The study of the phylogenetic relationships of Recent genera and species of insects and of their geographical distribution forms an important part of modern systematic entomology. It has been recognized almost universally that, no matter what the mechanism of evolution is, both morphological differentiation and geographical distribution are the results of historical processes. The reconstruction of these processes requires a carefully balanced evaluation of morphological and geological evidence. This balance is not easily achieved, however, and many unsatisfactory phylogenetic reconstructions have been proposed in which either the morphological or the geological evidence is over-emphasized, or in which details taken from morphology as well as from geology are combined to present a seemingly correct picture of the evolutionary history of the group which, however, is contradicted by other details of morphological or geological evidence. A further difficulty is that, in the desire of giving reality to ancestral forms supposed to have existed at certain times, Recent forms are often substituted for hypothetical ancestors in the phylogenetic trees. Chronologically no Recent form can be ancestral to any other Recent form, and it is desirable, therefore, to stress the hypothetical character of the ancestral forms incorporated in phylogenetic trees.
An attempt has been made by the present author at designing a method for the reconstruction of the phylogenesis of closely-related species and genera which, it is hoped, reduces the speculative character of this kind of research to a minimum. It is evident that only well-known groups exhibiting comparatively simple specific and generic characters afford any chance of elucidating the history of the Recent species. The material selected is a group of papilionid butterflies usually included in a comprehensive genus Troides, but here subdivided into Troides Hübner, Trogonoptera Rippon, Schœnbergia Pagenstecher and Ornithoptera Boisduval. They are found chiefly in the Australasian Archipelago, but extend to the Solomons and eastern Australia in the east, and to Ceylon, Peninsular India and north-west China in the west and north. 31 species with 123 subspecies are at present known. This group affords the following advantages: (1) the number of species is not likely to be increased much by future discoveries and their distribution is well known; (2) the specific characters are mostly non-adaptive and relatively simple (elements of pattern and coloration, differences in genitalia, secondary sexual characters); (3) the members of the group have formed numerous geographical subspecies; and (4) the geological history of the area of distribution is known and affords criteria for the inhabitability of certain areas at certain times. The method adopted in the present paper is that, instead of using all the available evidence simultaneously, phylogenetic trees are constructed by using points (2) and (3) only, and, in doing so, certain technical modifications are introduced in the trees. The trees which, thus, rely on morphology and Recent distribution only, are then tested by means of a critical comparison with the geological history of the area. Since the two lines of approach are both of a complex nature, a high degree of compatibility of the results suggests that the conclusions are likely to be essentially correct.
In Part I. of the paper the terms species, subspecies, species-group, subspecies-group and form are defined for the purposes of the present investigation. Examples are given for the evolution of subspecies during the Holocene and of species during the Pleistocene of Europe, and the evolution of taxonomic species from taxonomic subspecies is demonstrated, but since this subject is outside the scope of the present paper, it will be treated more fully elsewhere.
The conception of “phylogenetic stages” is introduced as being the period of time, or the equivalent amount of morphological differentiation, from any one point of a lineage up to another point at which certain characters have become differentiated so much that, taxonomically, the descendant has to be regarded as specifically distinct. In the special case of the Troides-group, this vague term gains considerably in value because of the connection of at least the two latest phylogenetic stages of most lineages with geological changes in the Australasian Archipelago. Though this term might prove to be applicable more generally, it is used here merely as a convenient terminus technicus.
In Part II. the species and subspecies of Troides, Trogonoptera, Schœnbergia and Ornithoptera are analysed from the morphological point of view. The closely-related species are combined in species-groups. The geographical distribution of the species-groups and subspecies-groups is considered. In order to ensure that the inter-relationship of the species be based on the safest evidence available, a special study is made of the venation and the structure of the scent-pouch of the hindwing, and of the genitalia of the males of all species.
