[[p. 339]] INTRODUCTION

    Students of both phytogeography, or plant distribution, and zoogeography, or animal distribution, must depend largely on the geologist to tell them how the land and sea were bounded in the past, but often they can help the geologist in this very matter. Studies of the relationships of living things, both recent and fossil, frequently point out that two islands or continents must have been connected by dry land at a certain geological time, for example the connection of Alaska with Siberia across Bering Strait. In the same way the close relationship or identity of various species of fishes and other marine animals on the Pacific and Atlantic coasts of Panama show that the two oceans were once connected over what is now the isthmus, and at no very distant period, geologically speaking. In fact, the geologist often depends to a considerable extent on the fossil remains of once-living creatures for information about his rock formations, their age and the climatic and other conditions under which they [[p. 340]] were deposited, as well as for hints as to probable land and sea connections in the past. Biogeographical studies, then, interlocked as they are with geology and paleontology, contribute to two different fields, primarily to a knowledge of how living things evolved and migrated and secondarily to a better understanding of the changes of land and sea areas and of climate during geological time.

    For many reasons, biogeographical problems are often difficult to solve, although some workers have been slow to realize this difficulty. The classical distributional researches of Darwin, Wallace, and their contemporaries threw a bright light on subjects that had been little thought about, and many of their followers lost sight of the fact that the most brilliant of the first results were drawn from those instances in which plain evidence was lying fallow to produce richly for those first in the field. Year by year the literature of biogeography grows, and as vast new stores of precise facts pile up, problems that formerly seemed simple become increasingly difficult to solve. Indeed, it is apparent that the data of the science are already too vast for any single person to digest, even in relatively circumscribed problems. Most biogeographers of today are of necessity narrower specialists than their predecessors; each has a large and growing body of fact in his own specialty to master and less time to acquire the geological background and knowledge of the literature of other fields that would enable him to see distributional problems in true perspective. It would seem that general biogeographical results of lasting value can be reached nowadays only through the cooperation of a group of competent specialists who can weigh apparently contradictory evidence and reach conclusions in accord with the soundest bodies of biological, ecological, paleontological, and geological fact.

ZOOGEOGRAPHICAL PROBLEMS OF THE AMERICAS

    Biogeographical problems are legion, ranging from questions of continental and oceanic relationships to minute details of the distribution of single species, but at this time I wish to speak only of a very few of the broader problems of the animal geography of North and South America.

    Generally speaking, zoogeographers, like Gaul, are divided into three parts--those who build bridges, those who do not, and the proponents of continental drift. To begin with the last first, it may be explained that Wegener and his followers believe that all the continents once formed a more or less compact land mass floating upon the deeper, heavier rocks of the earth's crust, and that sundry pieces of this land mass broke off from time to time and drifted away to form what we now call Australia, the Americas, and Africa. Despite the fulminations of those opposed to this theory, there is a considerable body of weighty evidence in favor of it, and I for one, should not be [[p. 341]] surprised to see it finally prevail. However, there is just as weighty authority against it, and since the fishes might be construed as favoring either side of the argument, I shall not now consider it further.

    The bridge builders are those zoogeographers who are prone to postulate vast continental connections, which they believe existed in what is now the deep sea, in order to provide bridges over which the land faunas could march to destinations to which they obviously got somehow. The more radical of this school frequently sink (in theory) whole halves or quarters of continents to the deep sea and raise vast areas of ocean bottom into new continents to explain their ideas of how living things evolved and migrated. Unfortunately, there is little geological evidence for the more bewitching of these schemes, and zoogeographers are coming to depend on them less and less. To my mind, the most sensible of these "bridges" are the narrow and not at all startling "isthmian links" of Willis (1932) and Schuchert (1932); they have the unique backing of what appears to be sound geological and bathymetric support, and in addition, some of them are still in existence either as true continental connections (Panama) or chains of islands (West Indies).

    Those who do not build bridges are the zoogeographers who, since the time of Wallace, have held sternly and perhaps a little too tenaciously to the theory of the "permanence of the ocean basins." They do build bridges, but very modest ones. They hold that, by and large, the only areas that have ever been dry land are those either now dry or a part of the continental shelves, that parts of continents have been flooded but never deeply, that no major portion of the really deep-sea bottom has ever been upraised into dry land, and that the small land bridges they postulate can account for all the migrations of land animals that have occurred. It must be admitted that this school has been the most cautious in examining the available evidence and the most erudite in its researches; of late years it has also been the most stiff-necked in considering contrary opinion. Its foremost recent exponent has been the late W. D. Matthew, whose Climate and Evolution (1915) has had a profound and overwhelming effect on nearly all recent American vertebrate zoogeographers, and but little on anybody else. There is no doubt, however, that Matthew's views deserve the most careful consideration, and it is unfortunate that he did not live to write the proposed enlarged and revised edition of his work.¹

    Matthew's thesis was, briefly, that "secular climatic change has been an important factor in the evolution of land vertebrates and the principal known cause of their present distribution" and that, in later geological epochs, most continental groups of animals have originated in the great northern land mass of Eurasia and North America [[p. 342]] (the Holarctic region) and migrated outward, radially, into South America, Africa, and the Indo-Australian region. On a globe, or a north-polar projection of the world, such evolution and migration is easy to visualize; Eurasia and North America, which have frequently been connected, form a preponderating land unit, from which the other continents radiate. I think no serious zoogeographer can disagree radically with Matthew's view that secular climatic change has had an enormous effect on animal distribution, but it is questionable that climate has directly influenced evolution itself.² In regard to his postulation of a northern origin of the faunas of South America and other southern continents, however, it should be observed that von Ihering (1907), Eigenmann (1909), and other eminent men have come to a directly opposed conclusion, which emphasizes the relationships of certain African and South American animals that they believe did not originate in the north, and which, therefore, would demand some sort of South Atlantic land bridge to account for their evident community of origin. The necessity for a South Atlantic bridge has also been strongly advocated by so eminent an authority as Regan (1922), whose thoroughly sound data on the ostariophysan fishes are of the greatest importance.

    In reviewing the distribution of American continental faunas in connection with some recent studies of West Indian fresh-water fishes, I have been struck with the amount of misinformation that passes as sound ichthyological evidence among zoogeographers. Part of this is the fault of the ichthyologists themselves. Papers on fish distribution written by competent ichthyologists and based on modern paleontological and ichthyological data are scarce, and the distributional information in two recent general textbooks on fishes (Kyle, MacFarlane) is scarcely to be relied on. On the other hand, the recent dependable papers that do exist have been neglected by most students of zoogeography. I have come across no modern paper that summarizes the broader geographical aspects of the fresh-water fishes of North and South America, and, in the belief that such may be of some general interest, I shall attempt to supply this want in very brief form. At the same time I shall speak about some of the implications of fresh-water fish distribution in the controversial question of West Indian paleogeography, but, in compliance with my thesis that biogeographical problems cannot be solved on the evidence of one group, I shall not pontify on matters I know I am not competent to settle.

