It may, however, be argued that it is not when at rest but during flight that the mimetic resemblance protects the mimic from attack. Actually this can hardly be true, for the mode of flight constitutes one of the most striking differences between model and mimic.P. hectorandP. aristolochiaefly much in the same way. They give one the impression of flying mainly with their fore wings, which vibrate rapidly, so that the course of the insect, though not swift, is on the whole sustained and even. The flight of all the different forms ofpolytesis similar and quite distinct from that of the models. It is a strong but ratherheavy and lumbering up-and-down flight. One gets the impression that all the wing surface is being used instead of principally the fore wings as appears inP. hectorandP. aristolochiae. The difference is difficult to put into words, but owing to these peculiarities of flight the eye has no difficulty in distinguishing between model and mimic even at a distance of 40 to 50 yards. Moreover, colour need not enter into the matter at all. It is even easier to distinguish model from mimic when flying against a bright background, as for instance when the insect is between the observer and a sunlit sky, than it is to do so by reflected light. I have myself spent many days in doing little else but chasingpolytesat Trincomalee where it was flying in company withP. hector, but I was never once lured into chasing the model in mistake for the mimic. My experience was that whether at rest or flying the species are perfectly distinct, and I find it difficult to imagine that a bird whose living depended in part upon its ability to discriminate between the different forms would be likely to be misled. Certainly it would not be if its powers of discrimination were equal to those of an ordinary civilised man. If the bird were unable to distinguish between say theAform of female andP. aristolochiaeI think that it would be still less likely to distinguish between the sameAform and the male or theMform of female. For my experience was that at a little distance one could easily confuse theAform ofpolyteswith the male. Except when one was quite close the red on theAform was apt to be lost, thewhite markings on the hind wing were readily confused with those of the male, and one had to depend entirely on the lighter fore wing. Unless the bird were keener sighted than the man theAform would be more likely to be taken in mistake for its unprotected relative than avoided for its resemblance to the presumably unpalatable model. On the other hand, if the bird were sufficiently keen sighted never to confuse theAfemale with the male form its sight would be too keen to be imposed upon by such resemblance as exists between theAfemale andP. aristolochiae.
These, however, are not the only criticisms of the theory of mimicry which the study of this species forces upon us.Papilio polytesis one of the few mimetic species that has been bred, and in no other case of polymorphism is the relation between the different forms so clearly understood. For this result we are indebted mainly to the careful experiments of Mr J. C. F. Fryer, who recently devoted the best part of two years to breeding the different forms of this butterfly in Ceylon[45]. Fryer came to the conclusion that an explanation of this curious case is possible on ordinary Mendelian lines. At first sight the breeding results appear complicated, for any one of the three forms of female can behave in several different ways. For the sake of simplicity we may for the moment class together theAandHfemales as the mimetic females, the non-mimetic being represented by theMor male-like females.The different kinds of families which each of the three females can produce may be tabulated as follows:—
(α) TheMform may give either:—(1)Monly.(2)Mand mimetics in about equal numbers.(3) Mimetics only.(β) TheAform may give either:—(1)Mand mimetics in about equal numbers.(2)Mand mimetics in the ratio of about 1:3.(3) Mimetics only.(γ) TheHform may give either:—(1)Mand mimetics in about equal numbers.(2)Mand mimetics in the ratio of about 1:3.(3) Mimetics only.
(α) TheMform may give either:—(1)Monly.(2)Mand mimetics in about equal numbers.(3) Mimetics only.
(α) TheMform may give either:—
(1)Monly.
(2)Mand mimetics in about equal numbers.
(3) Mimetics only.
(β) TheAform may give either:—(1)Mand mimetics in about equal numbers.(2)Mand mimetics in the ratio of about 1:3.(3) Mimetics only.
(β) TheAform may give either:—
(1)Mand mimetics in about equal numbers.
(2)Mand mimetics in the ratio of about 1:3.
(3) Mimetics only.
(γ) TheHform may give either:—(1)Mand mimetics in about equal numbers.(2)Mand mimetics in the ratio of about 1:3.(3) Mimetics only.
(γ) TheHform may give either:—
(1)Mand mimetics in about equal numbers.
(2)Mand mimetics in the ratio of about 1:3.
(3) Mimetics only.
The males are in all cases alike to look at but it must nevertheless be supposed that they differ in their transmitting powers. In fact the evidence all points to there being three different kinds of male corresponding to the three different kinds of female. But they cannot shew any difference outwardly because there is always present in the male a factor which inhibits the production of the mimetic pattern even though the factor for that pattern be present.
Returning now to the records of the females it will be noticed that although theMform may breed true the mimetics never give theMform alone. Where they give theMform among their progeny they produce mimetics and non-mimetics either in the ratio 1:1 or of 3:1. This at once suggests that thenon-mimetic is recessive to the mimetic forms—that the mimetics contain a factor which does not occur in the non-mimetics. If this factor, which may be calledX, be added to the constitution of a non-mimetic female it turns it into a mimetic. IfXbe added to a male such an individual, though incapable of itself exhibiting the mimetic pattern owing to the inhibitory factor always present in that sex, becomes capable of transmitting the mimetic factor to its offspring. Expressed in the usual Mendelian way the formulae for these different butterflies are as follows:—
whereXstands for the mimetic factor andIfor the factor which inhibits the action ofX. All males are heterozygous forI, but during the segregation of characters at some stage in the formation of the families only the male-producing sperms come to contain the factorI. It is lacking in all the female-producing sperms formed by the male.
♂ (1) does not contain the factor for the mimetic condition and gives only daughters of theMform when mated with anM♀. ♂ (2) on the other hand is homozygous for the factorX, and consequently all of his germ cells contain it. This is the male that gives nothing but mimetic daughters with whatever form of female he is bred. ♂ (3) is heterozygous forX; that is to say, one half of his germ cells contain it, the other half not. With theM♀ he must give equal numbersof offspring with and withoutX,i.e.half of his daughters will be mimetic and the other half non-mimetic. With a heterozygous mimetic female (iiXx), which is also producing germ cells with and withoutXin equal numbers, he may be expected to give the usual result, viz. dominants and recessives in the ratio 3:1; or in other words mimetic and non-mimetic females in the ratio 3:1.
