Jukes papers.

Kimura sent me a copy of his manuscript, but I didn't get around to studying it, so he went ahead and published it. It is probably just as well, for I didn't like his use of the Haldane cost of evolution argument along the assumption that all the DNA is genic. Sewall Wright received a copy at the same time and commented rather adversely, although I have forgotten what he said.

Science sent me a copy of the King and Jukes paper. I believe it had originally been rejected and I participated in the second round of reviewers. In any case I liked the paper and recommended publication. Whether my review was influential or not I don't know, but the paper was published as everyone knows.

I spent part of the summer of 1968 in Mishima with Kimura. Of course we talked a great deal about the neutral theory. By this time Kimura had much stronger arguments and these, along with those of King and Jukes made what I thought was a convincing case. I was also aware of the earlier work by Sueoka and Freese reaching the same conclusion from quite a different direction. By the end of the summer I was convinced that the theory had a great deal of merit. I had been invited to give the closing address at the International Congress of Genetics in Tokyo, and took the opportunity to say nice things about the neutral theory. This is published in the Proceedings of the XII International Congress of Genetics, Vol III, p. 105-113.

Although Kimura and I had worked together a great deal, I want to make clear that the neutral theory was entirely his. I had participated in the development of the "infinite allele model". The history of this is summarized in Genetics 121: 631-634, 1989.

-- James F. Crow, December 1, 2001

Questions for Tomoko Ohta
I think that Tomoko could add further information on the origins of Kimura's paper, since she did some of the background work for it. Also I think she may remember better than I the conversations that took place in Mishima in the summer of 1968.

-- James F. Crow, December 2, 2001

Dismay and disbelief
I don't recall saying this, but I have said a lot and don't doubt that I said this at some time. In any case I think it is a fair description of the way the neutral theory was initially regarded by most biologists, especially evolutionists. Some of these were simply knee-jerk reactions, but later there were thoughtful objections, too (e.g. Gillespie, Kreitman).

-- James F. Crow, December 3, 2001

Wilson Lab and Stony Brook
I left Alan Wilson's lab in August,1968 to take a position in Biochemistry at SUNY Stony Brook. The Wilson Lab was very happy with the neutral model as I recall. Alan and Vince Sarich's attempt to date the Chimp-human split rested on the protein sequence data suggesting constant rates of protein evolution. The Kimura and King and Jukes papers provided a firm genetic foundation for this observation.

After both papers came out I tried to test the neutral model. I recruited Charles Taylor who was then a graduate student in the Ecology and Evolution Department to help with the neutral theory. We published a test of the model in Nature (Arnheim and Taylor 1969 Nature 223:900-903) three months after the King and Jukes paper came out.

The crux of the proposed method was to compare the number of segregating neutral amino acid substitutions in each of two proteins in the same population. If both proteins evolved at the neutral rate, the theory at that time suggested that the number of segregating neutral amino acid substitutions at each locus should be the same. No estimate of Ne was needed since it will be the same for both loci.

For the test we used the 84 known amino acid variants at the human alpha and beta chain hemoglobin loci. The criteria for a neutral amino acid substitution was biochemical. We used a paper by Perutz and Lehman (Nature 1968, 219:902) who predicted from haemoglobin's three dimensional structure whether any of the known variant amino acids would have an effect on the function of the haemoglobin tetramer.

-- Norman Arnheim, December 3, 2001

Response to Dr. Crow
I joined the Kimura laboratory in the spring of 1967, and I remember the atmosphere when Kimura proposed the neutral theory. I helped, or confirmed the calculation on the rate of mutant substitution of hemoglobin, cytochrome c and another protein based on the reports in Evolving Genes and Proteins by Bryson and Vogel. Kimura and myself were much impressed by the Zuckerkandl and Pauling paper in this volume, which was the first to propose the molecular clock. In the next year Dr. Crow visited us and we had many lively discussions. Through these discussions, I noted Crow's comment that if the number of genes was much smaller than the total DNA amount suggested, protein evolution was in accord with the load concept. I remember also what Kimura told me about Wright's comment. Wright wrote to him that the pattern of protein evolution was not incompatible with the shifting balance theory of evolution.

