26It has been pointed out that the most general basis for maintaining polymorphism is that in which the heterozygote is at an advantage compared with both homozygotes. Moreover, this situation tends to evolve automatically when a major gene has become an asset, due to a change in the environment or in the gene-complex, and begins to spread through the population. There are two principal ways in which such 'heterozygous advantage' can arise, and these must often be combined.
In the first place, it will be realized that any gene which begins to be favoured must exist almost entirely in the heterozygous state during the earlier stages of its increase. That genotype must have gained some superiority over what was the 'normal' homozygote or it would not be displacing it. Recessive lethals and semi-lethals are a particularly common form of mutant, and they can accumulate in the neighbourhood of the major gene responsible for the new polymorphic phase because, as heterozygotes, they are sheltered from selective elimination. By the time the newly successful super-gene is becoming common in the population, its homozygote, which should then be appearing at a frequency no longer negligible, will tend to be handicapped by the harmful recessives with which it has become linked. In these circumstances segregation in a 2:1 ratio, or an approach thereto, must take place where a 3:1 ratio is expected. Yet the fact that the heterozygote must have an advantage over the original or 'normal' homozygote will tend to restore 3:1 segregation although the other homozygote is rare or absent. Many instances in which the homozygous dominant is deficient have been encountered. Thus R. A. Fisher (1927) has demonstrated this condition in females of the Oriental butterfly Papilio polytes, bred by Fryer (1913), 27 when segregating for genes controlling the sex-limited polymorphism of mimetic and non-mimetic forms, because the homozygote is handicapped by the linkage just mentioned.
Secondly, attention has already been drawn to the fact that major genes, at least, always have multiple effects ant that their mutation occurs at random relative to the needs of the organism. Consequently if one of the features for which a mutant is responsible gives it an advantage, the others are very likely to do so too [i.e., be random relative to needs]; and indeed they will almost certainly be deleterious. As Sheppard (1953) has pointed out, selection will then tend to make the beneficial effect dominant and the harmful ones recessive. In these circumstances, the homozygotes will, relative to the alleles in question, have both advantages and disadvantages while the heterozygote will have advantages only, so ensuring its superiority over the other two genotypes and therefore establishing polymorphism.
This situation is, for example, illustrated by the work of Caspari (1950) on the moth Ephestia kuhniella. In that species, the alleles responsible for the production of red or brown testis-colour control a number of other features in addition. Some of these are advantageous and dominant, or nearly so, while others are disadvantageous and recessive.
Further instances of a comparable kind have already been analysed. One of these is provided by ebony body-colour (e) in Drosophila melanogaster. Its influence on pigment production, giving rise to very dark flies, which also are rather weakly, is completely recessive; but certain of its effects improve viability, and these are operative in the heterozygote. Thus 'ebony' is not eliminated when introduced into a wild-type stock in the laboratory but, after some generations of adjustment, the two alleles stabilize at a value depending upon environmental conditions: that is to say, a polymorphism is established due to heterozygous advantage (Kalmus, 1945). The e gene reaches a greater frequency at low than at high temperatures while dryness tends also to favour it, though to a less degree. Starting with cultures containing 25 percent of phenotypically ebony flies, 28 Kalmus found that this form reached an equilibrium of about 8 per cent at 26 deg. and 27 deg. C and 30 per cent or over at 12 deg. to 15 deg. C. The experiment was continued for periods varying between 400 and 460 days, stability being attained at different times in the various stocks, depending upon the environment. Several other genes behaving in an approximately similar fashion have been studied, and in some of them the mutant homozygotes are so heavily eliminated as to be rare or absent. Thus it will be apparent that though a disadvantageous recessive character cannot itself be polymorphic, the gene controlling it may be so.
Sheppard (1953) points out that heterozygous advantage can also result from an inversion. This will generally carry many genes, some of which will be dominant in effect and others recessive, respectively favoured and opposed by selection. Thus the inversion may at once produce heterosis; if not, this can arise owing to dominance-modification on the lines already indicated.