Antarctica as a Faunal Migration Route

by George Gaylord Simpson (1940)

Editor Charles H. Smith's Note: A paper presented at the Sixth Pacific Science Congress. Original pagination indicated within double brackets. Notes are numbered sequentially and grouped at the end, with the page(s) they originally appeared at the bottom of given within double brackets. My thanks to the Pacific Science Association for permitting the reprint. Citation: Proceedings of the Sixth Pacific Science Congress of the Pacific Science Association (Berkeley & Los Angeles: Univ. of California Press, 1940), Vol. 2: 755-768.

    [[p. 755]] Every student of faunal and floral distribution has noticed that many nonmarine plants and animals are disjunctively distributed. A given form or group may find its apparently closest allies not in an adjacent region, but in some distant land, often separated by long stretches of sea or by other barriers. Some of the most striking of these disjunctive distributions occur in the Southern Hemisphere. To account for the distribution of life in the southern continents it is necessary to assume that various groups of plants and animals have somehow managed to migrate between these land areas without leaving any clear or permanent sign of their passage.

    As long ago as 1847 the botanist J. D. Hooker, in working up the results of the voyage of the "Erebus" and "Terror," suggested that these anomalies might be explained by the former existence of a continuous land connection between the other southern continents by way of Antarctica. This possibility has had a powerful effect on the scientific imagination. A recent partial bibliography (Wittmann, 1934) includes 350 titles, and an exhaustive list would not fall much short of 1000. The idea of an Antarctic migration route has been opposed as strongly as it has been supported, and it has become one of the major disputed problems of earth history and biogeography.

    So multiple and varied are the facts and conjectures that have been published in this field that judicious selection and emphasis of them can be made to support almost any opinion not completely irrational. No one person can hope to know at first hand all the pertinent data, and a general review of the literature leaves one with the feeling that it can be taken to prove any of a dozen conflicting theories and that it therefore proves nothing.

    The discussion seems to have reached a stalemate. New facts might break this stalemate at any time. For instance, the discovery of a single fossil mammal tooth in Antarctica could at once settle some of the most disputed aspects of the problem, although of course no single fact could solve it integrally. Such a decisive discovery has not been made and apparently no one is willing to await it before forming some sort of opinion. Indeed it may be futile to expect any such simple solution, and the nondiscovery of such evidence may well be caused by its nonexistence.

     In the meantime some progress can be made toward reducing this chaos to order and toward evaluating the various theories. This cannot be done simply by the literary device of amassing all the facts, supposed facts, and opinions, and attempting to strike a balance. This has repeatedly been attempted, most recently by Wittmann (1934, see references), whose condensed but fairly thorough review makes repetition by this method unnecessary at present. It is true that Wittmann, like most of the other compilers, reaches a conclusion or at least expresses a preference, but his own citations show that his preference does not really follow clearly or necessarily from his data. More progress can perhaps be made in the following ways:

    1. By exact definition of the problem, which includes critical selection of the instances of disjunctive distribution to be explained and statement of the various hypotheses that might explain them.

    2. By consideration of the question of authority and discarding of incorrect or inadequate data and the opinions based on them in favor the best-informed and most logical authoritative views in each field of evidence.

     [[p. 756]] 3. By bringing into the discussion all new and pertinent facts as they are discovered.

    4. By reconsidering the means of interpreting these facts and revaluing them in the light of consistent and scientific interpretive methodology.

    A brief review along these lines, without lengthy exposition of the data available in each field, must in part be critical in a destructive sense. By weeding out the material not truly pertinent and the authority and opinions considered least reliable, a constructive and positive conclusion can nevertheless be reached if the result is to leave, as I believe it will, a consensus and a set of hypotheses that may be taken as definitely more acceptable than those adversely criticized.


     The whole question of Antarctica as a migration route arises from attempts to explain examples of disjunctive distribution of groups known only, or mainly, from the Southern Hemisphere. A point of departure, at least, is well established: types of land plants and animals do occur in two or more southern regions, for instance, in Australia and South America, and appear to be more closely similar or related to each other than they are to plants and animals of other (i.e. of northern) regions. This is an apparent factual anomaly and requires explanation. The principal biological hypotheses that arise are:

    1. Structurally closely related forms of life may arise spontaneously, by processes other than those familiar as convergence and parallelism, in distinct places. I do not propose to discuss this ancient hypothesis seriously, although it has recently reappeared in somewhat more subtle form among certain European students who do not appear to me to have any grasp of the facts that they attempt to explain.

    2. The apparently related forms may not be such in fact, or not to such a degree as to be anomalous under the given circumstances. They may merely be convergent or parallel forms that have arisen independently in the two regions from separate stocks or from a common ancestry that existed in intervening, northern lands. Many of the supposed items of evidence for Antarctic migration are now known to be of this sort and hence not to demand or even to suggest such a route. Examples of animals that are often used to support the theory of an Antarctic route, but that probably represent convergence and parallelism from stocks that do not suggest any such route, will be mentioned later. Despite its wide emphasis, such evidence really has no value for this problem and may be discarded in our consideration of it. It is, however, true that by no means all the apparent instances of southern disjunctive distribution can be so explained.