-- Tomoko Ohta, December 3, 2001

Response to Comment on King and Jukes
About King and Jukes

You asked me to comment on my response to the 1969 paper by King and Jukes.

Initially I thought that their arguments were so obviously wrong that the paper would sink into decent obscurity, but a year later there were authors declaring, "As King and Jukes have shown", and the like. A reply seemed to be needed. Rollin Richmond must have felt the same, because he, too, published a critique in 1970 (Nature 225: 1025-1028).

Thirty-one years later, most of the criticisms still stand, and some are greatly strengthened. We have clear evidence of selection on codon use, and of differences between lineages in the evolutionary rates of single proteins, but we remain ignorant about the importance, or otherwise, of three- dimensional structures in messenger RNAs. Statistical evidence of selection on amino acid sequences continues to accumulate, even though the tests have very low power. Experimental evidence of selection on protein polymorphisms also accumulates, particularly through differential susceptibilities to disease.

Where does this leave the neutral theory? The answer may depend on the kind of variation. The original argument was about amino acid replacements. It is reasonable (though not yet obligatory) to suppose that the majority of such replacements have been driven by selection. This does not deny the existence, or indeed the importance, of random genetic drift. It is theoretically possible, as Fisher pointed out in 1930, for most evolutionary replacements to be selected while most polymorphisms are neutral. It is much more difficult to imagine a world in which replacements are neutral while polymorphisms are selected.

In some proteins, virtually all amino acid replacements must have been selected. It is hardly credible to suppose that histone 4, in which a single amino acid has been replaced since the common ancestor of the pea and the cow, arrived at its initial state by random drift.

When we consider non-coding nucleotide replacements, the assumption of neutrality can be more plausible, but often it may just reflect our ignorance about the activity or function of the region concerned.

The neutral theory was appealing because it overcame two apparent difficulties of selection, the problem of high genetic loads when many loci are changing, and the problem that the evolutionary rates of individual proteins seemed to be constant. Both difficulties have turned out to be illusory. Truncation selection, epistasis, density dependence and other factors can minimise loads (as Warren Ewens has remarked), and evolutionary rates are not constant.

Another advantage of the neutral theory was that it simply and quantitatively explained both evolutionary replacements and polymorphisms. Nowadays the simplicity has gone. To accommodate known facts, the theory must accept varying degrees of purifying selection, the widespread substitution of mildly deleterious genes, generation-time effects, variable mutation rates, background selection, and associations with selective sweeps. In these circumstances, it seems quixotic to deny, on the grounds of economy, an important role for directional selection in driving replacements.

You enquired why, in responding to King and Jukes, I described myself as a ‘classical evolutionist’. Their paper had made slighting remarks about such people, showing more than a hint of molecular triumphalism. I took the term ‘classical evolutionist’ to mean someone who had been educated in a relatively molecule-free environment. My first degree was in zoology, which in those days meant comparative anatomy, evolution, ecology, behaviour. palaeontology and physiology (most of which I missed). My doctorate was in ecological genetics, and my job at the time was to teach ecology. This history seemed to fit King and Jukes’ designation pretty well, though it must be admitted that I was already interested in molecular biology, and had avidly read Bryson and Vogel’s ‘Evolving Genes and Proteins’.

About amino acid replacements

In the early days of the neutral theory, Motoo Kimura proposed that about 50% of mutations were disadvantageous, and the remainder neutral. I plotted Sneath’s measure of the chemical difference between each pair of amino acids against the frequency, in a range of proteins, of evolutionary replacements between the members of the pair. The analysis included only pairs whose codons were one mutational step apart. There was a clear negative correlation. By extrapolating to a chemical difference of zero, one could estimate the average rate of replacement for completely neutral changes. Comparing this value to the overall average rate gave an estimate of the minimum proportion of mutations that were disadvantageous. This was about 90%. Kimura accepted this value.

