40 It is evident that polymorphism may be of advantage in some areas but not in others while the phases may not be balanced at the same frequencies in different parts of a species' range so giving rise in certain circumstances to polymorph ratio-clines. This is an obvious deduction from the theory of the subject; so too is the conclusion that even if established originally by balanced selection for visible qualities, which may or may not be combined with physiological effects, the unit controlling polymorphism will tend to evolve into a super-gene and to acquire heterozygous advantage. It will be instructive first to examine a few of the instances in which this does not seem to have occurred and then to consider different illustrative examples in which evolution of this kind has taken place.

Panaxia dominula, family Arctiidae, is a brilliantly coloured, day-flying moth, 4 to 5 cm across the extended wings. Its distribution comprises Continental Europe, excluding the more northerly regions, and the southern half of England. It forms local colonies, often widely separated, in which it may be abundant and to which it is closely restricted, so that there is very little interchange between them. Its habitats are of two types: principally marshes, but occasionally short stretches of hedgerows.

The fore-wings are blackish with a green, sometimes blue, iridescence, and bear a number of white or yellowish spots. The hind pair are scarlet, marked with black. The insect is therefore conspicuous as it flies actively in the sunshine. 41 In England it has but one generation in the year: the imagines are to be found for about a month, beginning in early July. The larvae will feed on a wide range of plants, including Urtica and Rubus, but they show a decided preference for Symphytum officinale, which is restricted to damp ground.

A polymorphism of Panaxia dominula has been the subject of intensive study (Ford, 1964). It occurs in one locality only; a marsh of about 15 acres at Cothill, 5 miles from Oxford. The condition is controlled by a single pair of alleles. The rare homozygote, bimacula, differs greatly in appearance from the typical form (dominula), which represents the other homozygous phase. All but the two basal spots on the fore-wings are obliterated while the amount of black on the hind pair is greatly extended. The heterozygotes, medionigra, are in appearance much nearer to the normal phase, but they can be distinguished from it: the central white spot on the fore-wings is reduced or absent while an extra black dot is present in the centre of the hind pair. Thus the gene-frequency of this polymorphism can be ascertained by direct observation.

The Cothill marsh is isolated from any similar habitat by woods, a village and agricultural land. It is of an unusual type, consisting of calcareous peat which is of the earliest Atlantic age almost to the present surface, and, no doubt in consequence of this, supports a fauna and flora unusual for that part of England. Thus the polymorphism of P. dominula may well be connected with the exceptional nature of the habitat, while there is no evidence that the condition is maintained by heterozygous advantage; quite the contrary indeed. A detailed study of it has been carried out in the field for twenty-six consecutive years (1939-64) and the total captures (nearly all of which have been marked and released into the colony again, Ford, 1964) amount to 15,784 dominula, 1,221 medionigra (the heterozygotes) and 28 bimacula. Assuming equal viability of the genotypes, 1,209 heterozygotes are to be expected, with the rarer gene occupying 1,277 out of 34,066 loci. Thus the results obtained in the field do not suggest heterozygous advantage, nor is there any indication of this in the extensive broods which have been reared in the laboratory. 42 It is indeed likely that such heterosis has not yet had time to evolve, for the polymorphism at Cothill may well be fairly recent. Several features suggest that conclusion. Thus, as already mentioned, it is known only in this place; moreover the character has not yet become dominant or recessive, though significant changes in both directions have been produced by only three generations of selective breeding.

The imagines have been proved violently distasteful to a wide range of potential predators (Fraser and Rothschild, 1960). They are warningly coloured in flight; also if disturbed at rest in cold weather when, instead of attempting to escape, they thrust the forewings forward, exposing the scarlet hind pair. On the other hand, when on vegetation, especially upon the flower heads of Eupatorium cannabinum to which they are much addicted, both the dominula and medionigra phases are cryptic; bimacula is less well concealed, though it is too rare for that consideration materially to influence the gene-frequency. However, the gene has other effects upon which selection must operate. These influence mating-habits through scent-production, fertility and survival in the earlier stages of the life-history.

Sheppard (1952) found that dominula, medionigra and bimacula females all prefer to mate with the males belonging to one of the other forms, though the males themselves show no such predilection. When, as in this instance, a gene promotes pairing between unlike types that aspect of its action must often pass unrecognized. Yet this instance does not provide the only one of the kind recorded in animals: it is said that the ebony and vestigial males of Drosophila melanogaster tend to pair with females of the opposite type (Rendel, 1951). Indeed such a tendency may well be widespread, for this is a genetic situation that alters the mating system, a condition likely to be subject to strong selection. It is clearly to be compared with the incompatibility mechanism in plants, including the heterostyle-homostyle device (pp. 45-53), in which, however, the effect is more extreme since it tends to promote cross-fertilization with in-breeding, respectively. In animals, the preferential mating of unlike genotypes must also increase heterozygosity; and it must produce, and maintain, polymorphism. 43 In the example under consideration, that of Panaxia dominula, the effect is heightened by the sexual behaviour; for the scent of a virgin female attracts numbers of males between which, therefore, there is competition, so tending to favour the individual which stimulates her most rapidly.