WHAT ARE FRESH-WATER FISHES?

    The importance of fresh-water fishes to students of geographical distribution depends primarily on two facts. Firstly, certain families of fishes possess an ancient physiological inability to survive in salt sea [[p. 343]] water, which binds them to the land as securely as any known animals. Secondly, on the land, they are inescapably confined to their own particular drainage systems and can migrate from one isolated stream basin to the next only through the slow physiographical change of the land itself (stream capture, etc.). Throughout the world the migrations of fresh-water fishes over extensive continental areas have generally been excessively slower than those of almost any creature that can creep, crawl, walk, or fly, however closely that creature may have been bound by its ecological tolerances. This is exceptionally well illustrated by Central America, where the interpenetration of North and South American faunas has proceeded in many groups practically to the limit of climatic tolerance, but where no truly Neotropical (South American) fresh-water fish has gotten farther north than Texas or New Mexico, and none truly Nearctic (North American) farther south than Nicaragua.

    There are, of course, exceptional methods by which fishes may be transported. "Rains of fishes" are sufficiently well known and authenticated to make it certain that cyclonic winds, in passing over bodies of water, sometimes pick up small fishes and deposit them at a distance, still alive. It is possible, too, that a fish eagle or gull might drop or disgorge alive a newly caught fish after having carried it over a divide between two distinct river systems. But the frequently made statement that the eggs of fishes are dispersed by adhering to the feet of wading birds in flight should cease to trouble zoogeographers; such a method of transportation is possible, but almost no fish eggs are sufficiently resistant to survive drying in the air more than a very few minutes. The main fact to keep in mind is that fish distribution is much more regular and understandable than it would be if these unusual methods of transportation were of much importance.

    One fact that some zoogeographers who have dealt with fishes have neglected is that the fishes to which the adjective "fresh-water" is applied differ widely in the extent of their tolerance of salt water. There are fresh-water fishes which never swim into salt water, some which occasionally do but can survive it only for a short period, some which habitually frequent estuaries and other brackish waters and frequently enter the sea, and some which migrate back and forth between river and sea either continually or periodically. It is plainly evident that a fish which can swim through sea water from one river mouth to another is not of much use in studies of terrestrial zoogeography.

    The only fresh-water fishes that need especially concern us at present are those of the first two of the categories I have just mentioned. These two categories I shall distinguish as a primary division whose members are very strictly confined to fresh water, and a secondary division whose members are generally restricted to fresh water [[p. 344]] but occasionally enter the sea voluntarily for short periods. The other groups not specifically placed in divisions grade off into wholly marine forms. Finally, there are a number of species and genera of salt-water families that have taken up more or less permanent residence in fresh water. Most of them return to the sea to spawn.

THE GROUPS OF FRESH-WATER FISHES

    The fresh-water fishes of the primary and secondary divisions tend to group themselves on natural family lines. In other words, all or nearly all species of one family usually show a similar tolerance to salt water. This, together with the fossil evidence available, leads us to believe that most of the families of the primary division have carried down their physiological inability to survive in the sea, as family characters, from early times and probably since the origin of the groups concerned.

    The fresh-water fishes of the primary division are of diverse relationships among fishes generally. A few of them are small, relict groups of primitive organization, such as the two living families of lung fishes (Ceratodontidae, with one surviving species in Australia, and Lepidosirenidae, with three surviving species of the genus Protopterus in Africa and one of the genus Lepidosiren in South America). Of similar relict distribution are the two living genera of the primitive paddle-fishes, Psephurus in the Yangtse River in China and Polyodon in the Mississippi. Higher on the scale of fish evolution are the paleoniscidlike bichirs of Africa and the family of bowfins, with one living North American species. Somewhat similar to these last are the "ganoid" garpikes of North America, a family which I place, somewhat hesitantly, in the secondary division of fresh-water fishes.

    The bulk of living sea fishes are of more specialized organization than these primitive families, and belong to the great subclass of bony fishes, or Teleostei, and this is true of the fresh-water families as well. The lowest order of teleosts is that of the herringlike fishes (Isospondyli) and to it belong several families of my primary fresh-water division; among them the hiodonts or mooneyes of the Mississippi; the strange phractolaemids, pantodonts, kneriids,³ and mormyrids of Africa; and the ancient osteoglossids, which were probably well-nigh cosmopolitan in Eocene fresh waters.

    More than half of the true fresh-water fishes of the world belong to a single order, the Ostariophysi,⁴ distinguished from the herrings by the peculiar chain of Weberian ossicles connecting the air-bladder with [[p. 345]] the auditory apparatus. The ostariophysans are primarily a fresh-water order and must have been since their pre-Eocene origin; out of nearly 30 families and 4,500 or more species of them only 2 specialized families (the ariid and plotosid catfishes), comprising not over 150 species, are marine, and these beyond reasonable doubt are descended from ancient fresh-water ancestors. To the Ostariophysi belong the hordes of carps or minnows, suckers, loaches, characins, gymnotid eels, and bewhiskered catfishes that swarm throughout the rivers and lakes of all the continents except Australia. The only fresh-water Ostariophysi of Australia and Madagascar are ariid and plotosid catfishes that have reinvaded the rivers from the sea.⁵

    The remaining fresh-water groups of the primary division are quickly enumerated. The rather herringlike order Haplomi contains the pikes or pickerels, the small mud minnows, and the Alaskan and Siberian brackfish. To some extent transitional from the foregoing, more primitive, bony fishes to the more highly developed, spiny-rayed types, are the strange percopsids and pirate perches of North America. Of the cyprinodonts I consider that only one family, the North American cave fishes, belongs to the primary division. Of the fully developed, perchlike groups we have the North American sunfishes or basses, the true perches of North America and Eurasia, and the tropical nandids. Ending the series are the Old World tropical labyrinth fishes and the isolated spiny-eels.

    The secondary division of fresh-water fishes is composed of families which, in general, behave like true fresh-water fishes, but whose members show a less sharp restriction to fresh water. Among the more important of them are the North American garpikes, the synbranchid eels, the cichlids, and the various families of topminnows or cyprinodonts. Garpikes are known to enter the sea along the Gulf of Mexico coast. Most cichlids can survive several hours or days in the sea, and one species of Tilapia, collected in brackish water in Mozambique, has been kept for months in sea water at the New York Aquarium. Many cyprinodonts do not seem to be inconvenienced greatly by salt water. Mollienisia latipinna enters the sea freely and multiplies in brine ponds about Manila Bay, where it was accidentally introduced. Metzelaar reported Rivulus from tide pools on Curaçao, and certain species of Fundulus and Cyprinodon live permanently on the seacoast beyond reach of estuaries. The Challenger even caught a Fundulus in a mid-Atlantic pelagic haul! It is evident that many species of this secondary group might easily survive a short sea journey. This is borne out by distributional fact. It is only members of the secondary [[p. 346]] group that have succeeded in crossing to the east of Wallace's Line⁶ in the East Indies, and in reaching Madagascar.⁷

    After this second group comes a succession of families of catadromous or anadromous fishes such as the fresh-water eels, the salmon and trout, the sturgeons, the galaxiids and the aplochitonids, and others which, for one reason or another, enter the sea freely and are in most cases not of use in studies of the distribution of fresh-water fishes. Lastly we find primarily marine families such as the sharks, herrings, gray mullets, sea basses, and gobies, a few of whose members have invaded the rivers and sometimes developed into purely fresh-water species.

