Discontinuous variation of this polymorphic kind, in which intermediates are nearly or completely absent, must be maintained by some type of switch-mechanism producing alternative forms. This, it might be thought, could be provided by environmental stimuli; yet it seems hardly ever to be so, doubtless owing to the difficulty of obtaining clear-cut and contrasted phases by such means. It is true that very distinct seasonal forms of butterflies are evoked by length of day or other external differences at a critical stage after pupation, as in Araschnia levana or Precis sesamus. However, in these circumstances, the whole of each generation is similar. Far greater difficulties, requiring an extremely delicate adjustment to external conditions, 12 would be encountered in producing in this way contrasted and discontinuous types within a single brood. Moreover, their proportions could not be accurately adjusted to meet changes in the ecological situation: a drawback fatal to the evolution of a polymorphism which, moreover, often arises automatically owing to the development of heterozygous advantage.
On the contrary, therefore, the control of polymorphic phases is almost always genetic. It may be imposed by the segregation of a single pair of major alleles, or else by the corresponding members of a super-gene. This latter may consist of a few closely linked units, of an inversion (whether 'short' or 'long', pp. 21-2), or even of a whole iso-chromosome.
The total selective effect of a mutant may change from one time or place to another, but in any given situation it must be either harmful, of neutral survival value, or advantageous to the organism. These three situations need at the outset to be considered briefly with reference to their evolutionary significance.
In the first place, a consistently disadvantageous gene or other genetic unit, can never spread beyond the status of a rare mutant. It is true that it may occasionally be injected into a population by immigration. This, however, must be very exceptional since, to be other than an occurrence of extreme improbability, such immigrants must come from some locality where the gene in question has an advantage; but they will constantly be eliminated by counter-selection on arrival in a habitat where it is harmful.
Secondly, it is a fact that the different phases of a polymorphism are often distinguished by features of an apparently trivial kind perhaps of no importance to the organism, such as certain slight changes in colour-pattern or superficial structure. It might be thought therefore that the genetic distinctions which these advertise have no effect upon survival-value and that the spread through the population of whichever is the more recent of them has been of a purely random kind. Such a conclusion is erroneous, and for several contributing reasons.
13 At the outset, R. A. Fisher (1930b) calculated that the balance of advantage and disadvantage between a gene (or other comparable unit, such as an inversion) and its allele must be extraordinarily exact if the two are to behave as neutral in respect of one another. That situation must therefore be very rare. There is also clear experimental evidence for this conclusion.
It appears that genes always have multiple effects: a point which may be illustrated by the species which has been subject to a more complete genetic analysis than any other; that is to say, Drosophila melanogaster which, in this respect, appears to be typical of organisms in general. The majority of the mutants studied in this fly, apart from lethals and semi-lethals which are commonest of all, are distinguished by quite trivial features: slight differences in wing-neuration, in eye-colour or in the bristles of the body. Yet there has never been one of them that does not affect viability; altering the length of life, capacity to survive in unfavourable conditions, male fertility or the number of eggs laid per unit of time. That is to say, the genes in question, insignificant as are their visible effects, have an important influence upon the physiology of the organism, modifying profoundly the individual as a unit upon which selection operates. It is another issue that their action is almost invariably disadvantageous: a point which can be dealt with in passing. Mutations are random events relative to the needs of the organism, though they are not so from other points of view: the chemical, for instance. But random changes can seldom promote the harmonious working of an organized system, such as the gene-complex of a plant or animal.
We have further evidence for the selective value of genes with apparently trivial effects. This is derived from Drosophila (and other) cultures when maintained in the laboratory. For seemingly unimportant features, such as bristle-number, do not vary at random in the stock bottles, as they would do were they selectively neutral.
Another step in this general argument was taken also by R. A. Fisher when he wrote The Genetical Theory of Natural Selection (1930a) and showed that the spread of a 'neutral' gene, necessarily rare as we have seen, must be exceedingly slow:14 so slow indeed that, if it derived from a single mutation, the number of individuals possessing it in any population cannot much exceed the number of generations since its origin. That is to say, long before the frequency of such a gene has increased to any appreciable degree, the delicate balance required to maintain its neutrality (in comparison with its allele) will have been upset by changes in the environment or in the gene-complex of the organism; while mutation is too rare for its recurrence materially to hasten the process. Indeed a unifactorial character must be polymorphic if found even in 1 per cent of a considerable population, amounting perhaps to 500 individuals or more, when random genetic drift may reasonably be excluded as unimportant.
It will be apparent, therefore, that the more recent phase (or phases) of a polymorphism must have been an asset to the organism in order to have reached a frequency greater than that attainable merely by mutation-pressure. Yet a gene (or other switch-unit) which possesses and preserves an overall advantage must spread through the population by selection until its former normal allele is reduced to the status of a rare mutant. A temporary or 'transient' polymorphism (Ford, 1940a) is generated meanwhile; one which is annulled when the process is complete. Evidently a permanent diversity cannot arise in this way. That can only be maintained if a genetic character possesses an advantage when rare; one, however, which wanes, is neutralized, and then converted into a disadvantage as the gene responsible for it becomes commoner. Such a situation, it might be supposed, is of a highly specialized and unusual type: specialized indeed it is, but not unusual. On the contrary, it represents the course of events to be expected when a major gene gains some advantage, owing to changes in the environment or in the gene-complex of the organism, and begins to increase in frequency. The various conditions which produce a lasting or 'balanced' polymorphism of this kind (Ford, 1940a) must first be studied briefly before the properties with which it is associated can be described and illustrated by examples.
It is clear that there are certain situations which automatically promote discontinuous diversity. Of these sex is an obvious instance falling, as it does, within the definition of balanced 15 polymorphism; for any tendency for females to increase at the expense of males or the reverse must normally be opposed by selection. A somewhat similar situation is encountered in Batesian mimicry, when polymorphic; a condition described in greater detail later (pp. 32-6, 53-5). It is one in which some or all of the phases are 'mimetic'. That is to say, they obtain protection by resembling other species, their 'models', which are relatively free from attack. Such immunity is due to poisonous qualities, the possession of a powerful means of defence such as a sting, or to an unpleasant taste; attributes that are advertised by conspicuous 'warning' colour-patterns easily recognized by potential predators. The frequency of a mimic rises by virtue of the benefit which its deceptive resemblance confers. Yet, with increasing numbers, the advantage thus acquired diminishes, reaches neutrality, and is converted into a disadvantage. The latter state is attained when such a palatable form, one necessarily conspicuous since it must resemble a warningly coloured species, becomes associated by its enemies with edibility rather than inedibility.
Several other 'ecological' conditions promote diversity. One of them arises from differential mating-preferences (p. 42). For there are instances in which each form of a species prefers to mate with some other type rather than its own, a tendency which always favours the rarest of them. De Ruiter (1952) first drew attention to another situation having a similar effect. He pointed out that many predators (birds and mammals) learn the colour-pattern of their prey, so that they tend to hunt for a form resembling the one they last found to the exclusion of others. The result of this is to handicap the more abundant phases and to increase the less numerous ones, whichever these may be, so opposing uniformity.
Such methods of maintaining polymorphism are of a somewhat exceptional kind; a list of them has been given by Williamson (1958). The one normally responsible for doing so is that in which the heterozygote is advantage compared with both homozygotes: a state which evidently ensures the co-existence of distinct phases in a population. 16 It can in fact arise in two ways, the consideration of which requires a brief discussion of super-genes and their formation. It is necessary at this stage to turn aside to look at that aspect of chromosome mechanics before considering the evolution, and analysing the effects, of 'heterozygous advantage'.