It has, in addition, been shown by Sheppard (1953) that the heterozygous (medionigra) males are slightly less fertile than the normal form. Also that in certain environments the early stages of medionigra are eliminated to a disproportionate extent, giving a 75 per cent survival-rate compared with those of typical dominula. The latter discovery was made by the method of establishing artificial colonies, which is proving of great value in the study of ecological genetics. These can be founded with known proportions of polymorphic phases (or of forms determined on some other genetic basis; for instance, polygenically), to enabling differential selection to be detected in nature. A corresponding device consists in estimating the size of a naturally occurring and isolated population (generally by means of the marking, release and recapture method: Fisher and Ford, 1947; Ford, 1953a) and adding to it a known number, and therefore a known proportion, of a particular mutant or polymorphic gene. This has the additional advantages in that the site has already proved suitable for the species.

These two techniques make it possible to conduct controlled experiments in natural conditions. They have thrown much light upon the ways in which polymorphism may be maintained. These have been discussed in detail with reference to P. dominula by Sheppard and Cook (1962). Three artificial colonies of this species have now been established starting with the medionigra gene at a high frequency, and in all of them it has been quickly reduced to a much lower level (from a gene-frequency of 25 to 7.3 per cent in 10 generations, from 25 per cent to 14 per cent in 3 generations, and from 50 to 43 per cent in one generation, respectively). On the other hand, the gene has been introduced at a very low level into an already existing and large population in the Oxford neighbourhood (Sheepstead Hurst, Berkshire). 44 A thorough investigation had shown that medionigra was absent there. It was then added in such numbers that the gene occupied about 0.02 per cent of available loci. This was in 1954 and it had increased 65 times by 1963, having risen to 1.3 per cent. Thus we have rather clear evidence that though heterozygous advantage is here absent, this type of polymorphism in P. dominula can nevertheless be maintained because medionigra is at a disadvantage when common and at an advantage when rare. For in the latter situation its relatively poor viability is balanced by the asset due to its disassortative mating, the value of which is reduced as the proportion of the less common form increases.

The question arises, and is not easily answered, whether the various effects associated with the medionigra form are in fact controlled by a super-gene; a situation which, in the absence of polytene nuclei, can only be demonstrated by crossing-over within the unit concerned at a frequency above the extreme rarity of the intra-cistronic type. Should that occurrence not be too rare, it can be detected if distinct visible characters are involved (e.g. the male and female parts of the flower in Primula, p. 47; background-colour and banding in snails, p. 56) but hardly when differences in pattern are associated with changes in habit or survival-rate.

On the whole, it seems more likely that we are here faced with the multiple effects of one gene, perhaps adjusted by selection, rather than those produced by the evolution of a compound unit. As already suggested in another connection, there does not seem to have been time for such a process to have taken place. It may be objected that in Panaxia dominula the chance combination of such courtship stimuli with the other medionigra characters is an extremely improbable one. On the other hand, it must be remembered that the difference in sexual response seems in fact to be the only one favoured by selection, the other features associated with it being irrelevant or else disadvantageous.

It appears, then, that the medionigra polymorphism of Panaxia dominula has evolved neither heterozygous advantage nor super-gene control. 45 The latter step has been taken, apparently without the establishment of heterosis, in respect of the heterostyle-homostyle mechanism of the Primulaceae and other plants. It will be necessary only to recapitulate briefly the main features of that condition in order to illustrate the general principles to be developed here. A more extended account of the subject, together with an entry into the relevant literature, may be obtained from Ford, 1964 (pp. 172-85).

The heterostyle polymorphism has been studied most fully in the Primrose, Primula vulgaris, a species which bears two types of flowers upon separate plants. In both of them the male and female parts are widely separated. The 'pin' form has the stigma, borne at the end of a tall style, at the top of a long corolla tube, while the anthers are half way down it. These positions are reversed in the 'Thrums'. Plants bearing one or the other type of flower approach equality in wild populations, with a slight excess of pins (a ratio of about 1 : 1.15 is usual).

As Darwin (1877) originally showed, this distinction in floral structure permits out-breeding since insects visiting the flowers tend to transfer the pollen between the alternate types. The interchange thus achieved is, however, only moderately effective since some selfing is bound also to occur and tends to be produced also by the Thrips and small Coleoptera which frequently inhabit the corolla tubes. Yet it is now known that the importance of these latter agencies is much reduced since the stigma of each form is accessible, owing to the opening of the bud, and accepts pollen before its anthers have dehisced; a situation which strongly favours intercrossing.

