Evolution Icon Evolution

In Terror of Chipmunks: A Response to Joseph Felsenstein


I’m now back in my hometown of Cologne after a trip to Mexico, where I had been studying several native plants and animals. On returning, I had the opportunity to consider a recent post by University of Washington geneticist Joe Felsenstein1 at Panda’s Thumb: “A devastating critique of population genetics? The Discovery Institute thinks so.” Felsenstein responds there to my post at Evolution News, “Randomness in Natural Selection and Species as Islands in a ‘Vast Sea of Conceivable Arrangements.’” Seldom have I seen a piece of scientifically inspired writing like Felsenstein’s that is so far off the mark. In fact, he quotes just one sentence of my post here, itself taken from my encyclopedia article on natural selection. He disregards my second post on the same topic entirely.

In the encyclopedia article mentioned by Felsenstein (but to which he provides no link), I unmistakably emphasized I am among those who accept “natural selection as a real process in nature.” In reply, Felsenstein kicks in an open door, as the saying goes, and pretentiously preaches to the converted.

I added in a later paragraph of that encyclopedia article:

Furthermore, survival of the fittest evidently takes place, for example, in cases of alleles and plasmids with strongly selective advantages, as in the cases of multiple resistance in bacteria and resistance to DDT in many insect species. After pointing out that Darwin knew hardly any cases of natural selection, Mayr asserts (1998, p. 191): “Now, there are hundreds, if not thousands, of well-established proofs, including such well-known instances as insecticide resistance of agricultural pests, antibiotic resistance of bacteria, industrial melanism, the attenuation of the myxomatosis virus in Australia, the sickle-cell gene and other blood genes and malaria, to mention only a few spectacular cases.”

Joseph Felsenstein (J.F.) also refers to a podcast conversation with Paul Nelson and myself. In contrast to the view of total non-randomness as the essence of natural selection, held by Dawkins and many others in line with Darwin himself, I explained in the podcast that a comment by Litynski (“…that natural selection is nothing but blind mortality which selects nothing at all”2) “is the other extreme in the opinion on natural selection.” This implies, of course, that the truth is to be found somewhere between these extreme views. I also emphasized that “we have just touched some of these things and there is much more to be said,” referring listeners to the encyclopedia article.

In two posts on randomness in natural selection, my starting point was the perpetually asserted “antithesis” between mutation as a random process and natural selection as a totally nonrandom process — the “directional ordering mechanism” — in the neo-Darwinian theory of evolution. That’s why I stressed Julian Huxley’s verdict referring to and supporting Waddington and Mayr:

The frequent assertion that biological evolution is based on chance is entirely untrue. “Chance” events furnish its raw material but the process itself is directional, self-steering, but automatically steering itself in a definite direction. This is because…natural selection is not a random but an “ordering” mechanism.

The absolute distinction here is between random/chance events in mutation versus unrestricted non-randomness in natural selection putatively governing all biological processes in the wild. The latter view was implied once more by Richard Dawkins in a comment on Stephen Meyer’s debate with Lawrence Krauss: “…natural selection is a NONRANDOM process” (capital letters by Dawkins).

Moreover, since some authors even assert that “natural selection comes close to omnipotence” (Avise 1999), and Exley (2009) is convinced that “both the beauty and the brilliance of natural selection are reflected in its omnipotence to explain the myriad observations of life” (emphasis added in both cases), I supposed that it would not be wrong to put this infallible, unerring Goddess into a new perspective. I thus indicated that there is a strong element of randomness involved in the very process of natural selection (in Darwin’s and most of his followers’ encompassing sense) itself.

This fact is clearly evident from the following calculations by several population geneticists themselves and other authors referred to in my encyclopedia article, as I also mentioned in the podcast with Dr. Nelson:

Fisher, perhaps the most important forerunner of the neo-Darwinian theory, has calculated (1930) that new alleles with even 1% selective advantage (i.e., more than is usually expected by neo-Darwinian theorists), will routinely be lost in natural populations. According to these calculations the likelihood of loosing a new allele with 1% advantage or no advantage is more than 90% in the next 31 generations (Fisher, 1930/1958; Dobzhansky, 1951; Schmidt, 1985; see also ReMine, 1993; Futuyma, 1998; Maynard Smith, 1998). Considering genetic drift, i.e. random fluctuations of gene frequencies in populations, Griffith and colleagues state in agreement with these authors (1999, p. 564):

Even a new mutation that is slightly favorable will usually be lost in the first few generations after it appears in the population, a victim of genetic drift. If a new mutation has a selective advantage of S in the heterozygote in which it appears, then the chance is only 2S that the mutation will ever succeed in taking over the population. So a mutation that is 1 percent better in fitness than the standard allele in the population will be lost 98 percent of the time by genetic drift.