    3. The forms of life in question may really be specially related and may have migrated more or less directly between the various southern lands, but may have done so by means other than a definite land connection. It is an established fact that both land plants and land animals cross considerable stretches of water by floating, by carriage on natural rafts, or by wind (plant spores and seeds, or animals that fly). There are also animals, including many fishes and some reptiles and mammals, that normally live in fresh water or on land, but that are capable of prolonged sojourn in the sea and are therefore capable of being carried across the ocean without a land bridge. Some of the best instances of southern disjunctive distribution belong to this class, for example the fish Galaxias, the meiolaniid turtles, and the sea cows. Some of the adherents of the Antarctic route complain that this process, however possible, is improbable, that repeated improbabilities may amount to a [[p. 757]] practical impossibility, and that their opponents use this sort of distribution as a deus ex machina to solve all their difficulties. To some degree this is true; but the process is often less improbable than it seems--for many forms of life, even highly probable. Evidence of plants and animals that might be placed in this class is to be considered as pertinent to the Antarctic problem, but the probabilities of such means of distribution must be reconsidered and it must be decided whether there are any plants and animals of this group for which Antarctic migration is definitely more probable than nonterrestrial migration.

    4. The common ancestors of the disjunctively distributed forms of life may have lived but may now have become extinct in regions forming a continuous connection between the areas where the forms now occur. This is, of course, the real basis of the problem of the Antarctic route, consideration of which is very much clarified by recognition of the fact that much of the supposed evidence is not really of this crucial type but belongs rather to the nonpertinent group (2) or to the dubious group (3) of this list.

    Animals that migrate by land and that cannot, within reason, be distributed in any other way are very numerous, but among them clear-cut examples of southern disjunctive distribution are rare. This is in itself a suspicious circumstance: that so few of the apparent examples of disjunctive distribution absolutely require a land bridge, and that so few of the animals that absolutely require a land bridge are disjunctively distributed.

    That the southern faunas as a whole do require land connections of some sort is obvious. As to the nature of these connections, there are four main hypotheses:

    1. That the southern lands (at least South America and Australia and possibly also Africa and New Zealand) were once connected by land bridges to Antarctica, and hence with each other through that continent.

    2. That they were once all part of the same land mass, which also included Antarctica and which has since been split up by continental drifting.

    3. That they were connected with each other more directly by transoceanic bridges, not necessarily involving Antarctica.

    4. That they were connected separately to the northern continents, Australia with eastern Asia, Africa with Europe and western Asia, and South America with North America, and the northern continents were united with each other.

    These hypotheses are not mutually exclusive. It would be possible for two or more of them to be true successively or simultaneously.

    Only the first two hypotheses are being examined directly here, since only they involve Antarctica, and of these the first is of primary importance. The second, which is that of Wegener, seems to me and to a consensus of geologists ill supported and improbable. In any case, for present purposes it suffices to say that the faunal distributional evidence for it should be of the same sort as that for the first hypothesis, but should be decidedly stronger and more obvious because such an intimate connection should produce not merely some faunal similarities, as Antarctic land bridges might, but practically identical faunas. The scattered and incomplete nature of the evidence for either view thus does not make the first impossible, but is almost conclusively opposed to the second.

    The third hypothesis, of transoceanic, non-Antarctic bridges, is peculiar in that even if a conclusive case for southern faunal migration routes were established, the absence of direct evidence in Antarctica itself would make this hypothesis as adequate to explain the facts as the hypothesis of Antarctic connections. Thus, the biological evidence alone can hardly permit an absolute choice between these two, and some adherents of the southern land bridge [[p. 758]] theory (e.g., von Huene, 1929) have rejected Antarctica as part of the supposed bridge. In common with most geologists and zoölogists, however, I feel that the idea of a transpacific bridge from Asia and Australia to South America is so nearly impossible on other grounds that if a southern bridge were established as probable I would assume, until the appearance of further evidence to the contrary, that it involved Antarctica.1

    The fourth hypothesis, of northern connections, is in part no longer hypothetical, but a well-supported theory or indeed a fact. Obviously, South America and Africa are so connected at present, and it is all but certain that they have been so connected at various times in the more remote past. The connection of Australia with Asia has been denied, but is also almost certain--perhaps not a continuous land connection, but at least insular steppingstones that permitted the entrance of some Asiatic animals and plants into Australia. The former and frequent land connection of Eurasia and North America is definitely proved by the fossil record. Since this series of connections is almost certain and is required in any case, the principle of parsimony demands that no additional hypothesis be considered necessary or acceptable unless it clearly explains facts that are difficult or impossible to explain by these connections alone.

    In further delimitation of the problem, it should be stated that the question of Gondwana Land, a South American-African-Southern Asiatic connection in the Mesozoic and earlier, does not involve Antarctica, is a different or additional and not an alternative hypothesis, and so does not enter into the present discussion. The question of possible very remote Antarctic connections, in the Jurassic or earlier, is also excluded from discussion here because almost all the usual evidence is from groups of plants and animals that must have migrated in the later Mesozoic or Tertiary. The probability that Antarctica was connected with South America, at least, at some period without serving as an important land-migration route is also beside the point here.

