Front Cover
 Back Cover

Title: Fossil Sirenians and Desmostylids from Florida and elsewhere
Full Citation
Permanent Link: http://ufdc.ufl.edu/UF00027852/00001
 Material Information
Title: Fossil Sirenians and Desmostylids from Florida and elsewhere
Physical Description: Book
Creator: Reinhart, Roy Herbert,
Publisher: University of Florida,
Publication Date: 1976
Copyright Date: 1976
 Record Information
Bibliographic ID: UF00027852
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: electronic_aleph - 003318788
electronic_oclc - 60779478

Table of Contents
        Page 184
    Front Cover
        Page 185
        Page 186
        Page 187
        Page 188
        Page 189
        Page 190
        Page 191
        Page 192
        Page 193
        Page 194
        Page 195
        Page 196
        Page 197
        Page 198
        Page 199
        Page 200
        Page 201
        Page 202
        Page 203
        Page 204
        Page 205
        Page 206
        Page 207
        Page 208
        Page 209
        Page 210
        Page 211
        Page 212
        Page 213
        Page 214
        Page 215
        Page 216
        Page 217
        Page 218
        Page 219
        Page 220
        Page 221
        Page 222
        Page 223
        Page 224
        Page 225
        Page 226
        Page 227
        Page 228
        Page 229
        Page 230
        Page 231
        Page 232
        Page 233
        Page 234
        Page 235
        Page 236
        Page 237
        Page 238
        Page 239
        Page 240
        Page 241
        Page 242
        Page 243
        Page 244
        Page 245
        Page 246
        Page 247
        Page 248
        Page 249
        Page 250
        Page 251
        Page 252
        Page 253
        Page 254
        Page 255
        Page 256
        Page 257
        Page 258
        Page 259
        Page 260
        Page 261
        Page 262
        Page 263
        Page 264
        Page 265
        Page 266
        Page 267
        Page 268
        Page 269
        Page 270
        Page 271
        Page 272
        Page 273
        Page 274
        Page 275
        Page 276
        Page 277
        Page 278
        Page 279
        Page 280
        Page 281
        Page 282
        Page 283
        Page 284
        Page 285
        Page 286
        Page 287
        Page 288
        Page 289
        Page 290
        Page 291
        Page 292
        Page 293
        Page 294
        Page 295
        Page 296
        Page 297
        Page 298
        Page 299
        Page 300
    Back Cover
        Page 301
        Page 302
Full Text


[year of publication] Florida Museum of Natural History [source text]

The Florida Museum of Natural History, formerly the Florida State
Museum, holds all rights to the source text of this electronic resource on
behalf of the State of Florida. The Florida Museum of Natural History shall
be considered the copyright holder for the text and images of this

Under the Statutes of Florida (FS 257.05; 257.105, and 377.075), the Florida
Museum of Natural History (Gainesville, FL) publisher of the Bulletin of the
Florida State Museum, as a division of state government makes its
documents public (i.e.,published) and extends to the state's official agencies
and libraries, including the University of Florida's Smathers Library, limited
rights of reproduction.

The Florida Museum of Natural History has made this publication available
to the University of Florida, on behalf of the State University System, for
purposes of digitization and Internet distribution.

The Florida Museum of Natural History reserves all rights to this
publication. All uses, excluding those made under "fair use" provisions of
United States of America copyright legislation (U.S. Code, Title 17, Section
107), are restricted. Contact the Florida Museum of Natural History, a
division of the University of Florida, in Gainesville, Florida, for additional
information and permissions.

of the
Biological Sciences

Volume 20 1976 Number 4




SCIENCES, are published at irregular intervals. Volumes contain about 300 pages and
are not necessarily completed in any one calendar year.

RHODA J. RYBAK, Managing Editor

Consultants for this issue:



Communications concerning purchase or exchange of the publications and all manu-
scripts should be addressed to the Managing Editor of the Bulletin, Florida State
Museum, Museum Road, University of Florida, Gainesville, Florida 32611.

This public document was promulgated at an annual cost of $3803.89
or $3.804 per copy. It makes available to libraries, scholars, and all
interested persons the results of researches in the natural sciences,
emphasizing the Circum-Caribbean region.

Publication date: July 9, 1976 Price: $3.85



SYNOPSIS: Florida is the only part of the world in which extensive sirenian remains
have been recovered from each epoch, beginning with the Eocene and continuing
into the Recent. The evolutionary stage of the Mio-Pliocene sirenians from Florida
parallels that of similarly aged European sirenians, in which a puzzling complex of
specimens assigned to Metaxytherium, Halianassa, and Felsinotherium exists. From
this study a proposal for synonymy of the above genera into the genus Metaxytherium
is made. Florida metaxytheres are subdivided into three species with retention of
trivial names now in use. Amplification and clarification of these species are made
through a description of a number of new specimens.
Changes in interpretation of the cranial morphology of Hesperosiren are sug-
gested, with the result that this form appears to be closely related to Metaxytheriuin,
rather than the highly aberrant form it has been considered.
The presence of Halitherium in continental North America is how established
beyond question in the new species H. olseni from Florida. Although of a similar
evolutionary stage, this specimen is from beds of Middle to Upper Miocene, as com-
pared to the Middle to Upper Oligocene European Halitherium.
Ribs of Middle Eocene sirenians have been substantiated from a number of Flor-
ida localities, but no diagnostic elements have been found.
Comments are made on Sirenia from areas other than Florida. These include
a review of Sirenia from Java, the second report of a dugongid from South America,
the validity of the Argentine trichechid Ribodon, and a report on the nomenclatorial
status of Manatus and Trichechus.
Desmostylid remains have been authenticated only from the circum-North Pacific
region, but reports on fragmentary teeth, incorrectly identified, have been attributed
to Desmostylus from various localities. Two such proposals, Cryptomastodon from
Java and Desmostylus from Texas, are discussed. Reports on new specimens of
Comwallius from the Aleutian chain, and Desmostylus from California and Florida
are made, the latter being the first desmostylid authenticated outside of the Pacific.-
Discovery of Desmostylus in Florida extends its upper stratigraphic range from
Late Miocene to Mio-Pliocene or to Middle Pliocene (Hemphillian), according to
the age designation given to the Bone Valley Formation.

SThe author is a Research Associate of the Florida State Museum and Professor of Geology.
Miami University, Oxford, Ohio. His current research includes systematics, paleogeographical
distribution, and evolution of the mammalian Orders Sirenia and Desmostylia. Manuscript
accepted 15 October 1973.

REINHART, ROY H. 1976. Fossil Sirenians and Desmostylids from Florida and Else-
where. Bull. Florida State Mus., Biol. Sci., Vol. 20, No. 4, pp. 187-300.


570. 832



INTRODUCTION ---....------.------------------- 188
ACKNOWLEDGMENTS ..-----------..-----------------.............-------------------- -------- 190
MATERIAL EXAMINED _._------- --- -----_---- 191
ABBREVIATIONS USED IN FIGURES _---- ------- --------.. ----------- 191
Metaxytherium ossivalense SIMPSON 1932 .---- ------- ...- -----. --- 199
Metaxytherium floridanum HAY 1922 _-...-__.. ..-------- 216
Metaxytherium calvertense KELLOGG 1966 ------.. ..--------- 220
Hesperosiren SIMPSON 1932 ---__----... ---....--------_---------- 229
GENUS Halitherium KAUP 1838 ...--........----_--------- ----- ----------. ---- 236
Halitherium olseni new species .--..-.--.-----------------------------------.. 238
EOCENE SIRENIA FROM FLORIDA .----.-.........------.. ------------------------..--------- 262
JAVA SIRENIANS .-----.--..-.--........---.-.. __-......-- ---- ------- 266
AUSTRALIAN SIRENIAN ......------.---...------...... --------......-------.......------------ 268
INCISOR ANOMALY IN DUGONG --.....-------------- --------..----.------------ ---- 269
PRESENCE OF Metaxytherium IN ARGENTINA. Metaxytherium,
SPECIES INDETERMINATE ..--.-------.. --_--....------ --..........------------ 272
TRICHECHIDAE, VALIDITY OF THE GENUS Ribodon ..--.- __......---_-.-..... 279
BRAZILIAN Trichechus ---...........---......-------------.--------------.- ----------- 280
Manatus IN SOUTHEAST EUROPE ------......-------.--.----------....... ------- __.-- 281
ALASKAN DESMOSTYLID --...--..................-- ---- ------------------------- 283
FLORIDA Desmostylus --.___---_..-.-----......-.......... ---_....--..----. 284
CALIFORNIA Desmostylus. Desmostylus hesperus MARSH 1888 ........--._---...--. 288
BIBLIOGRAPHY ---____----___--________ ._.. ..__ -------------________ .. 295

Recent finds of Sirenia in the western hemisphere show that some
modifications of our vague, but generally accepted, concepts of this order
are necessary. As a carryover from older European literature, the origin
of the Order Sirenia is often speculated to have occurred in the Middle
Eocene Fayum region of Egypt because of the relative abundance of
primitive Sirenia from this area, plus the earliest known representatives
of the closely allied Order Proboscidea. A Middle Eocene sirenian with
equally prototypic structural characteristics is also found in Jamaica,
while less diagnostic specimens from Florida, Alabama, and possibly
Java suggest a wide geographic distribution during this epoch, with no
positive evidence for initiation of the group in any single region. The
ancestral breeding ground is therefore unknown, as is true with most
mammalian orders. The general paucity of early sirenians from much
of the world likely results from insufficient investigation of Eocene ma-
rine and estuarine beds, rather than from their true absence. An interest-
ing fact, but one that casts no information on ordinal origin, is that
Florida is the only area with evidence of continuous sirenian habitation





i / 7 Ct,\

'''^p ^-/ _-' ''" o
""*" /' "*'" "/ >

FIGuRE 1.-Sirenia and Desmostylus localities in Florida.

from Middle Eocene to Recent time. If Europe and Egypt are com-
bined, a similar time range exists for this more geographically-widespread
region, with the exception of an absence of Pleistocene and Recent forms.
Higher in the geologic column, the evolutionary stage of Oligocene
as well as Mio-Pliocene sirenians from Florida parallels that of European


sirenians that are similar in age. Existence of a stratigraphic problem is
immediately obvious when these almost identical representatives are
compared. European specimens are invariably placed in a slightly older
unit of geologic time than their North American counterparts. Whether
this reflects a true picture of the genesis of the order or a fault of inter-
continental correlation is a matter of conjecture.
Also included in this study is new information on the Sirenia of South
America, where recent discoveries have revealed their presence in the
western Atlantic region, from Colombia to Argentina. The find of a
lower Miocene trichechid in Brazil (Couto 1967), in addition to the
Middle to Upper Miocene dugongid from Argentina reported in this
study, indicates trichechids existed with dugongids in the Caribbean
Sea and in South America throughout the Miocene, and possibly through
Pliocene time. Not knowing of any Trichechidae from areas other than
the Caribbean and the coastal rivers of the South Atlantic is an enigma.
Because the Dugongidae are geographically widespread from Middle
Eocene to Recent time, one would expect the Trichechidae to have a
somewhat greater range than the narrow distribution evidenced by the
fossil record. The considerable cranial and postcranial differences be-
tween Pleistocene and Recent Trichechidae and Dugongidae, plus evi-
dence of similar separation back through Lower Miocene, suggests the
possibility of a separate phyletic line for each since Eocene time.