One of Fryer's experiments may be given here in illustration of the nature of the evidence upon which the above hypothesis depends.
Inheritance in Papilio polytes
Families were reared from the two wildHfemales of whom nothing was known either as to ancestry or husband. The first family contained 10Mand 7Hfemales. Hence the original wild mother was probablyiiXxand had mated with a male of the constitutionIixx. The family from the second wildHfemale contained 26Hand 7Mfemales;i.e.the ratio in which these two forms appeared was not far from 3:1. Hence the wild female was probablyiiXxand her husbandIiXx. If this were so some of the 26 ♂♂ should receive theXfactor from both parents and consequently beIiXXin constitution. This was almost certainly so in the case of the single male in this brood tested by mating with anMfemale from the other brood. All of his 12 daughters were of theHform, as should have been the case had his constitution beenIiXX. Supposing this to be so, all his offspring, of both sexes, must be heterozygous forX. Consequently any pair mated together should give bothHandMfemales in the ratio of three of the former to one of the latter. In Mr Fryer's experiment two males and two females chosen at random were mated together. In the one case sixHand oneMfemale were produced, in the other tenHand twoMfemales. As was expected both classes of female appeared, and the looked-for ratio of threeHto oneMwas, in view of the smallness of the numbers, not departed from widely in either instance.
In the experiments selected as an illustration, the mimetic females happen to be all of theHform. In other experiments, however, both theHform and theAform occurred. As the result of his experiments Mr Fryer came to the conclusion that here again the difference is one of a single hereditary factor. All mimetic females contain theXfactor, but theHfemales contain in addition a factor which we may callY. The function of theYfactor is to carry the change made by theXfactor a step further, and to turn theAform of female into theHform.Yis a modifier ofX, but unlessXis presentYcan produce no effect. All the different individuals which are to be found amongP. polytesin Ceylon may be represented as follows:—
In this way is offered a simple explanation in terms of three Mendelian factors which serves at once to explain the various results of the breeding experiments, and the fact that intermediates between the different forms of female are not found.
The only other experiments comparable with these onP. polytesare some made by Jacobsen onPapilio memnonin Java[46]. Here again there are three forms of female, one of which,laomedon, is something like the male, while the other two,agenorandachates, are quite distinct. Of these threeachates, unlike the male and the other two females, is tailed, and resemblesthe speciesPapilio coonwhich belongs to the same presumably distasteful group asP. aristolochiae. These experiments of Jacobsen's are not so complete as the series onP. polytes, but Professor de Meijere and Mr Fryer have both pointed out that they are capable of being interpreted on the same simple lines.
Another instance of experimental breeding involving polymorphism and mimicry in the female sex is that of the AfricanPapilio dardanus, but the case is here complicated by the greater number of female forms (cf. pp.30-33). The data, too, are far more scanty than in the other two cases, but so far as they go there is nothing to preclude an explanation being eventually arrived at on similar lines[47].
And now we may consider briefly the bearing of these experiments on the theory of mimicry. Throughout the work no individuals intermediate between the three well-marked forms ofpolyteswere met with. There is no difference in appearance between the heterozygous and the homozygous mimetic insects, whether they belong to theAor to theHform. The factorX, whether inherited from both parents, or from one only, produces its full effect, and the same is also true of the action of the factorY. Now the most generally accepted hypothesis as to the formation of these mimetic resemblances supposes that they have been brought about through the gradual operation of natural selection accumulating slight variations.Professor Poulton, for example, a prominent exponent of this school, considers that theAform of female was first evolved gradually from theMform, and later on theHform came by degrees from theAform. If this be true we ought, by mingling theMgerm plasm with theHgerm plasm and by subsequently breeding from the insects produced, to get back our series of hypothetical intermediates, or at any rate some of them. We ought as it were to reverse the process by which the evolution of the different forms has taken place. But as is shewn by the experiment of Mr Fryer, which was quoted above, nothing of the sort happens.
From experiments with cultivated plants such as primulas and sweet peas, we have learnt that this discontinuous form of inheritance which occurs inP. polytesis the regular thing. Moreover, we have plenty of historical evidence that the new character which behaves in this way is one that has arisen suddenly without the formation of intermediate steps. The dwarf "Cupid" form of sweet pea, for instance, behaves in heredity towards the normal form as though the difference between them were a difference of a single factor. It is quite certain that the "Cupid" arose as a sudden sport from the normal without the intervention of anything in the way of intermediates. And there is every reason to suppose that the same is true for plenty of other characters involving colour and pattern as well as structure, both in the sweet pea, the primula, and other species. Since the forms ofpolytesfemale behave in breeding like the variousforms of sweet pea and primula there is every reason to suppose that theyarosein the same way, that is to say, as sudden sports or mutations and not by the gradual accumulation of slight differences.
But if we take this view, which is certainly most consonant with the evidence before us, we must assign to natural selection a different rôle from that which is generally ascribed to it. We cannot suppose that natural selection has played any part in theformationof a mimetic likeness. The likeness turned up suddenly as a sport quite independently of natural selection. But although natural selection may have had nothing to do with its production, it may nevertheless have come into play in connection with theconservationof the new form. If the new form possesses some advantage over the pre-existing one from which it sprang, is it not conceivable that natural selection will come into operation to render it the predominant form? To this question we shall try to find an answer in the next chapter.
It was suggested in the last chapter that if a new variation arose as a sport—as a sudden hereditary variation—and if that variation were, through resemblance to a different and unpalatable species, to be more immune to the attacks of enemies than the normal form, it was conceivable that the newer mimetic sport would become established, and in time perhaps come to be the only form of the species. We may suppose, for example, that theAfemale ofP. polytesarose suddenly, and that owing to its likeness to the presumably distastefulP. aristolochiaeit became rapidly more numerous until in some localities it is the commonest or even the only form. However, before discussing the establishing of a mimetic form in this manner we must first deal with certain general results which may be expected to follow on a process of selection applied to members of a population presenting variations which are inherited on ordinary Mendelian lines.
