The paucity of freshwater fishes is due, in the first place, to the arid climate and the deficiency of water in the Australian continent, as well as to the insignificant size of the freshwater courses in the smaller islands. Still this cannot be the only cause: the large island of Celebes, which, by its mountainous portions, as well as by its extensive plains and lowlands, would seem to offer a favourable variety of conditions for the development of a freshwater Fauna is, as far as has been ascertained, tenanted by seven Freshwater fishes only, viz. 2 Arius, 2 Plotosus, 1 Andbas, 1 Ophiocephalus, 1 Monopterus, all of which are the commonest species of the Indian region. New Guinea has not yet been explored, but, from the faunæ nearest to this island, we expect its freshwater fishes will prove to be equally few in number, and identical with those of Celebes and North Australia; a supposition confirmed by the few small collections which have reached Europe. Finding, then, that even those parts of this region, which are favourable to the development of Freshwater fishes, have not produced any distinct forms, and that the few species which inhabit them, are unchanged, or but slightly modified Indian species, we must conclude that the whole of this area has remained geologically isolated from the other regions of this zone since the commencement of the existence of Teleostei; and that, with the exception of Ceratodus and Osteoglossum, the immigration of the other species is of very recent date.

Fossil remains of Ceratodus have been found in Liassic and Triassic formations of North America, England, Germany, and India; and it is, therefore, a type which was widely spread in the Mesozoic epoch. Although it would be rash to conclude that its occupation of Australia dates equally far back, for it may have reached that continent long afterwards; yet it is evident that, as it is one of the most ancient of the existing types, so it is certainly the first of the Freshwater fishes which appeared in Australia. Osteoglossum, of which no fossil remains yet have been found, is proved by its distribution to be one of the oldest Teleosteous types. There must have been a long gap of time before these ancient types were joined by the other Teleostei. All of them migrated through the intervening parts of the ocean from India. Most of the Plotosina, some of the Arii, Dules, and the Atherinichthys, also Nannoperca (allied to Apogon), were among the earliest arrivals, being sufficiently differentiated to be specifically or even generically (Cnidoglanis, Nannoperca) distinguished; but some others, like Anabas scandens, Lates calcarifer, Dules marginatus, must have reached the Australian continent quite recently, for they are indistinguishable from Indian specimens.

In South-western Australia a mingling of the scanty fauna with that of the southern temperate parts takes place. Oligorus macquariensis (The Murray Cod), which has a congener on the coast of New Zealand, ascends high up the Murray river, so that we cannot decide whether this Percoid should be located in the Tropical or Temperate part of Australia. Several Galaxias also extend to the confines of Queensland, and will probably some day be found members of this region.

In the smaller Pacific islands the Freshwater fishes exhibit a remarkable sameness: two or three species of Dules, several Eels, an Atherine, or some Gobies, Mullets, and other fishes which with equal readiness exchange fresh for salt water, and which would at once reach and occupy any streams or freshwater lakes that may be formed on an island.

The Sandwich Islands are the only group among the smaller islands which are tenanted by a Siluroid, a species of Arius, which is closely allied to Central American species, and, therefore, probably immigrated from Tropical America.

II. Northern Zone.

The boundaries of the Northern Zone coincide in the main with the northern limit of the Equatorial Zone; but at three different points they overlap the latter, as has been already indicated. This happens in, and east of, Syria, where the mixed faunæ of the Jordan and the rivers of Mesopotamia demand the inclusion of this territory into the Northern Zone as well as the Equatorial; in the island of Formosa, where a Salmonoid and several Japanese Cyprinoids flourish; and in Central America, where a Lepidosteus, a Cyprinoid (Sclerognathus meridionalis), and an Amiurus (A. meridionalis) represent the North American fauna in the midst of a host of tropical forms.

