by Walter ReMine
April 5, 2006
Summary: Joe Felsenstein reviewed my paper, Cost Theory and the Cost of Substitution – a clarification, for a science journal, and later published that review on the Internet.1 This is my response. Felsenstein's review is erroneous and demonstrates commonplace fallacies that remain unchallenged in the evolutionary literature. Felsenstein's review demonstrates deep contradictions with other leading evolutionary geneticists, and shows why a paper like mine is sorely needed. Felsenstein's review also demonstrates how the review process sometimes fails, especially over topics with controversial implications, such as Haldane's Dilemma, or the origins debate. My paper, and Felsenstein's review are now both public and can be compared at your convenience.
Renowned evolutionary geneticists, Warren Ewens and James Crow also reviewed my paper,2 and acknowledge it is correct. So keep in mind that Felsenstein's technical criticisms contradict Ewens and Crow.
Starting in his early papers (1971 & 1972), and continuing to the present day, Joe Felsenstein has promoted the following mistaken notions, which are embraced still by many evolutionists:
My paper overturns those errors.
But Felsenstein continues those errors, and his review is an example of it. Felsenstein's review contains confusion that throws you off, so his key error gets overlooked. In particular, he uses a population size that fluctuates dramatically. First I quote his full argument in context:
ReMine's treatment ... is also inadequate internally. Take the idea that substitution requires reproductive excess. Suppose that we have a (haploid) species with no reproductive excess. Suppose that the environment changes so that all individuals have 20% less reproduction, except for 10% of them who have a particular allele, and those continue to barely replace themselves. A little consideration will show that the replacement will happen, and that the population will end up 90% smaller. But if there is even a slight reproductive excess, then with enough time the population will ultimately recover its numbers. There is then no lower limit on the reproductive excess needed. (Felsenstein's review)
Felsenstein chose a scenario that makes the substitution rate exceedingly slow.3 This is obvious when he says:
the population will end up 90% smaller. But … with enough time the population will ultimately recover its numbers. (Felsenstein's review)
By making the substitution rate exceedingly slow, he is making the required reproductive excess exceedingly small – which sets up for what comes next, in the form of a contradiction. The contradiction seems slight, at first, and thus is easily overlooked – until you recognize that it goes to the core issue. I next expose the contradiction:
But if there is even a slight reproductive excess, ... There is then no lower limit on the reproductive excess needed. (Felsenstein's review)
On the contrary, if there is "a slight reproductive excess", then the lower limit on it is, in fact, "a slight reproductive excess" – not zero reproductive excess! His argument fails over that self-contradiction. It slips passed you unnoticed, because the difference seems so "slight". In his scenario, reproductive excess is required – not zero – contrary to his long-standing claim.
*
Felsenstein chose an exceedingly slow substitution rate (in order to make the required reproductive excess exceedingly small). However, you cannot solve Haldane's Dilemma with an exceedingly slow substitution rate. That's the irony of Felsenstein's example. He chose a super-slow substitution rate – in order to establish his illusion (that the required reproductive excess is zero or infinitely approaches zero), then he pretends it remains that way all the time, regardless of substitution rate. In reality, the higher the substitution rate, the greater the required reproduction rate. Felsenstein garbles this fundamental principle.
*
Felsenstein's scenario includes a second layer of confusion. I'll describe it explicitly (though it is implicit in Felsenstein's discussion). By tradition, the cost literature focuses on allele frequencies, (and on allele "fixation", which is merely another way of saying "an allele frequency of 1"). The literature focuses on 'change in allele frequencies' and ignores times when allele frequencies remain constant. In short, by focusing exclusively on changing frequencies, the literature ignores the reproductive requirements when the frequencies remain constant – such as during population growth. That omission is a mistake corrected in my paper.
Felsenstein's above scenario takes advantage of that traditional mistake. Thus, under his reckoning, the fixation occurs "with no reproductive excess", then he completely ignores (as though irrelevant!) the periods where reproductive excess is required. By this chicanery, Felsenstein is ready to play hide-the-ball with you: Hiding the reproductive requirements: hiding the costs. I've seen many evolutionists use that type of argument.
*
I say, focus on the simplest cases first, and establish a firm foundation for the principles involved. However, many evolutionists (and especially Felsenstein) focus first on needlessly contrived, complicated cases that create confusion and establishes their erroneous conclusions. In Felsenstein's example, the confusion is easily eliminated by taking a constant population size.
Here is another irony. With the exception of Felsenstein, virtually all other evolutionary authors define the cost of substitution only when the population size is constant. (This is because they define it based upon genetic death or genetic load.) They cannot apply their cost concept to a non-constant population size even if they wanted to. For most evolutionists, a constant population size is the first and only case study. In other words, Felsenstein chose a scenario that can't even be studied under other evolutionists' cost concept. You'd think such a fundamental contradiction would be noticed, and commented upon, by evolutionists. But it wasn't.
*
Felsenstein did not deal with my paper's arguments. He merely used the traditional confusions he has promoted for decades, and ignored what my paper said about those.
My paper begins with a simple example, specifically constructed to eliminate confusion and illuminate the fundamental issue. Take a haploid species, where some number of individuals contain an allele of interest, and suppose a scenario claims that one generation later that number has increased by, say, 25 percent. This requires a reproduction rate of at least 1.25. In other words, it requires a reproductive excess of at least 0.25, and that amount is not reduced by any of the following factors:
Here is the fundamental principle: If a scenario claims an allele increases in number of copies (through reproductive means), then that requires reproductive excess. Felsenstein is wrong to insist otherwise.