The writer is indebted to many people for their kindness and help in the prepa-
ration of this paper. Colleagues who cooperated by sending specimens are William
Turnbull, Field Museum of Natural History; C. Lewis Gazin, U. S. National Museum;
Malcolm McKenna, American Museum; Craig Black, Museum of Comparative
Zoology at Harvard; Theodore Downs, Los Angeles County Museum; John Wilson,
University of Texas; and Rosendo Pascual, Museo de la Plata, Argentina. Informa-
tion concerning some of several papers and a reading of the final manuscript was
provided by Frank Whitmore of the U. S. Geological Survey. The late John Conrad
Hansen, formerly staff artist at the Field Museum of Natural History, prepared the
illustrations of Desmostylus and willingly aided on many illustrations. Lee Dutton,
in charge of the Miami University Library, was most cooperative in obtaining refer-
ences throughout the study. My indebtedness is also offered to F. R. Berquist of
the American Agricultural Company, Pierce, Florida, for full cooperation in field
Robert O. Vernon, chief of the Florida Geological Survey, was particularly help-
ful in my securing the excellent specimen of Florida Halitherium and provided funds
for its preparation. J. C. Dickinson, Jr., director of the Florida State Museum, aided
in numerous ways, and, with William Reimer, former chairman of the Department
of Natural Sciences of the same museum, was in charge of financial aid received
from the University of Florida to make the study possible. Later funds were re-
ceived through the office of Walter Auffenberg. H. Kelly Brooks of the University
of Florida, Gainesville, proved to be an indefatigable and enthusiastic field worker
whose time and knowledge of Florida stratigraphy was given generously. In the
closing stages of the study David Webb contributed comments and his efforts most


To Clayton Ray of the U. S. National Museum, Washington, D. C., (and form-
erly of the Florida State Museum), must go my deepest thanks for initiating the
project and providing stimulation throughout the study. Both Ray and Stanley
Olsen of the Florida Geological Survey sent many specimens, provided field data,
and continually added encouragement in our many meetings. For the varied exhaus-
tive help from these two colleagues I am most grateful.

Materials Examined
Material examined is from the following collections: American Mu-
seum Natural History (AMNH), Field Museum of Natural History
(FMNH), Florida State Geological Survey (FGS), Los Angeles County
Museum of Natural History (LACM), Museum Comparative Zoology
Harvard, (MCZ), National University of Cordoba, Argentina (NUC),
University of California Museum Paleontology (UCMP), University of
Florida, Gainesville (UF), United States National Museum (USNM),
University of Texas, Austin (UT). All measurements throughout the
text are in millimeters.

Abbreviations Used in Figures
Abbreviations used in figures are as followed: DENTAL: Anterior
cingulum (A.c.), Hypocone (Hy), Metacone (Me), Metaconule (Ml),
Median valley (M. vi), Paracone (Pa), Posterior cingulum (P.c.), Pro-
toconule (PI), Protocone (Pr); CRANIAL: Exoccipital (Ex), Frontal
(Fr), Jugal (Ju), Lacrymal (La), Maxillary (Mx), Nasal (Na), Parietal
(Pa), Premaxillary (Pmx), Supraoccipital (So), Supraorbital processes of
frontals (S. or Pr.), Squamosal (Sq), Zygomatic arch (Zyg).
There is no doubt that the evolutionary stage of the Mio-Pliocene
sirenians from Florida parallels that of the European sirenians of a
similar age. A puzzling complex of European specimens were originally
ascribed to the genera Metaxytherium, Halianassa, and Felsinotherium.
These genera have never been clearly defined. Herein lies the problem
of comparing specimens from Florida, assigned to these three genera,
with European forms, which are not distinctly defined nor differentiated
The close similarity of Metaxytherium and Halianassa has long been
recognized, and Simpson (1945), in his excellent taxonomic paper, re-
garded Metaxytherium Christol (1840) as a junior synonym of Halianassa
Meyer (1838). The writer and most recent students have followed the
work of Simpson; however, Kellogg (1966: 68-70), in his detailed review
of the problem, showed that the name Halianassa Meyer (1838) was not
originally accompanied by a description and is therefore a nomen nudum.
Kellogg's well documented work traces the type species of this genus to
Halianassa studeri Studer (1887), thereby giving Metaxytherium Christol


(1840) precedence. This has long been a difficult nomenclatorial prob-
lem, but on the basis of documentation by Kellogg I am changing my
usage of Halianassa to Metaxytherium.

The frustrating problem of generic distinction of the Mio-Pliocene
Metaxytherium-Felsinotherium complex has been emphasized by a num-
ber of authors, including Gregory (1941: 40), who stated, "Possibly the
most feasible solution of the taxonomic problem would be to regard the
differences between Miocene and Pliocene Halitheriinae as too slight
and variable to be worthy of generic recognition, and to place the species
now referred to Felsinotherium in Metaxytherium, recognizing that some
species of that genus showed certain evidence of advancement over their
earlier relatives." The problem of separation of the genera Metaxyther-
ium and Felsinotherium was further emphasized by Simpson (1932: 479)
when he stated in his discussion of two species (F. ossivalense and M.
floridanum) of Florida felsinotheres: "Even if neither species of Felsi-
notherium be directly derivative from any known species of Metaxyther-
ium, the differences are really so slight and unimportant that no
thoroughly satisfactory diagnoses separating the two genera have been
proposed. Capellini (1872), in fact, considered the genera to be 'sy-
Deperet and Roman (1920: 48) distinguished Felsinotherium from
Metaxytherium by the presence in the former of a slightly more advanced
reduction and more quadrate shape of M1, M2, and P4, plus an inflection
slightly more pronounced in the base of the premaxillary rostrum. Of
two species referred to Felsinotherium, the large F. forest has slightly
more quadrate teeth than does F. serresi. This was previously deduced
by Abel who, on the basis of this difference, placed F. serresi in the genus
Metaxytherium. If the degree of premaxillary flexure is regarded as a
valid distinction between the two genera, one finds a very similar ex-
panded downturned rostrum in F. foresti2, F. serresi, and M. cuvieri, as
opposed to a less deflected and less expanded rostrum in Metaxytherium
jordani (= Halianassa vanderhoofi) from California. Expansion of the
rostrum is most pronounced in the old, tusk-bearing adults. While not
necessarily old individuals, the European Sirenia mentioned bear well
developed tusks, as opposed to the California metaxytheres, which have
nonfunctional incipient incisors. Degree of rostral flexure, however,

' Other authors have pointed out this statement but fail to comment on the meaning of "a
slightly more advanced reduction." To me it means fewer cheek teeth, but in all cases I know
there appears to be a maximum of five cheek teeth in all adult Sirenia attributed to Metaxy-
therium and to Felsinotherium. If this is the meaning, I fail to see a difference between the
2 Dorsal portion of rostrum missing.


appears to be somewhat greater in the European Sirenia mentioned than
in M. jordani.
In a recent generalized, but completely summarized, review of Meta-
xytherium and Felsinotherium, Dechaseaux (1958: 353) recognized the
difficulty of generic distinction, but nevertheless differentiated Felsi-
notherium in this manner:

"II differe de Metaxytherium par une accentuation des tendances
evolutives marquees par le genre miocene plutot que par la
possession de nouveaux caracteres, aussi est-il tres difficile de
donner ses caracteristiques; plusiers auteurs considerent meme
ces deux genres comme synonymes.
"I y a d'assez grandes variations dans les caracteres du crAne
et de la dentition selon les esp6ces (voir fig. 33); les molaires
peuvent 6tres simples, primitives, quoique de type bunodonte,
ou bien poss6der des tubercules accessoires et un arrangement
en series des cuspides."

The difficulty of working with two genera that can be differentiated
only by vague criteria by the many competent students who have com-
pared them leads me to propose that the species of Felsinotherium be
placed in the genus Metaxytherium. This procedure is not suggested
as a matter of convenience to end the confusion experienced by many
students of the subject, but rather it is an attempt to frame the many
isolated specimens attributed to a bewildering number of species into
what appears to be a natural genetic group. Animal affinities are there-
fore considered primary, but the secondary result of cosmopolitan com-
munication is also highly important and can no longer be ignored in this
taxonomic problem.
The principal source of Mio-Pliocene metaxythere specimens in Flor-
ida is the Bone Valley district. The name "Bone Valley" is derived
from exposures of fossiliferous phosphatic gravels in Polk and Hills-
borough counties in south-central Florida. Prior to the use of the term
formation, exposures had been referred to as Bone Valley gravel, "land
pebble phosphates," or "pebble phosphates." Matson (1915: 36) de-
scribed the Bone Valley sediments thus:

"The Bone Valley gravel consists of rounded pebbles of
phosphate embedded in a matrix of sand or clay overlain by
varying thicknesses of loose or semi-indurated sand. The maxi-
mum thickness of this formation is probably more than 50 feet,
but only about one-third of this thickness should be assigned
to the phosphate.


"The phosphate-bearing portion of the Bone Valley forma-
tion is a gravel containing rounded and subangular pebbles of
phosphate of varying degrees of coarseness intermingled with
more or less sand and clay. In general the deposit shows dis-
tinct stratification, some beds being wonderfully persistent over
a distance of several hundred yards though others are lenticular
and extend only short distances. There are numerous altera-
tions of coarse and fine material and many of the layers are
distinctly undulating-the entire formation bears evidence of
having been deposited in shallow water where conflicting cur-
rents gave rise to irregular bedding and rapid alterations of
sediments of varying coarseness."

Most exposures of the unit are the result of phosphate mining opera-
tions, rather than natural outcrops, with the result that lithologic de-
scriptions are largely limited to regions where phosphatic concentrations
are relatively thick.
A general consensus of opinion is that the Bone Valley sediments are
in part estuarine and in part shallow marine in origin, an environment
with which most fossil and living sirenians have usually been associated.
There are also stratigraphic indications that periods of time existed when
meandering streams crossed lowland and swampy areas. The Bone
Valley was considered Pliocene in age by the United States Geological
Survey (Cooke 1945: 197-210), much of the evidence being based on
Simpson's determination of the age of fossil vertebrates found in the unit.
Because the methods of phosphate mining often involve mixing of the
underlying formations (usually Middle Miocene Hawthorne Formation),
as well as overlying Pleistocene material, care must be taken to determine
the faunal elements contemporaneous with deposition. Based on the land
mammals determined by Simpson (1930: 180) to be indigenous to the
Bone Valley, this stratigraphic unit is referred to the Hemphillian by Stir-
ton (1936) and Wood et. al. (1941). More recent work by Olsen (1959:
40) and the Florida Geological Survey sums up the present questionable
use of the term "formation" in relation to the Bone Valley sediments:
"Perhaps no area in Florida has caused the concern, in regard to dating
the vertebrate fauna it contains, as has the Bone Valley of Polk County.
The term "Bone Valley" should be used to define a geographic bound-
ary, rather than a stratigraphic unit, as beds ranging from Upper Mio-
cene through the Lower Pleistocene are known to occur in the "Bone
Valley" as defined by earlier workers. These beds, in some cases, show
a lithologic change and are not clearly mapable units." Olsen (1959:
42) states the Bone Valley fauna is Mio-Pliocene in age, corresponding


to the upper Barstovian and lower Clarendonian. Recent study of a
faunule collected from the Bone Valley Formation in Manatee County
indicates a Hemphillian age (Webb and Tessman 1968).
The first important record of a sirenian from the Bone Valley Forma-
tion of Florida is that by Matson (1915), in which he referred a portion
of an upper right maxilla to a "manatee." This specimen was later re-
figured and described by Hay (1922) as a new species, Metaxytherium
floridanum holotypee: USNM 7221), a taxonomic assignment that cor-
rectly removed it from the Trichechidae. Allen (1923) subsequently
referred a humerus and a number of vertebrae and ribs that had been
collected in the "land-pebble phosphate deposits" near Mulberry, Flor-
ida, to M. floridanum. The posteranial material was not associated with
teeth or other diagnostic cranial elements; therefore, the taxonomic as-
signment remains in some doubt. Simpson (1932) compared the hu-
merus and vertebrae with other more complete specimens and referred
Allen's material to Felsinotherium. Other sirenians from the sedimen-
tary unit were described by Simpson under the name Felsinotherium
ossivalense, the holotype (AMNH 26805) of which is a relatively un-
worn but fragmented M3. An important specimen referred to this species
by Simpson is the anterior half of a cranium in the Florida Geological
Survey collection (FGS V5454) that lacks the anterior end of the rostrum
and the zygoma (Simpson 1932: 452, fig. 12 A-C). The dentition of this
cranium was later described more fully by Gregory (1941), and certain
cranial elements were further amplified. The only other figured speci-
men from the Bone Valley sediments (the posterior half of a cranium)
is present in the Florida Geological Survey collection (FGS 3211), and
lacks teeth, rostrum, and the anterior half of the frontals. Another
specimen (FGS 3232), consisting of a well-preserved, fused parietal-
supraoccipital element, was discussed by Simpson (1932) but not figured.
With the fossil material found to date, the most feasible method of
differentiating Florida sirenians is to correlate two of the more common
diagnostic elements, the cranial caps with upper cheek teeth. To date
the only figured specimen from the Bone Valley beds in which these
elements are found associated is a relatively complete cranium (FGS
V5454). This paper will add three more specimens in which various
cranial elements and teeth are associated, as well as a number of isolated
cranial caps, individual upper and lower teeth, mandibular specimens
bearing teeth, and varied postcranial elements. On rare occasions
where postcranial elements have been found definitely associated with
a cranium this relationship is described; however, because such elements
appear to be more conservative in change than the cranium, they have
limited value in determining relationships. An exception to postcranial


conservatism is the sacral elements, but these bones are unknown in the
Florida Mio-Pliocene sirenians, undoubtedly due in part to their reduc-
tion in size and difficulty of recognition.

