Affinities and Origin of the Antillean Mammals1

by William Diller Matthew (1918)

Editor Charles H. Smith's Note: A paper read before the Paleontological Society on 1 January, 1918. Original pagination indicated within double brackets. Notes are grouped at the end, with the page(s) they originally appeared at the bottom of given within double brackets. Citation: Bulletin of the Geological Society of America 29 (1918): 657-666.

[[p. 657]] Contents

Limitations and relationships of West Indian mammal faunas, living and extinct . . . 657
      In general . . . 657
      The insectivora . . . 658
      The rodentia . . . 659
      The edentata . . . 660
      Bats and birds . . . 661
      Reptiles . . . 661
Summary of affinities and probable origin of the vertebrate groups . . . 662
Is the incomplete and unbalanced character of the fauna real or only apparent? . . . 663
Conclusions as to former geographic relations and manner of colonization . . . 664


Limitations and Relationships of West Indian Mammal Faunas, Living and Extinct

In General

    The indigenous land mammals of the West Indies consist of three groups: (1) Insectivora, (2) hystricomorph rodents, (3) gravigrade edentates. No perissodactyls (horses, rhinoceroses, tapirs, etcetera), no artiodactyls (peccaries, deer, antelopes, etcetera), no proboscideans (elephants, mastodons, etcetera), no true carnivores (dogs, cats, raccoons, mustelines, bears, etcetera). Nor are there any sciuromorph, lagomorph, or myomorph rodents, shrews, moles, hedgehogs, or opossums. All these large groups, most of them abundant and varied in Tertiary North America, are wholly absent. Nor do the Insectivora, rodents, or edentates include anything at all nearly allied to any North American members of [[p. 658]] the order or derivable from anything known to have inhabited North America in the later Tertiary.

    Most of these North American groups invaded South America in the Pliocene and are part of its later fauna. In the Miocene and early Tertiary they are not found in South America, but their place is taken by a number of other groups. In place of perissodactyls, artiodactyls, and proboscideans were a number of groups of hoofed animals peculiar to Tertiary South America--the toxodonts, typotheres, litopterns, homalodontotheres, astrapotheres, pyrotheres. In place of the true Carnivora is a variety of marsupial carnivores (Borhyænidæ) paralleling the true carnivores in structure and taking their place in the fauna.2 All of these abundant and varied groups of ungulates and pseudo-Carnivora are lacking from the Antillean fauna, nor do the rodents represent more than two or possibly three of the numerous hystricomorph rodent stocks of Miocene South America, while the edentates represent only one group of the ground-sloths, the three or four other ground-sloth groups, as well as the several kinds of armadillos and the glyptodonts, being quite unrepresented.

The Insectivora

    It appears to be reasonably certain that the Antillean rodents and edentates came from South America and from Tertiary South America. Hystricomorph rodents and edentates are unquestionably South American Tertiary types, which invaded North America when the two continents were joined, toward the end of the Tertiary. The insectivores, however, are more probably derivable from North American sources; [[p. 659]] for, with a single somewhat doubtful exception, the entire order of Insectivora is absent from the Tertiary faunas of South America, while they were many and varied in North America, especially in the older Tertiary. The two Antillean insectivores are not nearly related to each other, nor to any other genera of the order; they are placed in families by themselves. It has been repeatedly stated that Solenodon is related to the Malagasy Centetidæ, but in fact the affinity is a very distant one. Nesophontes is equally peculiar, and while it has some affinities with the Soricoidea (moles and shrews), they are very distant. The nearest relatives of both--collateral ancestors, perhaps--are certain imperfectly known insectivores of very primitive type in the North American Eocene and Oligocene.

The Rodentia

    The rodents are clearly of South American affinities. They are all hystricomorphs--a group chiefly South American since the middle Tertiary (if not before). The only North American hystricomorph is the porcupine, Pleistocene and Recent, and whose ancestors have been recognized in the South American Tertiary. No traces of the hystricomorphs have been discovered in the Tertiary of North America.3 There are certain Old World hystricomorphs, the Hystricidæ and certain Octodontidæ, and the early Tertiary Theridomyidæ of Europe have been considered ancestral to the group, but they have no significant relations to the Antillean genera and their affinities are disputed, so that they may be passed over for the present problem.