Interpreting the negative correlation was more contentious. Kimura, King and Jukes regarded it as evidence of the neutral theory (to them, fast rates represented neutral evolution, slow rates represented the elimination of large chemical changes by purifying selection). I pointed out that the selective theory made exactly the same prediction. Fisher had shown, by a geometrical argument, that the larger the phenotypic effect of a mutation, the less likely it is to be advantageous. Therefore we should expect a negative relation between the size of the chemical difference and the evolutionary rate. Kimura’s reply, in his book, was somewhat disingenuous. He emphasised, quite correctly, that the magnitude of the disadvantage and the slowness of the evolutionary rate are not the same thing. Making some plausible assumptions, he obtained an expected relation between the size of phenotypic effect and the evolutionary rate. The relation was a curve with a positive slope when the phenotypic effects were very small, but a negative slope when they were larger. He implied that this invalidated my argument. However, if you observe the part of the curve where 90% of the mutations are disadvantageous, the slope is very clearly negative.

General Remarks

Those who favour the overriding importance of random drift in evolution have shifted attention from replacements in amino acids, where their case has weakened, to replacements in non-coding nucleotides, where the issue is still unclear. The debate between ‘neutralists’ and ‘selectionists’ has been quiet for decades, not because there has been a clear resolution, but because people got tired of the battle. Meanwhile, for many purposes there has been comfort in assuming neutrality because it is mathematically tidy, and because it absolves one from having to do difficult experiments. There is still a lot of territory between the tenable extremes. Recent developments, particularly in analysing DNA and studying microbial populations, are likely to narrow the available range of argument.

-- Bryan Clarke, December 10, 2001

hemoglobin variability
Unusual to be asked about an article which I have written 30 years ago! But indeed we were sequencing hemoglobins at that time in the MPI in Munich and wondered about the variability it displayed in different species. To sum up what is known today, only 2 of the about 150 amino acid residues are invariant in this protein family. Thus the papers by M. Kimura (1968) and J. King and Th. Jukes (1969) gave an elegant and quantitative explanation but unfortunately bothered us with the term `Non-Darwinian Evolution` and even today in your e-mail outline I read that the theory lead to `a powerful new critique of Darwinism`. I think, that is not true: It has been shown, that already Darwin has discussed the matter of neutral changes in his ``The Origin if Species....``(1859) including the possibility of them being established by chance (Arnheim, 1973). Neutral variation is well in the concept of (Neo-)Darwinism; an extension of the theory, as it may come up in science when a new level of investigation-in this case the molecular- has been reached.

Another aspect behind the Neutral Theory, however, has not gained much attention: the dependence of the variability of nucleic acids/proteins from the general structure of these informative macromolecules. The elements of both are built from stabile backbones, providing the phospho-ribosyl/peptidyl-connections, which are not subject of the normal variation (pointmutation) but conserve the conformation and thus prevent destructive all-or-non effects of mutations, enabling a spectrum of variation (substitution), including the nearly neutral, and thus open a variety of possible adaptations in a population.

Gerhard Buse, Aachen, Germany

-- Dr. Buse Gerhard, December 17, 2001

King and Jukes
King and Jukes seemed to be spoiling our chances of seeing evolution in action in the metabolic enzymes of hybridizing house mice in Denmark. I was there doing field work in 1969 when Robert Selander sent me the K-J article. Our intention was to find alleles that differed across the narrow hybrid zone and to reveal natural selection in their patterns of introgression. The idea that selection might not distinguish among those still hypothetical genes was depressing, and even when striking interlocus differences in introgression did emerge a few months and a few hundred gels later, we wondered how big a part chance (especially linkage effects) had played with the distribution of those alleles. Were they coadapted parts in two fine watches (corresponding to the two mouse semispecies), or were those alleles just generic equivalents in a sea of ho-hum variation? The clincher was our discovery of five new alleles associated with the hybrid zone. The obvious explanation was that tight coadapted structure characterized the two parental gene pools, but faltered within the area of mixing (Hunt and Selander, Heredity 31:11-33).