    In the present state of knowledge the best form in which to state the question appears to be this: Are there any known facts of the distribution of recent and fossil plants and animals that make it necessary or preferable to suppose that Antarctica was involved in a land-migration route between the other southern lands, or can all these facts be adequately and more probably explained by other routes for which there is more direct evidence?


    It is not possible, and in view of the summaries of Wittmann and others it is not desirable, to attempt a review of all the evidence here. Some of the supposedly crucial evidence from each class of the vertebrates will be very briefly discussed, less for its own sake than to illustrate the conflicts of authorities and of interpretive methods and to show how a more detailed review (which I have made and on which these notes are based) can point definitely to one conclusion despite these conflicts.


    The eel-gudgeon Galaxias, a predominantly freshwater fish, occurs in South America, South Africa, Australia, and New Zealand. As early as 1905, Tate Regan pointed out that Galaxias freely enters salt water and can live in it indefinitely, that it is probably of marine ancestry, and that it therefore gives no evidence of land connections. This is excellent authority and this opinion has been shared by a clear consensus of ichthyologists ever since. Nevertheless, [[p. 759]] Galaxias is usually cited by adherents of Antarctic bridges as evidence for their view.

    So far as any reason can be given for this disregard of authority and consensus, it is to be found in the fact that there is a partially conflicting authority, that of Eigenmann (1909), who agreed that Galaxias might migrate by marine routes but held this to be highly improbable. This sort of judgment of probability occurs again and again in dealing with the present problem and so merits comment.

    It is indeed "highly improbable" that a given fish (or pair) should cross an ocean and colonize waters on the other side at any given time. The chances of occurrence (at a single trial) are extremely small, but probability does not depend solely on chances of occurrence, but also on opportunities for occurrence. The chances of throwing five aces with five dice in one throw are negligible, but if the opportunities for occurrence are increased, for instance by throwing one hundred dice instead of five or by throwing ten thousand times instead of once, this "highly improbable" event becomes probable and may even become certain, for all practical purposes. So with difficult migrations, such as that of Galaxias across the ocean. The great number of individual animals involved, usually thousands or millions, and the long span of time involved, often millions of years, give so many opportunities for occurrence that the "improbable" event becomes highly probable as long as the basic chance is real and finite, as it is granted to be for Galaxias. There is never an absolute certainty that the migration will be accomplished, and its time of occurrence is random--peculiarities that have a definite bearing on animal history, as will be mentioned again in discussing mammals.

The So-Called Leptodactylid Amphibians

    Certain amphibians, toothed toads, are frequently placed in a single family, Leptodactylidae or Cystignathidae, which occurs only in the southern continents. These forms have been claimed as having no indication of northern origin or relationships and hence as proving the existence of a purely southern land-migration route, especially by Hewitt (1922) and Metcalf (1923, 1929). Noble (1925, 1926, 1930) has claimed, however, that the supposed Leptodactylidae are merely members of the Bufonidae, a world-wide or essentially northern group, that happen to have retained teeth independently in Australia and South America. Noble's papers appear to be based on more intimate and broader knowledge of the groups concerned and to give his authority, on this particular point, greater weight. Moreover, Noble has actually identified a fossil toothed bufonid, or Leptodactylid, in the Eocene of Asia, a discovery that brilliantly confirmed his earlier zoögeographic conclusions and that also showed how precarious was the argument from absence of these animals in any northern continent.2

    The amphibia and several analogues will doubtless continue to be cited in proof of Antarctic migration, but they cannot now be taken very seriously in this connection. Granting them all possible force, it still must be clear that alternative views of their affinities are at least as probable as those demanded by this argument.

[[p. 760]] Meiolania

    One of the standard examples of disjunctive distribution and one cited by practically everyone in favor of an Antarctic or Pacific bridge is that of the fossil turtle Meiolania (usually incorrectly called Miolania), which is supposed to occur in Patagonia and in the Australian region but nowhere else, and hence to indicate a more or less direct southern connection between the two.3 This argument is mentioned here chiefly to illustrate three failings common throughout discussions of this whole problem: faulty identification from inadequately studied materials, poor selection of authority, and temerarious leaping over enormous gaps in knowledge.

    As I have elsewhere shown (1938), the Australian and South American horned turtles are not really congeneric, Meiolania being exclusively Australian and Niolamia South American. This was noticed by Tate Regan long ago (1914), and yet practically every student of zoögeography continued to state the occurrence of the same genus in the two widely separated areas as if it were a basic fact. It is not even well established that the two are especially related, although they may be, and they are tentatively placed in a single family. The only known possible ancestors of these forms are found in the Northern Hemisphere, although nothing like a direct phyletic sequence can be cited for either. Meiolania is found with aquatic and indeed marine associations. It was not a specialized marine turtle, but there is every reason to think that it could swim and did enter the sea. Niolamia and its ally Crossochelys are found with aquatic, but freshwater, associations, and beyond much doubt were also strong swimmers with no probable physiological barrier against entering the sea on occasion. Thus there are two alternatives--separate descent from northern ancestry and migration without land connection--each more probable than the widespread insistence on a southern land connection.