    The Antillean hystricomorphs are clearly related to the South American types, but it is equally clear that the relationship is not close. There are apparently three groups. One, including Amblyrhiza of Anguilla, Elasmodontomys and Heptaxodon of Porto Rico, is related to the chinchillas, but not closely related. Anthony4 places them in separate subfamilies. Miller5 states that they are more nearly related to the extinct Megamys and its allies than to the living chinchillids. These genera (Megamys, Tetrastylus, etcetera) are found in the Entrerian, Rio Negran, Hermosan, and Araucanian formations of Argentina accompanied by a fauna which is closely related to the Pampean, but contains a few little altered survivals from the Santa Cruz Miocene and comparatively few of the North American invading types. All should, in my [[p. 660]] judgment, be referred to the Upper Pliocene; they are certainly much later than the Santa Cruz.

    Although Megamys and Tetrastylus are without doubt more nearly related than any modern type to the Antillean chinchillids and are considerably older, they can not be regarded as ancestral. The common ancestral stock is to be found in the Santa Cruz chinchillids, all of which are of small or medium size. These Santa Cruz species are much more primitive, and the precise relationships will require more careful study.6

    The remaining Antillean rodents are of South American type and broadly derivable from Santa Cruz rodents, but their more exact affinities are disputed and require more thorough and critical consideration and, if possible, more complete material. A thorough revision of the fossil rodent faunas of South America is an almost necessary groundwork for a correct estimate of their affinities. It is clear, however, that with the exception of Capromys they are not very closely related to the South American rodents, the common ancestral stock dating back probably to Pliocene or late Miocene, as Miller believes. Capromys (and Geocapromys) would seem to be an exception, being quite close to the Venezuelan Procapromys.

The Edentata

    The edentates include four quite distinct genera from Cuba and a fifth from Porto Rico, all referred to the family Megalonychidæ, but not closely related to any of the mainland forms. The largest, Megalocnus,7 is about the size of a black bear; the smallest, Microcnus, about the size of a cat, and there are two of intermediate size, Mesocnus, with a rather long, narrow muzzle, and Miocnus, with a broad, square muzzle. The Porto Rican genus is related to Miocnus, both having heavy triangular tusks like the modern Cholœpus; the other three genera have large tusks, but of a peculiar dished shape, with a tendency to approach toward each other like the incisors of rodents. (They are not at all of the scalpriform gnawing type, however.) While these ground-sloths are sufficiently [[p. 661]] related to the North American genus Megalonyx to be placed in the same subfamily, they are quite evidently not descended from it, but contemporaneous specializations from the primitive Megalonychidæ of the South American Miocene, as represented in the Santa Cruz fauna. Among the known genera of this fauna there is only one, Eucholœops (including Megalonychotherium), which can be regarded as ancestral either to the Cuban ground-sloths or to Megalonyx. The others all have the caniniform teeth much reduced or vestigial and in series with the cheek teeth. To the best of my judgment, the anatomical evidence is not decisive as to whether the five Antillean genera are descended from one or from two or more nearly allied Upper Miocene or Lower Pliocene genera, but leads to the conclusion that the common ancestor or ancestors was very close to or identical with the Upper Miocene ancestor of Megalonyx, and was either the genus Eucholœops or one or more genera closely allied thereto. The Antillean genera represent, therefore, only the megalonychine division of the Megalonychidæ. The other families of ground-sloths--Mylodontidæ, Scelidotheridæ, and Megatheriidæ--are not found; nor are there any armadillos, glyptodonts, or anteaters.

Bats and Birds

    In addition to the terrestrial mammals, bats are numerous in the cave deposits, and a number of birds, lizards, crocodiles, and turtles have been found at the Ciego Montero locality and elsewhere.

    Concerning the bats, there is very little to say. Most of them are nearly allied to or identical with species now living on the islands. The Antillean bats include a number of peculiar genera, besides others common to the continental parts of tropical America. Their relations are much the same as those of the birds, and in either case it is obvious that the intervening seas would act as a hindrance to migration, but not as an absolute barrier, and would be more of a hindrance in some groups than in others. The result would be the presence of a number of peculiar types, preserved by relative isolation and specialized in adaptation to the peculiarities of their habitat, along with other widely ranging forms closely allied to or identical with those of the mainland. The distribution of the birds has been carefully studied by Chapman and others.


    The distribution of the lizards has been recently studied by Dr. Thomas Barbour, and his conclusions as to the paleogeography are sharply at variance with mine, owing to different methods of interpreting the data. I shall not take this part of the problem up at present.

