This book has two purposes. One is to portray the largely forgotten ideas of George Romanes on 'physiological selection' as a major contribution to our understanding of speciation. The other is to present Donald Forsdyke's own ideas on [a] variety of evolutionary topics, with an emphasis on speciation, but also including such questions as the evolution of dominance and dosage compensation. The connection between the two is that Forsdyke believes that his ideas on speciation are a molecular version of Romanes'. He also believes that he has hit on the true explanation of speciation, which has eluded the community of evolutionary biologists. Judging from the barbed comments scattered throughout the book, Forsdyke clearly has little respect for this community, especially theoretical population geneticists. Fisher, Haldane and Wright are dismissed with the comment 'their approach, and that of their followers, was largely genetical, with mathematical and rhetorical overtones which sometimes tended to obscure rather than enlighten' (p. 89).

    Needless to say, this belief is likely to be viewed with scepticism by mainstream evolutionists, especially since there is a vast literature on the evolutionary genetics of speciation which is barely mentioned by Forsdyke (there is only one reference to Dobzhansky and two to Mayr, as opposed to 32 to Forsdyke), which has led to very different conclusions from his. After reading this book, I am still not entirely clear what Romanes had in mind by 'physiological selection', but it seems to mean much the same as reproductive isolation. The quotations from Romanes on pp. 52-53 indicate that he regarded this as a much more important evolutionary principle than natural selection, which no doubt explains the hostility towards his ideas displayed by Huxley and Wallace. It is also quite unclear from these quotations whether Romanes viewed reproductive isolation as a by-product of evolutionary divergence in allopatry, or whether he envisaged some kind of sympatric speciation. 

[The chapter summary (p. 63) states" "Physiological selection progressively increases the reproductive isolation of the incipient group, thus creating conditions favourable to the preservation of non-adaptive or adaptive phenotypic variations. The latter would constitute a target for natural selection. The spontaneous 'intrinsic' gonadal variations provide a general basis for the origin of species, and are distinct from 'extrinsic' variations affecting phenotypic characteristics such as time of pairing." Thus Romanes saw physiological selection as a way of achieving reproductive isolation sympatrically and independently of natural selection, but he had no way of knowing at that time what might be the basis of the 'intrinsic' gonadal variations. DRF] 

If Forsdyke's account of Romanes' ideas is accurate, it is difficult to disagree with the remarks of Ernst Mayr (cited on p. 214) that 'Romanes ... made no clear separation of geographical and reproductive isolation ... and often dealt with speciation as if it was the same as natural selection.'[This was part of a paragraph designed to show that Mayr's reading of Romanes had been superficial.DRF]

    What about Forsdyke's own ideas? He focuses on hybrid sterility as the key problem in speciation. This in itself seems unfortunate, since there are many different types of isolating barriers which can separate good species.

[Early in the book (p. 21) it is stated: "Formation of a new species requires reproductive isolation. Any shape or form of reproductive isolation will suffice. ... We seek a fundamental form of initial reproductive isolation which has operated in the general case of the majority of species, both past and present." DRF]

Indeed, comparative work by Coyne and Orr (1997) shows that behavioural isolation in Drosophila often evolves more quickly than post-zygotic isolation, and that full sterility of both male and female hybrids comes relatively late. Furthermore, as has been pointed out by Klinman et al. (2001), the mechanism proposed by Forsdyke is in contradiction to a large body of data (none of which is mentioned in his book). His idea is that hybrid sterility arises as a result of evolutionary divergence in GC content of silent and synonymous DNA sequences. This is claimed to lead to sterility in hybrids between parents whose GC content is sufficiently diverged, although the precise mechanism involved is never spelt out. [The mechanism is spelt out (p. 114) accompanied by evidence derived from bioinformatic analysis of sequences. DRF] No evidence is presented that such divergence is indeed causally involved in hybrid sterility, and genetic studies that point to individual genes contributing to reproductive isolation are ignored. [The inadequacy of genic factors is the subject of Chapter 6. DRF] Studies of codon usage show that the mean GC contents of synonymous sites are often nearly indistinguishable between related species (Kliman et al. 2001). This difficulty is recognized by Forsdyke, who suggests that GC content first diverges and then converges, thereby rendering his theory untestable.[Further evidence is provided by studies or retroviral genomes (p. 113), and polyploids (p. 115). See p. 125 for "an ideal experiment" to test the theory. DRF]

   In fact, application of a little mathematics shows that random drift or selection are likely to produce only a very slow change in the GC content of a genome; the proportion of the genome that is GC has a denominator of the order of the genome size, unless the states of different sites are highly correlated. The variance in GC content between individuals in a population is thus of the order of the variance in GC content at an individual site, divided by the genome size. This is clearly an almost vanishingly small quantity; since both drift and selection operate at speeds that are limited by within-population variance (as was well-known by the despised trio of Fisher, Haldane and Wright), it will take many tens of hundreds of millions of generations for GC content to be significantly changed in evolution. This is inconsistent with the relatively recent dates of divergence of many closely related species. Forsdyke's theory is thus not just untestable, it is unworkable.

BRIAN CHARLESWORTH
Institute of Cell, Animal and Population Biology,
The University of Edinburgh

References

Coyne, J. A. & Orr, H. A. (1997). "Patterns of speciation in Drosophila" revisited. Evolution 51, 295-303.

Kliman, R. M., Rogers, B. T. & Noor, M. A. F. (2001). Differences in (G+C) content between species: a commentary on Forsdyke's "chromosomal viewpoint" of speciation. Journal of Theoretical Biology 209, 131-140.