THE FRESH-WATER FISH FAUNAS OF THE AMERICAS

1. Limitation to Tertiary

    In the mammals, the history of many of the modern orders has been traced back to their very different Eocene ancestors and considerable has been done in connecting these up with Mesozoic forms. The story of the continental fishes is very different. The Ostariophysi, as well as nearly all other groups of fresh-water bony fishes, are known only as far back as the early Tertiary, and the few known Eocene fossils usually turn out to belong to still living genera or their very close relatives. It would seem that the evolution, from very primitive types into practically modern forms, of a whole series of the major groups of fresh-water (and marine) bony fishes occurred between the end of the Cretaceous and the earlier Eocene. What little we know of Paleocene fishes is mostly from marine deposits, and the derivation and evolutionary lines of the Ostariophysi and most other fresh-water groups remains a closed book. The locating of early Eocene and Paleocene fresh-water beds and the working out of the history of the fishes remains one of the greatest untouched problems in vertebrate paleontology, and one in which American paleontologists, at least, have taken strangely small interest. It is evident that we can draw no definite conclusions regarding pre-Tertiary geography from the fresh-water bony fishes, and but little about post-Mesozoic from such primitive relicts as the lungfishes and bichirs.

2. North America

    The greater part of the Tertiary and recent fresh-water fish fauna of North America is divisible into two sections. The first forms the old, endemic fauna composed of paddlefishes, garpikes, bowfins, osteoglossids, suckers, ameiurid catfishes, primitive cyprinodonts, and [[p. 347]] pirate-perches. All of them are known to have been present in the Nearctic Eocene and all survive in North America today save the osteoglossids and primitive cyprinodonts; the last are probably represented by the existing cave fishes. The paddlefishes are recorded from the English Chalk and now exist in China; the bowfins are as old as the Jurassic and are found in the Mesozoic rocks of Europe, Brazil, and North America; and the garpikes occur in the Cretaceous of Europe and the Lameta beds of India. All three groups are too old to enter decisively into our discussion of the Tertiary fishes. The suckers are found in the Eocene of Wyoming, and Hussakof (1932) has reported isolated gill covers from the Eocene of Mongolia; there is one specialized genus now living in China and one North American species has invaded Siberia, probably recently. The ameiurid catfishes are represented in the Green River Eocene by Rhineastes and appear to be a purely North American group related to the Old World bagrids.⁸ A supposed living Asiatic species based on a Chinese painting is almost certainly mythical. To the old section probably belong the North American sunfishes or basses,⁹ and the peculiar percopsids, the geological history of which is nearly or quite unknown. The bulk of the North American fresh-water fish fauna of today is made up of the members of the carp or minnow family. Fossil carps are unknown in North America until the Miocene, at which time the family evidently arrived from Eurasia to form the second or younger section of the fauna. This invasion was made up entirely of the second most primitive (Leuciscinae) of the three great subfamilies of Cyprinidae (Regan, 1922). Lack of clear paleontological evidence does not allow us to say whether or not the Eurasian-American true perches are autochthonous North Americans. Genera supposed to be perches are recorded from both the American and European Eocene. If Voigt (1934) and Berg (1936) are correct in associating the short-jawed middle Eocene Palaeoesox with the pikes and mud minnows, the order Haplomi probably originated in Asia and may have invaded America with the carps. The mud minnows,¹⁰ unknown as fossils, have two genera and three American species. The pikes have several North American species¹¹ and have existed in Eurasia at least from the Oligocene to the present.

[[p. 348]] 3. South America (and Africa)

    In South America we have a very different picture.¹² The known fossil record is so fragmentary as to be of no particular importance, but the living fishes are exceedingly instructive. Of the living true fresh-water fishes of South America, not a single species, genus, or family is identical with truly North American groups. In fact the South American fishes show greater dissimilarity to those of North America than they do to those of any continent except Australia, which has no fresh-water fishes at all of my primary division save for one lungfish and one osteoglossid.

    Certain apparent similarities are at once disposed of. The primitive bony fishes of the family Osteoglossidae, which appear to have become extinct in North America in the Eocene, exist today in South America. But they also exist in Africa, the Malay Archipelago, and Australia, and, as I shall show a little later, I believe that their mere presence loses any great significance in Tertiary zoogeography. Single species of the South American characins and cichlids have reached Texas, but in each case the migrant is a generalized, aggressive fish, the recent migration path of which is clearly evident. The cyprinodonts form a different type. The viviparous cyprinodonts of the family Poeciliidae are common from Delaware, Illinois, and Arizona to western Ecuador and northern Argentina. Unknown as fossils, they may have developed in Central America, possibly from ancestors similar to the goodeids, which are autochthonous in the Mexican plateau. The oviparous cyprinodonts of the family Cyprinodontidae likewise occur from central North America to Argentina, but it has recently been shown (Myers, 1931) that the dominant South American genera are closely related to African rather than to North American forms.¹³ At any rate, all the cyprinodonts except the North American cave fishes belong to the secondary class of fresh-water fishes.

    South America is the richest of all the continents in fresh-water fishes. The largest section of the fauna is formed by Ostariophysi. Five families of characins are found of which only one, the most generalized, is shared with any other continent (Africa). There are 11 families of catfishes, as well as the gymnotid eels, none of which are found anywhere else save for a few aggressive migrants which [[p. 349]] have pushed up into Central America. The only other South American fresh-water fishes of the primary class are the lungfishes (one species)¹⁴ and the nandids (two genera and two species).¹⁵

    Of the secondary class of fresh-water fishes, the most numerous in South America are the cichlids; the genera seem to be more diverse than the African ones, though the number of species is smaller. The cyprinodonts have already been mentioned.