    [[p. 662]] The fossil crocodiles have been examined by Dr. Barbour, who informs me that they are all referable to Crocodilus rhombifer, a species still living on the island of Cuba.8 The origin of this species might be either North or South American; but too little is known of the phylogeny and distribution of the Tertiary Crocodilia for any conclusions to be drawn as to the time or method of its arrival.

    The fossil chelonians have not been carefully studied, but they include two species--one a giant tortoise, Testudo cubensis Leidy, which, like the giant tortoises of the Galapagos and Indian Ocean islands, has the carapace much thinned out, so that the plates are apparently more or less discontinuous. There is one North American Pliocene species, T. pertenuis Cope, from Texas which has a remarkably thin carapace, but apparently not discontinuous. The precise significance of this species in the paleogeographic problem must also await more careful study. The genus Testudo occurs sparingly in South America, and is recorded as a fossil in the Pliocene and Pleistocene formations--not earlier, so far as I know. On the other hand, species of Testudo are the most abundant of fossils in the Oligocene to Pliocene formations of North America; in the Pleistocene and Recent their range is restricted to the Southern States and Mexico. The indications point, therefore, preferably to North American origin for this Cuban tortoise, although not decisively.

    The second fossil chelonian is one of the Emydidæ, or marsh-turtles; it appears to be a species of Graphemys, probably the same as the still existing Cuban species, which is said to be a close ally of G. scripta of the Southeastern States. Whether the two are specifically distinct has been questioned. The discovery of this species (if it be the same) fossil in the Ciego Montero locality removes any doubt as to its being indigenous to the island, as it carries it back into the Pleistocene, and probably to a time before the arrival of man. Its close relationship with G. scripta and limitation to the western islands, Cuba and Haiti, is a strong indication of its having come from Florida, and the time of its arrival probably would be not earlier than Pleistocene or at most late Pliocene.

Summary of Affinities and Probable Origin of the Vertebrate Groups

    Summing up the indicated sources of the fossil and recent vertebrate fauna, we find it to be as follows:

    [[p. 663]] 1. The Insectivora, Solenodon and Nesophontes, of very ancient arrival, probably early Tertiary, and apparently of North American origin.

    2. The ground-sloths, Megalocnus, Mesocnus, Miocnus, Microcnus in Cuba and Acratocnus in Porto Rico, of moderately ancient arrival, probably late Miocene or early Pliocene, and of undoubted South American origin. The rodents, with the exception of Capromys, fall also into this category.

    3. The peculiar groups of birds, bats, and lizards are also no doubt of comparatively ancient arrival, but their source is unknown, as we know nothing of the Tertiary distribution of related groups on the mainland. The modern distribution of such related groups can not be relied on, for we know that among terrestrial mammals various groups which are today exclusively or chiefly Neotropical were Nearctic until the end of the Tertiary and unknown in South American faunas until the late Pliocene. Presumably corresponding changes in distribution have occurred among the bats, birds, and reptiles, but we have no records as to what groups were affected. The Cuban crocodile may also be placed in this category.

    4. Of the two chelonians the giant tortoise may be placed as more probably of North American than of Central or South American origin, but its time of arrival can hardly be estimated until its relations to the continental Tertiary species are known. The terrapin is almost certainly of North American origin, derived from the Southeastern States, and of comparatively late arrival, probably late Pliocene or Pleistocene.

    5. Capromys and Geocapromys are undoubtedly of South American derivation, like the other rodents; but, so far as may be judged from their comparatively near affinity with the Venezuelan Procapromys, are of later arrival, perhaps late Pliocene or Pleistocene.

    It appears, therefore, in sum, that the vertebrate fauna, fossil and recent, represents only a few selections from the continental faunas of either North or South America; that it falls into several groups of diverse origin, and judging from their degree of differentiation, of diverse times of arrival.

Is the Incomplete and Unbalanced Character of the Fauna Real or Only Apparent?

    In the absence of hoofed animals, which form the greater part of all continental faunæ, in the tendency of races normally of small size to assume relatively large size and importance, in the relative fragility, so to speak, of the fauna, leading to its early disappearance when man invades the region--in many further points of detail--it parallels the faunas of those islands which lie beyond the continental shelf, and differs [[p. 664]] from those islands which lie within the shelf. In particular, the parallelism with the Madagascar fauna is made much closer by the recent discoveries. On the other hand, the contrast with the fauna of continental islands such as Borneo or Sumatra is a marked one. On these islands the fauna, although considerably specialized by isolation in Borneo, less so in Sumatra, is a fairly representative one. It includes all or nearly all of the important mammalian groups of the mainland save those which there is reason to believe are of too recent arrival or of unsuitable habitat to be present.