    Moreover, Niolamia is either Cretaceous or Eocene in age and Crossochelys is Eocene, whereas Meiolania is Pleistocene. Coupled with the uncertainty of degree of relationship, this enormous time gap entirely removes the supposed disjunctive spacial distribution from the sphere of sober evidence into that of pure speculation. Anderson, who is beyond question the best-informed firsthand authority on Meiolania, pointed out the worthlessness of its evidence in this respect in 1925, yet the great majority of students of the land-bridge problem have continued to prefer or to give equal weight to older and less informed authorities or second- and thirdhand citations that are not, on this point, authorities in any proper sense.

The Ratite Birds

    The so-called ratite birds, including ostriches, rheas, emus, cassowaries, Apteryx, and the extinct moas and Aepyornis, are often believed to form a natural group derived from a common flightless ancestry and explicable only on the basis of southern land connections. Even in the last century Fürbringer, a great authority, suggested that the assemblage may really be unnatural and that the various ratites may have been derived from different flying ancestors. [[p. 761]] This has been widely accepted, and, as for Antarctic connections, about as far as most cared to go was to agree with Blaschke (1904) that if the Antarctic route were considered probable on other grounds, then its possible effect on flightless bird distribution would have to be considered. More recently, Lowe (1928) has asserted that the group is natural and that its ancestors never were capable of flight; but Gregory (1934) and Murphy (1934) have shown that Lowe's excellent observations are more logically interpreted as favoring the opposite conclusion. It is also known that every large land area has at some time developed flightless birds and that several quite distinct ancestral types have so been modified (for general review and exhaustive references see Lambrecht, 1933). The present situation is that the flightless birds cannot be accepted as a valid example of disjunctive terrestrial distribution and cannot be used as evidence for or against the Antarctic route.


    Most of the fishes, most of the amphibians, most of the reptiles, and most of the birds not only do not give evidence for an Antarctic connection, but definitely oppose this. The argument for such a connection is sustained not by considering the fauna as a whole, but by selecting a few apparently anomalous forms, like Galaxias, the "leptodactylids," the "ratites," or Meiolania, and considering them as isolated things. Usually when these are examined more carefully, it is found either that their peculiarity is wholly and more simply explicable on some different basis (e.g., Galaxias) or that they really are not exceptional but could really have had the same geographic history as most of the fauna (e.g., the "leptodactylids"). Similarly, among mammals the marsupials are usually singled out for attention and their associations with a large number of other animals that cannot have had the history claimed for marsupials alone is ignored.

    As for the marsupials, space will not be taken to argue the question, and only the categorical (but very well-grounded) statement may be made that no connection between South American and Australian marsupials has been demonstrated or is at all likely beyond common descent from undifferentiated types such as are known to have occurred in Holarctica at an appropriate time.

     Australia.--The native Australian mammalian fauna consists of monotremes (of entirely unknown but surely very ancient origin), marsupials (with no known allies in Asia, but related to ancient North American and, to a less degree, European forms and with parallel or convergent relatives in South America), rodents allied to the rats (clearly of Asiatic origin), bats (which require no land connections), the dingo (of Asiatic origin and probably introduced by man), and man. For able summaries of this fauna see Anderson (1925 and 1936).

    The classic explanation of this fauna was that marsupials (and monotremes) entered Australia before placentals existed and formed the whole mammalian fauna there until the few placentals were incidentally introduced at a much later date. Now it appears that placentals must have been in existence when the marsupials entered Australia,4 and a serious problem arises, which might be solved by one of these hypotheses:

     [[p. 762]] 1. That placental mammals did enter Australia with the marsupials, but became extinct there.

    2. That a route practicable for marsupials, but not for placentals, existed.

    3. That a route practicable for both, but at any one time unlikely for either, existed, and that marsupials happen to have been the ones to cross it, or to cross it first.

    The first hypothesis is entirely possible but is unlikely. Numerous marsupials are known to have become extinct in competition with placentals, but no placentals are known to have become extinct in competition with marsupials. On this basis, if on no other, the mooted South American connection is most improbable. If marsupials came from South America to Australia over a practicable bridge, placentals should have come with them, and placentals that were entirely capable of surviving along with these marsupials. If marsupials went from Australia to South America, they must have done so on a strictly one-way bridge and have eliminated the earlier native carnivores of South America. Either is just barely conceivable, but neither is acceptable as a serious theory on present evidence.

    The second hypothesis would clear up the difficulties just expressed, but there is no evidence whatever that such a route ever existed, and I for one am at a loss to draw up specifications for one. The Antarctic bridge, if it had any climatic or environmental selective action on animals passing over it, would probably have the opposite effect. It is hard to conceive of its being more easily passed by marsupials than by placentals, and the reverse is much more likely.

