1. Tropical American region.
2. Australian region.
In the chapter on geographical distribution in Bronn's Thierreich, Vögel, Systematischer Theil, p. 296 (1893), and in my Classification of Vertebrata (1898), due attention had been paid to the Amphibia as well as to the other classes of Vertebrata. It will be seen in the following pages that my arrangement is well applicable to the Amphibia so far as fundamental principles are concerned.
It cannot be sufficiently emphasised that any attempt to form the various faunas of the different classes of animals into one scheme must necessarily be a petitio principii. The time-honoured six zoo-geographical regions established by Sclater and Wallace represent fairly well the main continental divisions: North America, South America, Africa, Australia, and the large northern continental mass of the Old World, with India as a tropical appendix. There is no correlation and no subordination in this scheme. Huxley's division (1868) into Notogaea and Arctogaea (see p. [74]) is of fundamental importance. The next improvement was the combination of the Palaearctic and Nearctic "regions" into one, an advance originally due to Professor Newton, carried out by Heilprin (1887) as the Holarctic region. I have, in 1893, substituted for it the more appropriate term Periarctic, meaning the whole mass of land which lies around the indifferent Arctic zone. The want of further co-ordination and subordination required the combination of the African and Oriental or Indian countries into a Palaeotropical region (1893); the Ethiopian or African and the Indian or Oriental regions of Sclater and Wallace thereby assuming their proper subordinate rank of subregions.
The two primary divisions Notogaea and Arctogaea are fundamental. The four secondary divisions, namely the Australian and Neotropical, Periarctic and Palaeotropical regions, also stand the test of application to the various classes and main groups of Vertebrata; but naturally, under the present configuration of the world, the Palaeotropical region is nothing but the Southern continuation of the Eastern half of the Periarctic mass of land. This is especially obvious so far as India is concerned. There is, however, that broad belt of desert, sand, and salt-steppes, which extends from North-West Africa to Manchuria, and this belt is one of the most important physical features of the Old World. It is complicated by the system of mountain-chains which, broadly speaking, centre at the Pamirs, and radiate westwards through the Caucasus and Alps into Spain, eastwards through the Himalayas into China, and north-eastwards to Kamtschatka; interrupted by Bering's Sea, it is continued as the backbone of both Americas to Patagonia.
The tertiary divisions, the subregions, have no real existence. They depend upon the class, or even order, of animals, which we happen to study. The faunistic distribution of the Urodela is not that of the Anura, and both follow separate lines of dispersal, different from those of the various orders of Reptiles, Birds, and Mammals. This must be so. There is no doubt that the distribution of land and water was totally different in the Coal Age from what it is now. The face of the globe at the Jurassic Age can scarcely be compared with the aspect which the world has assumed in the Miocene period.
This leads to another consideration, often neglected. We know that the various classes, orders, families, etc., of animals have appeared successively upon the stage. A group which arose in the Coal Age followed lines of dispersal different from one which was not evolved until Jurassic times, and post-cretaceous creatures could not avail themselves of what assisted their ancestors, and vice versâ. The Amphibia are bound absolutely to the land and to fresh water; transportation across salt water is not excluded, but must be accidental, and is not a case of regular "spreading." Speaking generally, the older a group, the more likely is it to be widely distributed. If it appears scattered, this may be due to extinction in intermediate countries or to submergence of former land-connexions.
There is great danger of arguing in a circle. It is one of the most difficult tasks to decide in cases of great resemblance of groups of animals between their being due to direct affinity or to heterogeneous convergence, or parallel development. It is the morphologist who is ultimately responsible for the establishment of faunistic regions, not the systematist, least of all he who accepts an elaborate classification, and then mechanically, mathematically, by lists of genera and species, maps out the world. Let us take an example. The Neotropical region and Madagascar, but not Africa, are supposed to be faunistically related to each other. In both namely occur Boa and Corallus amongst snakes, Dendrobatinae amongst Ranidae, and of the Insectivora Solenodon in Cuba, Centetes in Madagascar. More cases can no doubt be found which would strengthen this resemblance, perhaps in support of the startling view that Madagascar and South America have received part of their fauna from the famous Antarctica. But the value of the Insectivores has been disposed of by their recognition as an extremely ancient group, or as a case of convergence, and the two genera are no longer put into the same family as Centetidae. The Dendrobatinae (Mantella in Madagascar, the others in South America) are decidedly not a natural group, but an instance of very recent convergence (cf. p. [272]). About the members of the ancient Boidae we do not feel quite so sure.
It is therefore advisable to eliminate for zoogeographical purposes groups about which there can be any reasonable doubt, otherwise we may argue that certain genera must constitute a very old family, because they are now restricted to widely separated countries, or on the strength of their distribution we may conclude that the genera in question cannot be related to each other, and do not belong to the same sub-family or family as the case may be. Such groups are the Engystomatinae and the genus Spelerpes; amongst reptiles the Eublepharidae, Helodermatidae, Anelytropidae, Ilysiidae, Amblycephalidae.
It is customary to represent the various regions and sub-regions as if they had boundaries as fixed as political frontiers. Such limitations are quite arbitrary, and what is of more importance, they differ in reality according to the class or order of animals with which we happen to deal. Moreover, there has been, and is probably still going on, an exchange or overlapping of faunas. Such debatable grounds are Central America and the highlands of North-western South America. The famous Wallace's line, between Borneo and Celebes, Java and Lombok, is absolutely inapplicable to the Anura. From their point of view the Austro-Malayan countries, Papuasia and Polynesia do not form a sub-region of the Australian, but rather of the Palaeotropical region. Concerning the Urodela, the division into Palae- and Ne-arctic sub-regions is unjustifiable since Eastern Asia has emphatically American affinities (cf. also p. [96]). The Sahara and the rest of Northern Africa are intimately connected with Arabia, Persia, Afghanistan, and Northern India, just as equatorial Africa and Madagascar possess strong faunistic relationship with Southern India and the Malay islands.