Fisher set out as a convinced Darwinian and Mendelian, and his first tasks were concerned with expressing the well-established results of the biometricians, such as phenotypic correlations among relatives, in terms of Mendel's irreducible particles. This he did in his startlingly original 1918 paper, already alluded to above. In so doing, he laid much of the foundation of modern biological statistics (e.g. he recommended variance - s2 - squared deviations from the mean - as a measure of variation rather than the standard deviation, s, since variances are additive when their sources are independent) Of direct relevance here, he recognised not only that all variation can be partitioned into genetic and non-genetic causes, but that, while the correlation among siblings is an expression of additive (selectable) genetic variance, there is also a genetic residue, including the effects of linkage, dominance and epistasis. Thus, in a single paper, nearly 80 years ago, Fisher set up the statistical model of quantitative inheritance that we still use today.
Fisher concluded that significant evolutionary change took place only in large populations, almost exclusively by selection on near-independent loci
In another seminal paper in 1922 (note 1), he turned to a consideration of the fate of variability in natural populations. Beginning from Hardy-Weinberg conditions, he investigated the effects of selection, dominance, mutation rates, random loss of alleles, and assortative mating. He concluded that the main stuff of long-term evolution concerned large populations, given their vast stores of genetic variability, and their consequent insurance against extinction through failure in the evolutionary race against a hostile external world. This being so, he considered random effects to be of no long-term importance, while even tiny selection differentials (which he saw as typical) could have important consequences, given enough time. In such populations, therefore, the deterministic results of selection acting upon the small phenotypic effects of single genes reigned supreme, since these effects would be, to all intents and purposes, randomized across virtually the entire population genome. Therefore, the alleles at different gene loci could be sorted more or less independently of one another. This is the so-called "bean-bag genetics." He looked explicitly at the fate of neutral alleles, that is, at rates of change under genetic drift, and found that in the large populations which he saw as typical of nature, such changes would be extremely slow. He concluded that chance was not a potent factor in directing evolutionary change. He developed the same attitude to the role of chance at the level of mutations: though they were, of course, important in providing grist for selection, mutation rates were so low that they could not direct evolutionary change in themselves. Also, he regarded beneficial mutations of large effect (producing saltations) as very rare at best.
This all led him to the view that deterministic selection was the mostimportant factor in evolutionary change, that selection generates adaptations, and that these adaptations are probably very precise, because he believed the effects of most alleles to be small.
It also led him, as suggested above, to conclude that small population size provided a very grave risk of extinction, since small populations held little genetic variation with which they might respond to novel evolutionary exigencies, and besides were always at risk of chance obliteration by natural catastrophe. In short, he disregarded the evolutionary importance of the effects of chance (mutation and drift), and exalted the role of selection. All this is not to say that Fisher was a naive pan-selectionist - believing that all variation is adaptive and that all adaptation is perfection - to the contrary, he saw populations as continually tracking, with a time-lag, some moving environmental target.
Fisher saw the dominance relationships among alleles as subject to evolutionary modification
He envisaged the effects of genes as being capable of evolutionary modification, and accordingly, interpreted dominance/recessiveness relations among alleles as a phenomenon evolved by the selection of 'modifier genes.' He used this to explain why most obviously deleterious mutants are recessive: because they are deleterious, selection will maintain them at low frequencies, and so they will exist almost entirely in a heterozygous state; any genetic variation in the rest of the genome which has an effect on the expression of the deleterious alleles, making their effects less extreme, will clearly confer an advantage; as the effects become less and less manifested, recessiveness evolves. He also applied this concept to the matter of mimicry with great effect. A further consequence of this dominance evolution would be the common fitness superiority of heterozygotes: over time, any allele's deleterious effects would evolve into recessiveness, while the beneficial effects would become dominant. Thus a heterozygote would have none of the deleterious effects of either allele, but would manifest all of the benefits. Heterozygote advantage was expected to be common in natural populations and therefore that widespread allelic polymorphism would be the rule. According to modern empirical findings, it seems that, while the latter point seems to be true, the suggestion of heterozygote advantage as the global explanation seems wide of the mark.