A separate Arctic Zone does not exist for Freshwater fishes; ichthyic life becomes extinct towards the pole as soon as the fresh water remains frozen throughout the year, or thaws for a few weeks only; and the few fishes which extend into high latitudes, in which lakes are open for two or three months in the year, belong to types in no wise differing from those of the more temperate south. The highest latitude at which fishes have been obtained is 82° lat. N., whence the late Arctic Expedition brought back specimens of Charr (Salmo arcturus and Salmo naresii).

The ichthyological features of this zone are well marked: the Chondrosteous Ganoids or Sturgeons, and the families of Salmonidæ and Esocidæ are limited to, and characteristic of, it; Cyprinoids flourish with the Salmonoids, both families preponderating in numbers over the others, whilst the Siluroids are few in number and in variety.

The two regions in which this zone is divided are very closely related to one another, and their affinity is not unlike that which obtains between the sub-regions of the Southern Zone. The subjoined list will show their close agreement with regard to families as well as species. Several of the latter are common to both, viz.—Acipenser sturio, A. maculatus, Perca fluviatilis, Gastrosteus pungitius, Salmo salar, Esox lucius, Lota vulgaris, Petromyzon marinus, P. fluviatilis, and P. branchialis; and all recent investigations have resulted in giving additional evidence of the affinity, and not of the diversity of the two regions.

In Europe and temperate Asia, as well as in North America, mountain ranges elevated beyond the line of perpetual snow would seem to offer physical conditions favourable for the development of a distinct alpine fauna. But this is not the case, because the difference of climate between the mountain districts and the lowlands is much less in this zone than in the Equatorial. Consequently the alpine freshwater fishes do not essentially differ from those of the plains; they are principally Salmonoids; and in Asia, besides, mountain-barbels and Loaches. Salmo orientalis was found by Griffith to abound in the tributaries of the Bamean river at an altitude of about 11,000 feet.

A. The Europo-Asiatic (Palæarctic) Region.—Its western and southern boundaries coincide with those of the Northern Zone, so that only those which divide it from North America have to be indicated. Behring’s Strait and the Kamtschatka Sea have been conventionally taken as the boundary, but this is shown to be artificial by the fact that the animals of both coasts, as far as they are known at present, are not sufficiently distinct to be referred to two distinct regions. As to the freshwater fishes those of North-western America and of Kamtschatka are but imperfectly known, but there can be little doubt that the same agreement exists between them as is the case with other classes of animals. The Japanese islands exhibit a decided Palæarctic fish-fauna, which includes Barbus and Cobitioids, forms strange to the North American fauna. A slight influx of tropical forms is perceived in the south of Japan, where two Bagrina (Pseudobagrus aurantiacus and Liocassis longirostris) have established themselves for a considerable period, for both are peculiar to the island, and have not been found elsewhere.

In the east, as well as in the west, the distinction between the Europo-Asiatic and North American regions disappears almost entirely the farther we advance towards the north. Of four species of the genus Salmo known from Iceland, one (S. salar) is common to both regions, two are European (S. fario and S. alpinus), and one is a peculiarly Icelandic race (S. nivalis). As far as we know the Salmonoids of Greenland and Baffin’s Land they are all most closely allied to European species, though they may be distinguished as local races.

Finally, as we have seen above, the Europo-Asiatic fauna mingles with African and Indian forms in Syria, Persia, and Afghanistan. Capoëta, a Cyprinoid genus, is characteristic of this district, and well represented in the Jordan and rivers of Mesopotamia.

Assuming that the distribution of Cyprinoids has taken its origin from the alpine tract of country dividing the Indian and Palæarctic regions, we find that this type has found in the temperate region as equally favourable conditions for its development as in the tropical. Out of the 360 species known to exist in the Palæarctic regions, no less than 215 are Cyprinoids. In the countries and on the plateaus immediately joining the Himalayan ranges those mountain forms which we mentioned as peculiar to the Indian Alps abound and extend for a considerable distance towards the west and east, mixed with other Cyprinina and Cobitidina. The representatives of these two groups are more numerous in Central and Eastern Asia than in Europe and the northern parts of Asia, where the Leuciscina predominate. Abramidina or Breams are more numerous in the south and east of Asia, but they spread to the extreme north-western and northern limits, to which the Cyprinoid type reaches. The Rhodeina are a small family especially characteristic of the East, but with one or two off-shoots in Central Europe. Very significant is the appearance in China of a species of the Catostomina, a group otherwise limited to North America.