Evaluation of cranial and dental characters, including their assign-
ment to genus and species, age of individual, or dimorphic status, has
long plagued European and North American students of Sirenia, par-
ticularly when the study concerned the relatively abundant specimens
of Mio-Pliocene age. Among the more common specimens are cranial
shields composed of ankylosed supraoccipital and parietals. Two major
shield types are readily distinguished, one of which consists of swollen
temporal crests that are lyriform (i.e., approach each other in the mid-
line), and the other being a more squared-up cranial shield. Temporal
crests in the latter are essentially lacking, not swollen, and widely
separated, with lateral boundaries varying from a gentle rounding to a
prominent upraised sharp edge.
Deperet and Roman (1920) considered Metaxytherium as having two
parallel ramifications. M. cuvieri with lyriform temporal crests and of
Vindobonian age represents one stock. The second, more abundant,
stock is represented by forms with separated crests: M. krahuletzi and
M. beaumonti of Burdigalian age and M. petersi and M. aff. peters of
Vindobonian age. Sickenberg (1928), on the basis of a complete young
specimen of Tortonian age from the Vienna basin, referred M. petersi to
a new genus, Thalattosiren. M. aff. petersi is then to be referred to T.
aff. peters; however, I would refer this specimen, as illustrated by
Deperet and Roman (1920: pl. VII, figs. 3 and 4), to M. cuvieri because
of geographical and geological proximity and osteological similarity.
T. petersi of the Vienna basin is of similar age as the Mont-de-Marsan M.
aff. petersi from France, but there are no good adult parietal specimens
of the Austrian T. petersi (no figured specimens known to me) with
which to compare the French specimen. The parietals of M. cuvieri
from France, on the other hand, are both reasonably abundant and well
preserved and do not differ from the Mont-de-Marsan specimen. Further-
more, DepBret and Roman (1920: 29) stated that the reference of the
Mont-de-Marsan cranial roof to M. aff. petersi was tentative.
Simpson (1932: 455) found a condition in the cranial roofs of speci-
mens from Florida, which he referred to the genus Felsinotherium,
similar to that in the European Metaxytherium. He described this as
follows: "The sagittal crests and general contour of the cranial roof
differ considerably in the several available specimens, but the degree to
which this variation is individual or specific is not clear." Later, in a


discussion of specimens V3211 and V3232 Simpson stated (1932: 455-
456): "In the former specimen the crests on the parietals are very
prominent, elevated and swollen, while on the latter they are very weak,
little more than the sharp angle between the superior and lateral parts
of the parietals. This distinction may be in large part due to age or sex,
or may characterize the two species."
The progressive development of crests and rugosities with increasing
age of individuals has been noted many times in various mammalian
groups. This process applies to the degree of development in the tempo-
ral crests of the two living genera, Trichechus and Dugong, and has been
recorded to some extent from the fossil record. In the young Trichechus
and Dugong the crests are virtually nonexistent, and therefore are de-
scribed as lacking to widely separated. Prominent crests progressively
develop on males and females of both genera, an indication that such
rugosities are probably attachments for muscle insertion and to a large
extent a factor of age. Neither temporal crests nor accompanying por-
tions of the cranial roof are seen to differ between sexes, and if this
condition is applied to metaxytheres, two species or other taxa are repre-
sented by the differentiation of cranial shields.

Tusks of a variety of marine mammals have been found in North
America since Cope's 1869 proposal of Hemicaulodon effodiens, but no
exacting dental study has been made of these isolated specimens. A
series of five specimens consisting of partial premaxillaries with tusks or
individual tusks are ascribed by Kellogg (1966: 78) to dugongids from
as early as Miocene. Following the well documented study by Ray
(1975) I consider the lost Hemicaulodon tusk to belong to the walrus
Odobonus rosmarus and the three individual tusks (USNM 9457; USNM
23110; AMNH 9852) studied by Kellogg (1966) to belong to physeterid
sperm whales. The two remaining specimens identified as Sirenia by
Kellogg, the type right premaxillary of Dioplotherium manigaulti Cope,
and a left premaxillary (YPM 21334) dredged from the Wando River,
South Carolina, may belong to this order, but the fragmented specimens
do not permit generic assignment. The only North American sirenian
with a complete cranium bearing tusk-like incisors is the Middle to Late
Miocene Halitherium olseni from Florida. Other cranial specimens such
as Caribosiren from the Middle Oligocene of Puerto Rico and Hespero-
siren from the Middle Miocene of Florida lack tusks, whereas Metaxy-
therium jordani (= Halianassa vanderhoofi) from the Late Miocene or
Early Pliocene of California reveals two small pits at the end of the snout,
which likely housed nonfunctional incisors.


Of the Recent Sirenia, the genera Hydrodamalis and Trichechus are
edentulous, each having two minor pits housing nonfunctional incisors,
whereas the genus Dugong possesses strong tusks in both sexes, with
those of the females being slightly smaller. These examples suggest
that the presence of tusks in sirenians does not indicate dimorphism,
since both sexes seemingly either bear or lack tusks. Unfortunately this
suggestion raises a paradox when one considers the following European
and North American metaxytheres, which have in common geologic age,
similar number of cheek teeth, almost identical cusp pattern of cheek
teeth, and similar cranial construction throughout. A comparison of
several pertinent cranial and dental characters can be seen in Table I.


Separated Lyriform
Species Crests Crests Tusks No Tusks
Felsinotherium forestii X X
F. serresi X X
Metaxytherium cuvieri X X
M. jordani ( = H. vanderhoofi) X Pits, nonfunctional
M. calvertense (UF 115791) X Pits, nonfunctional
1 Finest preserved cranium yet found in North America. From Florida, probably Late Miocene
or Early Pliocene age. Not reported upon in this paper.

From this small sample of closely related Sirenia it may be seen that
both types of cranial shields, whether separated or lyriform temporal
crests, may bear or lack tusks. The first three species are European
forms and bear tusks; the latter two tuskless species are from North
America. Thalattosiren, a closely related European sirenian of compar-
able geologic age, lacks tusks. It may be noted that in the European
Sirenia both types of cranial shields are associated with tusks, a feature
that can be interpreted as representing either two separate species or
merely sexual dimorphism. In the two North American specimens both
types of cranial shields lack tusks, which would result in an analogous
interpretation. Although the few well-preserved specimens suggest
tusked European versus tuskless North American metaxytheres, it seems
unlikely that the North American Mio-Pliocene forms would be virtual
structural duplicates of their European counterparts, except for the
absence of tusks. In a number of museums and small private collections
in Florida, fossil sirenian tusks have occasionally been noted, and chance
alone makes it seem unlikely that all belong, or can be assigned, to a
tusked genus such as Halitherium. Whether Dioplotherium Cope 1883,
a partial premaxillary with tusk, will eventually prove to be a metaxy-


there remains a question. Kellogg (1925: 59) placed Dioplotherium in
the synonymy of Metaxytherium, a reasonable assignment but one that
cannot yet be proved. Later Kellogg (1966: 91) re-elevated Dio-
plotherium to a distinct genus, which I also prefer until synonymy is

GENUS Metaxytherium CHRISTOL 1840
The Florida Mio-Pliocene metaxytheres are separable into three
species, the distinction being basically contingent upon the degree of
separation, size, and shape of the parietal temporal crests, coupled with
variances in the shape and complexity of the upper cheek teeth, par-
ticularly M3, which is always less worn than preceding teeth. Total
morphologic features are also naturally considered, but the cranial roof
and M3 are among the more common characters that can be used to
demonstrate relationship. The three Mio-Pliocene species are Metaxy-
therium ossivalense, M. floridanum, and M. calvertense.

Metaxytherium ossivalense SIMPSON 1932
DIAGNOSIs:-M3 obovate (egg shape) to subtriangular; M2 subquad-
rate to obovate; M1 quadrate; small cuspule blocks anterior cingular
valley on buccal side of M3; metaconule displaced forward but median
transverse valley relatively straight on M3, unblocked by cuspule on
either buccal or lingual side; low broad anteroposterior boss on frontals;
parietal with temporal crests thin, upraised, widely separated; supra-
occipital probably touched border of foramen magnum.
DIscussIoN.-Specimens assigned to this species by Simpson are the
type (AMNH 26805), which consists of part of the left upper maxilla
with broken M3 and alveoli, and the referred anterior two-thirds of a
cranium; FGS V5454, which lacks the downturned extremity of the pre-
maxillaries and the zygoma. The latter specimen is assigned by me to
Metaxytherium calvertense Kellogg 1966. Specimens referred by me to
M. ossivalense are FMNH P27227, an essentially complete posterior two-
thirds of a cranium broken vertically at a point immediately anterior to
M1; MCZ 4062, a parietal-supraoccipital cranial cap; UF 2847, an upper
left M1 or M2; UF 11575-77, three M3; UF 3596, consisting of a partial
rostrum, two fragmented unworn M3, a partial mandible with teeth, an
atlas, and a fragmented scapula. Other specimens tentatively assigned
to this species are UF 3281, an upper left M3; FMNH P27228, a palate
and maxillaries containing M1-2 and the alveolus of P4; FMNH PM155,




S- -----------





FIGcUE 2.-Correlation of Hawthorne and Bone Valley strata with North American
provincial ages. Bone Valley Gravels of Stirton (1936), Wood et al. (1941), and
Webb and Tessman (1968). 2 Bone Valley Fauna of Olsen (1959).
a plaster cast of right and left maxillaries containing on either side (ex-
ception left M') alveoli for P4, a worn M1 and M2 and partially erupted
M3, FGS V5953, a right M3.
PREMAXULLARIES.-No evidence of incisors is present in the premaxil-
laries of UF 3596. This feature is characteristic of all North American
metaxytheres reported to date, in contrast to European metaxytheres,


At" t

FIGURE 3.-Metaxytherium ossivalense (FMNH P27227): dorsal view of cranium.

which have well developed incisors. Owing to the general paucity of
rostral specimens, it remains a question whether future finds will con-
tinue to support present information concerning lack of all but incipient
tusks in North American forms. As stated previously, this lack of incisors
is also characteristic of the European Thalattosiren, the North American
Caribosiren, Hesperosiren, and Metaxytherium, the recently extinct Hy-
drodamalis, and the living Trichechus. A moderately broad, swollen boss
found on either side of the rostrum of M. ossivalense lacks the prominent
development present in Dugong. In an accentuated manner, similar to
the development in Dugong, a double arch exists in the anterior border


FIGURE 4.-Metaxytherium ossivalense (FMNH P27227): ventral view of cranium.
of the mesorostral fossa. The smaller anterior arch, also peculiar to
Thalattosiren petersi (Sickenberg 1928: 297, fig. 1) and Dugong, appears
to have its incipient development in Eotheroides. Such a double arch is
absent in Eocene Sirenia (exception, Eotheroides) and in the Oligocene
Halitherium, appears to be present but not prominent in Hesperosiren,
and is developed to a moderate degree in the Oligocene Caribosiren
(Reinhart 1959: 10, fig. 26) and the Mio-Pliocene metaxytheres. The
degree of arch development seems to be associated with a well-developed
and wide nasopalatine canal and most prominent in those genera having
wide grooves for the nasopalatine cartilaginous canal.