    In contrast with these two hypotheses, of which the first is improbable and the second barely short of impossible, the third appears to explain the facts in a simple and probable way, even though the evidence is incomplete. This hypothesis postulates that mammals did not enter Australia by a continuous and fully practicable land route, but by a route such that any migration of terrestrial animals over it was very difficult. A chain of scattered islands, such as actually exists between Australia and Asia, would provide such a route. This route would be passable to small animals and especially to arboreal forms, such as many or most of the primitive marsupials and placentals were. It would be highly improbable that any mammals would follow this route at any given time, but, given many opportunities for occurrence, it would be probable that some mammals would negotiate it sooner or later. The chances that two or more different groups of mammals, such as marsupials and placentals, would successfully traverse the route at the same time would be very slight. Which group happened to traverse it first would be largely the result of random chance, except that the chances would somewhat favor the more numerous group (which may well have been marsupial at the time when they did enter Australia) and the group which had more opportunities for being carried over the stretches of sea (which may also have been the primitive marsupials). It is not, however, necessary to postulate that marsupials were favored by these chances, for it could well happen that the less-favored group was in fact the first one to succeed in the journey.

    According to this view, the marsupials happened to be the first group to reach Australia and it would then probably be a long time before any other mammals succeeded in reaching that continent. When they did arrive, they would do so in impoverished and precarious condition and would find a continent already well stocked with diverse and well-adapted marsupials. Under these conditions placental invaders would have the odds against them, whereas under the ordinary conditions of migration the odds favor them in competition with marsupials. Eventually, no doubt, some placental group would get a [[p. 763]] foothold, and so in fact one did: the rodents. It is generally conceded that these Australian rodents are of Asiatic origin and reached Australia without a continuous land bridge; indeed, no other inference can be seriously supported. The whole problem of the Australian fauna is beautifully simplified if it be supposed that the ancestral didelphoid-dasyuroid stock reached that continent in the same way as the rodents, but at an earlier time.

    In this case the postulation of a difficult migration route, instead of being a disadvantage of the hypothesis, is really a great advantage. The difficulty of such a route makes the hypothesis more, and not less, probable, because this difficulty itself helps to explain facts that would be almost inexplicable if a continuous and easy route had existed and been used.

    As to where this route was, surely the simplest and preferable hypothesis is that it lay between Asia and Australia. Such a route has long existed in that position and it has been used by the rodents, by lower vertebrates, and evidently also by invertebrates and plants.5 The only strong argument against this is the fact that marsupials are unknown in Asia. All the indirect evidence, however, favors the belief that marsupials did exist there in the late Mesozoic and early Tertiary, and the negative direct evidence is practically valueless.

    Some students have argued from the distribution of marsupials within the Australian region that they must have come from the south. Most of the data used in this argument are really not pertinent to this problem and those that are could just as well be used to support northern origin. An example of the occasional complete absence of all logic or reason in such arguments is the serious statement that the occurrence in Tasmania of the only known Australian Tertiary marsupial, a late Tertiary specimen belonging to a specialized living group, favors southern origin for the marsupials in that continent.

     Africa.--The African mammals have no relationships at all with Australia, but mainly with the north, primarily with the Mediterranean region, especially its eastern part, and with Asia, especially its southern part. Some groups arose in Africa from earlier immigrants or from ancient autochthonous stocks, and others arose in the north and invaded Africa. For both, the land connections definitely suggested are northern, and, taking its fossil faunas as a whole, Africa seems historically to be rather another division of the Holarctic mass than a distinctly southern continent in the sense in which this is true of South America. At least, its mammals not only do not support, but also tend to oppose, the existence of an Antarctic route open to them. Possible connections of Africa with South America are mentioned below.

     South America.--Turning to South America, there is no mammalian faunal resemblance to Australia except for the marsupials and these are more consistent with connections through Holarctica than through Antarctica. For various ecological and other reasons it is very unlikely that placentals of South American type ever lived in Australia.

    The mooted resemblances between South American and African mammals are partly mythical. The South American pyrotheres were not proboscideans, and the South American primates have less African than North American affinities. The only good evidence is that of the hystricomorph rodents. Some of the South American and African forms do resemble each other very closely and do not have very close known relatives, fossil or recent, in the north. Here again, however, the question arises why a means of migration open to these rodents did not cause greater and less equivocal faunal resemblances in other [[p. 764]] respects. There is also the curious fact that these rodents are absent from the known early South American faunas and appear suddenly in the Oligocene, as if by adventitious entrance independent of the rest of the fauna. It certainly stretches the imagination to suppose that they came by a transoceanic, non-land route, but it is not impossible, and if any land mammals could use such a route it would be forms like these rodents. On the other hand, there is the more definite but still quite inconclusive fish evidence for a land route from Africa (see Myers, 1938). I cannot hazard more than a guess toward the solution of this mystery, which is perhaps the greatest involved in the problems of southern mammalian faunas. It is, however, evident that even this very dubious evidence of a South American-African connection is not strong, or even valid, evidence of an Antarctic connection. It is generally conceded that if such a connection existed it was probably transatlantic rather than Antarctic.