Let us suppose that we are dealing with the inheritance of a character which depends upon the presence of the genetic factorX; and let us also suppose that the heterozygous form () is indistinguishablefrom the homozygous form () in appearance. In other words the character dependent uponXexhibits complete dominance. With regard toXthen all the members of our population must belong to one or other of three classes. They may be homozygous (XX) forX, having received it from both parents, or they may be heterozygous (Xx) because they have received it from only one parent, or they may be devoid ofX,i.e.pure recessives (xx). An interesting question arises as to the conditions under which a population containing these three kinds of individuals remains stable. By stability is meant that with the three kinds mating freely among themselves and being all equally fertile, there is no tendency for the relative proportions of the three classes to be disturbed from generation to generation. The question was looked into some years ago by G. H. Hardy, who shewed that if the mixed population consist ofp XXindividuals,2q Xxindividuals andr xxindividuals, the population will be in stable equilibrium with regard to the relative proportions of these three classes so long as the equationpr = q2is satisfied[48].
Now let us suppose that in place of equality of conditions selection is exercised in favour of those individuals which exhibit the dominant character. It has been shewn by Mr Norton that even if the selection exercised were slight the result in the end would be that the recessive form would entirely disappear. The total time required for bringing this about woulddepend upon two things, (1) the proportion of dominants existing in the population before the process of selection began, and (2) the intensity of the selection process itself. Suppose, for example, that we started with a population consisting of pure dominants, heterozygotes, and recessives in the ratio 1:4:4. Since these figures satisfy the equationpr = q2, such a population mating at random within itself is in a state of stable equilibrium. Now let us suppose that the dominant form (including of course the heterozygotes) is endowed with a selection advantage over the recessives of 10%, or in other words that the relative proportion of the recessives who survive to breed is only 90% of the proportion of dominants that survive[49]. It is clear that the proportion of dominants must gradually increase and that of the recessives diminish.
At what rate will this change in the population take place? Mr Norton has worked this out (see App. I) and has shewn that at the end of 12 generations the proportions of pure dominants, heterozygotes, and recessives will be 1:2:1. The population will have reached another position of equilibrium, but the proportion of recessives from being four-ninths of thetotal is now reduced to one-quarter. After 18 more generations the proportions 4:4:1 are reached, the recessives being only one-ninth of the total; after 40 further generations of the process they become reduced to one-fortieth. In other words a selective advantage of 10% operating against the recessives will reduce their numbers in 70 generations from nearly one-half of the population to less than one-fortieth.
With a less stringent selective rate the number of generations elapsing before this result is brought about will be larger. If, for example, the selective rate is diminished from 10% to 1% the number of generations necessary for bringing about the same change is nearly 700 instead of 70—roughly ten times as great. Even so, and one can hardly speak of a 1% selective rate as a stringent one, it is remarkable in how brief a space of time a form which is discriminated against, even lightly, is bound to disappear. Evolution, in so far as it consists of the supplanting of one form by another, may be a very much more rapid process than has hitherto been suspected, for natural selection, if appreciable, must be held to operate with extraordinary swiftness where it is given established variations with which to work.
We may now consider the bearing of these theoretical deductions upon the case ofPapilio polytesin Ceylon. Here is a case of a population living and breeding together under the same conditions, a population in which there are three classes depending upon the presence or absence of two factors,XandY,exhibiting ordinary Mendelian inheritance. For the present we may consider one of these factors,X, which involves the proportion of mimetic to non-mimetic forms. It is generally agreed among observers who have studied this species that of the three forms of female theMform is distinctly the most common, while of the other two theHform is rather more numerous than theAform. The two dominant mimetic forms taken together, however, are rather more numerous than the recessiveMform. The most recent observer who studied this question, Mr Fryer, captured 155 specimens in the wild state as larvae. When reared 66 turned out to be males, while of the females there were 49 of the two mimetic forms and 40 of theMform, the ratio of dominants to recessives being closely 5:4[50]. Now as has already been pointed out the ratio 5:4 of dominants and recessives is characteristic of a population exhibiting simple Mendelian inheritance when in a state of stable equilibrium. The natural deduction from Mr Fryer's figures is that with regard to the factor that differentiates the mimetic forms from the non-mimetic, thepolytespopulation is, for the moment at any rate, in a position of stable equilibrium. This may mean one of two things. Either the population is definitely in a state of equilibrium which has lasted for a period of time in the pastand may be expected to endure for a further period in the future, or else the population is in a condition of gradual change as regards the numerical proportion of mimetics and non-mimetics, progressing towards the elimination of the one or the other, the present state of equilibrium being merely transitory and accidental. In this connection a few scraps of historical evidence are of interest. Of the various forms ofP. polytestheAform of female was the first to be described in 1758, and not long after (1776) theHform was registered as a species under the name ofPapilio Eques Trojanus romulus. Later on the female resembling the male found its way into the literature asPapilio pammon. From the fact that the mimetic forms were known before the non-mimetic, it is unlikely that they can have been scarce a century and a half ago. AsP. polytescertainly produces at least four broods a year in Ceylon this period of time represents something like 600 generations in the life of the species, and we have already seen that even if the mimetic forms have but a 1% advantage over the non-mimetic the proportion of the latter would decrease from nearly equality down to but 1 in 40 in about 700 generations. Actually forP. polytesthe decrease would not be so marked because the male is non-mimetic. Owing to this peculiar feature the rapidity of change in the proportion of the different forms is reduced to about one-half of what it would be if the males were also mimetic. Nevertheless the change from nearly equality to about one non-mimetic in 40 would have taken placeduring the timeP. polyteshas been known if a 2% selection advantage had operated during that period in favour of the mimetic. If there has been any appreciable selection going on during that time mimetics must have been far rarer when the species was first discovered, but the fact that both the mimetic forms made their way into collections before the non-mimetic tells distinctly against this supposition. Nor is there any reason to suppose that the non-mimetic form has been dwindling in numbers relatively to the mimetics during the last half century. Moore[51]in 1880 records an earlier observation of Wade's that "These three butterflies are very common, especially those of the first form; the second being perhaps least so." The first form alluded to is theMform, and the second is theAform, so that at the time Wade wrote the relative proportions of these three forms must have been very much what they are to-day. Even during half a century and with such a relatively weak selection rate as 2% in favour of the mimetics, the proportion of non-mimetics should drop from about 4:5 down to about 1:5. Therefore we must either infer that in respect of mimetic resemblances natural selection does not exist forP. polytesin Ceylon, or else we must suppose its force to be so slight that in half a century certainly, and perhaps in a century and a half, it can produce no effect appreciable to the necessarily rough method of estimation employed.
