We have to ask, therefore, what is the reason of this mutual sterility of species; whether it is a necessary outcome of the morphological differences between the species, or only a chance accessory phenomenon, or perhaps an absolutely necessary preliminary condition to the establishment of species.
The last was the view held by Romanes. He believed that a species could only divide into two when it was separated into isolated groups either geographically or physiologically, that is, when sexual segregation in some form is established within the species, so that all the individuals can no longer pair with one another, but groups arise which are mutually sterile. It is only subsequently, he maintained, that these groups come to differ from one another in structure. To this hypothetical process he gave the name of 'physiological selection.' This view depends—it seems to me—upon an underestimate of the power of natural selection. Romanes believed that when a species began to split up, even the adaptive variations would always disappear again because of the continual crossing, and that only geographical isolation or sexual alienation, that is, physiological selection, would be able to prevent this. But even the fact that there are dimorphic and polymorphic species proves sufficiently that adaptation to two or even several sets of conditions can go on on the same area. In many ants we find many kinds of individuals—the two sexual forms, the workers, and the soldiers, and these last are undoubtedly distinguished by adaptive characters which must be referred to selection. The same is true of the caterpillars, whose coloration is adapted to their surroundings in two different ways. If the individuals of one and the same species can be broken up into two or more different forms and combinations of adaptations, while they are mingling uninterruptedly with one another, natural selection must undoubtedly be able, notwithstanding the continual intermingling of divergent types, to discriminate between them and to separate them sharply from one another. Assuredly then a species can not only exhibit uniform variation on a single area, but may also split up into two without the aid of physiological selection. Theoretically it is indisputable that of two varieties which are both equally well suited to the struggle for existence, a mixed form arising through crossing may not be able to survive. Let us recall, for instance, the caterpillars, of which some individuals are green and some brown, and let us assume that the brown colour is as effective a protection as the green, then the two forms would occur with equal frequency; but though a mixed hybrid form which was adapted neither to the green leaves nor to the brown might occasionally crop up, it would always be eliminated. It would occur because the butterflies themselves are alike, whether they owe their origin to green caterpillars or to brown, and thus at first, at least, all sexual combinations would be equally probable.
I do not believe therefore in a 'physiological selection,' in Romanes's sense, as an indispensable preliminary condition to the splitting of species, but it is a different question whether the mutual sterility so frequently observable between species has not conversely been produced by natural selection in order to facilitate the separation of incipient species. For there can be no doubt that the process of separating two new forms, or even of separating one new form from an old one, would be rendered materially easier if sexual antipathy or diminished fertility of the crossings could be established simultaneously with the other variations. This would be useful, since pure and well-defined variations would be better adapted to their life-conditions than hybrids, and would become increasingly so in the course of generations. But as soon as it is useful it must actually come about, if that is possible at all. It may be, however, as we have said before, that the two divergent forms depend merely upon quantitative variations of the already existing characters; sexual attraction, whether it depends upon very delicate chemical substances, or on odours, or on mutual complementary tensions unknown to us, will always fluctuate upwards and downwards, and plus or minus determinants, which lie at the root of these unknown characters in the germ-plasm, must continually present themselves and form the starting-point for selection-processes of a germinal and personal kind, which may bring about sexual antipathy and mutual sterility between the varieties. I therefore consider Romanes's idea correct in so far that separation between species is in many cases accompanied by increasing sexual antipathy and mutual sterility. While Romanes supposed that 'natural selection could in no case have been the cause' of the sterility, I believe, on the contrary, that it could only have been produced by natural selection; it arises simply, as all adaptations do, through personal selection on a basis of germinal selection, and it is not a preliminary condition of the separation of species, but an adaptation for the purpose of making as pure and clean a separation as possible. It is obviously an advantage for both the divergent tendencies of variation that they should intermingle as little as possible. This is corroborated by the fact that by no means all the marked divergences of species are accompanied by sexual alienation, and that the mutual sterility so frequently seen is not an inevitable accompaniment of differences in the rest of the organism.
That this is not the case is very clearly proved by our domesticated animals. The differences in structure between the various breeds of pigeon and poultry are very great, and breeds of dog also diverge from one another very markedly, especially in shape and size of body. Yet all these are fertile with one another, and they yield fertile offspring. But they are products of artificial selection by man, and he has no interest in making them mutually sterile, so that they have not been selected with a view to sexual alienation, but in reference to the other characters. The segregation of animal species into several sub-species on the same area is probably usually accompanied by sexual antipathy, since in this case it would be useful although not indispensable. But the matter is different in the case of the transformation of a colony upon a geographically isolated region. 'Amiktic' forms, such as Vanessa ichnusa of Corsica, are hardly likely to be sexually alienated from the parent form; we have here to do only with the preponderance of a fortuitous and biologically valueless variation and its consequent elevation to the rank of a variety. The new form was not an adaptation, but only a variation, and as it was of no use, it was not in a position to incite any process of selection favouring its advancement.