It is found that Trogonoptera, hitherto considered as related to Ornithoptera, is a side-branch of Troides, and that Ornithoptera and Schœnbergia are, together, a homogeneous subgroup differing from the subgroup of Troides-Trogonoptera in the structure of the male harpe and of the scent-pouch, which are described in detail, as well as in other respects.
In Part III. the evidence of inter-relationship provided by Part II. is used in constructing phylogenetic trees. Where justified by morphological evidence, the assumption is made that each subspecies-group has evolved from a common ancestral form in comparatively Recent times, and that each species-group was, at some remoter period, a subspecies-group and, therefore, in turn, evolved from some earlier common ancestor.
All the known species and subspecies of Troides and Trogonoptera can be derived from a common ancestral form by assuming that the process of splitting into subspecies which, in the course of time, became species, occurred at least five times. This subgroup appears to have originated in Sundaland and thence spread mainly eastwards over the Archipelago in successive waves.
The Schœnbergia-Ornithoptera-group also can be reduced to a single ancestral form by means of five successive phylogenetic stages. It evolved in the Moluccas, Solomons and New Guinea. It has not spread westwards beyond the Moluccas and overlaps with the Troides-Trogonoptera-group in these islands only, if one neglects the recent arrival of a single form of Troides in New Guinea. Only one species of Ornithoptera has spread during the latest phylogenetic stages far and wide over the rain-forest areas of all these isles and of eastern Australia. The distribution of Ornithoptera during the earlier stages is discontinuous (Moluccas and Solomons), and the present geographical link, New Guinea, occupied by Schœnbergia.
In Part IV. the geological history of the area is considered according to geographical units, and the results of Part III. confronted with it in each case. Climatic changes and fluctuations of the sea-level are discussed, and it is shown that, during the Pleistocene, the shallow seas between the Malay Peninsula, Java, Borneo were repeatedly dry land (Sundaland) and that the same applies to the seas between New Guinea, the Aru Isles and Australia (Sahul Shelf). The other islands, from the Philippines, Celebes and the Lesser Sunda Isles to the Moluccas, were never connected with these shelves or with one another. The formation of the Recent geographical races of many species of Sundaland and the Sahul Shelf coincides with the rising of the sea-level after the low phase of the Last Glaciation.
The species and races found on the islands of the deep-sea area between the Asiatic and Australian shelf-areas conform closely to the pre-Pleistocene history of these islands. In particular, most parts of the chain of the Lesser Sunda Isles are geologically very young (late Pliocene and early Pleistocene); accordingly, only one species, T. haliphron, has become differentiated there in the course of the Pleistocene. This instance suggests that the rate of evolution of a taxonomic species in the Malay Archipelago is roughly the same as in Europe, i. e., about equal to the duration of the Pleistocene.
Other Recent species of the Troides-group appear to have evolved from ancestral species in the course of the Pleistocene. All phylogenetic stages previous to the last two therefore fall within the Tertiary.
Several independent lineages appear to have passed more or less simultaneously through certain stages of species-formation. It is suggested that geological changes which occurred at certain definite times, or were of a periodical nature, afforded at certain more or less regular intervals new chances for dispersal or new ecological conditions, resulting in the more or less simultaneous appearance of new forms.
The evolution of the Troides-group and its distributional history are summarized. There is complete agreement between morphological phylogeny and geological history, though two distributional problems remain unsolved unless recourse is taken to the theory of continental drift. These are (1) the slight degree of overlap of Troides and Schœnbergia-Ornithoptera in the Moluccas, and (2) the discontinuous distribution of ancestral Ornithoptera. Both problems would find an easy explanation if the claims of Wegener, du Toit and other geologists could be substantiated that New Guinea and Australia had drifted from the south into the forming island arcs of the Archipelago in the course of the Tertiary and Pleistocene.
In Part V. the values for the duration of the Pleistocene obtained by various geological methods are applied to Troides. It appears that in this genus, as in others, at least half to one million years were required for the specific characters to become stabilized in some of the species.