    In order that the reader may appreciate the evident relationship with Africa displayed by many South American fishes, it is necessary to add a few words on the Ethiopian fish fauna. This is definitely composed of two main elements, one a group of primitive and undoubtedly ancient fish families that are unknown outside Africa either fossil or recent and that in all likelihood evolved there, and a preponderating fauna of migrants which reached and flooded the continent in later times. To the old group belong the bichirs, pantodonts, phractolaemids, kneriids,¹⁶ and mormyrids. Undoubtedly existing there with them were the old ubiquitous osteoglossids, which may have given rise to the pantodonts and of which a single species still persists today. To the later invaders belong the teeming present-day African characins (of two families, one held in common with South America), cichlids, labyrinth fishes, carps, and certain catfishes. The carps undoubtedly arrived only in the middle or late Tertiary from southern Asia; their many species have mostly not yet become generically distinct from their Asiatic relatives. The characins and most of the catfishes were undoubtedly present in Africa before the arrival of the carps. Some of the families of catfishes (electric catfishes, mochokids, and amphiliids) and one of characins (citharinids) probably evolved there, the latter certainly out of the Characidae. The Characidae, cichlids, nandids, rivuline cyprinodonts, lungfishes, and osteoglossids are held in common with South America, but it has already been noted that the last-mentioned family is probably of no great importance in this connection, and the same is possibly true of the lungfishes. It is of especial importance to note that not one of five primitive, autochthonous, African families, that were in all probability in Africa before the characins, cichlids, and carps, are held in common with South America.

[[p. 350]] 4. Central America

    The Central American fish fauna is a strange mixture of North and South American types.¹⁷ Of the purely North American fishes, the garpikes have gone farthest south, having reached Lake Nicaragua, and their invasion was likely a fairly old one. The ameiurid catfishes and the suckers have gotten as far as Guatemala, the carps to central Mexico, and the perches and sunfishes only to northern Mexico. Of the South American families, the loricariid, astroblepid, callichthyid, and pygidiid catfishes, and four of the five families of Neotropical characins have gotten only a slight hold in Panama and Costa Rica. The gymnotid eels range north to Guatemala, the pimelodid catfishes to southern Mexico and Yucatan, the cichlids to Texas and the fifth family of characins to New Mexico. None of these northern or southern invaders save the cichlids has produced many startling endemic forms in Central America. The genera are few in number, almost all identical with those in North or South America, and represent merely a few of the most aggressive frontiersmen of dominant families.

    In the cichlids, however, there has been an extremely rich flowering of species of the South American genus Cichlasoma, some of them distinct enough to be placed in different genera. But the cichlids are probably among the youngest of the coterie of fresh-water fish families, and their active recent evolution in the African lakes makes it seem probable that their numerous, closely related, Central American representatives are rather young. Certain it is that the South American Cichlasomas are less specialized than the Central American.

    It is among the top-minnows or cyprinodonts that we find the most distinctive Central American fresh-water fishes. The most primitive living genus of the oviparous family Cyprinodontidae (Profundulus) is confined to southern Mexico and Guatemala, and from ancestors not very different from it probably arose the peculiar viviparous family Goodeidae, which is practically confined to the Rio Lerma Basin in Mexico. Returning momentarily to the Cyprinodontidae, a peculiar genus (Garmanella) related closely only to a Florida form (Jordanella) occurs in Yucatan, and the remarkable Oxyzygonectes is found only in Pacific Costa Rica; both are related to northern types, as is Profundulus. Of the southern rivuline group, Rivulus alone is found, as far north as Yucatan. Of the more advanced viviparous cyprinodonts (Poeciliidae), Central America has a great profusion of endemic species and genera, many of them very peculiar. At least one subfamily (Poeciliopsinae) is practically confined to the region, overflowing slightly into Arizona and Colombia. The most highly specialized genera, as well as some of the most generalized, occur in Central [[p. 351]] America, and the poeciliid fauna is richer than either the North or South American. I suspect the family to be of fairly recent origin and it seems probable that the forms now present in the southern United States are late immigrants from the south. They like warm water too much to have enjoyed Pleistocene Florida in the company of pike or muskellunge.

5. General

    There is not a scrap of factual evidence, fossil or recent, on which to postulate a North American origin of the present South American fresh-water fishes. At this seemingly pontifical pronouncement, I can see some of the proponents of the northern-origin-for-everything theory cast their eyes toward that good old standby, the Green River Eocene of Wyoming. I am perfectly aware that the common occurrence of representatives of the existing South American family Osteoglossidae, and of the percomorph genus Priscacara, in that and other Eocene formations, has been held time and again to denote where South America got its osteoglossids and cichlids.¹⁸ I have already said that I do not believe the osteoglossids can give us much information on the Tertiary distribution of the bony fishes, because of their age and probable wide Eocene or pre-Eocene distribution. Australia was cut off from Asia before the origin of any of the families of dominant fresh-water bony fishes, yet it has a still living osteoglossid which is closer to the Green River Phareodus than the latter is to the living South American genera. This single species of Scleropages forms the total living true, fresh-water, bony fish fauna of Australia and Papua and it finds its only very close relative in another living species of Scleropages west of Wallace's Line--and this genus is the only example of the primary division of fresh-water fishes that exists on both sides of this ancient barrier. Both species of Scleropages are extremely ancient relicts of the days before Ostariophysi existed. This will, I think, make it clear why I reject the mere presence of an osteoglossid in a North American Eocene formation as evidence of a northern derivation of the South American Ostariophysi.¹⁹

    Priscacara of the Green River Eocene was described by Cope as a cichlid and is considered to be one by two students of South American fishes (Haseman, 1912; Pearson, 1937). Our foremost authority on the cichlids (Regan, 1908, p. xiv), however, decided that Priscacara was not a cichlid, and I agree with him. To me it seems most likely that the priscacarids were either sunfishes, as Regan (1916) decided [[p. 352]] they were, or a parallel group, now extinct, that possibly arose from the same marine ancestors as the cichlids. Moreover, even if Priscacara was a cichlid, it should be remembered that even today this group is not particularly averse to sea water. The genus had for its companions in the Green River lake some fishes of marine groups like Diplomystus,²⁰ gonorhynchids, and rays that frequently come up tropical rivers today. In any event, the Green River shales give us a pretty picture of the early North American fresh-water fauna before the Miocene invasion of carps.

    If the South American fishes did not come from North America, where did they originate? The Ostariophysi in particular certainly had a common origin somewhere. I must confess that I do not know, nor do I believe that anyone can unravel the history of this order without the fossil evidence still locked in pre-Tertiary rocks. Without this evidence, speculations on the earlier dispersal of the order are useless. Naturally, it can be argued that all we have against the early presence of South American groups in the north is negative evidence, but if the southern families never existed in the north, negative evidence is all we can expect to find there. Moreover, the Tertiary record of North American fishes is not a blank, and if characins and similar old and aggressive groups were present in North America during any part of the Tertiary, it is exceedingly strange that not a single fossil has come to light. There is every probability that most of the dominant endemic South American ostariophysan families are at least as old as the Eocene, and if they were present in North America at one time, some remnant of them ought to show up in such formations as the Green River. The European and Asiatic fossil evidence of the northern origin of the older African fishes is as nonexistant as the North American.

    Eigenmann, the most eminent student of the South American fishes, believed that there was a very definite pre-Tertiary continental connection between South America and Africa, to account for the similarities plainly seen in the characins, cichlids, nandids, and others, but I cannot accept his gigantic land bridge (or some of his South American paleogeography) without better geological evidence than I have seen. Regan (1922) postulates a somewhat similar bridge.