It may, however, be argued that even an exceedingly low selection rate is able to bring about the elimination of one or other type provided that it acts for a sufficiently long time. This is perfectly true. A selective rate of .001% would reduce the proportion of recessives to dominants from 4:5 down to 1:40 in the course of about 1,400,000 generations where the mimetic resemblance is already established. Such a form of selection entails the death of but one additional non-mimetic in 100,000 in each generation. If, however, the mimetic resemblance is not fully established and the mimic bears only what supporters of the mimicry theory term a "rough" resemblance to the model, it is clear that it will have far less chance of being mistaken for the model. Its advantage as compared with the non-mimetic form will be very much less. Even supposing that the slight variations concerned are inherited, an intensity of selection which would produce a certain change in 1,400,000 generations where a mimetic resemblance is already established must be supposed to take an enormously greater time where an approach to a model has to take place from a "rough" resemblance.
From the data as to the relative proportions of the polymorphic females ofP. polytesduring the past and at present, and from the behaviour of their different forms in breeding, the following conclusions only can be drawn. Either natural selection, from the point of view of mimicry, is non-existent for this species in Ceylon, or else it is so slight as to be unable in half acentury to produce an appreciable diminution in the proportion of non-mimetic females. For even if the mimetic resemblance brings about but the survival of one additional protected form in 100 as compared with the unprotected, this means a marked diminution in the proportion ofMfemales in 50 years—a diminution such as there are no grounds for supposing to have taken place.
It has been argued that in populations exhibiting Mendelian heredity even a relatively low selection rate must bring about a rapid change in the constitution of a mixed population. Have we any grounds for supposing that populations of this sort can undergo such rapid changes? In cases where mimetic resemblances are involved we have no examples of the sort. But some interesting evidence as to the rate at which a population may change is to be gathered from the study of melanism in certain moths. It is well known that in some parts of England the common peppered moth,Amphidasys betulariahas been almost entirely supplanted by the darker melanic formdoubledayaria. It first made its appearance near Manchester in 1850, and from that centre has been gradually spreading over northern England, the Midlands, and the south-eastern counties. At Huddersfield, for instance, fifty years ago only the type formbetulariaexisted; to-day there is nothing butdoubledayaria. In Lancashire and Cheshire the type is now rare. On the continent, too, there is the same story to be told. The melanic form first appeared in RhenishPrussia in 1888; to-day it is much more abundant than the older type. There, too, it is spreading eastwards and southwards to Thuringia, to Saxony, to Silesia. What advantage this new dark form has over the older one we do not know[52]. Some advantage, however, it must have, otherwise it could hardly supplantbetulariain the way that it is doing. From our present standpoint two things are of interest in the case of the peppered moth—the rapidity with which the change in the nature of the population has taken place, and the fact that the two forms exhibit Mendelian heredity,doubledayariabeing dominant andbetulariarecessive[53]. Moreover, mixed broods have been reared from wild females of both sorts, and so far as is known the two forms breed freely together where they co-exist. This case of the peppered moth shews how swiftly a change may come over a species[54]. It is not at all improbable that the establishing of a new variety at the expense of an older one in a relatively short space of time is continually going on, especially in tropical lands wherethe conditions appear to be more favourable to exuberance of variation and where generations succeed one another in more rapid succession. At present, however, we are without data. A form reported by an old collector as common is now rare; a variety once regarded as a great prize is now easily to be found. Such to-day is the sort of information available. For the solution of our problem it is, of course, useless. The development of Mendelian studies has given us a method, rough perhaps but the best yet found, of testing for the presence, and of measuring the intensity, of natural selection. Much could be learned if some common form were chosen for investigation in which, as inP. polytes, there are both mimetic and non-mimetic forms. Large numbers should be caught at stated intervals, large enough to give trustworthy data as to the proportions of the different forms, mimetic or non-mimetic, that occurred in the population. Such a census of a polymorphic species, if done thoroughly, and done over a series of years at regular intervals, might be expected to give us the necessary data for deciding whether the relative proportion of the different forms was changing—whether there were definite grounds for supposing natural selection to be at work, and if so what was the rate at which it brought the change about.
The theory of mimicry demands that butterflies should have enemies, and further that those enemies should exercise a certain discrimination in their attacks. They must be sufficiently observant to notice the difference between the mimetic and the non-mimetic form; they must be sufficiently unobservant to confuse the mimetic form with the unpalatable model. And, of course, they must have enough sense of taste to dislike the unpalatable and to appreciate the palatable varieties. What these enemies are and whether they can be supposed to play the part required of them we may now go on to consider.
Butterflies are destroyed in the imago state principally by three groups of enemies—predaceous insects, lizards, and birds. It is known that monkeys also devour butterflies to some extent, but such damage as they inflict is almost certainly small in comparison with that brought about by the three groups already mentioned. In view of the very different nature of these groups it will be convenient to consider them separately.