Moreover, the differences between the Thrum and pin types are by no means limited to the position of the sexual parts but include stigma-shape (flattened, with short papillae, in Thrum; rounded, with long papillae, in pin) and pollen-size: the pin grains have 2/3 the diameter of those produced by Thrum anthers. However, by far the most important distinction between the two phases is a physiological one. Each class of pollen proves to be largely incompatible upon its own type of stigma, whether that of the same flower or borne upon another plant. Thus the cross Thrum x Thrum is almost completely infertile while pin x pin, though less heavily handicapped, has much reduced fertility compared with Thrum x pin. 46 This check to the setting of seed even when pollen does reach a stigma of like is of two kinds. Thrum pollen germinates upon a Thrum stigma but usually fails to penetrate its surface. Pin pollen, on the other hand, both germinates upon a pin stigma and penetrates it. However, it is then subject to retarded growth compared with Thrum pollen, so that in competition the latter reaches the ovules first. The result is that out-crossing is almost universal in ordinary mixed populations, though in special circumstances (p. 50) selfing can take place between pins but not between Thrums, and normally does so to a slight extent.

Heterostyly is controlled by a super-gene the members of which normally behave as if they were a pair of alleles (denoted S and s) with the Thrum characters dominant to pin. It will be appreciated therefore that the out-breeding system ensures that mating takes the form of a series of back-crosses, Ss x ss, so that the Thrum plants are always heterozygotes.

The genetic control of heterostyly involves a maternal effect. No difficulty arises in relation to the pin pollen, which is genetically homogeneous and grows more rapidly down a Thrum than a pin style. The situation is more curious when we consider the behaviour of the Thrum type, which uniformly fails to penetrate a Thrum stigma and uniformly succeeds upon a pin one. For Thrum pollen is not genetically homogeneous; on the contrary, though half of it carries S half carries s and this latter should behave as pin pollen, which it does not do. Thus the growth of the Thrum gametophyte must be controlled by the type of sporophyte upon which it developed.

This is not due to maternal inheritance but to Mendelian segregation delayed one generation. That occurrence is overtly very rare, though, suspiciously enough, it has been detected in extremely dissimilar forms: thus in addition to heterostyle plants it has been found in the control of Gastropod torsion (Diver, 1925). As a result of studying the description of crosses between widely dissimilar species in Echinodermata, I long ago reached the conclusion that a maternal effect of this type is widespread and fundamental in organisms 47 (Ford, 1931, pp. 65-67), controlling early cleavage until the gene-complex of the embryo could determine its own development; for this does not seem feasible for it until the nucleo-cytoplasmic ratio returns to normal, which happens as the number of blastomeres increases. This general proposition seems never to have been tested by further observation or experiment, partly owing to the great difficulty of obtaining hybrids between species so different that they are recognizably distinct as early blastulae.

The Ss locus responsible for the heterostyle polymorphism is in fact a super-gene consisting of two main units which control respectively the position and characteristics of the male and female parts of the flower together with their incompatibility reactions. Dowrick (1956) has indeed shown that each component of the locus itself consists of several genes responsible for different features of the heterostyle distinction: an analysis carried out upon other species (Primula viscosa and P. hortensis).

As I have pointed out elsewhere (Ford, 1964, p. 176), the two major parts of the heterostyle super-gene may conveniently be regarded as G for the dominant Thrum gynaecium (the short style and associated features), allelic with g for the recessive pin form. Also A for the dominant Thrum androecium (typified by the anthers placed at the top of the corolla tube) and a for the recessive pin condition. Thus the super-gene producing the Thrum phase may be represented Ss=(GA)/(ga). On rare occasions, crossing-over occurs within it, producing (Ga)/(gA). This gives rise to gametes for short homostyles (Ga)=s^s and long homostyle (gA)=s^l (often denoted as s' in the literature).

Ernst (1933) and Dowrick (1956) have shown that both G and A represent a cluster of seven closely linked genes, crossing-over within which is normally responsible for the homostyle condition. A further and revealing situation is, however, found in Primula sinensis (Mather, 1950). In that species, additional genes, influencing the colour-pattern and producing double flowers, can give rise to homostyles. 48 These are unlinked with the S locus, which co-exists with them. Thus a, responsible for the recessive Primrose Queen eye, shortens the style, consequently aa pin is a short homostyle. On the other hand, another recessive, m, giving fertile double flowers, raises the position of the anthers and, as a result, mm pin is a long homostyle. It is of much interest that the combination ss, aa, mm corresponds structurally with a normal Thrum (S-A-M-), though produced by genetically distinct means. As Mather (l.c.) points out, the incompatability reaction of S. sinensis has been so weakened by horticultural selection that it is difficult to study. However, a certainly influences it, and in the Thrum direction, but no such effect has been detected in studying m. The fact that other genes, scattered elsewhere on the chromosomes, can control stigma-length and anther-height in P. sinensis indicates the type of raw material from which the normal S locus has been compounded.