    One fact alone prevents me from believing in a wide, open, continental connection across the South Atlantic during the life of the present South American and African families of fresh-water fishes. I have mentioned the occurrence in Africa of a most remarkable assemblage of undoubtedly old families of isospondylan teleost fishes, [[p. 353]] to say nothing of the vastly more primitive bichirs; I cannot but believe that these were probably present in Africa before the cichlids, carps, or characins. If there has been a wide Tertiary or even late Cretaceous bridge across the South Atlantic, why do we find not a single, solitary representative of any of them (save again the osteoglossids) in South America? I refuse to believe that competition in South America could have killed them off; they have survived just as acute competition in Africa.

    If we are to have a South Atlantic bridge in the late Cretaceous or earliest Tertiary, the only kind I can conceive as fitting the requirements of the fishes is one like Willis's (1932) Brazil-Guinea isthmus in the Atlantic, and from Schuchert's data, it may have still been partly in existence up to the very end of the Mesozoic. If our fishes were present in the earliest Eocene, they may have been there at the end of the Cretaceous. Such a narrow isthmian connection, with its short, swift rivers, would not provide a broad highway for all the fresh-water fishes, but it would allow to pass the very same aggressive types that are now held in common by Africa and South America. The lowland, slow-water mormyrids, pantodonts, kneriids, and bichirs might very well be kept out of South America while more active, swift-water fishes passed. Again, the isthmus may not have been connected at both ends at the time the teleostean fishes came on the scene, and the transfer may have been alternate, or only one-way, but this seems unlikely. Any way one looks at it, the weak little lowland nandids, present in both continents, are problematic if one supposes a bridge that excluded mormyrids.

    In any event, the fish evidence indicates there has been no South Atlantic connection since the Eocene. I have carefully compared the entire external anatomy and osteology of the African and South American characins that are supposed to show the closest intercontinental relationship (the African Brycinus and Alestes and the American Brycon). My conclusion is that both the African and American fishes are closely similar only because each is a generalized, dominant, and aggressive animal of a type that has probably changed little since the more modern types of characins originated. Moreover, in my studies of the cyprinodonts (Myers, 1931) I have compared the difficult-to-separate South American Rivulus and African Aphyosemion. They both belong to a rather specialized group of genera (the tribe Rivulini), and are undoubtedly closely related, but it should not be overlooked that they belong to the secondary division of fresh-water fishes, and that a fortuitous marine dispersal of one or more species at some time in the Tertiary is not impossible, particularly if a Brazil-Guinea ridge remained for a time as an island chain.

    [[p. 354]] What I have said will, I believe, make it clear that our present knowledge of the fishes very distinctly favors a late Mesozoic or very early Tertiary South Atlantic land connection and makes a direct northern origin of at least the South American ostariophysans seem exceedingly unlikely. But I refuse to take a definite stand on these questions. It would be extremely presumptious, on the basis of the fishes alone, to attempt a flat contradiction of the Holarctic dispersal of the mammals, reptiles, and amphibians so ably advocated by Matthew (1915), Noble (1925), and Dunn (1923 and 1931). Our knowledge of fossil fresh-water fishes, especially in South America and Africa, is very meager and I realize how quickly the discovery of critical fossil evidence night change the picture. However, I do believe that, fossils aside, the real fresh-water fishes offer better, clearer, and vastly more conservative continental zoogeographical evidence than any one of the classes of quadrupeds, and their basic classification is far better understood than that of the frogs and perhaps of the reptiles. The mammals take precedence only because of the more abundant fossil record. My chief contention is that the proponents of Holarctic dispersal have given too little attention to contrary conclusions in other groups and have, perhaps, ridden along on the coattails of the mammal evidence a little more easily than the evidence of their own groups actually warrants.

    The true explanation of the apparent conflict between the fish and quadruped evidence may lie in a direction that I have not seen pointed out. If the epi-Mesozoic interval were a time of great uplift and denudation, of longer continental duration than generally supposed, it may have been long enough for the differentiation of characins and other primitive ostariophysans in the north, the southward dispersal of a few types of ancestral bagrid-pimelodid catfishes and generalized characins into Africa and South America, and the extinction of these in Holarctica. We do know that much of the record of this period has been lost. But this is pure speculation, and, in the words of Regan, "The * * * view that Ostariophysi originated in the north and spread southwards, involves so many improbabilities as to be almost unbelievable."

    Returning to the North American fishes, it may be said that there is no particular argument about them. They, at least, seem to have originated in the north, either in North America or the often-connected Eurasia. The South American ones are a problem, and I leave them thus, up in the air, so to speak, for it will be remembered that I promised to come to no conclusions.

WEST INDIAN ZOOGEOGRAPHY²¹

    The geological history of the land and fresh-water vertebrate animals of the West Indies before the Pleistocene is still almost [[p. 355]] unknown. It is therefore pertinent to point out that finely spun schemes of distribution based entirely on the living (or Pleistocene) fauna are likely to receive rough treatment when and if good fossil evidence is found. What is known of the fresh-water fishes of the islands, though instructive, will of itself answer no important question conclusively, and I shall therefore give only a brief and very sketchy summary of one or two of the problems which the fishes may help us to solve.

    Much has been published on West Indian zoogeography. For my purposes, I refer only to Matthew (1915 and 1916), Barbour (1914 and 1916), Anthony (1918), Scharff (1922), Schmidt (1928), and Dunn (1934). From the bibliographies of these papers some idea of the other literature may be gained.

    Except for Matthew, most of these writers are rather definitely in favor of a union of the Greater Antilles at some time during the Tertiary, and a continental connection of this mass or one of its elements with the North, Central, or South American mainland at some period from the late Mesozoic to the middle Tertiary. I admit that the evidence these men present is both enticing and impressive. The distribution of the amphibians is the most impressive to me, perhaps because I am more familiar with that group than with other quadrupeds. Since the time of Darwin the natural occurrence of amphibians on an island has been almost universally accepted as incontrovertible evidence that that island has had a continental land connection. Amphibians are delicate creatures, extremely sensitive to desiccation and to salt water, and since they do not possess wings, it is difficult to imagine how they could possibly cross a body of sea water by any natural means. The West Indies, especially the Greater Antilles, are extremely well supplied with several genera of amphibians. Indeed it would seem that one of the chief diversions of contemporary American herpetologists is the describing of new West Indian Eleutherodactyli! If I am not believed, I would refer my questioner to Dr. Barbour's recent list (1935).