The Cyprinoids, in their dispersal from the south northwards, are met from the opposite direction by the Salmonoids. These fishes are, without doubt, one of the youngest families of Teleostei, for they did not appear before the Pliocene era; they flourished at any rate during the glacial period, and, as is testified by the remnants which we find in isolated elevated positions, like the Trout of the Atlas, of the mountains of Asia-Minor, and of the Hindu Kush, they spread to the extreme south of this region. At the present day they are most numerously represented in its northern temperate parts; towards the south they become scarcer, but increase again in numbers and species, wherever a great elevation offers them the snow-fed waters which they affect. In the rivers of the Mediterranean Salmonoids are by no means scarce, but they prefer the upper courses of those rivers, and do not migrate to the sea.

The Pike, Umbra, several species of Perch and Stickleback, are also clearly autochthont species of this region. Others belong to marine types, and seem to have been retained in fresh water at various epochs: thus the freshwater Cottus (Miller’s Thumb); Cottus quadricornis, which inhabits lakes of Scandinavia, whilst other individuals of the same species are strictly marine; the Burbot (Lota vulgaris); and the singular Comephorus, a dwarfed and much-changed Gadoid which inhabits the greatest depths of Lake Baikal.

Remnants of the Palæichthyic fauna are the Sturgeons and Lampreys. The former inhabit in abundance the great rivers of Eastern Europe and Asia, periodically ascending them from the sea; their southernmost limits are the Yangtse-kiang in the east, and rivers flowing into the Adriatic, Black and Caspian Seas, and Lake Aral, towards the centre of this region. None are known to have gone beyond the boundaries of the Northern Zone. If the Lampreys are justly reckoned among Freshwater fishes, their distribution is unique and exceptional. In the Palæarctic region some of the species descend periodically to the sea, whilst others remain stationary in the rivers; the same has been observed in the Lampreys of North America. They are entirely absent in the Equatorial Zone, but reappear in the Temperate Zone of the Southern Hemisphere. Many points of the organisation of the Cyclostomes indicate that they are a type of great antiquity.

The remaining Palæarctic fishes are clearly immigrants from neighbouring regions: thus Silurus, Macrones, and Pseudobagrus from the Indian region; Amiurus (and, as mentioned above, Catostomus) from North America. The Cyprinodonts are restricted to the southern and warmer parts; all belong to the carnivorous division. The facility with which these fishes accommodate themselves to a sojourn in fresh, brackish, or salt water, and even in thermal springs, renders their general distribution easily comprehensible, but it is impossible to decide to which region they originally belonged; their remains in tertiary deposits round the Mediterranean are not rare.

B. The boundaries of the North American or Nearctic Region have been sufficiently indicated. The main features and the distribution of this fauna are identical with those of the preceding region. The proportion of Cyprinoid species to the total number of North-American fishes (135:339) appears to be considerably less than in the Palæarctic region, but we cannot admit that these figures approach the truth, as the Cyprinoids of North America have been much less studied than those of Europe; of many scarcely more than the name is known. This also applies in a great measure to the Salmonoids, of which only half as many as are found in the Palæarctic region have been sufficiently described to be worthy of consideration. North America will, without doubt, in the end show as many distinct races as Europe and Asia.