Conclusion: Felsenstein's review continues the errors he has promoted for decades. He did not seriously respond to the material in my paper. The main lesson from Felsenstein's review is that his errors continue to thrive (and are unopposed anywhere in evolutionary literature) – therefore, Ewens and Crow were wrong to suppress my paper and allow those traditional errors to thrive. (See the tale of peer review.)
Minor points from Felsenstein's review:The cost of neutral substitutions:
My paper's readers cannot tell my background, since my paper is neither creationary, nor anti-evolutionary. Rather, my paper is a neutral, inoffensive, clarification of the cost concept, and operates like putting on a clean, clear, pair of glasses. However, journals reveal the author's name to reviewers, so they soon became aware of my background. From previous encounters, Felsenstein knows I am an Intelligent Design theorist, and his prejudice shows in his irrelevant barb about "divine interventions". He was not the only reviewer to interject irrelevant comments against Intelligent Design into their rejection notices. It happened with three reviewers: Joe Felsenstein, Warren Ewens, and (most prejudicially) Reviewer #1 at the Journal of Theoretical Biology. These irrelevant pejoratives suggest that prejudice is playing some role in rejecting papers by outsiders. The beneficial substitution rate is limited by cost, but the neutral and harmful substitution rates are not, and my paper explains why. There are special mechanisms available to neutral and harmful substitutions, that, on average, are not available to beneficial substitutions. My paper identifies those mechanisms and makes that explanation, and there is nothing like it in evolutionary literature. Instead of commenting on it responsibly, Felsenstein garbled it. The problem for the population?
As noted in my paper, the cost of substitution is not a "problem for the population." Rather, cost is a problem for the scenario. That is, if the species cannot pay the cost of a scenario, then the scenario is not plausible, and the theorist must start over again. Felsenstein garbled this point. Historical treatment:Felsenstein claims my paper does not adequately address the academic history of this topic – concerning who said what, when. He cites two examples: Example 1:
The fact that Haldane called it the “cost of natural selection” and modern writers call it the “cost of substitution” is well-known and is a tedious point of little concern to most readers. Haldane's phrase "the cost of natural selection" was later known as "the cost of selection". That phrase was also then subdivided into various types of cost, which included the "cost of mutation" and the "cost of segregation", as well as Haldane's concept, now known as the "cost of substitution." Felsenstein’s requirement for such arcane historical focus is unreasonable in my paper. My paper is already quite filled with more important matters about the cost concept itself. Example 2:
Felsenstein mistakenly claims "Ewens put forward a cost [argument]". In reality, Ewens put forward a load argument. Warren Ewens made a correct argument, but with one fatal error: he cast his argument in terms of “substitutional load” (not the cost of substitution). Ewens argued (1972 & 1973) that neutral substitution causes a “substitutional load,” while Nei (1973) argued it does not. Their argument was indecisive (as seen by the fact that most geneticists now take Nei’s side). That is noted in my paper. My paper shows that neutral substitutions indeed have a cost, but they do not cause a fitness depression, therefore, by definition, they do not cause a substitutional load. Thus, Ewens and Nei were each partly right, and partly wrong – and it is not useful, in my paper, to disentangle their dialog and their particular mix of confusions about it. Again, Felsenstein's requirement for historical detail is not only unnecessary for my paper; it would detract from the conceptual issues, which are the focus of my paper. In General:There are many confusion factors about the cost of substitution, such as: genetic death, genetic load, environmental-change, extinction, and soft selection. Various researchers combined those confusion factors in a multitude of unique ways, and addressing each specific combination is a needless complication for most readers. My paper is complicated enough, and so is my treatment of history. It is unreasonable of Felsenstein to require further diversion into historical entanglements (such as between Ewens and Nei), especially when those do not lend insight into fundamental concepts of my paper. My paper already meets the technical and documentary requirements. *
Felsenstein has verified my claim that the cost literature is confused and never resolved. Also his review asserts "it is not at all clear that different researchers had the same objective in mind." Therefore, it is not essential to decode what objective the different researchers "had in mind." We are not mind-readers; we cannot read people's minds. Felsenstein demands an extended treatment of what various researchers "had in mind," which is an especially confused and evasive topic, and a detour of little use to readers of my paper. The important issue is how the cost concept is actually used in the field. Felsenstein (1971 & 1972) and many of his followers today: (1) call Felsenstein's cost concept the "cost of substitution," and (2) advanced the notion of that beneficial mutations inherently have "zero" cost of substitution, and (3) positioned that as a solution to Haldane's Dilemma. My paper shows that is false. Trivialities:
Felsenstein snipes against the widely used equation p + q = 1. I see no reason for his complaint. |
1 Felsenstein's review was for the Journal of Theoretical Biology. On his own, Felsenstein chose to publish his review. It was a bad decision, for his sake, to expose his mendacity to the world. But since he did, we may as well learn more about a review process then went terribly awry. Felsenstein (through his intermediary, Allan Force) published his review at the ARN website.
2 The review by Warren Ewens and James Crow was for the journal Theoretical Population Biology.
3 Our concern with "substitution rate" is, ultimately, the long-term rate, averaged over many substitutions. (Not the instantaneous rate of a single substitution or the rate in single generation, though those are of passing interest.)