FIGURE 5.-Metaxytherium ossivalense (FMNH P27227): posterior view of cranium.

NAsALs.-Because the nasal-frontal sutures were fused, the boundary
between these elements is indeterminate, and it cannot be ascertained
whether the nasals were separated or met in the midline. Closely re-
lated forms in the same geographical area and stratigraphically equiva-
lent sediments have nasals separated by a forward extension of the
frontals; a condition characteristic of the metaxytheres and the referred
felsinotheres. The only sirenian recorded from the North American-
Caribbean area that is believed to have nasals meeting in the midline
is the Florida Halitherium reported upon in this paper. Although the
nasal region is largely missing in this halitheriine, it is comparable in all
other features to the European genus and therefore considered to have
similar bony relationships in this area. Among European genera in
which the nasal-frontal relationship is known, only the metaxytheres and
referred felsinotheres are characterized by nasals that are not united in
the midline. An exception to this statement is Metaxytherium n. sp.,
which has nasals meeting in the midline (see illustration by Kaltenmark
[1942: 107]). All other metaxytheres that show the nasal region are in
disagreement with this reported condition. For a more complete dis-
cussion of the nasal-frontal relationship see Reinhart (1959: 59-62).
FRONTALS.-The frontals of metaxytheres vary from forms with
squared lateral boundaries and a relatively rectangular outline to those
exhibiting a gradual anterior expansion of these elements from the


capital cap.

is present in the anterior midline, a condition that appears to characterize

anterior end of the frontal boss, but the distal ends are broken. When'
r t I." .

Fircur 6.-Metaxytherium ossivalense (MCZ 4062): dorsal view, parietal-supraoc-
cipital cap.

parietal-frontal contact. M. ossivalense features elongate, squared fron-
tals bounded by weakly developed temporal ridges that parallel each
other back to the parietals. A low broad anteroposterior elongate boss
is present in the anterior midline, a condition that appears to characterize
all species of Metaxytherium, as well as referred felsinotheres. Gradual
expansion of the postorbital processes begins at a point lateral to the
anterior end of the frontal boss, but the distal ends are broken. When
the frontals and parietals are found separated, the frontal-parietal sutures
are marked by a complex series of slanting ridges, a condition often well
expressed in both young and mature individuals.
PARMETALS.-The configuration of the parietal-frontal suture, seen on


5- \

FIGURE 7.-Metaxytherium ossivalense (FMNH P27228): occlusal view of left maxil-
lary containing M1-2 and P4 alveolus.




FIGURE 8.-Metaxytherium ossivalense (FMNH PM155): occlusal view of M1-3 and
P4 alveolus (M3 partially erupted).

both dorsal and dorsolateral surfaces, is approximately the same as that
found in other metaxytheres and in all Sirenia in general (i.e. long nar-
row arms of the parietal project forward over the frontal on either side
of the midline). Lyriform temporal crests are relatively thin, sharply
upraised, and widely separated. Outline of the crests is similar to those
attributed to Metaxytherium aff. petersi Abel (Depiret and Roman 1920:
pl. 8, fig. 4). One element of distinction among the Florida metaxy-
theres is based upon variations in the degree of development and shape
of the parietal temporal crests, as found in adult specimens. The differ-
ence in shape of temporal crests is probably not due to age but may
instead be a sexual, or perhaps specific difference that developed in popu-
lations separated in geologic time. The last theory is advanced despite
the assumption that animals existing during the years encompassed by
deposition of the Bone Valley sediments are thought to be essentially


contemporaneous. Temporal crests are often moderately developed on
young adults, with growth progressively accentuated with increased age.
The general unworn quality of the crania would seem to rule out the
possibilities of removal of the crests by abrasion or solution, either of
which would leave some indication of such actions.
The parietal-supraoccipital element is stoutly fused into a single
bony unit, as is characteristic of the Sirenia. This ankylosed condition
is present even in the fetal stages of Dugong and Trichechus, but there
is no complete union between the supraoccipital and other adjacent
bones until the adult stage is reached. In FMNH P27227 the parietal-
supraoccipital complex meets at an angle of 101 degrees, which is com-
parable to the 105 degrees of similar elements in FGS V3232, referred to
M. floridanum. In this instance the slight difference in degrees between
the two species may be the result of preparation in the former specimen
or to age difference, a greater angle being characteristic of youth.
Measurements in the union of the parietal-supraoccipital shield in the
metaxytheres are in contrast to Dugong, in which these elements meet
at a much higher angle (122 degrees in the young specimen at hand).
Internally the parietals form the greater portion of the braincase in
the superior and posterior areas. The cerebral hemispheres are elongate,
laterally compressed, and separated by a very prominent longitudinal
median carina, which forms an internal occipital protuberance posteriorly
and ends abruptly, abutting against the supraoccipital. The median
carina divides into two separate ridges about 15 mm posterior to the
frontal-parietal suture in the braincase. These continue forward as two
low-rounded ridges, become progressively smaller anteriorly, and spread
out on either side of the crista galli. This same bony condition is true
in the large Metaxytherium jordani of the Pacific coast, in M. calvertense,
and in FGS V5947 (Metaxytherium indet.); in contrast to Trichechus
and Dugong, in which the median carina continues anteriorly to the
crista galli as a single ridge that spreads out just before contact with this
bony complex. The relationship of the parietals to other bony elements
within the braincase is about the same as that found in all Sirenia; how-
ever the elongate shape of the cerebral hemispheres differs from those
in certain other genera, which are more subquadrate in this region.
SUPRAOCCIPITAL.-In FMNH P27227 an unusually well developed
external occipital protuberance, resembling that of M. jordani, is bounded
on either side by deep grooves, which in turn are bordered by an oblique
rugose area. The prominent degree of development of tuberosities and
rugosities, as well as the large size of the parietal, suggests the cranial
specimens assigned to this species are adults, perhaps tending toward
old age. Specimen MCZ4062 displays a broadly rounded lambdoidal


crest, in front of which lies a small but noticeable boss in the midline
of the parietals.
ExoccIprrALs.-The condyles are semilunar in shape, with the arti-
cular surfaces convex both transversely and vertically. Condylar fora-
mina are similar in size to those of Metaxytherium jordani ( = M. vander-
hoofi) and are comparably much larger than those in Dugong. The
supracondylar fossa is broad and shallow and terminates in sharp crests
at its lateral borders, in contrast to the gently rounded borders of com-
parable areas in Dugong. Exoccipital borders are unusually prominent,
overriding the posterior end of the post-tympanic processes that abut
against it. Inferiorly, the paroccipital process is very thin and is char-
acterized by the deep concave surface formed by a border that curves
back toward the median line. Accentuation of the paroccipital process
is shown by their extension below the ventral surface of the condyles.
Sutures between the supraoccipital and exoccipitals are indeterminate
in FMNH P27227, but judging from the angle at which the exoccipitals
rise to form the dorsal portion of the foramen magnum, it is quite possible
that the supraoccipital closely approached or touched this foramen. If
the supraoccipital formed part of the dorsal border of the foramen mag-
num, it lacked the high dorsal indentation illustrated for "Metaxytherium
nov. spec." (Kaltenmark 1942: 105, fig. 1). The foramen magnum of
FMNH P27227 is a flattened transverse outline, with a suggestion of a
very slight rise in the dorsal midline. A dorsal indentation in the border
of the foramen magnum is present to some degree in practically all
sirenians, the only exceptions being those of Trichechus and of Felsi-
notherium serresi (D6peret and Roman 1920: 5, fig. 4), the later illu-
strated from a cast. On the same page, however, it is stated that the
occipital foramen is oval and higher than wide, which is in agreement
with the illustration but different from all other metaxytheres. Both
Abel (1904) and Simpson (1932) attributed considerable taxonomic
importance to the proximity of the supraoccipital to the dorsal margin
of the foramen magnum and to the sutural angle formed by the supra-
occipital and exoccipitals. I am in complete agreement with this con-
Beginning with Eocene sirenians in the family Dugongidae, there
has been an irregular progressive tendency for the ventromedial portion
of the supraoccipital to be near, or to form a minor part of, the mid-dorsal
border of the foramen magnum. An exception is the Eocene Protosiren
fraasi (Sickenberg 1934: pl. 1, fig. 2), in which a character of the recent
Dugong had already appeared (i.e. the supraoccipital touches the fora-
men magnum and the supraoccipital-exoccipital angle is approximately
160). This condition is in contrast to that of Trichechus and the Eocene


Eotheroides (=Eosiren) libyca (Andrews 1906: 198-199, pl. XX, fig. 16);
in Eotheroides the supraoccipital is separated from the foramen magnum
by 23 mm, and the supraoccipital-exoccipital angle is 1600. The Oligo-
cene Halitherium schinzi has a similar supraoccipital-exoccipital relation-
ship as is present in Eotheroides; the exoccipital angle is 1670 and the
supraoccipital approaches to 21 mm of the foramen magnum. Separa-
tion of the supraoccipital from the dorsal border of the foramen magnum
in Trichechus is considered retention of a primitive feature.
Exclusion of the supraoccipital from the foramen magnum has been
considered a prototypic arrangement, but participation of the supra-
occipital in the foramen margin of the otherwise structurally primitive
Middle Eocene Protosiren cannot be overlooked. Perhaps Protosiren,
as do most other Sirenia, bore a combination of osteological advances
and a retention of primitive phases in its construction.
Abel (1904) has given figures for Metaxytherium krahuletzi in which
the supraoccipital is 18 mm from the foramen magnum, and the exoc-
cipitals meet at an angle of about 1300. Metaxytherium jordani (= Hali-
anassa vanderhoofi) (Reinhart 1959: 26) is in essential agreement, with a
separation of 10 mm and an angle of 1360. Coupled with the California
Middle Miocene M. jordani is the immature Baja California Metaxyther-
ium reinharti, which bears a comparable 1370 exoccipital angle and an
estimated 5 mm supraoccipital-foramen magnum separation. According
to Kaltenmark's figure (1942: 105, fig. 1), the exoccipitals of Metaxyther-
ium nov. spec. meet at an angle of about 1300 and the supraoccipital
reaches the foramen magnum. In Felsinotherium forest Capellini, the
supraoccipital may just touch the foramen magnum and the exoccipital
angle is about 124. The supraoccipital may either touch or be ex-
cluded from the foramen magnum in Felsinotherium serresi Gervais.
Dep6ret and Roman (1920: fig. 2) illustrated F. serresi with the supra-
occipital excluded and an oval rather than the more characteristic cordi-
form outline of the foramen magnum. Simpson (1932: 457) described
two specimens that he referred to Felsinotherium; FGS V3211 with an
exoccipital angle of about 1150 and the supraoccipital not over 5 mm
from the foramen magnum', and FGS V3232 with an angle of about 1200
and the supraoccipital closely approaching the foramen magnum. From
the specimens at hand the supraoccipital of Dugong touches the foramen
magnum and the exoccipital angle is 1400. Comparable elements of the
adult Trichechus display a 12 mm separation and an angle of 1360.