-- Grainger Hunt, January 6, 2002

Response to The neutral theory - comment
Molecular Evolution and the Neutral Theory

When Kimura wrote his Nature paper, I had a rather close contact with him. I was a visiting researcher in his laboratory around that time and used to visit him in Mishima once in about two months. (At that time I was employed in the National Institute of Radiological Sciences, Chiba, Japan.) In my recollection, his original motivation of writing the Nature paper was to defend the classical (mutation-selection balance) theory of maintenance of genetic variability against the balance (overdominance) theory (Dobzhansky, Cold Spring Harbor Symp. Quant. Biol. 20:1-20, 1955). Around 1960 there was a fierce debate between the ?classical? camp and the ?balance? camp of population geneticists. The leaders of the former camp were James Crow and Motoo Kimura, whereas Theodosius Dobzhansky and Bruce Wallace were the leaders of the latter camp. Around 1966 when electrophoretic studies uncovered extensive protein polymorphisms in human and fruitfly populations, the ?balance? camp appeared to have won the battle because the balance theory had predicted a large amount of polymorphisms. However, there was a lingering problem annoying the balance camp; it was the unbearably high genetic load that is required to maintain polymorphism by overdominant selection.

There were two ways out of this dilemma. One was to assume that natural selection eliminates individuals homozygous for many loci disproportionately (truncation selection). The other one was simply to assume that the majority of allelic polymorphisms are more or less neutral and therefore do not produce any significant genetic load. This latter idea was apparently conceived independently by Alan Robertson and James Crow. I heard of this idea from the late Terumi Mukai (who was working in Crow?s laboratory), when I visited him in Madison in the summer of 1966. As soon as I returned to Japan, I told the story to Kimura. I do not remember whether he knew of it or not, but he seemed to be pleased with it. Now the problem was how to convince fellow evolutionists who now seemed to be sympathetic with the balance theory.

For Kimura, it was to compute the rate of nucleotide substitution from protein sequence data and argue that if all nucleotide substitutions in a mammalian genome (about 4 billion base pairs) were caused by natural selection the cost of natural selection as formulated by J.B.S. Haldane would become unbearably high for most mammalian species. This was soon challenged by John Sved and John Maynard Smith, who argued that the Haldane cost could be reduced substantially if natural selection operates in the form of truncation selection. For a number of reasons, I did not think that truncation selection applies to natural selection, but Crow (Haldane and Modern Biology, ed. K. R. Droniamraju, pp. 165-178, 1968) studied the same problem independently using genes (10,000 genes) as units of selection and reached the conclusion that the rate of amino acid substitution observed in hemoglobins, cytochrome c, etc. is compatible with the Haldane cost argument. Later, I also showed that mammalian species can tolerate a considerably higher level of cost than Haldane?s suggestion by changing the assumptions slightly (Nei, Molecular Population Genetics and Evolution, Elsevier, 1975).

Kimura sent me a copy of his Nature manuscript on the day he sent it to Nature. I did not think about his cost argument very carefully, but his final conclusion appeared to be correct, because by that time a substantial amount of molecular data supporting the idea of neutral evolution had been published (see the papers by Zucherhandl and Pauling, Margoliash and Smith, Freese and Yoshida, Sueoka, and others in ?Evolving Genes and Proteins? edited by Bryson and Vogel, 1965). This view was reinforced when King and Jukes (1969) published their Science paper, examining all available molecular data systematically. Their main conclusion was that as long as the function of a protein or RNA molecule remains the same most nucleotide substitutions occurs by random fixation of nearly neutral mutations. For me, this conclusion was easily acceptable, and I believe it is still valid, though details of the King- Jukes arguments have been modified.