    It may be objected that this difference is merely apparent; that the Pleistocene fauna of the Antilles was really of continental character, but because they are islands and not continents, it has been easily exterminated by man, and that the cave and spring deposits present only two distinct and very limited facies, not including the ungulates, carnivores, etcetera, which have been present. The best reply to this objection is to test it by comparison. Sumatra or Java are islands of comparable size to Cuba; Borneo is larger; Formosa or Hainan are comparable to Porto Rico. In none of these islands has the indigenous fauna been wiped out to anything like the extent necessary to obliterate its continental character, although all have been inhabited by man for a much longer time and in much larger numbers than the Antilles. The indigenous faunas of Great Britain or Ireland are far from exterminated, in spite of the great density of population and of modern civilization.

    Nor can we assume that a cave or a spring fauna is so limited in its facies as to disguise a continental fauna type. The faunas of numerous caves in Europe and North America have been examined, and wherever any considerable collection is obtained it is clearly representative of the continental type. Spring or bog faunas sometimes contain little except hoofed animals, but I never heard of one in which hoofed animals and carnivora were absent. I can not escape from the conclusion that the Pleistocene fauna of Cuba was not a normal fauna, but deficient in most of the more abundant groups and composed of a selection of a very limited number of types which had expanded to a disproportionate variety and importance, owing to the absence of the rest of the fauna.

Conclusions as to Former Geographic Relations and Manner of Colonization

    As to the diverse origin of the several groups and the varying time during which they have been isolated on the islands, I have stated my interpretation of the evidence. I do not feel, however, that evidence of this sort leads to positive and certain conclusions.

    [[p. 665]] As to the former connection of the Antilles with each other and with the mainland, my conclusions with the proviso just stated are as follows:

    1. That the Greater Antilles have probably been united with each other, as far east as the Anguilla bank, in the late Tertiary or Pleistocene. This I conclude from the near affinity of representative species of the same or closely allied genera and the general similarity of the fauna, so far as known, in the different islands.

    2. That they have not at any time during the Tertiary been united with North America. If they had been we should find North American ungulates, rodents, carnivores, etcetera, differentiated in accord with the length of subsequent isolation, but of clearly recognizable affinities, and it would be a balanced or representative fauna. We might object that such a fauna had perhaps existed, but been wiped out by subsequent submergence. But the presence of Solenodon and Nesophontes negatives that, for they represent a very ancient survival, and if there had been a representative fauna it is hardly credible that submergence would have spared just two insectivores and destroyed all the rest of the fauna.

    3. That they probably have not been connected with South America, either via the Lesser Antilles or via Central America, during the Tertiary; for if they had the fauna should be of continental South American type, with South American ungulate groups, marsupial carnivores, and a full representation of the rodents, edentates, etcetera.

    4. The mammalian fauna appears to me to be reducible to perhaps three primary rodent stocks, one or more primary ground-sloth stocks, and two Insectivora. These I conceive to have arrived at various times during the Tertiary, the rodents and ground-sloths from South or Central America, the insectivores from North America, by accidents of transportation, of which the most probable for the mammals would perhaps be the so-called "natural rafts" or masses of vegetation dislodged from the banks of great rivers during floods and drifted out to sea. The probabilities of this method I have elsewhere discussed.9 For birds and bats, for the smaller reptiles, amphibians, fishes, and invertebrates, the problem of oversea transportation is a much simpler one.10 That successful colonization in this way can occur is shown by their presence on nearly all oceanic islands; for it will hardly be maintained by reasonable men that every oceanic island has been joined to the mainland and has been continuously above water since its separation. Obviously, the larger the island and the nearer to continental land, the more often such colonization will occur.

    [[p. 666]] 5. The geology of the Caribbean region appears to me to afford no positive evidence against union of the Antilles either with South America or Central America; but neither does it afford any evidence that there ever was such union. Undoubtedly there is a line of disturbance and uplift along the Lesser Antilles, and another stretching through Haiti and Jamaica to Nicaragua; but evidence of similar and contemporaneous upheavals and similar sedimentation in two portions of this line of disturbance that are now separated by abyssal depths does not in the least prove that the intervening depths were formerly continuous land bridges. They may have been, but I do not see how any geologic evidence can prove that they were so. If we have evidence from some other source that there must have been a land bridge somewhere, then these lines of disturbance show its most probable location. That is all.