But even adaptive transformations on isolated regions from which the parent species is excluded are not likely to develop rapidly any sexual antipathy as regards the parent stock, and I should not be surprised if experiments showed that there is perfect mutual fertility between, for instance, many of the species of Achatinella on the Sandwich Islands or of Nanina on Celebes, or between the species of thrushes on the different islands of the Galapagos Archipelago, or between these and the ancestral species on the adjacent continent, if that species is still in existence. For there was no reason why sexual antipathy to the parent form should have developed in any of these adaptation forms which have arisen in isolation, and therefore it has probably not been evolved.
That our view of the mutual sterility between species, as an adaptation to the utility of precise species-limitation, is the correct one is evidenced not only by our domesticated races, but even more clearly by plants, in regard to which it is particularly plain that the sexual relations between two species are adaptational. We have already seen in what a striking way the sensitiveness of the stigma of a flower is regulated in reference to pollen from the same plant, that some species are not fertilizable by their own pollen at all, that others yield very little seed when self-fertilization is effected, and that others again are quite fertile—as much so as with the pollen of another plant of the same species. We regarded these gradations of sexual sensitiveness as adaptations to the perfectly or only moderately well-assured visits of insects, or to their entire absence. I wish to cite these cases as well as the heterostylism of some flowers as evidence in support of the conception of the mutual sterility of species which I have just outlined. But this only in passing. The point to which I chiefly wish to direct attention is the mutual fertility of many plant-species. In lower as well as in higher plants fertile hybrids occur not infrequently under natural conditions, and cultivated hybrids, such as the new Medicago media, a form made by mingling two species of clover, may go on reproducing with its own kind for a considerable time. A number of Phanerogams yield fertile hybrids, and in Orchids even species of different genera have been crossed and have yielded offspring which was in some cases successfully crossed with a third genus.
If these facts prove anything it is that the factors which determine the mutual sterility of species are quite distinct from their morphological differences, in other words, from the diagnostic characters of the specific type. For a long time the verdict on this matter was too entirely based on observations made on animals, among which mutual sterility arises relatively easily, even where it was not intended (sit venia verbo!). Even the pairing, but still more the period of maturity, the relations of maturity in ovum and sperm, and even the most minute details in the structure of the sperm-cell, the egg-shell, envelope, &c., have to be taken into account, and these may bring about mutual or, as Born has shown, one-sided sterility. We know, through the researches of Strasburger, that a great many Phanerogams, when pollinated artificially from widely separated species of different genera and families, will at least allow the pollen-tube to penetrate down to the ovule, and that in many cases amphimixis actually results. It follows that we must not lay too great stress upon the mutual sterility which occurs almost without exception among the higher animals, but must turn to the plants with greater confidence.
Among plants there is very widely distributed mutual fertility between species. I doubt, however, whether the observations on this point are sufficient to warrant any certain conclusion in regard to the importance of the phenomena in the formation of species. At least it is not easy to see why the mutual sterility of many species of plants should not have been necessary or useful in separating species, and why it was not therefore evolved. We may point to the fact that animals can move from place to place as the chief reason, and this factor does undoubtedly play a part, but the widespread crossing of plants by insects makes up to some extent, as far as sexual intermingling is concerned, for their inability to move from place to place. I do not know whether the species of orchid which are fertile with one another belong to different countries, so that we may assume that they originated in isolation, or whether fertile orchids from the same area are fertilized by different insects and are thus sexually isolated. This and many other things must be taken into consideration. Probably these relations have not yet been adequately investigated; probably what is known by some experts has not yet been made available to all. Future investigations and studies must throw more light upon the problem.
In any case, however, we can see from the frequency of mutual fertility among plants that mutual sterility is not a conditio sine qua non of the splitting up of species, and we must beware of laying too great stress upon it even among animals. Germinal selection is a process which not only forms the basis of all personal selection, but which is also able to give rise of itself, without the usual aid of sexual intermingling, to a new specific type. And we cannot with any confidence dispute that, even without amphigony, a certain degree of personal selection may not ensue solely on the basis of the favourable variational departures originating in the germ-plasm. It would be premature to express any definite views on this point as yet, but the diverse cases of purely asexual or parthenogenetic reproduction in groups of plants rich in species make this hypothesis seem probable.
The most remarkable example of this is probably to be found in the Lichens, the symbiotic nature of which we have already discussed, in which—now at least—neither the Fungus nor the Alga associated with it is known to exhibit sexual reproduction. If this is really the case, then the existence of numerous and well-marked species of lichens leads us to the hypothesis just expressed, and we must suppose that the unity of the specific type is attained in this case solely by a continual sifting of the useful from the useless variations of the determinants, and through purely germinal intensification of the surviving variational tendencies.