I.Predaceous Insects.Butterflies are known to be preyed upon by other insects of different orders, and a considerable number of observations have recently been gathered together from various sources and put on record by Professor Poulton[55]. These observations shew that butterflies may be devoured by mantids, dragon-flies, and blood-sucking flies of the families Empiidae and Asilidae. For mantids the records are scanty, but they have been observed to kill presumably distasteful forms as often as those which are considered palatable. An interesting set of experiments was made by G. A. K. Marshall on captive mantids in Africa[56]. Of the eleven individuals representing several species with which he experimented, some ate every butterfly offered, including the distasteful Danaines and Acraeines. Others, however, shewed some distaste of the Acraeines and would not devour them so freely as butterflies of other species. There are no grounds, however, for supposing that the mantids had any appreciation of the warning coloration of the Acraeines. Whether completely eaten or not the Acraeines were apparently sufficiently damaged to prevent their taking any further part in the propagation of their species. Warning coloration is not of much service to its possessor who has to be tasted and partially eaten before being eventually rejected. Even if some mantids shew distaste of certain unpalatable butterflies, that distaste is probably seldomexercised with a gentleness sufficient to ensure that the butterfly reaps the reward of its disagreeable nature. And unless, of course, the butterfly is allowed to do so the enemy can play no part in the production or maintenance of a mimetic resemblance.
What is true for mantids is probably also true for the other groups of predaceous insects. Dragon-flies and wasps have been recorded as attacking the distasteful as well as butterflies of unprotected groups. Among the most serious enemies of butterflies must probably be reckoned the blood-sucking Asilids. These powerful and ferocious flies seize butterflies on the wing with their strong claws and plunge their proboscis into the thorax. Apparently they inject some swift poison, for the butterfly is instantly paralysed, nor is there any sign of struggle. The Asilid flies off with its victim, sucking the juices as it goes. There can be no doubt in the mind of any one who has watched these creatures hawking butterflies that their natural gifts are such as to enable them to exercise discrimination in their food. Most insect life is at their mercy but they appear to exercise no choice, seizing and devouring the first flying thing that comes within easy reach. Certainly as regards butterflies palatability or the reverse makes no difference, and they are known to feed indiscriminately both upon the evil-flavoured and upon the good. Taking it all together the evidence is such that we cannot suppose predaceous insects to pay any attention to warning colours, and, therefore, we cannot regard them as playing any part in connection with mimetic resemblance.
II.Lizards.In those parts of the world where lizards of larger size are abundant there is plenty of evidence that certain species are very destructive to butterfly life. As might be expected this is especially true of forms which are either arboreal or semi-arboreal in habit. Among the reptiles of Ceylon, for example, are several species of the genusCalotes, of which two,C. ophiomachusandC. versicolor, are particularly abundant. In appearance and habits they are not unlike chameleons though far more active in their movements. Like chameleons, too, they are able to change colour, and the fact that they can assume a brilliant scarlet hue about the head and neck has probably led to their popular name of "blood-suckers." It is not impossible that the assumption of this scarlet coloration may serve as a lure to bring insects within range. These lizards have often been observed to seize and devour butterflies. Moreover, it is a common thing to find butterflies with a large semi-circular patch bitten out of the hind wings, and there is little doubt but that such injuries have been inflicted by lizards. There is, however, no evidence to suggest that they exercise any discrimination in their choice of the butterflies which they attack. This is borne out by their behaviour towards various species offered to them, both when at liberty and when caged. In an ingenious series of experiments Col. Manders brought various butterflies within reach of aCalotesby the help of a fishing-rod and a long line of fine silk, by this means simulating natural conditions as far as possible.He found that the lizards ate the so-called distasteful forms such asDanais chrysippus,Euploea core,Acraea violae, andPapilio hector, as readily as the presumably more palatable forms[57]. In captivity, too, they will take any butterfly as readily as another. Experiments by Finn[58]and by the writer[59]proved that they ate Danaids, Euploeas, andPapilio aristolochiaewithout any hesitation so long as the insects were alive and moving. When, too, a mixture of different species, some with and some without warning coloration, was given to them all were eaten, nor was there any discrimination evidenced in the order in which they were taken. The lizard simply took the first that came within reach and went on until the whole lot was devoured, wings and all.
Some experiments by Miss Pritchett on the American lizardSceleporus floridanuspoint to the same conclusion[60]. She found that it took without hesitation any butterfly offered to it including the presumably distasteful modelsDanais archippusandPapilio philenor(cf. pp.45and49). On the other hand, another species of lizard with which Miss Pritchett experimented,Gerrhonotus infernalis, refused all the butterflies offered to it, though it fed freely on Orthopterous insects as well as on spiders and scorpions.
It seems clear from these various observations andexperiments that certain lizards devour butterflies freely, but that they do not exercise any discrimination in the species which they attack. All are caught and devoured indiscriminately, so that in spite of the fact that such lizards are among the most serious enemies of butterflies we cannot suppose them to play any part in establishing a mimetic resemblance.
III.Birds.The relations which exist between butterflies and their bird enemies have for many years been the subject of keen discussion. It is generally recognised that if mimetic resemblances become established through the agency of discriminating enemies those enemies must be birds. Hence those interested in the question of mimicry have for some years past turned their attention to birds more than to the other enemies of butterflies. That many birds systematically feed on butterflies is a fact that does not admit of doubt. It is true that, as Mr Marshall points out in the valuable paper in which he has summarised the evidence[61], observations of birds eating butterflies are relatively scanty. Though, as he points out, this is equally true for other groups of insects besides butterflies, bird attacks on butterflies, owing to the conspicuous nature of the victim, are much more likely to attract attention than attacks on other groups. We are still without much information as to the extent to which birds destroy butterflies and as to whether they shew any decided preference for certain species over others. A careful examination of the contents of thestomachs of large numbers of insectivorous birds in a tropical area would go some way towards deciding the matter, but at present such information is lacking. We have to rely upon the existing observations of birds attacking butterflies in the wild state, and upon certain feeding experiments made with captive birds.
Observations on birds attacking butterflies where mimetic forms occur have been made almost entirely in certain parts of Africa, in India, and in Ceylon. For Africa, Marshall has collected some forty-six observations of which almost half are concerned with Pierines. The remainder include four instances of attacks on species ofAcraea, a genus which on the mimicry theory must be regarded as among the most unpalatable of butterflies.