Compare also J.F. (2015) as quoted later. So, far from offering a “devastating critique of population genetics,” I cited this branch of theoretical biology in support of a critique of the limits of natural selection. Now, how does Felsenstein deal with these explications?

Consider, please, the following points:

(1) Instead of focusing on my arguments for diploid populations, in his reply Felsenstein starts tutoring an obtuse W.-E. L. with a haploid population where the new mutation has already been established even to the point that exactly half of all the individuals are displaying it3. He states:

For example, if we have a population of a haploid species with N = 10,000 individuals with two alleles A and a at equal frequencies, each of them will produce a vast number of gametes, equal numbers from the two genotypes.

So, why does he start with a population with two alleles A and a at equal frequencies? We find the answer in J.F. 2015, p. 307: “[M]ost of the loss of advantageous alleles takes place while these alleles are still present in only a few copies” (see context below). And for the rest he thus skips the immense waiting time problem (see also below).

(2) And why did he choose a haploid species? Maybe Felsenstein starts with a haploid species because diploid ones are more complex to model in theoretical population genetics (see J.F. 2015, pp. 18, 49, 103, 108, 143).

(3) Equal numbers of gametes? Not necessarily so. Mutants often produce less. And by adding further presuppositions, you will find, of course, the outcome J.F. describes in his post.

(4) However, does the standard model to which he refers in his post, really consider “All the haphazard random mortality that Lönnig is worried about”? What about such terrible things like tsunamis, hurricanes, earthquakes, floods, heavy thunderstorms, famines and pestilences/epidemics? What about migrations? Assume that the two genotypes display different ecological preferences so that they separate from each other: A prefers to live in coastal areas and a in mountaineous regions. Then, after many generations, unexpectedly a tsunami occurs and unfortunately kills almost all A individuals. In the interim the a genotype adapted also to the coastal niche and settles as well there.

(5) And after all, always 10,000 individuals? Real population numbers can vary strongly sometimes even from one generation to the next.

(6) What about available area/space and the effects of population densities? “It is unlikely that relative fitnesses of genotypes will remain constant through time, since the environmental conditions, population density, and densities of other species will fluctuate, and these will in many cases affect the strength of natural selection” (J.F. 2015, p. 106).

(7) So, has the a genotype really forever a viability 1 percent lower over enormous numbers of generations even under extremely different environmental conditions? How, then, could the population of N=10,000 individuals ever come to have two alleles A and a at equal frequencies? A biologist well versed in population genetics stated his personal opinion on this question in a preliminary note to his mail (11 July 2016) as follows: “Using a 50 percent initial allele frequency is cheating — it is optimal for evolution for several reasons — but has almost no relevance to reality…”

It seems that the standard model functions only under idealistic, virtually paradisiacal conditions. Well, I know, of course, that for decades population geneticists have tried to develop models for many different biological conditions, premises and assumptions. J. F. himself has written an extensive overview on them (J.F. 2015). However, by his standard model and subsequent explanations the non-informed public may be deceived into a naive confidence, baseless trust and unreasonable faith in the putative omnipotence of natural selection in the wild.

(8) Among the many examples in my encyclopedia article I also drew attention to the following momentous discovery regarding earth history:

One of the major setbacks for the idea of a pervasive and the history-of-life-dominating process of natural selection has been the rise of what neo-Darwinians derogatorily call “neocatastrophism” (Hsü, 1986; Alvarez, 1998; Prothero, 1998). Darwin “postulated a single process, the biotic struggle of natural selection, that was uniform over all the time on the earth, proceeded always at the same rate, on a planet that ceaselessly changed in detail but never abruptly changed state” (Hsü, p. 47). Today, Darwin’s view is generally rejected by all informed scientists. The current question is not whether catastrophes have repeatedly interrupted natural selection worldwide, but which kinds of catastrophes are the most important ones in earth history.