1 Simpson's illustrations of this specimen (1932: 457, fig. 14B) differs from the text measure-
ments, and I agree with the illustration, which shows that the supraoccipital formed the dorsal
border of the foramen magnum. Sutures of the specimen are not distinct in the dorsal midline
of the foramen magnum; therefore this feature is a matter of interpretation.


There may be a slight discrepancy in angles as a result of variation in the
preservation or preparation of the fossils.
It is interesting to note the similar configuration of the parietal-
supraoccipital complex between the young Pleistocene trichechids and
certain older Mid-Oligocene-Pliocene sirenians. A series of water-worn
parietal-supraoccipital elements (UF 3965 and UF 14222-14226) re-
ferred to Trichechus sp. display a narrow cranium with convergent lyri-
form crests. The thin elongate roof distinguishes these specimens from
the larger metaxythere crania but suggests affinities with Hesperosiren
(MCZ 4432), Halitherium, and Caribosiren. Coupled with this im-
portant feature, a rugose sutural surface on the supraoccipital indicates
this element met the exoccipitals at an angle of 146 and was separated
from the foramen magnum. In crania where the supraoccipital ap-
proaches or takes part in the foramen magnum, the element usually
thins in the ventral midline and bears a small dorsal indentation. The
referred specimens thicken at this point and show no indentation; the
supraoccipital was therefore well separated from the foramen magnum.
The occiput is unknown in Caribosiren, but the lack of contact between
the supraoccipital and the foramen magnum differentiates the Florida
trichechids from Hesperosiren. The supraoccipitals of these specimens
are readily distinguished from H. olseni in the more posterior thrust of
the ventral midline and the much thicker medial ventral area, which lacks
the central fossa of Halitherium.
Unfortunately the only locality data given for UF 3965 is "Florida."
UF 14222-14226 were collected in Pleistocene sediments in the Wacca-
sassa River, Levy County, Sec. 16, R15E, T15S.
SQUAMOSAL.-The posterior portion of the cranial part of the squa-
mosal is not in contact with the exoccipital dorsally, the two bones being
separated by a narrow, elongate fontanelle which in its ventral portion
leaves the mastoid exposed. The corresponding squamosal-exoccipital
in Dugong is wider but of similar length. As in all sirenians, the squa-
mosal forms a small circle directly above the post-tympanic process and
around the periotic, thus leaving the latter bone exposed externally. In
contrast to Dugong, the post-tympanic process is well developed, with
the ventral portion directed anteriorly and terminating abruptly in a blunt
ovoid. The squamosal-parietal relationship shows a detailed similarity
to that of Dugong, particularly in the angulation of these bones along
the temporal wall of the braincase. The cranial wing is set into and
topped by a thin overlap of the parietals. The zygomatic root and arch
has the same general configuration as other specimens of Metaxytherium,
except M. jordani, which has a pronounced anterior inclination. The
zygomatic root is elongate, vertically thin, and terminates posteriorly in


a deep-U that is open in a posterior direction. Immediately behind,
and parallel to, either broken glenoid surface is an elongate transverse
depression, which is much deeper than in Dugong. A high narrow tri-
angle is formed by a cross-section of the sinous arch, with all borders
ending in thin crests except the ventro-anterior which is broadly rounded.
AUDITORY REGION.-All of the ear bones are of a dense composition
and have a swollen appearance, particularly the periotic and mastoid.
As in the Sirenia in general, the periotic lies in a hollow cup formed
in the base of the squamosal, with the more swollen and posterior mas-
toid lying in a shallow basin formed chiefly within the paraoccipital
processes. The tympanic ring is missing, but its attachments to the
petrosal are similar to those in Dugong. Two of the ossicles, the incus
and the stapes, are present. The incus is a stout bone, with complex
articular surfaces for the malleus on its inferior side and a convex smooth
surface on its superior side. The upper portion of the incus lies in the
deep groove of the fossa incudus. The crus breve is shorter and stouter
than that of Dugong dugong (=D. australis) (Doran 1879: pl. LXIII,
fig. 29); the crus longum is broken. The left stapes, which extends
inward and lies slightly below the incus, is a stout cylindrical bone per-
forated by a minute stapedial foramen near the base; the right stapes
appears to have been imperforate or nonfunctional. This ossicle is 12.2
mm long with the foramen 3.7 mm from the base. These measurements
are comparable to the stapes of M. jordani, which is 14.8 mm long with
the stapedial foramen 4.4 mm from the base, and to that of M. jordani
(=H. vanderhoofi), which is 12.6 mm long with the foramen also 4.4
mm from the base. The entire group resembles in detail that of Dugong
as illustrated by Doran (1879: pl. LXIII, fig. 30).
BASIOccIPrrAL.-There is little to distinguish the basioccipital from
similar elements in other metaxytheres or from Sirenia in general. The
ventro-anterior surface is greatly thickened, rugose, and roughly rounded;
the dorsal surface is flat, with a thin spatulate extension overlying the
basioccipital. The maximum thickness of this element at the anterior
end is 39 mm; the minimum width, 30.3 mm, is at the center.
BASISPHENOm.-The basisphenoid of FMNH P27227 is comparable
in size to that of Dugong, being larger only because of a larger cranium.
Apparently this element maintains a relatively constant size, depending
upon the size of the cranium. Between the base of the pterygoid
processes, the ventral surface of the basisphenoid is approximately 30.5
mm wide; from the ventral surface of the basisphenoid the pterygoid
processes are inclined laterally and measure about 68 mm between the
inner surface of their ventral tips. The ventral surface of the basisphe-


noid and basioccipital meet at an angle of about 1200. Unlike most
other sutures in the cranium, the basisphenoid-basioccipital suture is
not fused. This condition is the same in M. calvertense (FGS V5454),
whereas the suture is thoroughly fused in FGS V3211, also referable to
M. calvertense. Sickenberg (1931: 428) notes the general occurrence
of a nonfusion of this suture, discusses its relationship to other condi-
tions in the Sirenia, and attributes these to hypothyroid metabolism.
Nonfusion of this suture occurs in Caribosiren and M. jordani (=II.
vanderhoofi), but both specimens studied are young adults, as judged
by the general open sutures throughout the cranium of each. Dilg
(1909: 115) stated that the two bones are separated in Trichechus up
to the ninth year, indicating that the state of fusion revealed in FGS
V3211 marks it as a very old individual.
Endocranially, a prominent olfactory groove on either side of the
crista galli leads anteriorly into a perforated fossa, partially destroyed
in FMNH P27227 but well displayed in M. calvertense (FGS V5454).
Behind and ventral to these grooves are the optic foramina, separated
by a thin crest of the presphenoid. In a transverse line between the
optic foramen the bones are perforated by many minute foramina.
PTERYGOIDS.-A comparison of the pterygoids of M. ossivalense and
M. calvertense reveals no differences. Well-developed pterygoid proc-
esses begin at the posterior portion of the basisphenoid and slant anter-
iorly. They are characterized by stout bases and are deeply grooved
posteriorly from base to tip; the lateral and medial margins of the
grooves are more nearly equal than in Dugong, in which the medial is
much the more prominent of the two. The postero-lateral and medial

TABLE 2.-CRANIAL MEASUREMENTS OF Metaxytherium ossivalense (FMNH P27227).

Length, posterior end of occipital condyle to anterior end of M2 ------....-- 258.0
Length, posterior border of mesorostral fossa to posterior end
of external occipital protuberance _--..-...--- ... .------ ....------.. 202.4
Maximum width between lateral extremities of zygomatic arches --____._..... 252.2
Maximum width of exoccipitals at lateral margins __- ---------- ------176.5
Maximum width, foramen magnum ....--------...--........-----. .----------- 72.2
Maximum width, lateral border of occipital condyles --------___ --------123.0
Length, midline of parietal-frontal suture to posterior border of
external occipital protuberance .----------..---------................---- 110.0
Maximum width of parietal temporal crests at juncture with squamosal wing --- 97.2
Length, midline of parietal-frontal suture to posterior border
of external occipital protuberance ....-- ---__ -----__-__--_--. ..- 94.5
Maximum width of pariteal temporal crests at juncture with squamosal wing -_ 93.4



Width Length Width Length
Anterior Anterior Posterior Posterior
M3 Length loph loph loph loph
Type AMNH 26805 30.5 28.8e 1.06 17.2 1.77
FMNH P27227 30.0 25.7 1.17 21.5 1.40
UF 11576 30.5 24.0 17.0
UF 11575 32.2 24.0 18.3
UF 11577 31.0 26.7 20.3
FMNH PM155 29.2 23.5 1.24 20.0 1.46
UF 3596 25.0 22.0 1.14 20.6 1.21
UF 3281 29.3 22.9 1.28 19.0 1.54
FGS V5953 29.7 24.2 1.23 21.0 1.41

FMNH P27227 29.4e 25.9 1.14 23.0 1.28
UF 2847 20.5 17.9 1.15 15.0 1.37
FMNH PM155 21.8 24.2 right .90 18.6 1.17
21.8 left 1.00
FMNH P27228 25.0 24.0 1.04 20.0 1.25

FMNH PM155 20.2 20.2 1.00 18.2 1.11
FMNH P27228 19.7 18.8 1.11 17.7 1.11
e = estimated

margins of M. ossivalense are thin and have sharp crests, whereas the
medial margin of Dugong has a broadly-rounded crest and relatively
no crest at the lateral margin. On either side of and above the ptery-
goids is the large sphenorbital fissure, the anterior end of which descends
abruptly downward and continues by a broad groove into the anterior
portion of the base of the pterygoid process. In Dugong this fissure is
formed by the posterior margin of the palatines and the internal portion
of the alisphenoid, roofed by the inferior margin of the orbitosphenoid,
and connected postero-internally with the postero-dorsal edge of the
base of the pterygoid process. Although in an excellent state of preser-
vation, the bony relationship of the elements forming these fissures in
FMNH P27227 is indeterminate.
ORBITOSPHENOID.-The orbitosphenoid is essentially the same as in
Dugong. It is pierced in an antero-posterior direction by the optic canals
that open anteriorly just above and in front of the sphenorbital fissure.
The optic canals of FMNH P27227 (M. ossivalense) differ from those
in FGS V5454 (M. calvertense) only because they are larger.