Although I accepted the neutral theory without much trouble (Nei, Nature 221:40-42, 1969), I was not very happy about the fact that the neutral theory was powerless in explaining the evolution of morphological and physiological characters, in which most biologists were interested. I also thought that if adaptive evolution ever occurs in morphological or physiological characters it must be caused ultimately by some nucleotide changes. I was then interested in finding such evidence. I was therefore quite happy when Austin Hughes and I were able to show that the high degree of polymorphism in MHC loci is caused by overdominant selection. One of the important implications of the neutral theory was that protein and DNA molecules could be used for clarifying evolutionary relationships of different populations or species, and I then invested a substantial amount of my time for this purpose.

I suppose this forum is not for arguing the validity of the neutral theory. If we do this, we will have to examine extensive literature. However, since Brian Clarke strongly rejected the neutral theory, I would like to add a few comments. A reader, who wants to know detailed aspects of this issue, may refer to Li?s (1997) book Ã?Â?Molecular Evolution.?

The first thing I want to emphasize here is that if we are concerned with very minor fitness differences between alleles, say s = 0.0001, it would be extremely difficult to resolve the debate on neutrality. However, is it important to identify alleles with very small fitness differences? Early molecular evolutionists were interested in knowing whether the function of a protein is affected by amino acid substitutions or not. If the function did not change appreciably, the alleles were regarded as neutral. Examination of this problem was often done by studying whether a protein from a species functions normally with substrates from other species. We can now use site-directed mutagenesis and gene-knockout experiments for this purpose. Recent knock-out experiments have shown that a surprisingly large number of genes are dispensable in mice and yeast and that there are many different ways of achieving a particular biological function such as host defense from parasites. In my view this type of study is more important than the identification of alleles with minor fitness differences.

In recent years an extensive statistical study of detecting selection among polymorphic alleles has been done. The results indicate that it is generally difficult to reject the null hypothesis of neutrality. However, a substantial number of authors reported detection of slightly deleterious mutations. These results are interesting because they support the classical theory of maintenance of genetic variability, which was discussed in the 1950s and 1960s. However, these mutations would not be very important in long-term evolution, because accumulation of deleterious mutations in a gene is expected to result in the deterioration or malfunctioning of the gene eventually. Of course, if the fixation of slightly deleterious mutations is followed by that of slightly advantageous mutations, the function of the gene can be maintained in long-term evolution. However, this is essentially the same as the nearly neutral evolution as envisaged by King and Jukes (1969). Similarly, background selection may be important in understanding certain patterns of genetic polymorphism, but it is unlikely to play important roles in long-term evolution. It should also be noted that purifying selection is allowed in the neutral theory, and therefore the evolution of histone genes is in accord with the neutral theory (Rooney, Piontkivska, and Nei, Mol. Biol. Evol. 19:68-75, 2002). Furthermore, selectively important amino acid substitutions such as those for MHC loci have been identified in a number of genes, but they are a minority of the total amino acid substitutions.

Masatoshi Nei

-- Masatoshi Nei, January 8, 2002

Response to M. Nei
In general, my memories agree with Masatoshi's statements and I appreciate his input. I had not known of his conversation with Mukai, although I do remember his visit in Madison.

Although I entertained ideas about the possibility of neutral evolution and neutral polymorphism in 1965 and 1966, I did not, as I said earlier, participate in discussions with Kimura immediately prior to his 1968 paper. Much of the time we were in close contact if not in collaboration, but not in this time period. His manuscript was new to me, and as I said earlier, I didn't trust the Haldane argument for the whole genome, although as both Nei and Ohta have said, the difficulty disappears when the number of genes is considered rather than tht number of nucleotides. I was receptive to the neutral theory partly because of King and Jukes and because of early familiarity with the work of Sueoka and Freese.

Although Kimura was certainly associated with the classical view, I thought that immediate antecedents of his 1968 paper were molecular evolution rates rather than classical-balance arguments. But Nei was in much closer contact with him at this time than I was, as was Ohta. In any case, by the time I spent several days in the summer of 1968 in Mishima the neutral theory had taken on a life of its own.