    Evidence for a North American connection with respect to early South American mammals is much stronger.6 A very archaic North American mammalian fauna could have given rise to the South American fauna. The two continents shared condylarths, and all the other South American ungulates could have come from these or from similarly archaic ungulates like those of North America. The annectant steps are missing, for the most part, but this is a reasonable view and there is no equally probable alternative. There were also archaic primates, edentates, and marsupials in North America that strongly indicate union with the source of South American mammals, even if North America was not itself that source. Despite many troubling gaps in the evidence, gaps that necessitate caution in reaching definite conclusions, they show that North America might have been that source, and no other has been shown to be anything like as probable.

    This does not cast any light on the hystricomorph problem. If South American mammals came from North America, they did so before there were rodents in the latter place. Then the hystricomorphs are a real as well as an apparent exception and any theory must postulate for them a migration different from that of the majority of South American mammals and not necessarily from the same region. The other great mystery is negative: outside of South America there is no known source of placental herbivores unaccompanied by placental carnivores. But at present this is as great a mystery, whatever source be postulated for the herbivores, so that it is no stronger against North American than against any other origin. On the whole it argues against rather than for an Australian connection. South America did somehow get a wholly marsupial carnivore fauna accompanied by a placental herbivore fauna, and at present one can do no more than state the fact.


    Even when all instances are examined and it is shown that an Antarctic migration route is an unnecessary hypothesis, the fact remains that the most striking instances of disjunctive distribution are southern. This observation undoubtedly has given most zoögeographers a predisposition toward belief in lost southern land connections. The cumulative effect of these cases may appear to be impelling, even though it be granted that each separate item is exceptional or is explicable by purely northern routes.

     [[p. 765]] Several possible explanations of this predominance of disjunction in the south have been proposed and much of the literature attempting to establish the reality of a southern route has been devoted to attacking alternative explanations. Of these the most discussed and on the whole the most successful is that set forth in Matthew's Climate and Evolution (1915). Some students seem to feel that if they could demolish this theory they would, ipso facto, establish the reality of the Antarctic route. Of course this is not so. Matthew's theory depends on Holarctic origin and migration of the various southern groups, but the theory of Holarctic as opposed to Antarctic migration routes by no means depends on Matthew's theory, or even on that of Holarctic origin.

    Matthew's theory is that most forms of life have radiated from northern centers of evolution, that more advanced forms arise in such centers, that more ancient or primitive forms are peripheral in distribution, and that most disjunctively distributed southern animals are such peripheral ancient forms. Most criticisms of the theory have been given over to pointing out exceptions or in showing that the theory falls short of explaining all the facts of distribution. It may be granted that Matthew tended to oversimplify and also to extend the theory to some groups of animals not really distributed in accordance with it. It cannot, however, be denied that Matthew's theory does afford the best explanation so far proposed for many of the important and broad features of animal distribution. Both its critics and its adherents need to remember that the theory does not insist that the primitive stocks must survive in a southern area or that they cannot also give rise there to diverse and specialized forms different from those of the north. It does suggest that the southern animals are likely to have split off from the main line of evolution at an earlier date than the northern and that ancient and once cosmopolitan groups are likely to survive longer in the south. This is in accord with many known facts which this theory most successfully explains. To this extent it also helps to explain why some striking instances of disjunctive distribution are southern rather than northern.

    Regardless of the truth or falsity of this particular theory, there is an explanation of the apparent predominance of such disjunction in the south, an explanation so obvious that it seems almost puerile to state it and yet one that some students appear to overlook. The southern land areas are disjunctive in fact and have, in all probability, been still more so during the later part of geological history. The northern land areas are nearly connected now and clearly have been fully connected at many stages in geological history. Many genera of mammals are common to Asia and North America, but the explanation of this fact is so plain that these are not, like the far fewer and weaker but somewhat analogous instances of a common occurrence in the Southern Hemisphere, anomalies demanding special explanation. Generally speaking, anomalously disjunctive distribution is uncommon in the north simply because northern animals occupy what is, historically, a single land area, and it is less uncommon in the south simply because southern animals occupy what are historically separate land areas. If it be supposed that all the continents once had the same fauna--a gross oversimplification but one that is valid in illustration of the process--and that they then were affected by random extinction, some groups becoming extinct in one area and not in another, the result would necessarily be to produce disjunctive distribution in the south but not, or to less extent, in the north. No additional postulate with respect to the relative primitiveness or the origin of the various groups is necessary to explain this inevitable result.


    The idea is not new, but relatively little attention has been paid to the hypothesis that Antarctica was not merely part of a migration route but also was a center of evolution, particularly for the mammals. The question is different from the principal subject of this paper, but it is pertinent here and must be considered briefly.