The records from the Indo-Malayan region (principally India and Ceylon) are somewhat more numerous and here again more than one-third of them refer to Pierines. Among the others are records of the distasteful formsEuploea core,E. rafflesii,Acraea violae, andPapilio hectorbeing taken and devoured.
There is one interesting record which seems to suggest that Swinhoe's Bee-Eater (Melittophagus swinhoei) may exercise that discrimination in the butterflies it attacks which is demanded on the mimicry theory. Lt.-Col. Bingham on one occasion in Burma noticed this species hawking butterflies. He records that they tookPapilio erithonius,P. sarpedon,Charaxes athamas,Cyrestis thyodamas, andTerias hecabe, and probably also species of the generaPrioneris,Hebomoia(Pierines),JunoniaandPrecis(Vanessids). And he goes on to say: "I also particularly noticed that the birds never went for aDanaisorEuploea, or forPapilio macareusandP. xenocles, which are mimics of Danais, though two or three species ofDanais, four or five ofEuploea, and the two above-mentioned mimickingPapiliossimply swarmed along the whole road[62]."
Marshall also quotes a case of attack by a green bee-eater on aDanaisin which the butterfly was caught and subsequently rejected, after which it flew away. Little stress, however, can be laid upon this case in view of the more recent data brought together by Col. Manders and Mr Fryer. Discussing the attacks of birds on butterflies in Southern India and Ceylon, Col. Manders gives the following quotation[63]from a letter of Mr T. N. Hearsy, Indian Forest Service:
"Coimbatore, 6. 6. 10.... I have frequently seen the common green bee-eater (Merops viridis) and the king-crow (Buchanga atra) take butterflies on the wing, the butterflies beingCatopsilia pyranthe,C. florella,Terias hecabeandPapilio demoleus. The bee-eater I have also seen takingDanais chrysippusandDanais septentrionis, and I remember to have been struck with their taste for those latter...."
Col. Manders also brings forward evidence for these Danaids and Euploeas being eaten by Drongos and by the paradise flycatcher. Still more recently an interesting contribution to the matter has been made byMr J. C. F. Fryer[64]. The Ashy Wood-swallow (Artamus fuscus) had been recorded on two occasions as having attackedEuploea core. Mr Fryer was fortunate in coming across this bird in the gardens at Peradeniya, near Kandy, at a time whenEuploea coreandDanais septentrioniswere particularly abundant, and he watched a number of them systematically hawking these presumably unpalatable species. As he observes, "in Ceylon a resemblance to the generaDanaisandEuploeais doubtfully of value; in fact in the neighbourhood of Wood-swallows it is a distinct danger." Fryer also noted that the mimetic forms ofP. polyteswere taken as well as the non-mimetic.
For tropical Central and South America, that other great region where mimetic forms are numerous, there are unfortunately hardly any records of butterflies attacked by birds. Bates stated that the Pierines were much persecuted by birds, and his statement is confirmed by Hahnel, but exact observations for this region are remarkably scanty. Belt observed a pair of birds bring butterflies and dragon-flies to their young, and noticed that they brought no Heliconii to the nest although these swarmed in the neighbourhood[65]. On the other hand, Mr W. Schaus[66], from an experience of many years spent in the forests of Central America, considers that the butterflies of this region are hardly, if ever, attacked by birds.
For North America Marshall records over 80 cases of birds attacking butterflies. Among them is an interesting record of a bird seizing and rejecting a specimen ofAnosia plexippus(=Danais archippus), one of the few Danaines found in this region.
It must be admitted that the data at present available with regard to the attacks of birds upon butterflies under natural conditions are too meagre to allow of our coming to definite conclusions on the points at issue. It is safe to say that a number of species of birds have been known to attack butterflies—that a few out of the number feed upon butterflies systematically—that some of the most persistent bird enemies devour the presumably protected forms as freely as the unprotected—but that in a few instances there is some reason for supposing that the bird discriminates. Beyond this it is unsafe to go at present.
In attempting to come to a decision as to the part played by birds in the destruction of butterflies an evident desideratum is a knowledge of the contents of the stomachs of freshly killed birds. Unfortunately few systematic observations of this nature exist. G. L. Bates[67], when collecting in the Southern Cameroons, noted the stomach contents of a considerable number of birds. The remains of beetles were recognised in 213 cases: Orthoptera in 177: ants in 57 (mostly in stomachs of birds of the genusDendromus): other Hymenoptera in 8: coccids in 32: bugs in 19: white ants in 31: slugs and snails in 24: spiders in 85(mostly in Sunbirds): millipedes in 20; but in no single instance were the remains of butterflies found. More recently Bates' account has been criticized by Swynnerton[68]who comments on the difficulty of identifying butterfly remains as compared with those of beetles and grasshoppers. He states that the pellets ejected by captive birds after a meal of butterflies contain only fine debris which is very difficult to identify. Further, he found that of twenty small bird excreta collected in the forest no less than eighteen contained scales and small wing fragments of Lepidoptera.
Some attention has been paid to the relation between birds and butterflies in the United States, and under the auspices of the Department of Agriculture a large number of birds' stomachs have been investigated. Careful examination of some 40,000 stomachs of birds shot in their natural habitats resulted in the discovery of butterfly remains in but four. It cannot, therefore, be supposed that birds play much part in connection with such mimetic resemblances as are found in North America (cf. pp.45-49). Nevertheless, it is known that on occasion large numbers of butterflies may be destroyed by birds. An interesting case is described by Bryant[69]of an outbreak in North California ofEugonia californica, a close relative of the tortoiseshell. The butterfly was so abundant as to be a plague, and five species of birds took advantage of its great abundance to prey largely upon it. Fromhis examination of the stomachs Bryant came to the conclusion that some 30% of the food of these five species was composed of this butterfly. The stomachs of many other species were examined without ever encountering butterfly remains. Nor did field observations support the view that any species, other than the five specially noted, ever attacked these butterflies. The case is of interest in the present discussion as evidence that the identification of butterfly remains in the stomachs of birds is by no means so difficult as some observers suggest.