    Against this evidence, the late Dr. Matthew has stood out practically alone in maintaining that the West Indies are true oceanic islands in that they have never had a continental connection. He holds that the entire terrestrial fauna of these islands is a waif fauna--one gained entirely through fortuitous methods of dispersal such as winds and drifting debris. It may be said that many biologists are coming to appreciate that an island can pick up an astonishing number of vertebrate and invertebrate animals in this manner. Great tropical rivers continually carry floating masses of vegetation, sometimes of considerable extent, out to sea, and all sorts of creatures are known to have been floated away to no one knows what fate on rafts of this type. It is certain that the populations of these rafts [[p. 356]] usually meet their end in rough water or storms soon after leaving the rivers, but undoubtedly in rare instances natural rafts are deposited on distant shores with at least some of their crews in a viable state. It is entirely conceivable that even tree frogs, which form the greater part of the West Indian amphibian fauna, might occasionally become unwilling sailors. Nor is it only rafts from rivers that need be considered. Almost any kind of flotsam or jetsam of any size is liable to be tenanted by some land creature, hanging on grimly, if he is able, until the sea swallows him up. The millions and millions of years of geological time surely allow enough for the (comparatively) frequent washing up of rafts on distant shores and it is beyond question that many islands have gotten their present faunas in this way. But, I will remind you, nobody has ever attempted to explain the distribution of freshwater fishes by the raft method!

    Great paleogeographical interest attaches to the fresh-water fishes of the West Indies. Let us see what they tell us.

WEST INDIAN FRESH-WATER FISHES

    The most striking feature of the fresh-water fish fauna of the West Indies is the complete absence of members of the primary division of fresh-water fishes, in particular the Ostariophysi, which swarm in all the waters of North, Central, and South America. Every West Indian fish to which the adjective "fresh-water" could possibly apply belongs to my secondary division or to groups still more partial to sea water. In this, the West Indies closely parallel Madagascar, where not only is the primary division entirely absent, but also the secondary ones that are present belong largely to the same systematic groups as do those of the West Indies.

    To begin with, the rivers of the West Indian islands harbor a number of fishes, belonging to purely or partly marine families, which come up from the sea. There are gobies and eels and silversides and gray mullets, some of which enter fresh water only occasionally and some of which live in the rivers most of the time, returning to the sea only to spawn. One of the most conspicuous and widely-known of West Indian fresh-water fishes belongs to this category. This is the mountain mullet (Agonostomus), known in the Spanish islands as dajao or lisa. There are three West Indian species, the largest reaching a length of a foot, and others occur along the Caribbean coasts of South America, in Central America, and the Galapagos.²² A related genus (Joturus) lives in Cuba and Central America. It is not known whether the mountain mullets return to the sea to spawn, but we do know that they belong to the marine family of gray mullets, and that they and their companions, the eels, gobies, [[p. 357]] and others, do not obey the distributional rules of true fresh-water fishes. They are of no use to us in our present problem. Neither are the two blind brotulids of the Cuban caves. They are the only known fresh-water representatives of a primarily deep-water marine family.

    The fishes which really belong to our secondary division of fresh-water fishes are referable to five families, the garpikes, the synbranchid eels (Synbranchidae), the viviparous cyprinodonts (Poeciliidae), and the cichlids (Cichlidae). It may be best to survey them briefly to make clear our discussion of the faunas of the particular islands.

    One large garpike, Lepidosteus tristoechus, occurs in Cuba. It is generally supposed to be identical with the alligator gar of the southern United States, but this is questionable. If the Cuban gar is really different, it has probably been in the island a long time, but garpikes are reported to enter salt water.

    Of the synbranchid eels, only one species (Synbranchus marmoratus) is West Indian. It is the only American form of the family so far known and it ranges from Veracruz to Argentina on the mainland. A single species of the same genus occurs in Africa and another in India; the Indian one, at least, is known to frequent brackish water.

    I have already mentioned the cichlids in connection with South and Central American fishes. Three Antillean species are known, all closely related species of Cichlasoma (Myers, 1928; Tee-Van, 1935).

    If the two records of West Indian Fundulus are mythical, as I believe they may be, the oviparous cyprinodonts of the Antilles belong to only three genera. Cyprinodon is a genus of the southern and eastern United States, northern and eastern Mexico, and certain Caribbean islands. The seaboard continental species always occur in brackish or salt water, and the inland ones are partial to the alkaline waters of desert pools. The same habits are exhibited by the island forms. Rivulus, on the other hand, belongs to a South American group²³ that has a general inland distribution as far north as Yucatan and the Rio Papaloapan. It has been taken in tide pools at Curaçao. The peculiar Cubanichthys is found only in Cuba.

    The viviparous cyprinodonts form by far the largest proportion of the Antillean fresh-water fish fauna. They belong to two subfamilies, Gambusiinae and Poeciliinae, and of the first, Cuba possesses an endemic tribe composed of five distinctive genera.²⁴ The other Antillean Gambusiinae are all members of the widespread Central and North American genus Gambusia. Of the Poeciliinae there is one endemic West Indian genus, Limia, and three others (Poecilia, [[p. 358]] Lebistes, and Mollienisia) that are shared with the mainland of South or Central America.

    The Lesser Antilles may be said to possess no fresh-water fishes at all. I hasten to say that I do not include Trinidad. That island is merely a recently separated part of the mainland. Most of its many fresh-water fishes are specifically identical with Orinoco and Guiana species, and those that now appear to be peculiar may confidently be expected to turn up when the Orinoco delta is carefully fished. All the Lesser Antilles that have sizable streams seem to have the semimarine mountain mullets and gobies but the only fishes that we could really call fresh-water ones in the whole chain are two poeciliids, the "guppy" or "millions" (Lebistes) and Poecilia vivipara. The latter is recorded from Martinique and the "guppy" from St. Lucia and Barbados. It is probable that both occur in other islands of the Leeward group, but I am not wholly convinced that these two tiny and very prolific fishes were not introduced by man. In late years the "guppy" has been spread far and wide in antimalarial work. Both are admirably fitted for waif distribution through windstorms or even actual navigation of small stretches of ocean. They are tiny lowland fishes of the fresh-water tidal belt, and on the continent never occur far from the coast. I have observed their hardiness in strong salt water in aquaria on several occasions, although they cannot exist for any very extended period in sea water. The deposition of one pregnant female in an island stream would soon fully populate that island. It is therefore evident that there is no fish evidence to support a claim of continental union of the Lesser Antilles, or to give us a hint of the histories of individual islands.

    The Virgin Islands, so far as fresh-water fishes go, belong zoologically with the Lesser Antilles; there are no fresh-water fishes. A Fundulus has been described from them by Fowler from the old van Rijgersma collection, but the specimens are only doubtfully distinct from a common North American brackish water killyfish (Fundulus hereroclitus). If a Fundulus of this type were really present, it should have spread throughout a good part of the Antilles.