One may think, for instance, of the Permian extinction (up to 96 percent of all marine species and about 70 percent of terrestrial vertebrate species annihilated). If J.F. is fully convinced that he has considered all the haphazard random mortality that I am so worried about then you have to conclude that in theoretical population genetics omnipotent natural selection still functions extraordinarily well even in cases of the complete absence of any populations. Vive la sélection naturelle!

(9) It is also important to note that Felsenstein is not a naturalist, that is, “a person who is expert …in botany or zoology, especially in the field,” “a person who studies plants and animals as they live in nature,” or, similarly, “a scientist who studies living organisms.” Many naturalists, like Carl Correns, William Bateson, Oscar Hertwig, John Christopher Willis, Lucien Cuénot4 (and many others up to now) have vigorously confirmed their strong reservations regarding the omnipotence of natural selection. Their inferences were specifically due to their direct studies of species in the wild.

Evolutionary biologist Cuénot formulated the basic objection quoted in my post and article. Willis started as a convinced Darwinist, until he found the exact opposite in natural species “of tropical flora and the remarkable plant family Podostemaceae” (Cronquist) in Ceylon/Sri Lanka. Or in the words of Willis himself: “…selection could not be responsible for evolution.” Bateson and Cuénot discussed several examples of distinctly relaxed natural selection in the wild (natural selection simply not found to be functioning or explaining anything in these cases). As for Bateson’s arguments and those of many other biologists on the limits of natural selection, see the detailed discussion in my book on the evolution of carnivorous plants.

(10) Felsenstein explains humorously in an interview with Dr. Mary Kuhner for the Distinguished Faculty Interview Series at the University of Washington:

…by the time I got to high school, I was convinced I was going to go to into wildlife conservation, which is completely ridiculous because I am the world’s most timid field person. I’m always, when I’m camping out, I’m always terrified, that I’m going to be eaten by chipmunks. And such a person should not go out into wildlife conservation.

He is in terror of chipmunks? There is no doubt some backstory to that, and of course I understand he is trying to be funny. However, had Felsenstein been out in the woods during, say, the last two or three months, carefully studying plants, he would have noticed thousands of tree seedlings (probably billions in the Northern Hemisphere alone) of many different genera and families just germinating. Of those thousands, only a very small percentage — perhaps less than 0.01 percent on average — will ever reproduce or even become full-grown trees. In exploring nature, Felsenstein would have seen the evidence for Cuénot’s basic argument that, similarly, 99.99 percent and more of this juvenile generation will become extinct. The mutation rate per gene per generation of the surviving rest (i.e. of the less than 0.01 percent) is generally calculated to be between 10-5 and 10-7 (about 5 x 10-9 per nucleotide per generation in the 10-5 cases). More than 99 percent of these mutations with any effects on the phenotype are negative, i.e. deleterious or at least slightly deleterious, essentially constituting losses of function. And now, of all these mutants, again an extreme minority of perhaps less than 0.001 percent (1 in 100,000; cf. here and here) displaying a new allele due to a new beneficial mutation in the sense of Sanford et al. (2015) with 1 percent selective advantage the overwhelming majority of 98 percent will ultimately also be lost/become extinct simply due to stochastic events and/or population density.

For the latter point see also the calculations of J.F. (below) in agreement with virtually all other population geneticists. Thus, the question may be allowed: Is natural selection really “directional, self-steering, but automatically steering itself in a definite direction” or undoubtedly “close to omnipotence” or displays even the “omnipotence to explain the myriad observations of life”?

A biologist well versed in population genetics stated his personal opinion on this question as follows: “Using a 50 percent initial allele frequency is cheating — it is optimal for evolution for several reasons — but has almost no relevance to reality…”

Acer platanoides.jpg

Mostly Acer platanoides tree seedlings in a wood observed for several years and photographed (unretouched), April 29, 2016 by W.-E. L. The undergrowth can, of course, vary widely in different populations and environments as well as in different seasons.

Interestingly, since Felsenstein himself has stated (2015, p. 133) that “Genetic drift, which changes gene frequencies at random, may cause a favored allele to be lost,” he in principle has already long understood and even taught Cuénot’s argument.