Stages of life represented by the studied specimens range from ap-
proximately early adult to a point that precedes old age. Cusps of the
teeth are lophodont or essentially arranged in transverse rows, a condi-
tion typical of the Sirenia.
M3.-A perplexing number of minor cusp variations with transitional
gradations occur in the M3 of the metaxytheres. In the norm of the
specimens a prominent anterior cingulum is closed on the lingual side
by a juncture of the protocone and cingulum, and an accessory cuspule
blocks the buccal side of the anterior cingular valley. A variation from
this condition is found in UF 11577, where the anterior loph appears
to have been formed by two separated cusps. A second loph is formed
by a large protocone and a medium-sized protoconule and paracone.
A deep median valley separates the anterior and larger portion of the
tooth from the posterior part. Behind the transverse median valley the
lophodont character is modified as the middle cusp (metaconule) and is
displaced forward to block the valley, a feature characteristic of the
metaxytheres. The degree of anterior displacement of the metaconule
varies from the metaconule lying completely in front of the hypocone
and metacone to a very minor advance. A lingual hypocone is generally
confluent with the metaconule, and either cusp is usually slightly larger
than the metacone. Without exception the hypocone lies posterior to
the metacone. A posterior cingulum composed of two or three small
cusps is usually adherent to the hypocone, with a fossa between meta-
cone and accessory cusps opening on the buccal side. The border of
this cingulum is broadly rounded in all specimens but the type, in which
the round outline is much narrower. Two unworn M3's of FMNH
PM155 are in the process of eruption and extend to a point about half-
way down the crown of M2. The arc-like eruption of these teeth follows
the pattern of tooth succession of other Sirenia, in which the anterior
cusps come into use first and at eruption extend out of the jaw farther
than the posterior portion of the tooth.
M2.-Comparison of the ratio of length/width in M2-3 reveals that
M2 is more nearly subquadrate and is usually of shorter length but of
relatively greater width. Like M3 it is lophodont and divided subequally
by a transverse median valley into a large anterior and a smaller posterior
portion. A thin prominent anterior cingulum is separated by a deep
valley that is closed by a small accessory cuspule at the buccal side of
the tooth (see UF 2847), and is blocked at the inner margin by the
tightly adhering protocone and cingulum. A transverse loph is formed
by the protocone, protoconule, and paracone, of which the former may


be slightly larger than the other two cusps. As in M3, the metaconule is
closely appressed to the hypocone and is anterior to this cusp and the
metacone, which are on a transverse line. The transverse median valley
is near closure on the lingual side but widely separated at the buccal
extremity. One large cusp, in close contact with the hypocone, usually
forms the posterior cingulum. A fossa between the metacone and cingu-
lum opens at the outer posterior border of the tooth. One obvious dif-
ference between M2 and M3 is the lack of minor accessory cusps in M".
M1.-M1 has a quadrate outline and is much smaller than M2. The
available specimens are deeply worn, but it can be seen that the trans-
verse median valley divided the tooth into two equal parts. A thin
transverse anterior cingulum is inferred from the presence of a small
external enamel lake, which marks the buccal opening of the cingular
valley. A similar condition is present at the posterior end of the tooth,
where a fossa formed between the posterior cingulum and the hypocone
opens buccally. Because of the worn condition of the tooth, it is not
known if accessory cusps were present.
P3-4.-The crown pattern of these teeth is unknown in the Florida
specimens, but the alveoli for the teeth are present. FMNH P27228
bears alveoli for P4, but there is no evidence for P3. Dp3 may have
been present in a very young stage of this individual and later shed,
but there is no suggestion of this condition. Gregory (1941: 36) states
that ". . an indistinct smaller anterior region which lodged Dp3 may
be observed on the right side in FGS V5454." This specimen, originally
identified as F. ossivalense, is now referred to M. calvertense.
P4.-The manner of cheek tooth replacement from a posterior to
anterior position indicates some pressure is being applied from rear teeth
to those anterior. Evidence of this pressure is revealed by the noticeable
indentations in the enamel between teeth, and the manner in which the
most anterior teeth are worn and pushed out of the jaw. Such an an-
terior progression would suggest a possible slant to the root system, but
FMNH 27228 clearly reveals vertical P4 alveoli. Following the loss of
P4, the anterobuccal root of M1 moves forward into the alveolus of the
posterobuccal root of P4. The other two alveoli of P4 are then gradually
filled with cancellous bone and become indistinct.
ROOT SYSTEM.-M1-3 are firmly anchored in the maxilla by one large,
centrally-located lingual root and two transversely broad buccal roots,
which display incipient bifurcation. As is the usual case in metaxytheres,
the single lingual root when exposed is found to be sharply recurved
posteriorly at the base, which acts as a hook to lodge teeth in their
sockets. In contrast the two buccal roots are straight.




A % B


FIGURE 9.-Metaxytherium ossicalense occlusal views: (A) (UF 11575) left M3;
(B) (UF 11576) right M3; (C) (FGS V5953) right M3; (D) (UF 2847) left M1
or M2.

Metaxytherium floridanum HAY 1922

DIAGNOSIs.-M3 anterior cingulum and loph transversely expanded;
central portion of anterior border of tooth just forward, indicating in-
cipient development of an anterior cingulum separate from anterior loph;
anterior cingular valley deep, pronounced, not blocked buccally by
accessory cuspule; metaconule displaced forward, completely blocking
transverse median valley; posterior border transversely straight, com-
posed of variable number of supernumerary cusps; M2 subquadrate to
broadly egg-shaped, and although worn the M3 cusp description essenti-
ally applies (except for posterior border missing in M2); cranial roof
"squared up;" low, broad, elongate boss on frontals; temporal crests
widely separated, the lateral border terminating in sharp edge on parie-
tals, low and gently rounded in medial direction.
DISCUSSIoN.-I believe that the generic designation given by Hay
(1922) to Metaxytherium floridanum is correct, and that later references


FIGuRE 10.-Metaxytherium floridanum: (A) (FGS V3232) dorsal view of parietal-
supraoccipital cap; (B) (UF 2199) occlusal view of right Mi-3; (C) (UF 2199)
dorsal view of cranium.

of this species to Felsinatherium are incorrect. For the sake of accuracy,
it. should be added that in the holotype, the anterior border of M3 is
worn through contact with the preceding M3; therefore the length 26.5
mm would be slightly greater in an unworn specimen. The width of the
M3 given by Hay as 25 mm is the maximum width across the anterior
loph formed by the protocone, protoconule, and paracone. UF 2199 is
referred to this species and consists of portions of the cranial roof,
exoccipitals, squamosal, jugal, ear bones and maxillary, the last con-
taining an M3 on either side, an M2, and a fragmented M1. Also referred
is a cranial cap of fused parietals and supraoccipital (FGS V3232), a
specimen discussed as Felsinotherium sp, but not figured by Simpson
(1932: 455-458). Reference of UF 2199 to M. floridanum is done with
some reluctance, because the holotype is a worn specimen that does not
present the diagnostic qualities of the species as distinctly as would be


CRANIAL CAP.-Varied cranial remains of UF 2199 indicate that it
represents a large adult, but not a form as massive in construction as
the larger M. iordani of the California Miocene. The angle at which
the posterior end of the premaxillary arm joins the postorbital process
indicates a mesorostral fossa with a very wide posterior border. Nasal
boundaries cannot be determined because of nasal-frontal fusion and frag-
mentation. A low, broad, elongate frontal boss is present, as in M. ossi-
valense and M. calvertense, but unlike the latter the frontals are "squared
up" and essentially lack temporal crests. The undulatory lateral border
of the postorbital processes is almost identical to that of M. jordani.
Temporal crests on the parietals are widely separated and weakly de-
veloped in comparison to the condition present in other Florida metaxy-
theres, and end in a sharp lateral edge with a gentle medial rounding.
The almost complete lack of temporal crests on adult specimens may
prove to be a specific distinction differentiating this species from M.
ossivalense, which bears relatively narrow upraised sharp crests, and from
M. calvertense, which is characterized by broadly swollen temporal crests
that closely approach each other in the midline.
JUGAL.-The anterior end of the zygomatic arch is fragmented and
does not reveal the bony contacts of the maxilla and jugal. The jugal is
vertically expanded beneath and posterior to the orbit, with a gradual
diminution of this bone posteriorly. The malar process, which bears a
wide shallow groove on the lateral surface, extends back to a point about
17 mm anterior to the glenoid plate. This process is either rarely pre-
served or perhaps less readily recognized and collected as an isolated
bone than are most other cranial elements of fossil Sirenia. There are no
features that differentiate this element from those found in better pre-
served metaxytheres. When compared to the three recent genera, the
general configuration of this element most nearly approximates Hydro-
damalis; also, the outline differs little from the Eocene Protosiren.
An important element that suggests degree of evolution is the pro-
gressive approach of the ventral midline of the supraoccipital to the fora-
men magnum. In FGS 3232 the exoccipital-supraoccipital contacts are
diverted at an angle of 125, and the supraoccipital closely approaches
the foramen magnum. This is the general condition found in all Florida
M3.-One of the distinguishing qualities in UF 2199 is the great
width across the anterior loph, as compared to the length in M2 and M3.
The result is a length-width ratio that indicates a more squared-up tooth
than the usual elongate MS present in other metaxytheres. The width
across the protocone, protoconule, and paracone of 30 mm, in both M2
and M3, is by far the greatest recorded width for this portion of either





Sc P.c.

FIGURE 11.-Metaxytherium foridanum (UF 2199); right M3.

nounced, this feature being accentuated from M1 to M3; large well de-
veloped cuspule on lingual side blocking transverse median valley in M1-3,
this character not well developed on partially erupted, unused M3; vari-
able number of supernumerary cusps; incisors, if present, nonfunctional;
rostrum acutely downturned, as in all metaxytheres; nasals separated by
frontals in midline; frontals elongate; squared-up, low, elongate boss;
temporal crests on parietals swollen, rounded, and closely approaching
each other; supraoccipital touching dorsal border of foramen magnum.
REFERRED SPECIMENS.-A study of the relatively complete cranium
of the holotype of M. calvertense reveals no differences with FGS V5454,
an anterior two-thirds of a cranium lacking the ventral half of the rostrum.
FGS V5454 had previously been attributed to Felsinotherium ossivalense
by Simpson (1932) and by Gregory (1941). This specimen and the
holotype have already been adequately described, and further amplifica-
tion will be made only where information is added by comparative study.
Also here assigned to M. calvertense is an important half of a cranium
(FGS V3211), well described by Simpson (1932) and referred to Felsi-
notherium sp.; the undescribed MCZ 4218, consisting of a left portion
of the maxilla with M1-3 implanted; and a parietal (UF 11574). The
holotype is from the Middle Miocene Calvert Formation of Maryland,


FIGURE 12.-Metaxytherium floridanum (FGS V5454); occlusal view of (A) right
M1-2 and (B) left M1-3 from Gregory, 1941.

and the referred specimens are from Mio-Pliocene or Middle Pliocene
Bone Valley Formation of Florida.1
CRANIUM.-A review of the specimens suggests that a series of growth
stages is represented. The holotype (USNM 16757) is a very young
adult; FGS V5454 is a young adult, more advanced in age than the holo-
type; MCZ 4218 and FGS V3211 are mature adults. Incisors are absent
in the holotype and FGS V5454, the two specimens in which the rostrum
is preserved; but most of the anterior ends of the premaxillaries are
missing in the latter form. It is unknown if the tuskless condition is a
dimorphic quality in which only females lack tusks, or if M. calvertense
is a tuskless species. The rostrum is downturned, as in other metaxy-

' UF 11579, an excellent cranium, is referred here also but could not be included in this study.
(see Table 1).


Three large roots are present in M3. Beneath the protocone a large
long root is curved dorsally toward the buccal side of the maxillary.
The root beneath the paracone follows a dorsal course, with a strong
curve to the buccal side of the maxillary. This root is much shorter than
that beneath the protocone, displays an incipient bifurcation throughout
its length, and ends distally at the buccal border of the maxillary. Be-
neath the hypocone and the posterobuccal portion of the posterior cingu-
lum, a large root is strongly bent posterobuccal toward the elongate,
laterally-flattened, grooved knob formed by the ventroposterior extension
of the maxillary. Because of exposure on the buccal side of the maxillary,
it can be noted that the distal extension of the root is open, and during
life was covered by a very thin, moderately-porous covering of the maxil-
M2.-As in a number of sirenians, the M2 of UF 2199 has an equal
or greater width across each major transverse loph than has M3 and is
but slightly shorter in length. Only minor differences in cusp pattern
are present between M2 and M3, but in less worn specimens there is
invariably an increase in the number of accessory cusps from anterior
to posterior teeth. The anterior border of M2 is broadly rounded, prob-
ably is less undulatory than in M3, and the hypocone is proportionally
larger. Because of wear, it is impossible to determine the number of
cuspules in either the anterior or posterior cingulum. Three large roots
are present in a position similar to those in M3. A break in the specimen
between M2-3 indicates that the posterior root is essentially vertical.
As in larger Sirenia, the cheek teeth have a forward displacement
in relationship to the zygomatic-orbital bridge. The anterior border
of M2 of UF 2199, the M1 in young specimens of Metaxytherium and
the assigned felsinotheres, the transverse valley of P4 in Dugong, and
the anterior end (approximately) of the sixth cheek tooth in adult Tri-
chechus all lie lateral to the same part of the bridge. In the case of UF
2199, this is 9 mm anterior to the posterior border of the bridge.
MI.-Fragmented remains of a subquadrate tooth mark the presence
of M1. Like M. ossivalense the width appears to be greater than the
length, although both measurements in UF 2199 are estimated. The
anterior border of M1 lies lateral to the anterior border of the zygomatic-
orbital bridge in a well advanced position; this characteristic apparently
is age-dependent, the border being farther advanced in older specimens.