    Land union with Florida appears to be distinctly against the geologic evidence, as in this region we have extensive flat-lying Tertiary marine and littoral formations which indicate that there has been very slight movement during the Tertiary, and that the present limits of the continental shelf represent probably the extreme extension of the land in the Pleistocene. Dall has shown the evidence very clearly in the case of Florida. Apparently the conditions in Yucatan are partly similar, but Vaughan has shown that its tectonic relations to the Antillean ridges are more favorable to a former union.


Notes Appearing in the Original Work

1Manuscript received by the Secretary of the Society August 22, 1918. [[on p. 657]]
2A similar but distinct group of marsupial carnivores (Dasyuridæ and Thylacinidæ) developed in Australia in absence of true Carnivora and still survives there. NOTE.--In this paper no account is taken of animals which may have been brought to the islands by man, whether intentionally or by accident, in post-Columbian or pre-historic time. Some of these have evolved under insular conditions into races distinct enough to be recorded as species or subspecies. The majority are identical with species of North or South America, Europe, Africa, etcetera. Some are known to have been introduced; others may be so explained by reason of associations of one kind or another. The formation of distinct races, such as are classed as species by modern mammalogists, does not necessarily take many centuries under these conditions--as witness the Porto Santo (Madeira) rabbits and other instances. Some of these supposedly introduced forms may have been brought in through natural means and be truly indigenous, although not very ancient; but it is impossible to prove such cases. It seems better here to omit all this doubtful evidence and consider only the fauna that is proved to be indigenous either by occurrence in the Pleistocene cave and spring deposits, by its sharp distinctions from any continental relatives, or by the high improbability that the animal could have been transported through human agency. Dr. G. M. Allen has compiled an annotated list of the West Indian mammals which includes both introduced and indigenous types, and Dr. Thomas Barbour has done the same for the reptilia and batrachia. The evidence therein summarized will be discussed in a memoir on Cuban fossil mammals now in preparation. [[on p. 658]]
3Except Leidy's Hystrix (Hystricops) venusta, based on two teeth of doubtful affinities and uncertain geologic age. [[on p. 659]]
4Anthony: New fossil rodents from Porto Rico. Bull. Am. Mus. Nat. Hist., vol. xxxvii, 1917, pp. 185-186. [[on p. 659]]
5Miller: Bones of mammals from Indian sites in Cuba and Santo Domingo. Smithsonian Miscell. Coll., vol. lxvi, no. 12, 1916, p 3. [[on p. 659]]
6Miller apparently associates the Santa Cruz fauna of southern Patagonia with the very different and much later Entrerian fauna of northern Argentina, as he speaks of the two as though they were essentially one fauna, and refers to the "enormous extinct Pagatonian rodents" as the nearest relatives of the Antillean chinchillids. The largest Santa Cruzian rodents are species of Perimys, which is not nearly related to the Antillean genera. Most of the species are quite small. As between the Santa Cruz and the Hermosan (Pliocene) group of faunas, the tendency to rapid increase in size and specialization of numerous phyla is very marked, and is further emphasized in the Pampean (Pleistocene) group of faunas. [[on p. 660]]
7Megalocnus, Leidy, 1868. Proc. Acad. Nat. Sci. Phila., 1868, p. 179. Microcnus, etc., La Torre and Matthew, 1915. Bull. Geol. Soc. Am., vol. xxvi, p. 152 (names only; descriptions have been reserved pending the securing of more complete and better associated skeleton material). [[on p. 660]]
8Leidy's Crocodilus pristinus (1888, 1. c.) was based upon a vertebra not distinguished from C. rhombifer. The skulls obtained by La Torre and Brown represent, in Doctor Barbour's opinion, a series of growth stages of the modern species, the largest much exceeding any modern specimens. Part of a skeleton associated with one of the largest skulls equals or exceeds Leidy's type of pristinus in size. [[on p. 662]]
9Matthew: Climate and evolution. Annals N. Y. Acad. Sci., vol. xxiv, 1915, p. 206. [[on p. 665]]
10Tropical storms, as Wallace pointed out years ago, probably play a principal part in transportation of very small animals or their eggs. Mammals could hardly be carried that way nor survive if they were. [[on p. 665]]

*                 *                 *                 *                 *

Return to Home