Much of Fisher's view is synthesised in his Fundamental Theorem of Natural Selection, which has been much misunderstood
All of his views he eventually collected into a single statement: The Genetical Theory of Natural Selection, a book published in 1930. Here he stated, among much else, his Fundamental Theorem of Natural Selection. This theorem has been the subject of much discussion and evident misunderstanding (note 2), but, using modern terminology, it states that: the rate of increase in the mean fitness of any population of organisms at any time, due solely to the effects of allele substitution through natural selection, is exactly equal to the additive genetic variance in fitness in the population at that time.
It is clear that selection reduces any additive variance in traits related to fitness, and that there is more opportunity for evolutionary change through selection when there is more additive variance in the population. It is also clear that we would expect traits which are strongly associated with fitness to show typically small amounts of additive genetic variance (because selection will quickly remove it), and vice versa: traits with only weak ftiness correlates may have large amounts of additive variance. But it does not imply (as was thought by almost everyone, experts included, and as makes intuitive sense) that the population mean absolute fitness must always necessarily increase if there is genic variance in fitness.
However, it does succeed in capturing, in one succinct statement, "the essence of the way selection works, and encapsulates a great deal of evolutionary insight" (note 3). Fisher also maintained, by this theorem, that although the quality of the mean organism may be increasing as long as there is genic variance in fitness, this increase would, at least eventually, be offset by a deterioration in a complex of "external factors," such that the population may not increase numerically at all, though it be continuously evolving adaptively:
We must be very clear that Fisher's conclusions (and anyone else's) regarding the "balance of power" between the "forces" of selection and chance, and the probable consequences of selection's actions, stemmed almost entirely from assumptions about the characteristics of typical natural populations - in Fisher's case, that they were effectively very large, and that individual gene effects were typically small and finely graded. It is important to understand that this was not known empirically to be the case, indeed, the matter is still not settled, remaining under active investigation.
Having noted, several times now, this empirical near-vacuum in which theorists of Fisher's day were working, it is appropriate to begin some consideration of the sources of Fisher's hunches, opinions and convictions about matters of empirical fact relevant to his theories - after all, ideas have to come from somewhere! We may distinguish various kinds of non-empirical or peripheral influences that may have served to inspire and/or support (even 'confirm') ideas in Fisher's evolutionary theorizing; but, in fact, these several sources are probably interacting and inter-dependent. First, we must acknowledge the influence of the immediate context - the research tradition - in which he worked. We can next comment upon his concerns and interests in general intellectual life, elsewhere in science, and in regard to society at large. Finally, we can look at his beliefs.
One of Fisher's primary motivations was to reconcile modern genetics with Darwin's view of life
Fisher had life-long ties with Darwin's family and works. He received Darwin's complete works as his school-leaving prize, and it is clear that he found Darwin's work a powerful inspiration for his own work; indeed it was, as we have seen, his express intention to reconcile Darwin's view of life with modern genetics, and proceed to elucidate the consequences. He easily fitted, then, into the wholly selectionist, adaptationist, progressionist tradition which was powerful in England from Darwin's time up to the late 20s.