Metaxytherium calvertense KELLOGG 1966
CRANIAL DIAGNOSIS.-Four cheek teeth; M2-3 subquadrate; M1 quad-
rate; transversely straight anterior and posterior border on M1-2; borders
on M3 straight to slightly rounded; anterior cingular valley deep, pro-


TABLE 3.-CRANIAL MEASUREMENTS OF Metaxytherium floridanum.

UF 2199 Width between lateral borders of postorbital processes ....-..--__-__160.0e
FGS 3232 Maximum width of parietal temporal crests at juncture
with squamosal wing ___ -------- --___--- -----_ 83.0
Width Length With Length
Anterior Anterior Posterior Posterior
M3 Length loph loph loph loph
Type USNM 7221 26.5 23.1e 1.15 20.0 1.33
UF 2199 right 29.5 30.0 0.95 21.7 1.36
UF 2199 left 29.0 27.0 1.07 21.3 1.36
UF 2199 31.3 30.0 1.04 26.3 1.19
e = estimated

tooth in any North American sirenian. Hay (1922: 2) stated that the
paracone is larger than the protoconule in M3 of the type, but this is
definitely not correct. The reason for this belief evolved from the fact
that the protocone and protoconule of the type had been worn down to a
point where the two cusps merged, with no vestige of the former bound-
ary. As is true of the anterior loph in all species of Metaxytherium, the
protocone is the largest cusp, the protoconule is of medium size, and the
paracone of similar size to the protoconule or smaller.
Another distinctive feature of the M3 concerns the anterior border of
the cingulum, which juts forward from the center of the tooth and gives
indication of the development of a cingulum that might become separated
from the anterior loph. The anterior cingular valley, which extends about
halfway across the worn specimen, is transversely straight and lacks clo-
sure by an acessory cusp on the buccal side, although an incipient cuspule
is present on the left M3. The metaconule is displaced forward, block-
ing the transverse median valley, a condition characterized to some
degree in all metaxytheres. It is, however, more closely affixed to the
protocone and protoconule than in most metaxytheres. In this respect it
is most similar to UF 11575, 11577, and M. krahuletzi. Forward dis-
placement of the metaconule is noticeable to a lesser degree in Halither-
ium of Stampian age and is essentially absent in the bilophid Eocene
teeth. As in M. ossivalense, a valley formed between the hypocone and
the posterior cingulum remains open, in contrast to the closed valley in
the M3 of M. calvertense. The posterior cingulum, composed of three
or four cusps, is transversely straight, due in large part to a cusp affixed
to the posterolingual border of the metacone.



FIGuRE 13.-Metaxytherium calvertense: (A) (MCZ 4218) occlusal view of left
M1-3; (B) (MCZ 4218) lingual view of left M1-3.

there, but not so pronounced as in Halitherium schinzi, Caribosiren, or
In both the holotype and FGS V5454 the nasals are separated by
the frontals, a condition often not clearly detected in older specimens
due to a fusion of the nasal-frontal sutures, which obliterates the contact


between these elements. In the latter specimen the closest approach
of the nasals to each other at the posterior end of the premaxillary arms
is 10 mm, with this distance increasing slightly in width both anteriorly
and posteriorly. Separation of the nasals by the frontals is a condition
characteristic of Eotheroides aegyptiacum, E. libyca, Caribosiren, Hespe-
rosiren, and the metaxytheres. In Thalattosiren the frontals do not
separate the nasals (Kellogg 1966: 73), but a forward projection of the
frontals pushes into and partially overrides nasals in the midline. Sicken-
berg (1928: 200) stated that the nasals of T. petersi were joined medially
along a suture 11 mm in length, which lies along a line of continuation
of the sagittal suture. The frontals of M. calvertense are moderately
elongate, squared up, bear a low elongate boss, and have a slight ex-
pansion of the postorbital processes. Expansion of the postorbital proc-
esses is a factor of age, with progressive expansion coinciding with
growth of the frontals.
General size and configuration of the parietal temporal crests of FGS
V5454 (Gregory 1941: pl. 11) closely resembles Metaxytherium ( = Felsi-
notherium) forest Capellini (1872: pl. 2, fig. 1). Crests present in FGS
V3211 (Simpson 1932: fig. 13) are also swollen but project higher than
in the aforementioned specimens, and posteriorly exhibit a gradual, rather
than a sudden increase in expansion and swelling. Close approach of
the crests in the midline is reminiscent of Halitherium, but in this genus,
unlike metaxytheres, the parietal crests are thinner and extend posteriorly
to the lateral edges of the supraoccipital, rather than terminating near
the anterior half of the squamosal ring.
The outstanding structural condition in the rear of the cranium is
the prominent supraoccipital (which forms the dorsal border of the fora-
men magnum), a feature considered progressive in the metaxytheres.
The same essential relationship of posterior cranial elements is present
in FGS V3211, as well as in other Florida metaxytheres and Hesperosiren.
DENTITION.-In M3 the anterior cingular valley is open buccally
from the base of the crown; in M1 the initial opening is midway down
the crown; and in M2 the condition is intermediate between these two
teeth. The lack of an accessory cusp to block the anterior singular
valley is a feature shared by M. calvertense and M. floridanum, and
generally serves to distinguish fully erupted adult cheek teeth from
those of M. ossivalense. When present, the blocking cusp is normally
confined to M3. This observation must be tempered by the fact that in
the specimens available the crown of M2 is moderately worn, and the
crown of M1 is usually worn down to a point that the cusp, if present,
would be obliterated. It should also be noted that partially erupted
third molars often have crowded, indistinct cusps rather than clearly


TABLE 4.-MEASUREMENTS OF THE DENTITION OF Metaxytherium calvertense.

Length/ Length/
Width width Width width
Anterior Anterior Posterior Posterior
M3 Length loph loph loph loph
MCZ 4218 32.0 25.6 1.25 20.4 1.57
FGS V5454 30.1 24.2 1.24 22.0 1.37
USNM 16757 28.5 18.5 1.54 18.5 1.54
MCZ 4218 28.4 27.5 1.03 24.2 1.17
FGS V5454 28.4 26.2 1.08 23.0 1.23

MCZ 4218 23.5 23.5 1.00 21.2 1.11
FGS V5454 20.0-20.5e 22.+ .91 20.0' 1.00-
USNM 16757 18.5 19.6 .94 15.5 1.19
e = estimated

separated entities that characterize an M3 ready for use, or but slightly
worn. Many also retain abundant, fine, vertical striations on and be-
tween cusps and are pitted as the result of physiological effects on the
enamel. Both of these latter features quickly disappear with wear.
The transverse anterior loph is composed of the protocone, proto-
conule, and paracone as in all sirenians. Although all three are of similar
size, the protocone is largest and the paracone smallest. To a degree
comparable to M. ossivalense, the transverse median valley is blocked
by the forward displaced metaconule, but this cusp is not as closely
adjoined to the protoconule as in M. floridanum. The M3 in the large
M. jordani from California has a straight, open, median valley; a feature
shared to a degree by M. ossivalense (FMNH PM155). Confluent with
the protocone a large accessory cusp blocks the lingual side of the trans-
verse median valley in M1-3. This feature, coupled with the rounded
temporal crests on the parietals, distinguishes M. calvertense from other
North American metaxytheres. In MCZ 4218 a smaller cuspule, which
is appressed to the paracone, partially closes the same valley buccally
in M1 through M3. A separate accessory cusp situated lateral to the
metaconule and anterior to the metacone on M3 further blocks the valley
on the buccal side. Positions of the forward metaconule, the large,
closely-appressed hypocone, and the smaller metacone are typical of all
species of Florida Metaxytherium. The transversely straight posterior
boundary in M3 is formed by three distinct cusps, with that of M1-2
essentially forming one large transverse cusp. The posterior cingular


valley is open in M1-2 but closed by accessory cusps in M3. It must be
emphasized that supernumerary cusps are far more common on the large
posterior molars, so variance of such cusps must be treated conserva-
tively before establishing a new species on this basis. If accessory cusps
are present on M3 but lacking on anterior molars, such cusps are con-
sidered by the writer to be of no systematic value; but if accessory cusps
are a prominent part of all the molars, as in MCZ 4218, the features are
considered to be morphological differences of taxonomic value. Strong
vertical ridges and wrinkling are unusually well developed on this speci-
men. Such wrinkling is found to some degree on many unworn sirenian
teeth and is most prominent on the larger teeth.
Three root sockets, two on the buccal side and one on the lingual
side, mark the position of the missing p4 in MCZ 4218, and indicate a
tooth only slightly smaller than M1. Bone enclosing the terminus of the
root is thin, and a minimum of erosion will expose these elements, as
has occurred in five points on this specimen. In FGS V5454 the alveoli
are essentially filled with a cancellous bony tissue, and it is difficult to
ascertain if one or two premolars were present when this animal was im-


In this study the following limiting factors applied: (A) The small
number of specimens studied from Florida represents an accidental rem-
nant or a local morphologic sample range rather than a species range;
(B) The lack of stratigraphic data limits information on the former
extension of the sample in time and space and renders interpretation of
evolutionary trends difficult.
1. From a dental study of Florida's Mio-Pliocene Sirenia it cannot
be determined if the metaxythere group was evolving from simple buno-
dont to more complex hypsodont.
An increase in complexity of enamel due to infolding and growth of
accessory cusps has accompanied the history of many successful ter-
restrial herbivores and has usually been accompanied by a growth from
bunodont to hysodont teeth. The question that arises is: Were
metaxythere cheek teeth evolving a complex triturating surface by fold-
ing of enamel and growth of accessory cusps, or were cheek teeth of
this group becoming simplified? Evidence suggests both conditions of
dental change occurred, but each in a different geographic area. The
increase in complexity is indicated by the European metaxytheres, which,
on the basis of stratigraphic evidence, show an increase in accessory
cusps in geologic time.


2. There appears to be a development toward a strong anterior cingu-
lum that tends to become an independent transverse loph with loss of
attachment even to the protocone. To some extent an independent
anterior cingulum is found as far back as Eocene genera, as well as
in the living Trichechus, but in neither of these cases does the loph ex-
tend up to the height of the loph formed by the protocone, protoconule,
and paracone. This feature is well expressed in M. ossivalense (UF
11576-11577), M. calvertense (MCZ 4218 and FMNH PH155), and the
European H. krahuletzi.
3. Through Miocene and Pliocene time (Burdigalian through Astian),
the metaconule is gradually displaced forward to block the transverse
median valley and eventually lies completely anterior to the hypocone
and metacone. Expression of this feature is displayed in all North
American metaxytheres, as well as in European metaxytheres and as-
signed felsinotheres.
4. Closure of the lingual side of the transverse median valley is
accomplished in M. calvertense through growth of a prominent accessory
cusp posterolingual to the protocone. Incipient development of the
cusp appears in M. ossivalense. To my knowledge, no other metaxy-
there or assigned felsinothere of North America or Europe bears this cusp.
5. An accessory cuspule blocks the buccal side of the valley of the
anterior cingulum in M. ossivalense. This condition is believed to have
developed from an ancestry having the more typical open valley.
6. I believe that metaxytheres of Europe and the Atlantic coast of
North America during Miocene-Pliocene time slowly evolved a more
complex triturating surface, but conversely, during Pliocene time, a loss
of dentition developed in North American Pacific coast metaxytheres.
If one examines the entire history of the sirenians, it is evident that,
with the exception of the living Trichechus, there is a gradual loss in
number of functional teeth from the almost complete Eocene eutherian
dentition to the reduced number of simple peg-like teeth in Dugong and
the complete loss in Hydrodamalis. Until a recent study by Domning
(1972, unpub. Ph.D. disser.), no fossil adult with fewer than four func-
tional cheek teeth (16 teeth) had been found, and these failed to show
loss of accessory cusps or any dental degeneration that one would expect
in an ancestor of Dugong or Hydrodamalis. Domning, however, pro-
vided conclusive proof of the relationship of the geographically wide-
spread Metaxytherium to the recently extinct (ca. 1768) Bering Sea
Hydrodamalis gigas. A similar relationship had been proposed by
Simpson (1932), Vanderhoof (1941), and Reinhart (1959), but diag-







FIGURE 14.-Metaxytherium calvertense (MCZ 4218): occlusal view of left M1-3.

nostic evidence was lacking. A well-preserved new species of Hydro-
damalis, intermediate between Metaxytherium and H. gigas both in time
and morphological characteristics, clearly displays a close affinity to
these genera.
It should be noted here that Felsinotherium gunteri, which appeared
in Simpson's (1932: 449) review of sirenian metaxytheres, is a nomen
nudum, and should be referred to F. ossivalense (G. G. Simpson, pers.
comm., 14 December 1959).