Cyprinoids, belonging to genera living as well as extinct, existed in North America in the tertiary period. At present, Cyprinina, Leuciscina, and Abramidina are well represented, but there is no representative of the Old World genus Barbus, or of the Cobitidina[26]; Rhodeina are also absent. On the other hand, a well-marked Cyprinoid type is developed—the Catostomina, of which one species has, as it were, returned into Asia. Very characteristic is the group of Centrarchina, allied to the Perch, of which there are some thirty species; two Grystina. Of the Sticklebacks there are as many species as in Europe, and of Pike not less than seven species have been distinguished. Umbra appears to be as local as in Europe. Some very remarkable forms, types of distinct families, though represented by one or two species only, complete the number of North American autochthont fishes—viz., Aphredoderus, Percopsis, Hyodon, and the Heteropygii (Amblyopsis and Chologaster). The last are allied to the Cyprinodonts, differing from them in some points of the structure of their intestines. The two genera are extremely similar, but Chologaster, which is found in ditches of the rice-fields of South Carolina, is provided with eyes, and lacks the ventral fins. Amblyopsis is the celebrated Blind Fish of the Mammoth Cave of Kentucky: colourless, eyeless, with rudimentary ventral fins, which may be occasionally entirely absent.

A peculiar feature of the North American Fish Fauna is that it has retained, besides the Sturgeons and Lampreys, representatives of two Ganoid families, Lepidosteus and Amia. Both these genera existed in tertiary times: the former occurs in tertiary deposits of Europe as well as North America, whilst fossil remains of Amia have been found in the Western Hemisphere only.

It is difficult to account for the presence of the Amiurina in North America. They form a well-marked division of the Bagrina, which are well represented in Africa and the East Indies, but absent in South America; it is evident, therefore, they should not be regarded as immigrants from the south, as is the case with the Palæarctic Siluroids. Nor, again, has the connection between South and North America been established sufficiently long to admit of the supposition that these Siluroids could have spread in the interval from the south to the northern parts of the continent, for some of the species are found as far north as Pine Islands Lake (54° lat. N.)[27]

III. Southern Zone.

The boundaries of this zone have been indicated in the description of the Equatorial Zone; they overlap the southern boundaries of the latter in South Australia and South America, but we have not at present the means of exactly defining the limits to which southern types extend northwards. This zone includes Tasmania with at least a portion of South-eastern Australia (Tasmanian sub-region), New Zealand and the Auckland Islands (New Zealand sub-region), and Chili, Patagonia, Terra del Fuego, and the Falkland Islands (Fuegian sub-region). No freshwater fishes are known from Kerguelen’s Land, or from islands beyond 55° lat. S. The southern extremity of Africa has to be excluded from this zone so far as Freshwater fishes are concerned.

This zone is, with regard to its extent as well as to the number of species, the smallest of the three; yet its ichthyological features are well marked; they consist in the presence of two peculiar families, each of which is analogous to a northern type, viz. the Haplochitonidæ, which represent the Salmonidæ, Haplochiton being the analogue of Salmo, and Prototroctes that of Coregonus; and the Galaxiidæ, which are the Pikes of the Southern Hemisphere.

Although geographically widely separate from each other, the Freshwater fishes of the three divisions are nevertheless so closely allied that conclusions drawn from this group of animals alone would hardly justify us in regarding these divisions as sub-regions. One species of Galaxias (G. attenuatus) and the three Lampreys are found in all three, or at least two, sub-regions.

Freshwater Fishes of the Southern Zone.

But little remains to be added in explanation of this list; Percichthys is in Chili the autochthont form of the cosmopolitan group of Percina. Diplomystax, an Arioid fish of Chili, and Nematogenys seem to have crossed the Andes from Tropical America at a comparatively early period, as these genera are not represented on the eastern side of South America; the Trichomycterina occur on both sides of the Andes, which they ascend to a considerable height. Retropinna is a true Salmonoid, allied to, and representing in the Southern Hemisphere the Northern Smelt, Osmerus. In both these genera a part of the specimens live in the sea, and ascend rivers periodically to spawn; another part remain in rivers and lakes, where they propagate, never descending to the sea, this freshwater race being constantly smaller than their marine brethren. That this small Teleostean of the Northern Hemisphere should reappear, though in a generically modified form, in New Zealand, without having spread over other parts of the Southern Zone, is one of the most remarkable, and at present inexplicable facts of the geographical distribution of freshwater fishes.