Besides this evidence derived from observations upon birds in the wild state some data have been accumulated from the experimental feeding of birds in captivity. Of such experiments the most extensive are those of Finn[70]in South India. He experimented with a number of species of insectivorous birds belonging to different groups. Of these he found that some, among which may be mentioned the King-crow, Starling, and Liothrix[71], objected to Danaines,Papilio aristolochiaeandDelias eucharis, a presumably distasteful Pierine with bright red markings on the under surface of the hind wings (Pl. II, fig. 1). In some cases the bird refused these forms altogether, while in others they were eaten in the absence of more palatable forms. The different species of birds often differed intheir behaviour towards these three "nauseous" forms. The Hornbill, for example, refused the Danaines andP. aristolochiaeabsolutely, but ateDelias eucharis. Some species again, notably the Bulbuls (Molpastes) and Mynahs, shewed little or no discrimination, but devoured the "protected" as readily as the "unprotected" forms. Finn also states that "Papilio polyteswas not very generally popular with birds, but much preferred to its model,P. aristolochiae."
In many of Finn's experiments both model and mimic were given to the birds simultaneously so that they had a choice, and he says that "in several cases I saw the birds apparently deceived by mimicking butterflies. The Common Babbler was deceived byNepheronia hippia[72]and Liothrix byHypolimnas misippus. The latter bird saw through the disguise of the mimeticPapilio polites, which, however, was sufficient to deceive the Bhimraj and King-crow. I doubt if any bird was impressed by the mimetic appearance of the femaleElymnias undularis" (cf.Pl. IV, fig. 5). Finn concluded from his experiments that on the whole they tended to support the theory of Bates and Wallace, though he admits that the unpalatable forms were commonly taken without the stimulus of actual hunger and generally without signs of dislike. Certainly it is as well to be cautious in drawing conclusions from experiments with captive birds. The King-crow, for instance, according to Finn shewed a marked dislike for Danaines in captivity; yet Manders records thisspecies as feeding upon Danaines under natural conditions (cf. p.111).
A few further experiments with the birds of this region were carried out by Manders[73]in Ceylon. The results are perhaps to be preferred to Finn's, as the birds were at liberty. Manders found that the Brown Shrike (Lanius cristatus) would take butterflies which were pinned to a paling. In this way it made off with the mimetic females ofHypolimnas bolinaandH. misippus, as well as withDanais chrysippusandAcraea violaewhich were successively offered to it. Evidently this species had no repugnance to unpalatable forms. Manders also found that a young Mynah allowed complete liberty in a large garden would eat such forms asAcraea violaeandPapilio hector. As the result of his experience Manders considers that the unpalatability of butterflies exhibiting warning coloration has been assumed on insufficient data, and he is further inclined to doubt whether future investigations will reveal any marked preference in those birds which are mainly instrumental in the destruction of butterflies.
A few experiments on feeding birds with South African butterflies are recorded by Marshall. A young Kestrel (Cerchneis naumanni) was fed from time to time with various species of butterflies. In most cases the butterflies offered were eaten even when they were species ofAcraea. On the other handDanais chrysippuswas generally rejected after being partlydevoured. When first offered this unpalatable species was taken readily and it was only after it had been tasted that the bird rejected it. When offered on several subsequent occasions it was partly eaten each time, and the behaviour of the Kestrel did not suggest that it associated a disagreeable flavour even with this conspicuous pattern. Another young Kestrel (Cerchneis rupicoloides) was also used for experiment. At first it would not take butterflies and at no time did it shew any fondness for them. Indeed it is doubtful from the way in which they seem to have shaped at the insects whether either of these Kestrels had had any experience of butterflies before the experiments began.
A Ground Hornbill with which Marshall also experimented ate various species, includingAcraea, but, after crushing it, refused the onlyDanais chrysippusoffered. It is hardly likely that this large omnivorous bird operates as a selecting agent in cases of mimicry.
In an interesting paper published recently McAtee[74]discusses the value of feeding experiments with animals in captivity as a means of indicating their preference for different articles of diet. After reviewing the various evidence brought forward he concludes that the food accepted or rejected by captive animals is very little guide to its preferences under natural conditions. He points out that a bird in captivity not infrequently rejects what is known to form a main staple of its diet in nature, and that conversely it may eagerly accept something which, in the wild state, itwould have no opportunity of obtaining. Great caution must, therefore, be exercised in the interpretation of feeding experiments made with birds in captivity.
It appears to be generally assumed that colour perception in birds is similar to what it is among human beings, but some experiments made by Hess[75]render it very doubtful whether this is really the case. In one of these experiments a row of cooked white grains of rice was illuminated by the whole series of spectral colours from violet to deep red. Hens which had been previously kept in the dark so that their eyes were adapted to light of low intensity were then allowed to feed on the spectral rice. The grains illuminated by green, yellow, and red were quickly taken, but the very dark red, the violet, and the blue were left, presumably because the birds were unable to perceive them. Again, when the birds were given a patch of rice grains of which half was feebly illuminated by red light and the other half more strongly by blue light, they took the red but left the blue. Previous experiment had shewn that with ordinary white light the birds always started on the best illuminated grains. It seems reasonable to conclude, therefore, that in the red-blue experiment the feebly illuminated red grains were more visible than the far more strongly lighted blue ones. It might be objected that the birds had a prejudice against blue, but, as Hess points out, this is almost certainly not the case because they took grainswhich were very strongly illuminated with blue. Results of a similar nature were also obtained from pigeons, and from a kestrel which was fed with pieces of meat lighted with different colours.
On the whole these experiments of Hess convey a strong suggestion that the colour perceptions of birds may be quite different from our own, more especially where blue is concerned. Great caution is needed in discussing instances of mimicry in their relation to the bird, for we have no right to assume that the bird sees things as we do. On the other hand, it is a matter of much interest to find that in general blue plays relatively little part in cases of mimetic resemblance among butterflies; some combination of a dark tint with either red, white, brown, or yellow being far more common.