In the majority of the Primulaceae, the s^s and s^l super-genes behave as additional alleles at the s, S locus. Long homostyles have the dominant Thrum androecium and the recessive pin gynaecium, so that they behave as recessives to Thrums and dominants to pins. Both short and long homostyles occur as rarities in the wild Primrose population, but only the latter phase has established itself anywhere at high frequencies. Its structure and physiology favour in-breeding. For, though their relative positions vary slightly, the stigma is generally clasped by the anthers which dehisce on to it. Moreover, unlike the heterostyle flowers, the selfing of homostyles is compatible: it is comparable with a cross between pin and Thrum, but one in which the respective male and female parts of which exist independently within the same flower.

The rare cross-overs allow plants to pass from the heterostyle condition with predominant out-breeding to the homostyle one with predominant in-breeding, and the reverse. Here, then, is a mechanism that alters the breeding-system, from one promoting diversity to one promoting uniformity.

Now out-breeding has normally, and from the point of view of long-term evolution always, a great advantage over in-breeding since it provides the genetic variability upon which selection can operate. 49 There is, however, a situation in which selfing has at least a short-term superiority: when the organism is ecologically well adjusted. For the effects of segregation take place at random relative to individual needs, so that variation is less likely to be useful as a population becomes more closely adapted to its environment.

Thus we find all the British Primulaceae to be heterostyled plants (except P. scotica). This is true of the Primrose itself save in two small areas where it is principally homostyled. These are about 80 miles apart and of rather similar size: one approximately 12 miles in diameter on the Chilterns in Buckinghamshire; the other, a little larger (perhaps 16 by 14 miles), near Sparkford in Somerset.

It is not yet clear what advantage the homostyles have in these localities. It can only be said that almost the whole of the English countryside has been completely transformed from its primeval state, while great changes in farming practice, owing especially to the enclosures, and in forestry have taken place since the early 18th century. It is not altogether surprising therefore that in the new types of habitat thus created, there should be a few places where in-breeding proves to be an asset to the Primrose.

If the homostyled plants were entirely self-pollinating and were as viable as pins and Thrums, they would have a clear superiority, though a short-term one, over the heterostyled type. This would reside in their self-fertility compared with the hazards and delay of cross-breeding. With homostyles thus favoured, the whole Primrose population of Britain would be converted to the in-breeding form, which would sweep through the entire country once it had become established in some area. Crosby (1949) holds that we are witnessing the early stages of that event. There appear, however, to be a number of reasons for rejecting such a view. In the first place, it assumes that the necessary cross-over within the S locus must be of extreme rarity, otherwise all Primroses would have become homostyled long ago. How then can that process be starting today in two distinct localities at the same time? Moreover, it would be fatal to this view if true long homostyles occur as rarities here and there in the normal Primrose population. 50 But this is precisely what they do (Ford, 1964, p. 183). Furthermore, over 120 species of Primula are heterostyled. Now that Crowe (1964) has provided the survey of heterostyly that has long been needed, she finds that the condition is botanically extremely widespread and must have arisen independently a number of times, occurring as it does in 18 Orders of Angiosperms. If indeed it carries an automatic disadvantage compared with homostyly, we may marvel that it has survived and, moreover, so extensively.

In addition, the situation found in the two colonies of homostyle Primroses is opposed to the view that this condition is spreading from them to replace heterostyly throughout the species in Britain. That at Sparkford, where the plants are far the commoner, has been much the more fully analysed. Here the frequencies of homostyles do not give the impression that they are advancing from a single origin; for the higher values are scattered here and there throughout it; though Crosby (1960), using an electronic computer, concluded that this is due to random drift. Nor do homostyles comprise 100 per cent of the population, but reach a limit at about 80 at which the remaining 20 per cent are pins, Thrums being excluded owing to their disadvantages where the opportunities for out-breeding are reduced (p. 46). Moreover, Bodmer (1960) has shown that, far from increasing, the homostyles have become significantly less common since 1941 in two of the localities within the Sparkford area.

The supposed overriding advantage of that form was calculated on the assumption that it is always self-fertilizing, a proposition disproved by Bodmer (1958, 1960). As in the heterostyles, the flowers open and fertilization can be effective before the anthers dehisce; while the stigma is at this stage slightly above them and therefore easily accessible. They only enclose it a little later owing to a slight relative shortening of the style.

Some light may be thrown upon the spread of homostyle primroses in the two areas where they occur by growing them in a number of experimental plots differing in soil and ecological conditions. The use of field transplants is, however, likely to prove the more rewarding technique. 51 Important information may well be obtained when homostyles are introduced at different densities into heterostyle colonies of several distinct types and their success is studied in varying degrees of competition.