    All three of the northern continents, North America, Europe, and Asia, show two major and mysterious mammalian invasions. One of these was at about the beginning of the Paleocene. It was also this invasion, by whatever route, that reached South America. It included relatively few of the ancestors of modern mammals. A second invasion occurred around the beginning of the Eocene, and the recent faunas are in the main the highly modified descendants of this Eocene fauna, descendants of which straggled into South America during the Oligocene and flooded in at the end of the Tertiary. The second invasion cannot have come from South America or Australia, and almost surely did not as a whole come from Africa, although certain of its elements were probably elaborated there. Africa had shared also in the first invasion, although the evidence is indirect, and like South America it was a center of evolution for some (but different) elements of that group and later sent their modified descendants northward; but this suggests quite the opposite of its being the source of the second invasion. Although the source is really quite unknown otherwise, the second invasion almost surely did not come from the south, and Antarctica can be almost categorically ruled out on that score.

    The first invasion cannot be quite dismissed even in this unsatisfactory and negative way. If these mammals were in the Northern Hemisphere at the end of the Cretaceous, when their immediate and recognizable ancestry must have been somewhere, it is certainly strange that no trace of them has turned up any clearer than the few insectivore-carnivore skulls of the Mongolian Cretaceous. One would expect evidence of them among the many rich deposits of terrestrial reptiles of that time. There is room in the north for lack of discovery, because of facies or because of geographic isolation, and a northern origin is not impossible, but it is tempting to think that Antarctica, unknown paleontologically so far as land animals are concerned, might have been the mysterious source.

    Again the possibility cannot be absolutely ruled out, but again what evidence there is opposes it. If the source of this mammalian invasion was in Antarctica, the fauna must have moved north by routes involving one or all of the present southern continents, for to suppose Antarctica connected with the northern and not with the southern continents is too fantastic for consideration. If Australia was on the route, I cannot conceive why or how all the many basic and potent placental stocks died out there without the marsupials also dying out. If South America was on the route, it is difficult to see why its abundant early faunas do not give evidence of this. Their evidence, as now known, is definitely one way: all the South American mammals might, in a general way, have been derived from groups with known North American representatives, but the converse is decidedly untrue. Even in the first invasion of North America there are many different groups of mammals that could not be derived in even the broadest sense from groups represented in the early Tertiary of South America. As for Africa, there is little good evidence one way or the other. As pure speculation, it might itself have been the source of the first great mammalian invasion. Even in such a speculative way, however, there is little reason to involve Antarctica, for Africa alone would be adequate [[p. 767]] as such a source, spread from it would almost surely be northward, and the Antarctic connection would again have only the status of an unsupported and superfluous hypothesis. There is no special evidence of an Antarctic connection specifically with Africa, but only, for what it is worth, of Antarctica as a means of connecting two or more other southern land masses.


    The preceding revaluations of facts and arguments and collations of new and old data are only given as samples of reconsideration of the whole problem far too lengthy for presentation here. The conclusion reached is not based solely on what has been said here, but also on this more extensive background.

    It seems to me that a definite answer can now be made to the question given above as summarizing the Antarctic migration problem. There is no known biotic fact that demands an Antarctic land-migration route for its explanation and there is none that is more simply explained by that hypothesis than by any other. The affinities of the southern faunas as a whole are what would be expected from the present northern connections of those continents and from similar connections known, or with considerable probability inferred, to have existed at appropriate times in the past. There are certain troublesome anomalies and exceptions in the evidence, but none of these can be adequately explained by postulating an Antarctic connection. The general weight of the evidence is against such a connection.

    This does not prove than an Antarctic migration route did not exist; such negative proof is practically impossible. It cannot be denied that such a route may have existed or that its existence may some day be proved. But at present its existence is merely a complicating and unnecessary hypothesis additional to other hypotheses for which there is more and better evidence and which are adequate to explain the facts so far as these are now capable of explanation. In scientific theory the best-supported and most nearly self-sufficient hypothesis should be preferred and unnecessary additional hypotheses should be rejected or held in abeyance. On this basis the Antarctic migration route hypothesis remains simply a hypothesis with no proper place in present scientific theory.

Anderson, C.
1925a. "The Australian Fauna." Jour. Proc. Roy. Soc. New South Wales, LIX.
1925b. "Notes on the Extinct Chelonian Genus Meiolania, with a Record of a New Species." Rec. Australian Mus., XIV: 223-242.
1926. "The Origin of Australian Mammals." Australian Mus. Mag., October-December, 1936, pp. 133-137.

Blaschke, F.
1904. "Ueber die tiergeographische Bedeutung eines antarktishen Kontinents." Verh. zool.-bot. Ges. Wien, LIV.

Eigenmann, C. H.
1909. "The Freshwater Fishes of Patagonia and an Examination of the Archiplata-Archhelenis Theory." Rept. Princeton Univ. Exp. Patagonia, III: 225-374.

Gregory, W. K.
1934. "Remarks on the Origins of the Ratites and the Penguins." Proc. Linn. Soc. New York.

Gregory, W. K., and Simpson, G. G.
1926. "Cretaceous Mammal Skulls from Mongolia." Am. Mus. Novitates, No. 225, pp. 1-20.