He was also inspired by simple physical laws, and sought biological analogues for them
Oddly, given his early interest in, and commitment to, particular views on biological evolution, he went to university to study mathematics and physics, and, at least by the time he completed these studies, he had developed an affection for the deep explanatory power of simple physical laws, such as those of gravitational attraction, the behaviour of gases, and thermodynamics. We know from later writings that he was very much concerned to find a correspondingly simple biological law which would render the entire nature of biological populations intelligible, and he felt that he had this in his Fundamental Theorem. His whole approach was analytical and highly reductionist, disregarding the possible messiness of real organisms. His view of the behaviour of genes in populations was analogous to that of gases:
There are some very interesting correspondences (some would say causal connections) between Fisher's understanding of the central position of entropy (disorder) in physical systems, his views on populations and their evolutionary processes, and his own personal abhorrence of chaos and liking for order, progress, and optimism. He recognised and analysed (see note 1) the influences of random processes, as they manifested in random drift and in mutation, but, as we have seen, his assumption of very large panmictic populations (analogues of gases composed of freely- and randomly-moving molecules) reduced their effects to near-zero when compared to selection. Even tiny selection differentials among alleles could render mutation and drift effectively irrelevant. The ruling forces of selection could then, albeit perhaps slowly, build adaptive progress, and vanquish all tendency towards arbitrary degeneration and decay through the forces of chance. This picture of progressively adapting evolutionary populations was consistent both throughout Fisher's own life and with his general, religiously-based, life-long optimism about life, in particular about human life:
Fisher grew up in a time when the English intelligentsia were preoccupied by the loss of Empire and the decay of "the race." Such concerns, automatically assumed to be of a biological nature - a matter of "blood" - was also generalized to the problems perceived within English society: the upper classes were being threatened by extinction through their diminishing population size - they were in danger of losing the adaptive race with the profligately-breeding "lower elements." Hence eugenics: the restoration of natural processes in society (as the eugenicists saw them), to the greater benefit of society at large. Today, eugenics has a (largely justified) bad name, given its association with the odious practices of Nazism; it is commonly supposed to have an exclusive association with far-right political attitudes, but it is important to realise that this was not always so. Before World War II, people of all political stripes, from the most unreconstructed reactionary to card-carrying Marxists embraced some form of eugenics (though of course with very different goals in mind), and thereon hangs a very interesting point. At that time, almost everyone took it for granted that virtually every aspect of an individual's make-up - personality, morality, intelligence, you name it, had a biological (genetic) determination, and that determination was almost the same thing as destiny (if any of this is beginning to sound familiar, it should.)
It is possible that Fisher's social eugenecist notions coloured his science
In this connection, then, we should note that Fisher was a committed eugenicist years before he became a professional population geneticist. Some historians of science find this an extremely telling point (see, for example, Norton, in Grene, 1983), and they wonder about the extent to which his science informed his social ideas and vice versa. Michael Ruse tells me (in litt.) that it seems clear to him that "Fisher literally lived eugenicism...... He married a naive young girl, just seventeen, who was chosen deliberately because she would be good breeding stock. And then, despite severe financial hardships, he proceeded deliberately to have a large family of eight children, as an expression of his beliefs..... At the professional level, again and again it was the dog of eugenics which wagged the tail of biology...... Fisher's thinking about evolution was impregnated with his extra-scientific beliefs about society and its Progress, positive and negative." As an example of Fisher's thought on these matters:
Fisher has left a large & extremely rich legacy
Altogether, Fisher's work was enormously influential (and remains so),
and it formed the intellectual basis for much research in evolutionary
and ecological genetics, especially in England: a research effort that
sprang from a starting view giving primacy to natural selection in moulding
nature, minimizing the importance of random phenomena at all levels, and
seeing adaptation as precise and nearly ubiquitous. In a very real sense,
however, Fisher had almost as much influence on modern thinking through
misinterpretation of his Fundamental Theorem, via the works of Sewall Wright.
Unlike the fight over saltationism and gradualism, however, this fight
was highly stimulating to the participants in the field. It is to Wright
we now turn; he had a very different set of assumptions.
Note 1 Fisher, R.A. 1922 On the dominance ratio.
Proc. Roy. Soc. Edinburgh 42: 321-341. Return
to text location.
Note 2 Frank, S. A. & M. Slatkin 1992 Fisher's Fundamental Theorem and Natural Selection. Trends in Ecology & Evolution 7: 92.-95.
see also: Edwards, A.W.F. 1994 Fisher's Fundamental Theorem of Natural Selection. Biol. Rev. 69: 443-474. Return to text.
Note 3 Crow, J. 1990, Fisher's contribution to genetics and evolution. Theoret. Pop. Biol. 38:263-275. Return
Note 4 Van Valen, L. 1973 A new evolutionary law. Evolutionary Theory 1: 1-30. Return
Go to next section