It is evident that the heterostyle-homostyle system provides a mechanism for passing from relative out-breeding with variability to relative in-breeding with uniformity, or the reverse. The super-gene control of pins and Thrums may in its out-breeding aspect be compared with that of sex in animals (the best known of all polymorphisms) by means of the non-pairing of the X-chromosome, which also constitutes a super-gene. Here, however, there is no easy transition to the in-breeding situation.

It is to be noticed that the corresponding state in plants, that of dioecy, is relatively uncommon. Only 5 per cent of Angiosperm genera are wholly dioecious, though the condition occurs sporadically in about 75 per cent of the families (Crowe, 1964). We may speculate on this striking difference in the two Kingdoms.

Owing to the exposed nature of the sexual organs in plants, perhaps an outcome of their alternation of generations, many other out-breeding mechanisms are possible to them. Moreover, that in which male and female flowers are borne by different individuals is not primitive in the Angiosperms (Crowe, l.c.). In animals, on the other hand, where the nature and structure of the sexual organs generally makes self-fertilization impossible, a reduction in recombination-frequency must be achieved principally by means of super-genes, a method used extensively in plants also. Moreover, the individual mobility of animals reduces the advantage, compared with plants, of highly in-bred populations closely adapted to a particular environment. As A. J. Cain points out to me, the two animal groups which largely abandoned normal bisexual reproduction are also the least mobile of the terrestrial forms: that is to say, the Pulmonate Mollusca, which are hermaphrodite, and the Oligochaeta, which are largely parthenogenetic. 52 The latter group is the more immobile of the two and resembles plants the more closely owing to the frequency occurrence in it of polyploidy, otherwise rare in animals, which is an obvious consequence of reproduction without the necessity of cross-breeding. The hermaphrodite condition of the Pulmonata may be regarded as an intermediate situation between the bisexual and self-fertilizing states and indeed corresponds with that found in monoecious plants with self-sterility. Murray (1964) studied it in Cepaea nemoralis and showed that it provides a reservoir of variability during temporary extreme reduction in numbers.

Some further aspects of out-breeding in plants may be mentioned here. Many genera contain tristylic species: Lythrum, Narcissus and Oxalis provide examples of them. These produce flowers of three kinds each upon separate plants. The sexual organs are placed at three positions in the corolla-tube: basal, intermediate and distal. Any two of them are occupied by male parts and one by female. Consequently the style is either of the long, mid or short type. A fertile union is possible only when the pollen is derived from anthers growing at the same level as the stigma and, since these must be borne upon different plants, the situation promotes out-breeding.

Trystylic polymorphism has been studied most fully in Lythrum salicaria (Fisher and Mather, 1943). In that species it is controlled by two pairs of alleles which assort independently. One of them determines whether or not the style shall be short. The other has no effect upon short styles but decides whether non-shorts shall be long or mid. Long styles represent the double recessive and short ones the double dominant. Consequently, mid styles are dominant to long but recessive to short. Fisher and Mather find that the three phases vary in frequency from one wild population to another.

Both in di- and trystylic forms we are faced with heteromorphic incompatibility in which structural differences that favour out-breeding are combined with a physiological one, the illegitimacy mechanism. Clearly the latter can promote out-breeding without any corresponding morphological changes; giving rise, that is to say, to the cryptic polymorphism of homomorphic incompatibility. 53 This may be of several kinds (Bateman, 1955; Crowe, 1964). The mating-type of the pollen may be determined gametophytically and controlled by alleles at one locus (e.g., Oenothera) or at two (e.g., Festuca). Alternatively, the behaviour of the pollen may be imposed sporophytically (as in the heteromorphic condition) and controlled by one locus (e.g., Capsella, Brassica).

As Crowe (l.c.) points out, incompatibility systems with the mating-type of the pollen determined gametophytically may be transformed into those in which the determination is sporophytic owing to the familiar principle that genes can affect the time of onset and rate of development of processes in the body: a concept first elaborated by Ford and Huxley (1927).

Data on the evolution of heterozygous advantage seem lacking in these polymorphisms. In the Primrose the selfing of Thrums is almost completely excluded in the field, and it is so difficult to achieve in the laboratory that critical information on the viability of the SS and Ss genotypes does not appear to be available. As with sex, the frequency of the Thrum and pin types in heterostyle communities will be determined by the tendency to maintain them in optimum proportions, and this would override any advantage which the heterozygote may have compared with the dominant homozygote were the latter produced which, apparently, it is not. The slight deficiency of Thrums in normal Primrose populations is itself explicable (pp. 45, 46) and, as already indicated, the preponderance of pins becomes overwhelming where homostyle frequencies are high.