Hewitt, J.
1922. "On the Zoological Evidence Relating to Ancient Land Connections between Africa and Other Portions of the Southern Hemisphere." So. African Jour. Sci., XIX.

Huene, F. von
1929. "Los saurisquios y ornitisquios del cretáceo argentino." Ann. Mus. La Plata, Ser. 2, III.

[[p. 768]] Lambrecht, K.
1933. Handbuch der Palaeornithologie (Berlin: Gebrüder Borntraeger).

Lowe, P. R.
1928. "Studies and Observations Bearing on the Phylogeny of the Ostrich and Its Allies." Proc. Zoöl. Soc. London, I: 185-247.

Matthew, W. D.
1915. "Climate and Evolution." Ann. New York Acad. Sci., XXIV: 171-318.

Metcalf, M. M.
1923. "The Origin and Distribution of the Anura." Am. Nat., LVII: 385-411.
1929. "Parasites and the Aid They Give in Problems of Taxonomy, Geographical Distribution, and Paleogeography." Smithsonian Misc. Coll., 81, No. 8.

Murphy, R. C.
1934. [Discussion of Gregory, 1934]. Proc. Linn. Soc. New York.

Myers, G. S.
1938. "Fresh-water Fishes and West Indian Zoögeography." Smithsonian Rept. for 1937, pp. 339-364.

Noble, G. K.
1925. "The Evolution and Dispersal of the Frogs." Am. Nat., LIX: 265-271.
1926. "An Analysis of the Remarkable Cases of Distribution among the Amphibia,with Descriptions of New Genera." Am. Mus. Novitates, No. 212, pp. 1-24.
1930. "The Fossil Frogs of the Intertrappean Beds of Bombay, India." Am. Mus. Novitates, No. 401, pp. 1-13.

Pfeffer, G.
1927. Die Frage der Grenzbestimmung zwischen Kreide und Tertiär in zoogeographischer Betrachtung (Jena, Gustav Fisher).

Regan, C. Tate
1905. "A Revision of the Fishes of the Family Galaxiidae." Proc. Zoöl Soc. London.
1914. "Fishes." British Antarctic ("Terra Nova") Exp. Nat. Hist. Repts., Zoölogy, I: 1-45.

Simpson, G. G.
1938. "Crossochelys, Eocene Horned Turtle from Patagonia." Bull. Am. Mus. Nat. Hist., LXXIV: 221-254.

Wittmann, O.
1934. "Die biogeographischen Beziehungen der Südkontinente." Zoogeographica, II: 246-304.


Notes Appearing in the Original Work

    1. Moreover, von Huene's faunal evidence for his transpacific bridge seems to me to be invalidated by more recent discoveries. [[on p. 758]]
    2. Metcalf emphasized the occurrence of the parasite Zelleriella in the so-called leptodactylids and its probable absence in the northern bufonids. This argument has generally been granted great weight, perhaps because it is so esoteric from the point of view of students of the vertebrates, but I cannot see that it has any decisive value for his thesis. Aside from other possible explanations, the survival of a parasite in toothed southern survivors or relicts and its disappearance in more advanced northern members of the same stock would be a normal phenomenon. [[on p. 759]]
    3. As an example, one of many that could be ferreted out in this baffling literature, of accretionary errors in quoting authority, Hesse (Tiergeographie, p. 113) said, that fossil pleurodires occur in the Northern Hemisphere. Arldt (Naturwissenschaften, XII), reviewing Hesse, agreed and added, also correctly, that Meiolaniidae are not known in the north, assuming, incorrectly but with the agreement of the consensus of that time, that Meiolania was a pleurodire. Wittman (1934, p. 284) apparently read and misunderstood Arldt's review and did not read Hesse's book, for he cites Hesse as saying that Meiolania has been found fossil in the northern continents, which is not true and which Hesse did not say. Similarly, Wittmann incorrectly cites Hesse as claiming that fossil Chelyidae are found in the north. [[on p. 760]]
    4. The work of Pfeffer (1927), often given as authority in discussion of this problem, is an instructive example of how the psychology of scientists can interfere with the solution of scientific problems. He supported the outmoded view that the Australian fauna was marsupial because Mesozoic, placentals first evolving at the beginning of the Tertiary. While he was working, unquestionably placentals were found in undoubted Mesozoic beds in Asia (Gregory and Simpson, 1926), absolutely disproving his already generally discarded thesis. Pfeffer was aware of this discovery, but instead of discarding or modifying the theory that the fact disproves, he discarded the fact. [[on p. 761]]
    5. The Wegnerians alone would strongly combat the belief that an island bridge lay between Australia and Asia in, say, late Cretaceous times, but they base their argument in part on the very fact of which such a connection is the best explanation and their theory is also dubious in many other respects. [[on p. 763]]
    6. Of course the many recent South American mammals, like the camelids, peccaries, placental carnivores, tapirs, etc., that are known to have come from North America at the end of the Tertiary do not enter into this problem. [[on p. 764]]

*                 *                 *                 *                 *

Return to Home