It will probably be admitted by most people that the evidence, taken all together, is hardly sufficient for ascribing to birds that part in the establishing of a mimetic likeness which is required on the theory of mimicry. That birds destroy butterflies in considerable numbers is certainly true, but it is no less true that some of the most destructive birds appear to exercise no choice in the species of butterfly attacked. They simply take what comes first and is easiest to catch. It is probably for this reason that the Wood-swallow feeds chiefly on Euploeines and Danaines (cf. p.112). It is probably for this reason also that such a large proportion of the records of attacks on butterflies under natural conditions refer to the Pierines; for owing totheir light colour it is probable that the "Whites" are more conspicuous and offer a better mark for a bird in pursuit than darker coloured species.
Mammals.Apart from man it is clear that only such mammals as are of arboreal habits are likely to cause destruction among butterflies in the imago state. Apparently there are no records of any arboreal mammal, except monkeys, capturing butterflies in the wild state, nor is there much evidence available from feeding experiments. But such evidence as exists is of considerable interest. As the result of feeding butterflies of different sorts to an Indian Tree-shrew (Tupaia ferruginea) Finn[76]found that it shewed a strong dislike to Danaids and toPapilio aristolochiaethough it took readilyPapilio demoleus,Neptis kamarupa, andCatopsilia(a Pierine). It is fairly certain that if the Tree-shrew is an enemy of butterflies in the wild state it is a discriminating one.
The other mammals with which experiments have been made are the common baboon, a monkey (Cercopithecus pygerythrus), and a mongoose (Herpestes galera)—all by Marshall[77]in South Africa. The mongoose experiments were few and inconclusive, nor is this a matter of much moment as it is unlikely that this mammal is a serious enemy of butterflies.
The monkey ate various forms ofPrecis(a Vanessid), after which it was givenAcraea halali. This distasteful form was "accepted without suspicion, but whenthe monkey put it into his mouth, he at once took it out again and looked at it with the utmost surprise for some seconds, and then threw it away. He would have nothing to do with anAcraea caldarenawhich I then offered him[78]."
The experiments with the baboons were more extensive. Two species ofAcraea,halaliandaxina, were recognised when first offered and refused untasted.Danais chrysippus, on the other hand, was tasted on being offered for the first time, and then rejected. This species was twice offered subsequently and tasted each time before being rejected. When offered the fourth time it was rejected at sight. The baboon evidently learned to associate an unpleasant taste with thechrysippuspattern. At this stage it would have been interesting to have offered it some well-known mimic ofchrysippus, such as the female ofHypolimnas misippusor thetrophoniusform ofPapilio dardanus, but this experiment was unfortunately not made. Marshall did, however, offer it at the same time a specimen each ofByblia ilithyia(a Vanessid) and ofAcraea axinato which it bears a general resemblance. The baboon took the former but neglected the latter altogether. The general resemblance between the two species was not sufficiently close to deceive it.
These experiments with mammals, though few in number, are of unusual interest. Should they be substantiated by further work it is not impossiblethat, as a factor in the establishing of a mimetic likeness, a stronger case may be made out for the monkey than the bird. The monkey apparently eats butterflies readily[79]: owing probably to a keener sense of smell it shews far less hesitation as to its likes and dislikes: its intelligence is such that one can easily imagine it exercising the necessary powers of discrimination; in short it is the ideal enemy for which advocates of the mimicry theory have been searching—if only it could fly. As things are its butterfly captures must be made when the insect is at rest, probably near sunrise and sunset, and this leads to a difficulty. Most butterflies rest with their wings closed. In many of the well-known cases of mimicry the pattern on the under surface of the mimic's wings which would meet the monkey's eye is quite different from that of its model. It is difficult in such cases to imagine the monkey operating as a factor in establishing a resemblance between the upper surfaces of the wings of the two unrelated species. On the other hand, some butterflies,e.g.Papilio polytes, rest with wings outspread, and there are rare cases, such as that ofP. laglaizei(p. 27), where the most striking point about the resemblance is only to be appreciated when the insects are at rest with their wings closed. In such cases it is conceivable that the monkey may play a part in the elimination of the non-mimetic elements of a palatable species which at the same time possessed a mimetic form closely resembling another species disagreeable to the monkey's taste. As has been pointed out earlier (p.96) even a slight persecution directed with adequate discrimination will in time bring about a marked result where the mimetic likeness is already in existence. It is not impossible therefore that the establishing of such a likeness may often be due more to the discrimination of the monkey than to the mobility of the bird.
It is clear from the last few chapters that the theory of mimicry in butterflies with its interpretation of the building up of these likenesses by means of natural selection in the form of predaceous birds and other foes is open to destructive criticism from several points of view. The evidence from mimicry rings makes it almost certain that in some cases the resemblance must be founded on an initial variation of such magnitude that the mimic could straightway be confused with the model. Till the mimic can be mistaken for the model natural selection plays no part. The evidence from breeding suggests strongly that in certain cases (e.g.Papilio polytes) the likeness arose in the form in which we know it to-day. In such cases there is no reason for supposing that natural selection has had anything to do with the formation of the finished mimic. Considerations of this nature may be said to have destroyed the view, current until quite recently, that in the formation of a mimetic resemblance the exclusive agent was natural selection. During the past few years it has come to be admitted by the staunchest upholders of the theory of mimicry that naturalselection would not come into play until the would-be mimic was sufficiently like the model to be confused with it under natural conditions[80]. The part now often attributed to natural selection is to put a polish on the resemblance and to keep it up to the mark by weeding out those which do not reach the required standard. It is supposed that if natural selection ceases to operate the mimetic resemblance is gradually lost owing to the appearance of variations which are no longer weeded out. An interesting case has recently been brought forward by Carpenter[81]and explained on these lines: The NymphalinePseudacraea eurytusis a polymorphic species found in Central Africa. In Uganda it occurs in several distinct forms which were originally supposed to be distinct species. Three of these forms bear a marked resemblance to three species of the Acraeine genusPlanema.