Or consider Felsenstein’s treatment of random genetic drift in his book Theoretical Evolutionary Genetics (2015), contrasting random genetic drift with natural selection defining random genetic drift as a “new force” (p. 339), or “another force,” setting, together with the mating system and the mechanisms of recombination, “the context within which natural selection takes place” (p. 133). He also affirms in the context of mutations that (p. 139) “it is genetic drift, not mutation, that is the random force.” He explains that (p. 271) “Genetic drift is the one force which can act as the “thermal noise” in the evolutionary machine. The relative strength of this “noise” compared to the nonrandom forces will determine to what extent the random effects of genetic drift will override other evolutionary forces.”

Also, please see and check J.F. 2015, pp. 306-307, and compare the following quotation with that of the other population geneticists quoted from the encyclopedia article above:

When s is small, clearly 2s is close enough to the probability of survival to serve as a working rule of thumb. It is worth considering how small a probability of survival this is. When s = 0.01, only one new mutant in 50 will succeed in spreading, despite the fact that all are advantageous. Even with s as large as 0.1, large enough to guarantee fairly rapid change in gene frequencies in the deterministic case, only one new mutant in six will establish itself. Obviously, genetic drift is a powerful force when only a few copies of an allele are in existence. Only rarely will an allele, even if advantageous, escape from the risk of loss due to the randomness of births and deaths, and of Mendelian segregation.” […] “So most of the loss of advantageous alleles takes place while these alleles are still present in only a few copies…. This in turn must be during the first few generations. An allele present in only one or a few copies is constantly at risk of being lost and could not last long in that state. If it survives many generations it must therefore be fortunate enough to have drifted to a larger number of copies.”

Thus, Felsenstein is obviously in agreement, at least to a certain extent, with Cuénot and other biologists in doubting the omnipotence of natural selection.

Let me emphasize again: In contrast to what Felsenstein calls a “devastating critique of population genetics,” this branch of theoretical biology, imperfect as it is (more on that in a further post), really supports a well-founded and compelling critique of the limits of natural selection by many famous biologists.

Hence, altogether, can there be any doubt that the process of natural selection includes, in fact, “an inescapable element of randomness” (Nelson)? Or, in other words, that randomness/chance events necessarily play a substantial and really far-reaching part for the survival of the fittest in the wild?

And yet, in my encyclopedia article I have also considered the surviving individuals of, for instance, the tree example above (“nevertheless, it appears that if such a mutation [one percent better in fitness] occurred at a constant rate in a large population, it would have a fair chance to become established after an average occurrence of about 50 times”).

Following Felsenstein’s post, a reader, “Ravi,” offered the following comment:

It is a real flaw in Darwinist thinking — some organisms just don’t reproduce in sufficient numbers for a slightly beneficial change to be “selected.” The unrealistic cost of replacement, of course, is the dilemma faced by Haldane.

This is Haldane’s dilemma in the clear words of Theodosius Dobzhansky referring to Crow and Kimura (1977, pp. 163-164):

Crow and Kimura (1970) give the following example of gene substitution: “if the typical allele has a initial frequency of 10-4, a population of one million individuals will have to have nine million genetic deaths each generation if it is to substitute an average of one allele per generation.” […] Granted that most species produce numbers of progeny far in excess of those needed to have the population survive, it is difficult to understand how evolution can happen at such an enormous cost in genetic deaths. Haldane saw clearly that he was confronted by a dilemma.

The simplest way to escape the dilemma would be “to postulate that only a few genes are substituted in racial, specific, and transspecific evolution by natural section,” all the others being neutral, thus constituting “the strongest argument in favor of the panneutralist position”(Dobzhansky). Such a solution, however, has not been substantiated so far. Many regulatory and target genes seem to be involved in most cases. For an intriguing recent paper on the waiting time problem associated with Haldane’s dilemma, see Sanford et al. (2015). But how can we explain the origin of a species definitely not producing numbers of progeny far in excess of those needed to have the population survive, such as Eleutherodactylus limbatus (“Female frogs have a single ovary and lay one egg at a time”) and others?