    Cuba has the most distinctive fauna.²⁵ There are the garpike and Synbranchus, neither of which is known from any of the other islands. A distinctive cichlid, related rather closely to several species in southern Mexico, is very common. Among the oviparous cyprinodonts there is one Rivulus, related to Central American forms, and one or two subspecies of a Cyprinodon which is common in brackish water in Florida. The most interesting member of the group is the little endemic Cubanichthys cubensis which finds its only near [[p. 359]] relative in the Florida fresh-water cyprinodont Chriopeops goodei. With the gar and Cyprinodon, this forms the only element in the entire West Indian fresh-water fish fauna that points toward a North American connection. Cyprinodon certainly came by sea or wind, and the gar is not really good evidence until we know more about it. I consider that the weak little Cubanichthys and Chriopeops, forming by themselves a group not particularly close to other genera, may have had relatives, now extinct, in Central America. The most striking element in the Cuban fish fauna are the five genera which form the poeciliid tribe Girardinini.²⁶ They form a very distinctive group, distantly related to Central American types, and must have arisen in Cuba a long time ago. Gambusia, with two Cuban species, belongs to the same subfamily, but to a widespread tribe. There is one Cuban Limia, a West Indian genus of another subfamily.

    Hispaniola (Haiti and Santo Domingo) has only cichlids and cyprinodonts.²⁷ The single well-known cichlid is almost identical with the Cuban one; a doubtful species (not yet named) is known from a single specimen. But there is a fossil cichlid from the Miocene differing from the common living one only by a couple of vertebrae--the only fossil of the present Antillean fresh-water fishes yet discovered. It would seem, therefore, that cichlids have been in Hispaniola since the Miocene. The only oviparous cyprinodonts reported from the island are a large Cyprinodon, living in the Haitian salt lakes, and a Rivulus based on one specimen captured under peculiar circumstances on Saona Island off the southeast coast. The viviparous forms belong to the ubiquitous genus Gambusia, to Limia, which is known outside Hispaniola only by the Cuban and Jamaican species, and to Mollienisia, a genus known elsewhere only on the mainland. There are three Gambusias, related to Cuban and Jamaican species. Limia, with eight species in the island, probably originated there; it is close to Poecilia. The Hispaniolan Mollienisia is an isolated form, not at all close to the Florida one, and perhaps finds its closest relationship in the species of the coast of Colombia or Panama.

    Jamaica has only viviparous cyprinodonts. Single old records of a South American catfish and a Central American cichlid are probably mistakes, but I should not be surprised to see a real cichlid turn up there. The cyprinodonts are four Gambusias and two Limias related to Hispaniolan ones.

    Grand Cayman has one Gambusia.

    [[p. 360]] Puerto Rico is the most remarkable of the Greater Antilles. In spite of the fact that its higher vertebrate fauna is very comparable to that of Hispaniola, its sole and only fresh-water fish is the viviparous cyprinodont Poecilia vivipara, which is common in South America and the Leeward Islands. That this strange absence of anything we might expect is real and not merely apparent is shown by Dr. Hildebrand's recent extensive survey of the Puerto Rican streams.²⁸ I am strongly of the opinion that the Poecilia got there by fortuitous means, or by the hand of man. Many years ago Cuvier and Valenciennes described a Fundulus from Puerto Rico, but it has remained known from the single type specimen in Paris, since the other type specimens were shown to be a Gambusia. I do not believe that a Fundulus exists in the island. Where Nichols (1929) got his figure of it I cannot say.

WHAT DO THE ANTILLEAN FRESH-WATER FISHES INDICATE?

    It is difficult to say just what the fishes indicate. The distribution of several of them is so peculiar as to suggest that their arrival has been fortuitous and not dependent on dry-land connections. Mollienisia in Hispaniola, Poecilia in Puerto Rico and the Leewards, and Cubanichthys in Cuba are examples in point. The arrival of one Limia in Cuba and the presence of scarcely different cichlids in Cuba and Hispaniola might signify ocean navigation by these fishes, and I think this not unlikely. However, the presence of girardinine poeciliids in Cuba, their absence in Hispaniola and Jamaica, and the close relationships of the Gambusias and Limias of the last two islands is instructive. I do not think Cuba and Hispaniola have been united during the probably long existence of the girardinines, but Jamaica and Hispaniola may have been. Puerto Rico, from the fish evidence, seems to have had a long separate history.

    Of course the most remarkable thing of all is the entire absence of Ostariophysi in the West Indies. On the face of it, the fish evidence therefore points strongly toward a lack of any mainland connection during a considerable part of the Tertiary. Personally, I am of the opinion that some northern and a number of southern Ostariophysi have been in the process of penetration of Central America for a long time, and that their absence in the islands is significant. But another alternative must not be lost sight of. Scharff, Schmidt, and others have mentioned the fact that the land area available seems to have a direct bearing on the survival of at least parts of a land fauna--in other words that an island, through partial submergence, might become too small to permit the survival of certain species. I do not at the moment recall having seen this idea applied to the fresh-water fishes of islands, and I think it will repay us to digress a bit to discuss it.

    [[p. 361]] Although ichthyologists have never particularly remarked it, it is a fact that all of the primitive, relict forms of fresh-water fishes that have persisted to modern times are inhabitants of relatively sluggish (and usually large) lowland water systems. The late Professor Eigenmann overlooked this when he journeyed to the Guiana plateau in 1908 to search for possible survivors of the fish fauna of his "Archiguiana"; he found none. Changes in the physical features of mountain areas are too swift to permit survival of any except fishes that have become peculiarly adapted to life in swift water, and the few relicts of a dying race are not the ones that become adapted to such a hard life. Peneplaination of a mountain range means the extinction of the highly adapted hill-stream fishes, since (despite Regan's case of the astroblepids) evolution is not exactly reversible and specialized swift-water fishes could scarcely be expected to revert to a slow-water habitat deficient in oxygen.

    The only instances I know of true relict fishes living in mountainous areas concern species existing in mountain lakes that have been elevated bodily, or which are part of a once extensive plateau lake or river system, and where the certain draining of the lake in the not far distant geological future will sound the death knell of the relict. An especially good example is seen in the remarkable fish Chaudhuria in the Inlé Lake, Burma.

    It may be expected that a considerable proportion of a mainland fish fauna living on land masses (islands) subsequently cut off from a continent will be lowland fishes not particularly well adapted to swift water. If the island were to be partially submerged within a comparatively short geological period, the lowland fish fauna might be entirely annihilated. This would be all the more likely if the original continental connection had been so low as to permit the entrance only of lowland types and the submergence occurred before these lowland fishes had had time to evolve hill-stream types.

    We do know that many of the Antilles have experienced considerable changes of elevation. Barbados in particular appears to have been very badly treated by orogenic or other forces. If we can believe the evidence of its sedimentary deposits, it sank 6,000 to 10,000 feet below the sea between the Eocene and Miocene, and in the Pliocene bobbed up again as an island. Naturally, complete subsidence below the sea would destroy the fresh-water fauna, and the Greater Antilles have experienced no such devastating changes of level, but even relatively slight subsidence might have a profound effect on the river fishes.