It has been pointed out that heterozygous advantage tends normally to evolve when a major gene (or super-gene) spreads through a population and polymorphism supervenes (pp. 26-8). This must be true even when the condition is maintained independently, due to an 'ecological' need for diversity, a good example of that situation is provided by polymorphic mimicry. There is evidence for differential viability of the genotypes in two forms, both mimics, of Hypolimnas misippus (Ford, 1953b); but the subject may also be examined in Papilio dardanus, which has been studied more thoroughly from the aspect of ecological genetics than has any other mimetic butterfly.

54 The existence of heterozygous advantage in that species is established by the occurrence of mimetic polymorphism in the Polytrophus race, which inhabits the mountains of central Tanganyika and Kenya. Here the models are very rare. Indeed we have an estimate of their frequency based upon large random collections in which every butterfly, of all species, seen was caught both at Nairobi, within the range of Polytrophus, and at Entebbe (the race Meseres) where normal conditions obtain and models are seventy-three times the commoner relative to their mimics (Ford, 1953b) as shown in the following table:

That the models are too rare to maintain the mimicry of Polytrophus is evident since in that race the P. dardanus females, so constant elsewhere, become highly variable and their mimicry breaks down, although 'unifactorially' controlled: itself a demonstration of the way in which each resemblance has been attained by selective modification of the effects of a super-gene (pp. 35-6). Here therefore the polymorphism is preserved solely by heterozygous advantage, which has evolved under the cloak of mimicry. Such heterosis is of course responsible for the occurrence of non-mimetic phases together with the mimetic ones in certain instances (as in the Antinorii race of P. dardanus, in the oriental Papilio polytes, and in many other species). When models become ineffective, owing to their rarity, the various phases of their mimics may thus be maintained as in ordinary non-mimetic polymorphism. That condition has often been analysed; for example, in Colias by Remington (1954) who pointed out (p. 421) that it depends upon heterozygous advantage.

As already mentioned (p. 34), polymorphic mimicry generally involves adaptations in a number of distinct features. Consequently the switch-mechanism responsible for it is likely to take the form of a super-gene. 55 Precise proof for that conclusion is difficult to obtain, for it requires the detection of a cross-over between the loci concerned, and this may well be reduced to a frequency approaching, though certainly not equalling, the rarity of the intra-cistrionic type. However, strong circumstantial evidence for this type of control is available in several of the instances which have been examined genetically.

In the first place, no less than eleven of the polymorphic phases of Papilio dardanus behave as if determined by a single series of multiple alleles. This is to be anticipated if they are, in reality, substitutions within a super-gene (pp. 17-25) but, considering the diversity of the characters involved, it is otherwise a situation of extreme improbability.

Moreover, though the occurrence of a cross-over within the controlling unit has not been detected in the P. dardanus breeding experiments, Clarke and Sheppard (1960b) point out that two of the phases are most easily interpreted as the result of such cross-overs. These are niobe, and the Polytrophus forms in which white is replaced by fluorescent yellow pigment of a type present in the yellow male-like females that exist at a low frequency in that race.

As mentioned on p. 17, a 'single gene' converts one of the tail-less non-mimetic forms of the oriental Papilio memnon into a tailed mimic. The alteration in colour-pattern is considerable and the probability that this should be combined fortuitously with a unifactorial change in wing-shape, appropriate to copying the model, seems extremely small. It is far more likely that a gene responsible for tail-formation and one affecting the colours and markings have been brought together to form a compound unit, while its evolution by duplication seems excluded when such diverse effects are involved; and this must be true of the control of co-adapted genes in general (pp. 17-18).

Whatever its colour, the shell may bear bands, usually of a blackish shade. They may be absent or there may be any number of them on each whorl, up to 5 while, for the sake of clarity, absences are always indicated by the use of 0. Thus, 00000 and 12345 represent the bandless and the fully banded types, while the expression 00345 shows that the two upper bands are missing. These are the only ones visible when the shell is seen from above. Consequently the 00000 shells and those lacking bands 1 and 2 can be combined as 'effectively unbanded'.

The bandless condition (00000) is dominant to all kinds of banding, the varieties of which are controlled by modifiers. Thus the 00300 and 00345 bands are each simple dominants, but it will be understood that the genes giving rise to them work only on the banded genotype: that is to say, they may be present, though inoperative, in bandless snails.

The alleles controlling shell-colour, and those responsible for the presence or absence of banding, are included together within a super-gene. Indeed the linkage between them at first appeared to be absolute, but Cain, King and Sheppard (1960) have now detected a few instances in which it has broken down; giving in two families a cross-over value of about 2.25 per cent. It should here be remarked that such crossing-over, differing in frequency from one family to another, had previously been reported by Fisher and Diver (1934). 57 However, as pointed out by Lamotte (1954), their results were invalid since they used specimens which could have been fertilized previously; for these snails store living sperm for long periods (three years or more) after copulation. The breeding work of Cain et al. is, however, free from this objection.