Lucien Cuénot sums up his position on natural selection as follows (1951, pp. 401-402):

Il y a égalité de chances pour les 800.000 oeufs pondus anuellement par une Carpe et abandonnés au hazard, et de 200 oeufs de l’Épinoche abrités dans un nid surveille par le mâle, entre les 120.000 oeufs (1) de nos Grenouilles et Crapauds, et les 9 oeufs portés sur le dos de la femelle du Leptodactyle Ceratohyla (Équateur) (fig. 155), ou l’oeuf unique de Sminthillus [Eleutherodactylus limbatus] dont le développement ne comporte pas la phase de têtard libre. La mortalité intraspécifique n’a donc, à part les quelques morts du début, aucun caractère sélectif ; lorsqu’il n’échappe que 2 individus sur 120.000, comme chez la Grenouille verte, comment admettre que ces deux sont élus en raison de petits avantages anatomique ou physiologiques5? C’est tout a fait invraisemblable, puisque ce sont les jeunes qui sont sacrifiés alors qu’ils n’ont pas leur perfection definitive. Enfin les observations positives montrent que la selection suppose n’a nullement la rigeur et l’infaillibilité que requiert la conception darwinienne ; il ne pas rare du tout de trouver à l’état sauvages des animaux handicapés par des malformations ou de mutilations, et qui cependent se maintiennnent comme les intacts.

For the latter, Cuénot mentions the following cases among others:

…il n’est par rare du tout de trouver à l’état sauvage des animaux handicapés par malformations ou de mutilations, et qui cependant se maintiennent comme les intacts. Kellicot ayant examine 450 exemplaires de Bufo lentiginosus, trouve que 8% d’entre eux sont anormaux, blesses ou amputés (Science, 1908, 855). J’ai peché plusieurs fois le Bassins d’Arachon des Poissons (Atherina presbyter, Mugis labrosus) à colonne vertébrale ondulée (fig. 156), moins bons nageurs que leurs congénères normaux (Sigalas compte 14 Athérines à rachis déformé sur 122 exemplaires); il est habituel de rencontrer des Orthoptères sauteurs qui ont perdu une patte sauteuse depui longtemps, des Lapin, Chamois, etc. qui ont réparé tant bien que mal des fractures ou amputations graves…

Relaxed natural selection may also be involved to a large extent in processes of degeneration in the wild (see the details here). In contrast, considering the possibility that advantageous mutants even under strong selection can exactly have this superior quality of being advantageous, beneficial, and helpful due to losses of function, as Michael Behe has so well pointed out and argued for in “Experimental evolution, loss-of-function mutations, and ‘the first rule of adaptive evolution’” (Behe 2010, see also Lönnig 1971, 1993, 2015), Darwin’s ensuing assertion on natural selection of 1859 is simply wrong:

It may be said that natural selection is daily and hourly scrutinising, throughout the world, every variation, even the slightest; rejecting that which is bad, preserving and adding up all that is good…

I will address some other inaccuracies in Felsenstein’s post in another entry.


(1) Joseph Felsenstein is Professor of Genome Sciences and of Biology and Adjunct Professor of Computer Science and Statistics at the University of Washington.

(2) Cuénot was more cautious, declaring: “Enfin les observations positives montrent que la selection suppose n’a nullement la rigeur et l’infaillibilité que requiert la conception darwinienne” (see full quotation above).

(3) Of the commenters on Felsenstein’s post, “Ravi” seems to have been the first who noted a discrepancy here.

(4) All great names in the history of biology. For more on Cuénot, see the Wikipedia article.

(5) Note, please, that by this statement Cuénot is not denying natural selection totally or suggesting that if it does not exist at all. He nevertheless strongly limits “l’infaillibilité que requiert la conception darwinienne.


Cuénot, L. (1951) L’Évolution Biologique. Les Faits. Les Incertitudes. Masson et Cie Éditeurs. Paris.

Dobzhansky, T. (1977). Chapter in T. Dobzhansky, F.J. Ayala, G.L. Stebbins and J.W. Valentine (1977): Evolution. W.-H. Freeman and Company. San Francisco

Felsenstein, J. (2015). Theoretical Evolutionary Genetics.

Lönnig, W.-E. Publications from 1971 to 2016, see here, here, and also here.

Sanford, J. Brewer, W. Smith, F. Baumgardner (2015). The waiting time problem in a model hominin population. Theoretical Biology and Medical Modeling 12: 1-28.

Photo at top: Eastern chipmunk, by Gilles Gonthier [CC BY 2.0], via Wikimedia Commons.

Wolf-Ekkehard Lönnig

Wolf-Ekkehard Lönnig worked for 25 years as a research scientist at the Max Planck Institute for Plant Breeding Research in Cologne, Germany. He is now retired but still writes often on the topic of Darwinism and intelligent design.