    In spite of this, I feel confident that had aggressive Ostariophysi ever been able to reach the Greater Antilles, nothing short of almost complete submergence of the larger islands would have entirely destroyed them. Characins and carps in particular would have been [[p. 362]] especially adaptable to hill-stream life, and their vigor and fecundity, especially in the absence of competitors, would have ensured their survival under almost any natural conditions short of complete desiccation or immersion in sea water.

    If this view be granted, the only conceivable continental connection of a Greater Antillean land mass is one with Central America, at a time when neither the North American nor the South American Ostariophysi had invaded much of Middle America. I have already shown that we cannot date these invasions. Finally, if we are to suppose that all the South American Ostariophysi originally wended their way southward through Central America, I believe we should have to push any such continental bridge back into the Mesozoic, if indeed it ever existed at all.

LITERATURE

VOIGT, ERHARD.
1934. Die Fische aus der mitteleozänen Braunkohle des Geiseltales, mit besonderer Berücksichtigung der erhaltenen Weichteile. Nova Acta Leopoldina, Halle, Neue Folge, Bd. 2, Heft 1-2, pp. 21-146, pls. 1-14.

WILLIS, BAILEY.
1932. Isthmian links. Bull. Geol. Soc. Amer., vol. 43, pp. 917-952, pls. 25-29.

EXPLANATION OF PLATES

Plate 1 [[unnumbered page]]

Plate 2 [[unnumbered page]]

Plate 3 [[unnumbered page]]

_________________________

Notes Appearing in the Original Work

¹For which I had gathered the ichthyological information. [[on p. 341]]
²Except as one of the factors of natural selection. [[on p. 342]]
³The Nilotic Cromeriidae seem to be larval Kneriidae. [[on p. 344]]
⁴See Regan (1922). A few characins, carps, and fresh-water catfishes enter brackish water in estuaries, but none is known to be able to survive in the sea for more than a few hours; none could breed there. The only exception of which I know is the carp Acahara hakonensis, which Prof. J. O. Snyder tells me is occasionally taken in the open sea off southern Japan. [[on p. 344]]
⁵It seems probable that the supposed Bourbon catfish Laimumena, close to certain South American trachycorystids, is based on mislabeled type specimens. [[on p. 345]]
⁶It seems probable that the presence of the air-breathing climbing perch (Anabas) east of Wallace's Line is due to human introduction. Anabas is much carried about as a food fish. As to the osteoglossids, see further on. [[on p. 346]]
⁷Etroplus, the only Indian cichlid, is confined to the southern Indian Peninsula and is closely related only to certain Madagascan species; it perhaps arrived in India by sea via the Seychelles-Chagos-Maldive chain. [[on p. 346]]
⁸Regan (1922). Hussakof (1932) records fin spines of Rhineastes from the Pliocene of Mongolia. If these are really generically identical with the skulls from the Bridger and Green River, they indicate that the ameiurids were a Holarctic group. [[on p. 347]]
⁹Since this paper was written, Schlaikjer (1937) has described a sunfish from an unplaced Tertiary formation in Alaska. His attempt to see similarities to the sunfishes in certain Asiatic fresh-water serranids is scarcely convincing, and his supposed fossil sucker, of which practically no description is given, is probably a carp. If Regan (1916) is correct in referring Priscacara to the sunfishes, the family has existed in North America since the Eocene. [[on p. 347]]
¹⁰For the present purpose I include in the Umbridae the peculiar relict Novumbra, recently discovered by Schultz in western Washington. The Alaskan and Siberian blackfish is of little importance to our discussion. [[on p. 347]]
¹¹I have examined a large jaw of Esox lucius or E. masquinongy from the Pleistocene of Florida (specimen in U. S. National Museum). [[on p. 347]]
¹²The peculiar, depauperate, Patagonian fauna is not here considered. Its only close relationship, as shown by the lampreys, galaxiids, and aplochitonids, is with New Zealand and South Australia, and this points to either an Antarctic connection or sea migration, which I do not think impossible for any of the three groups. [[on p. 348]]
¹³Except for the semimarine subfamily Cyprinodontinae of southern North America and the Caribbean, of which Carrionellus is known from the Tertiary of Ecuador. The exact relationship of the autochthonous Orestiatinae of Lake Titicaca is unknown. [[on p. 348]]
¹⁴Lepidosiren. The only other living species of Lepidosirenidae are the three Protopterus of Africa. The Australian Neoceratodus is very different. [[on p. 349]]
¹⁵Polycentrus and Monocirrhus are closely related to the Nigerian Polycentropsis and the Indian Nandus. The Indian Badis is very distinctive and probably should form a separate family. It is close to neither Nandus nor Pristolepis. [[on p. 349]]
¹⁶Following up Pellegrin's recent comparison of the peculiar Nilotic cromeriids with larval albulids, and his hesitant suggestion of kneriid relationship, I shall be greatly surprised if Cromeria is found to be anything but a larval Kneria. [[on p. 349]]
¹⁷See especially Regan (1908) for a discussion. [[on p. 350]]
¹⁸See especially Matthew (1915, p. 298). Matthew's reasoning, and information, is here very faulty. Lepidosteus, which he in some way imagined to be a Neotropical group, is Holarctic. Arius is a marine ariid not related either to the South American pimelodid Phractocephalus or to Rhineastes of the Bridger, which belongs with the living North American Ameiurids (see Regan, 1922). [[on p. 351]]
¹⁹Fossil osteoglossids are known, outside the Green River, from the probably Eocene Mergelschiefer (Sanders, 1934) of Sumatra (Musperia and Scleropages). Brychaetus is known from the marine Eocene London Clay and may not be an osteoglossid. [[on p. 351]]
²⁰The rough-backed Eocene herrings of the genera Diplomystus and Knightia have living marine relatives on the coasts of Chile (Ethmidium) and Australia (Hyperlophus). Diplomystus should not be confused with the Chilean fresh-water Diplomystes, the most primitive known catfish. [[on p. 352]]
²¹I am entirely incompetent to handle the invertebrate evidence and have not referred to it. [[on p. 354]]
²²Regan (1908) has revised the American species. [[on p. 356]]
²³See Myers (1931). [[on p. 357]]
²⁴See Hubbs (1924). The new Puerto Rican genus and species, Neopoecilia holacanthus, described in this paper, was later synonymized with Poecilia vivipara by Hubbs. [[on p. 357]]
²⁵See especially Eigenmann (1903). Eigenmann's several forms of Cuban cichlids have been synonymized by Hubbs (see Myers, 1928). Hubbs (1924) has revised Eigenmann's classification of the viviparous poeciliids, and more recently (1934) described the peculiar Quintana, now known from both Western Cuba and the Isle of Pines. Further revisional work by Dr. Hubbs and Dr. Howell Rivero is in preparation. [[on p. 358]]
²⁶Hubbs (1924 and 1934). [[on p. 359]]
²⁷See Tee-Van (1935) and Myers (1928 and 1935). [[on p. 359]]
²⁸Hildebrand (1935). [[on p. 360]]

*                 *                 *                 *                 *

Return to Home