It is interesting to notice that the genes already mentioned which modify potential 5 banding to the 00300 and 00345 types are not included in this super-gene. Nor, as pointed out by Cain, King and Sheppard (l.c.) in an able discussion of the subject, is this to be expected. Since these modifiers can operate only on the banded form, which shows very close linkage with colour, they behave much as if linked with colour too. Indeed any tendency to bring them into the same super-gene could be effective only if promoted by extremely powerful selection.

The genetic control of colour and banding in Cepaea nemoralis thus presents a clear picture. These characters are polymorphic in nature and the mechanism which keeps them so must briefly be outlined. We may first of all consider the normal situation in rural England. To the human eye the yellow shells, (greenish when occupied by the snail) are the less conspicuous in a green habitat, such as ordinary grassland, while the pinkish-browns are so on a litter of dead leaves, as among deciduous trees; the browns are at a marked advantage upon the brown and the blackish floor of a beech wood. Similarly, the banded shells are the less easily seen upon a diversified background such as that provided by a mixed hedgerow, as are the unbanded upon a relatively uniform one such as stretches of beech-leaf litter.

The proportion of the various colour and banding phases differ greatly from one locality to another, the commonest phenotypes in each population (except those exhibiting 'area effects', pp. 60-1) being those that appear to human judgement to be the least conspicuous on the prevailing background (see Cain and Sheppard, 1954, Fig. 1). Moreover, we have evidence that they are so to some at least of the snail's predators. Thus the Song Thrush, Turdus philomelos, one of the most important of these, fortunately picks up all but the smallest specimens, which no doubt it swallows, and carries them to convenient stones on which to break them. 58 There the fragments accumulate, so that it becomes possible to determine if the birds destroy a random sample of the population. This they do not do. On the contrary, they kill an undue proportion of the more conspicuous forms whichever these may be.

Thus at Marley Bog, Wytham Woods, Berkshire, a habitat presenting a somewhat uniform background, Cain and Sheppard (1954) found that 264 out of 560 living Cepaea nemoralis examined were of the effectively banded type, which amounted to 486 out of 863 destroyed by thrushes. The significance of the excess of banded forms among those killed by the birds is measured by P<0.02.

A more striking situation was demonstrated by Sheppard (1951) when studying shell-colour in two localities also in Wytham Woods. Here the thrushes eliminated first an excess of snails with yellow shells and later of the dark types as the spring advanced and the background changed from brown to green (see the following table, in which the collections from the two localities have been combined as they prove to be homogeneous). The result was not due to reducing the proportion of yellow during the earlier period; for the yellow shells amounted to 24.2 per cent (80 out of 330) in a sample of living snails examined on the 14th April and to 27.9 per cent (57 out of 204) on the 26th May.

Cain and Currey (1963) point out that, in general, deciduous woods are greenest at ground level in summer but that the snail samples seem to be the best adjusted to the spring background.59 This, and other evidence provided by Cain and Currey, suggests that the common thrush, and perhaps other birds, prey upon Cepaea nemoralis chiefly in the spring when they need an especially large supply of food during the breeding season. Since within that period, owing to the change of background-colour in mixed woodland, visual selection is reversed, its total effect there throughout the year is probably very small and could be approximately zero in such places, but not of course in those that remain more uniform throughout the season.

If p and q are the frequencies of two alleles responsible for three genotypes whose selective advantages are a, b (the heterozygote), and c, then at equilibrium

In each habitat, then, certain phenotypes are eliminated year by year, those less well adapted to the local conditions, yet the population does not become uniform. That is to say, the polymorphism has evolved heterozygous advantage, so that it is maintained on a physiological basis but adjusted to particular frequencies by selective predation.

To this, there is, in theory, one alternative possibility; that put forward by de Ruiter who, as already mentioned (p. 15), showed that some birds that prey upon protectively coloured species tend to hunt for specimens resembling one they have recently found. This, on the average, puts the commonest phase at a disadvantage whichever that may be, so maintaining polymorphism, even in the absence of heterosis. It is, however, clear that the effect to which de Ruiter draws attention does not apply to the predation of Cepaea nemoralis, or does so to a negligible extent. 60 One of the reasons for adopting that view has already been mentioned. For Sheppard (1951) found that thrushes removed an excess first of yellow shells and later of dark ones in a deciduous wood as the spring advanced, while the data show that the birds were not concentrating upon the commonest form.

There is as yet no direct evidence that the polymorphism of Cepaea nemoralis, though generally adjusted by selection, is maintained by heterozygous advantage, yet that conclusion seems inescapable. It is in accord with the situation in other organisms, and it derives strong support from the work of Sedlmair (1956) and Lamotte (1959), who find differences in survival between the phenotypes when exposed to adverse conditions.