I agreed with Nei at the time about truncation selection, although I did mention Sved's analysis in my paper at the Tokyo congress. My view changed later when Kimura and I showed that a rather loose approximation to truncation had rather similar consequences.

-- James F. Crow, January 9, 2002

Response to Dr. Nei
I would like to defend the nearly neutral theory. In the early 1970s, I was hoping to find out the true processes or the mechanisms of molecular evolution, because I could not be satisfied by the neutral theory due to the following three points. 1) Molecular clock. Why was the rate of molecular evolution year dependent rather than generation time dependent? I thought that the mutation rate depended on generations. 2) Why was the heterozygosity measured by electrophoresis so uniform among the species? 3) Mutants effects on fitness should be continuously distributed around neutrality. Then one would expect population size effects on the rate of molecular evolution. I was much puzzled by these questions. Then I realised that the nearly neutral theory assuming the prevalence of slightly deleterious mutations could explain these problems, and published it. There were disagreements on the significance of near neutrality between Kimura-Nei and myself and we had many hot discussions. Dr. Crow agreed to my idea to some extent but not quite, as I understood. Jukes liked the idea and sent me Conglaturations when the Nature (1973) paper was published, but he did not make any statements until the last one. King completely agreed with the near neutrality concept and we published the joint paper(Genetics 79, 681-691). As to the relationship of Mukai viability polygenes and nearly neutral mutations, there should be overlap between the two classes. The mean effect on fitness should be smaller for nearly neutral mutations than for viability polygenes. The overlap is an important problem for the future study. The effect of such weakly selected mutations should not be constant for various environmental conditions and genetic backgrounds, and selection intensity in a local area is thought to be larger than Nei's estimate. His value comes from molecular evolution data which is the average effect. So nearly neutral mutations are not totally noise, and may be related to some quantitative characters. In any case, clarification of interaction of drift and selection is important for understanding mechanisms of molecular evolution.

-- Tomoko Ohta, January 10, 2002

A reply to Masatoshi Nei, and some comments
We cannot intelligently consider the King and Jukes paper without discussing the value of their arguments. This may not be the place for a full-blooded debate, but at least we must state our own views clearly. Otherwise, this exchange becomes just a set of pleasant reminiscences.

Many of the K-and-J arguments were silly, but Kimura's neutral theory was not. Some nucleotide substitutions must surely be neutral, or nearly so. However, the case that most amino-acid substitutions are neutral is weak.

Of course histone 4 is subject to intense purifying selection. It sits on a very sharp adaptive peak. The question is how it got there. To suppose that it wandered randomly over a flat landscape, and then a peak happened to appear beneath it, stretches credulity.

In fairness, I should confess some failings of my paper criticising King and Jukes (Science, 168, pp.1009-1011, 1970)

I wrote that the Treffers mutator (mut-T) allele would probably be disadvantageous in natural circumstances. Since then, the work of Cox and others has shown that it can sometimes be favoured. In a changing environment, the short-term advantage of producing beneficial mutations at other loci can allow mut-T to spread by hitch-hiking, despite its long-term disadvantages. Thus, selection on the mutation rate is not always downwards, a fact that can explain the pressure on GC ratios. The results of Sueoka do not necessarily imply that the widespread changes in codons have been neutral. There are two opposing selective forces.

I failed to point out that the natural selection of synonymous codons to minimise mutational damage may be effective only where just a few somatic mutations can produce severe problems, as with oncogenes. Selection against errors in translation is probably more common.

I argued for a selective interpretation of geographical patterns in allozyme frequencies but did not discuss the possible roles of historical events and gene flow. I should have done so even though I felt that they had already received too much emphasis. It is still true that selection of molecular variants by ecological factors is rarely tested, and often completely ignored. When a test is made, evidence of selection is frequently found. There is still an uncritical tendency to invent historical events a posteriori as explanations of current geographical patterns.

-- Bryan Clarke, January 17, 2002