Population genetics revisited

John Maynard Smith¹

In 1987, just 100 facsimile copies of this book were produced. The book has now been typeset and made available to a wider audience, with a new preface by the author. In it, Fred Hoyle rediscovers many of the classical results of population genetics. Although he has already read Ronald Fisher, J. B. S. Haldane, Sewall Wright and Motoo Kimura, Hoyle has clearly rediscovered some of the main results. This is apparent from some of the less familiar ideas that he reports: for example, Müller's ratchet, Orgel's catastrophe theory of ageing and, more relevant to Hoyle's main thesis, the idea that a favourable mutation is unlikely to become established in a population unless genetic recombination occurs in the face of recurrent deleterious mutations — a point made in more detail by Joel Peck.

Has Hoyle reached other correct conclusions that are also new? This question is difficult to answer, because the book carries few references explaining which of Hoyle's findings are new, and which are merely confirmations of findings by others. However, I did spend some time on one particular topic, to which Hoyle devotes a whole chapter. This is the notion of a "cost of selection" associated with the establishment of a favourable mutation, of which he says "the claims are illusions". If true, this would be important. The idea of a cost originated with Haldane, who calculated that the number of individuals that must die selectively during the establishment of a favourable mutation lies between 10 and 100 times the population size.

Kimura used this as an argument for the neutral theory of molecular evolution. In effect, he estimated the rate at which molecular changes have occurred during evolution, and argued that the total cost of bringing about such changes was greater than the population could bear. He therefore concluded that most of the changes must be selectively neutral. My own contribution to the debate was to point out that Kimura's conclusion depends on the assumption that the costs associated with different genes can be added. This is correct if the effects of genes on fitness are independent, so that fitness effects are multiplicative. If gene effects are synergistic, the total cost is much lower.

When I first read Hoyle's discussion of this topic, I wondered whether he had rediscovered my objection, but this is not so. He does not object to the additivity of costs; it is the cost of a single substitution that he regards as an illusion. I am baffled by what he says. I can see nothing wrong with the algebra, but the conclusion that costs are an illusion does not follow from it — it is merely an assertion. I may be mistaken, and so I hope others will look at Hoyle's argument. But my attempt to find something novel and true in his book has therefore failed. Others, however, might profitably look for nuggets of gold. After all, when Manfred Eigen and Peter Schuster reinvented evolution by natural selection, they came up with the novel and important idea of an error threshold. Hoyle is a very clever and creative scientist; it would be surprising if he has laboured entirely in vain.

Above all, however, Hoyle reaches an unjustified and ridiculous conclusion. He accepts that, given genetic recombination, natural selection can produce detailed adaptation of species to their ways of life. But he argues that it cannot be responsible for the major changes required for the emergence of new orders, classes or phyla. Something else is therefore needed. This, Hoyle asserts, is the introduction of new genetic material from outer space. He is not suggesting merely that life was first seeded on Earth from space — the theory of panspermia — an idea I find unappealing but not irrational. Hoyle argues for a continuing process of introduction.

He claims that the origin of new major groups is impossible without such intervention because, "What mutations cannot do is to find improvements which demand the simultaneous change of several base pairs". Evolutionary biologists would agree that a change requiring a number of base changes, each of which is without value until all are present, cannot occur by natural selection. They have therefore concluded that the origin of major groups has been a stepwise process, with each genetic change being an advantage on its own (although they would probably mention symbiosis as an exception). If there is no stepwise path up the mountain, natural selection won't climb it. Much thought has been given to the nature of the intermediate steps.

Hoyle gives no reason why such intermediate steps could not have existed; it is merely an assertion. In his preface, it is clear that he does not actually believe it. He says his preferred scenario is that "all genes in present day organisms were here already in the metazoans that invaded the Earth 57 million years ago at the beginning of the Cambrian era, making the subsequent story of terrestrial evolution into one in which genes have been called into operation as ecologic conditions permitted them to be so". Thus, most of these genes would have remained unexpressed because they would have been useless until appropriate conditions arose. Hoyle accepts that unexpressed genes would accumulate damage, but argues that, if populations were large enough, the deleterious mutations would not be fixed. Therefore, when conditions were right, selection plus recombination would restore the original genetic message. But this is to assume that selection can act simultaneously at many loci, which requires that each of the multitude of genetic changes should be separately advantageous. This is precisely the assumption he has rejected in order to conclude that intervention from outer space is needed.

It is sad to see such a creative mind devoted to reaching such an absurd conclusion. It might have been better for the reputation of a great scientist if this book had been left to the decent obscurity of a facsimile edition.

To read this story in full you will need to login or make a payment (see right).