Cain and Currey (1963) have detected and studied what they describe as area effects in Cepaea nemoralis. That is to say, the occurrence of certain phenotypes at a high frequency over large and diversified regions of a special type, to the different components of which their appearance is not adjusted though they are subject to visual selection. Here, then, the physiological aspect of the polymorphism overrides that determining colour-pattern. It does so on several high chalk downs in southern England. In these localities there are not only unusual areas where the yellow mid-banded forms predominate, but others in which 5 banded or dark (pink and brown) shells are abundant, though ill adjusted to their background and subject to predation by thrushes. Thus it seems indeed that in an environment of this rather extreme type, selection for physiological characters proves particularly important. Moreover, at these sites on the English chalk where 'area effects' are evident, the snails often appear in such abundance that birds probably destroy them less selectively than elsewhere.

A. J. Cain tells me that in such places it is still uncertain to what environmental features particular phases are adjusted. Similar 'area effects' to those occurring today are evident in holocene material from these localities, so the condition has apparently been maintained in spite of considerable climatic changes. It is indeed possible, as Cain suggests to me, that such factors as trace-elements in the soil may here be decisive. 61 Evidently genes determining physiological qualities are incorporated into the same super-gene with those controlling colour and pattern, which could otherwise be selected independently of them. It seems that the selective effect of visual features, normally powerful but never overriding the polymorphism, is greater in luxuriant conditions of woodland and meadow and relatively less in certain more extreme downland localities.

It has indeed been possible to assess the effect of heterozygous advantage with precision in numerous instances. Two further examples of the kind may be considered here; the second (pp. 64-6), which introduces the subject of chromosome polymorphism, being far the more detailed and extensive of them.

A marine Copepod, Tisbe reticulata, has colonized the brackish water of the lagoon of Venice. It occurs there in four polymorphic forms differing in their colour-pattern. These have been studied by Battaglia (1958) who has shown that they are controlled by three alleles, V^V, V^M and v. The first two give rise to the phases violacea and maculata respectively, which are fully dominant to trifasciata (vv). Individuals of the genotype V^VV^M are, however, phenotypically distinct, for they show the characteristics of violacea and maculata combined.

62 Tisbe reticulata has been examined also at Roscoff, Brittany, by Bocquet (1951), where a larger number of forms is found. Not only do their frequencies differ considerably in the two localities but there are other adaptive differences between them. Thus violacea comprises 5 per cent of the population at Roscoff but 30 to 37 per cent at Venice, where also the species is adapted to live at much lower salinities than usual. Indeed crosses between these French and Italian communities produce offspring in which the sex-ratio is disturbed, indicating that considerable genetic differences have accumulated between them. Battaglia has studied the relative survival of the Venetian genotypes in overcrowded, intermediate and favourable conditions; the latter being obtained when the animals are kept at a low density. He principally worked with the F2 generation from crosses between the violacea and maculata forms, in which all three genotypes can be scored visibly. His results are summarized in the following table. It will be seen that while the heterozygotes always survive better than the homozygotes the difference becomes more marked as the conditions deteriorate, so demonstrating its selective importance. Battaglia also obtained evidence for a similar superiority of the heterozygotes in respect of V^Vv and V^Mv, in which that genotype is not visibly distinct.

63 Genetic polymorphism is most frequently maintained by heterozygous advantage, as illustrated in the foregoing instances. Under the powerful selection-pressures now known to be operating in nature, that condition can evolve rapidly.

For a suitable example of this important fact we may return to the phenomenon of industrial melanism in the moth Biston betularia (pp. 38-9). It has already been mentioned that the black form carbonaria (C-), which is unifactorial and dominant to the typical pale one (cc), originally appeared in Manchester in 1848. There, and in many other industrial areas, it first increased rapidly as a transient polymorphism (p. 14). However, it later acquired heterozygous advantage, for it has already occupied about 96 to 98 per cent of the population by 1895. At that value, somewhat lower in other places, it stabilized and it is no commoner today. From an analysis of Kettlewell's data, including extensive work on the present frequencies together with past records which he has brought together, Haldane (1956) finds that the balanced melanism of B. betularia is consistent with a population-structure in which the fitness of cc is about half, and of CC about 90 per cent, that of the heterozygotes.

Here then we have evidence of heterozygous advantage, and Kettlewell (1957) has shown that this has evolved during the spread of carbonaria. In the course of much experimental breeding, he obtained back-cross broods from industrial areas where carbonaria is somewhat less common than in Manchester. Some of these were reared between 1953 and 1956 on polluted and unwashed food and they produced 65 typicals and 108 carbonaria when, of course, equality is expected. He was also able to obtain information on families, also back-crosses, bred from London specimens between 1900 and 1906 at a time when carbonaria was still spreading there. They produced 255 typicals and 217 carbonaria. The difference between the two sets of results is heavily significant (X²₍₁₎ = 13.05, P < 0.001). Clearly the advantage of the heterozygous phase, so marked at the present day, had not evolved in the metropolitan district at the beginning of this century.

Next

Contents

References