Fr boss Ju

I l^ 'IPa

S= Sq
FIGURE 15.-Hesperosiren crataegensis (AMNH 26838): dorsal view of cranium.
Modified from Simpson, 1932.

Hesperosiren SIMPSON 1932
As is often the case in the taxonomic history of the Sirenia, the rela-
tive scarcity of diagnostic remains results in many monotypic genera,
whereas a larger number of specimens available for comparison may
indicate that synonymization of some of these genera is called for.
Hesperosiren is regarded as a rare sirenian, possibly because it actually
is rare or, of equal likelihood, because its bony elements are not readily
distinguished from the more common metaxytheres found in the same
geographical area or in sediments of similar geologic age. Despite the
moderate number of sirenian remains discovered in Florida since 1932,
only two new specimens are referable to Hesperosiren. These are MCZ
4432, a relatively complete posterior half of a cranium, and FGS V4250,
a well preserved specimen of the mid-cranium. The latter specimen
was designated a plesiotype of Felsinotherium ossivalense by Simpson
(1932: 454). Also tentatively referred to Hesperosiren is USNM 181550,
a parietal-supraoccipital complex from an old individual that features
an unusual vertical thickness. Although the holotype cranium is com-
plete, its crushed condition requires judgments to be made concerning
the configuration and contacts of a number of bony elements as they



Fr. boss

FIGURE 16.-Hesperosiren sp. (MCZ 4432): dorsal view of cranium.



AMNH 26838
(Holotype) MCZ 4432 FGS 4250
Width across postorbital processes 140 186e
Minimum separation between inner
border of nasals, posterior end 20.2 30e 21.4
Width of frontals across middle
of boss 75 74 64
Height of frontal boss 10 10 8
Anterior midline of parietals to
posterior border of nasals 75 107 76.5
Posterior border of external
occipital protuberance to
anterior midline of parietals 75 94.7
Minimum separation of temporal
crests on parietals 41 17.7
e = estimated

existed in a pre-crushed condition. The referred specimens, therefore,
while displaying some variatoin from the holotype, may aid in the clari-
fication of the interpretations. Whether these variations will eventually
be found attributable to age, sex, individual differences in the same
species, or to different species, is a matter that can only be determined
by more specimens. Based on the present information, I prefer to retain
all the specimens in a single species.
On the dorsal cranial surface the rounded posterior borders of the
nasals are inserted under the overriding frontals and are separated in the
midline by the same bones. Thin anterolateral extensions clasp the
posterior end of the premaxillary arms in the same fashion as in the
metaxytheres. Within the narial passage the inner half of the nasals
is thin, compressed laterally, extends deeply into the frontals, and thins
out anteriorly as a long extension lying against the premaxillary. Various
turbinals lie immediately adjacent to the inner side of the nasals, but
the relationship is not clear. The size and shape of the nasals are the
same as in metaxytheres of similar size.
A notable difference between the type of Hesperosiren and the re-
ferred specimen may be seen in the width between the lateral extremities
of the postorbital processes of the frontals, which is estimated to be
140 mm for the holotype and 186 mm for the referred MCZ 4432. The
great lateral divergence of the postorbitals in the latter specimen is
similar to that found only in the very largest sirenians, such as Metaxy-
therium jordani (198 mm) and Hydrodamalis (192 mm, but will vary
with age). The distance from the posterior margin of the parietal to
the anterior angle of the postorbital process of the frontal is 316 mm


in the specimen of Hydrodamalis at hand, and 248 mm in MCZ 4432. In
the referred Hesperosiren the posterior measurement was taken from
the midline of a small transverse ridge immediately anterior to the ex-
ternal occipital protuberance. Because of the complete fusion of the
supraoccipital and the parietals in all fossil and living sirenians, except
during the fetal stage, it would be difficult to determine the exact pos-
terior boundary of the parietals. A comparison of the width across the
postorbital processes is made to demonstrate that Hesperosiren has a
width in this area comparable to that of such large sirenians as M.
jordani and Hydrodamalis, and yet it has a much shorter skull with much
smaller dimensions throughout its posterior half. Certainly one of the
outstanding features of Hesperosiren must be the widely divergent post-
orbital processes and, possibly in conjunction with this feature, a wide
mesorostral fossa. Anterolateral divergence of the frontal processes is
more exaggerated in the two specimens referred to Hesperosiren than in
the holotype, because of the distinctive narrowing of the frontals pos-
teriorly, which culminates in a pinched-in parietal-temporal area. Lateral
divergence of the postorbital processes may be most prominent in FGS
V4250, as suggested by the sharply branching base of the processes, and
in this respect it shows some resemblance to Halitherium and to Mana-
All specimens of Hesperosiren bear an elongate, prominently up-
raised dorsal median boss on the frontals, a feature similar to Dugong
but completely unlike Trichechus. In Dugong the bone joining the boss
is not thicker than the bone immediately bordering the upraised area
but is merely arched upward. Hesperosiren (FGS V4250) reveals a con-
dition suggesting that the bone forming the boss was thicker, whereas
Metaxytherium floridanum (UF 2199), which clearly displays the area,
shows a low broad boss composed of unthickened bone. A question
arises as to whether the boss is formed solely by the frontals, or if
other cranial elements add to its formation. A lateral section in Metaxy-
therium (UF 11573) displays a spine that rises above the crista galli
and becomes deeply entrenched to within 4 mm of the dorsal surface
of the frontals at a point just behind the midline of the boss. Another
specimen (FGS V5947), which essentially lacks a boss, also has a spine
that rises to within 3 mm of the dorsal surface of the frontals. Lepsius
(1882: table 5, fig. 58) reveals a cross-section of the frontal-ethmoid
contacts in Halitherium, but no entrenchment of the ethmoid is shown.
Despite its proximity to the boss, the ethmoid bone does not appear to
be a factor in the elevation of the bone in the median frontal region.

' Sickenberg (1934: 274) refers the genus Manatherium delheidi to Halitherium schinzi f.


Another aspect to be considered is why the frontal boss is so prominent
in Hesperosiren, but not in other genera. Conceivably it is prominent
in young individuals, then spreading out with growth of the frontals,
and disappearing in adults. Dilg (1909), working with the growth stages
of Trichechus inunguis, has shown that in young sirenians the frontals
are short, and the parietals long. Further growth in the cranium takes
place to a great extent in the frontals, but parietal growth is relatively
limited. In MCZ 4432, which appears to be an adult, the boss remains
well elevated above the cranial roof. Formation of the boss may be due
to a telescoping of bones in the nasal region; a condition that frequently
occurs in the evolution of aquatic mammals as they become better
adapted to a water environment. Such a suggestion may be plausible
in the case of Dugong, but not for Trichechus, which has a broad con-
cavity in the frontals, bordered by high squared-up crests. Sirenians in
general show a reduction in size of the nasal bones from Eocene to
Recent forms.
Lyriform temporal crests on the parietals converge moderately in
the midline of these elements, with an estimated minimum separation of
about 41 mm on the holotype. This is in essential agreement with
metaxytheres, which vary somewhat in the degree of separation, although
most are more squared-up in this region. Unlike the holotype, MCZ 4432
has a minimum separation of well developed parietal temporal crests of
only 17 mm, and in this respect greatly resembles the convergent crests
characteristic of Halitherium. FGS V4250 also shows indication of bear-
ing strongly convergent crests, but inasmuch as only the anterior ends
of the parietals are present the degree of convergence is unknown.
Separation of the temporal crests on FGS V4250 at the posterior border
of the frontals in the midline is 22 mm. The crests are tapered so that
they undoubtedly came closer together at a more posterior point. Al-
though some degree of relationship is shown with Halitherium in this
individual character, it would necessarily be with a more conservative
member of this genus. A young Halitherium sp. individual from the
island of Madagascar has parietal crests in contact for a distance of 55
mm. In Halitherium schinzi the parietal crests are in close proximity,
but not touching, for an estimated 40-42 mm. Dorsally the anterior ends
of the parietals in Hesperosiren extend forward to a point midway across
from the center of the frontal boss (Fig. 15). Although the degree of
parietal crest convergence in the two specimens referred to Hesperosiren
is quite similar to that in Halitherium, a number of other cranial rela-
tionships reveal important differences. In Halitherium the nasals are
larger and meet in the midline; in Hesperosiren the smaller nasals are
separated in the midline by these bones. The supraoccipital is separated


from the foramen magnum by the exoccipitals in Halitherium, but it
forms a small portion of the dorsal-central border of the foramen mag-
num in Hesperosiren. There is no prominent upraised median boss
present on the frontals of Halitherium, as is present in Hesperosiren.
From a dorsal view the cranial roof of the referred MCZ 4432 is
superfically like that of Metaxytherium cuvieri, the resemblance being
based largely on a thin elongate cranial cap with swollen temporal crests.
The pronounced medial boss on the frontals, characteristic of Hespero-
siren, is lacking in M. cuvieri. A low broad boss or convexity is present,
however, as in all Miocene and Pliocene metaxytheres. The minimum
degree of separation of the temporal crests of M. cuvieri ranges from
that of Flot' (1886), who gives a figure of 92 mm, to the type of M.
cuvieri (listed as 73 mm), to the 70 mm on the excellent specimen de-
scribed by Cottreau (1928), to a specimen from Maine-et-Loire region
with a minimum separation of 69 mm. Cottreau believed the separation
of the temporal crests to be a character that varies with both age and sex.
Dilg (1909), working with growth stages of Trichechus, showed that in
the younger stages the temporal crests are less distinct and more widely
separated than in the older stages. I agree that crests, as well as other
rugosities, are less distinct in the younger forms, and because of the lack
of development the relatively flat cranial roof bears such small crests that
indeed there is a wide separation of these elements.
Again, in reference to the dorsal surface of the cranial roof, the holo-
type bears a superficial appearance to Thalattosiren, rather than to M.
cuvieri. It differs (a) from Thalattosiren by the division of the nasals
by the frontals, (b) from M. cuvieri in having a less elongate cranium
and less prominent temporal crests, and (c) from both genera by the
presence of a prominent frontal boss.
The general configuration of the supraoccipital is more rounded in
Hesperosiren than in Metaxytherium and Dugong, but less so than in
Trichechus. Essentially all fossil Sirenia, but not recent forms, have a
prominent external occipital protuberance in the dorsal midline of this
element. In Hesperosiren (MCZ 4432) the protuberance is a very broad
triangle, with the apex ventral. This shape contrasts in several ways to
that present in Dugong, Trichechus, Halitherium, and adult metaxy-
In Hesperosiren, as with many crushed specimens, various interpre-
tations can be made concerning the correction of distortions. In the case
of the holotype, the only illustrations are those from Simpson (1932:

',Abel (1904: 13) places Halitherium fossil, depicted by Flot (1886: pls, 26-27), in the synonymy
of Metaxytherium cuvieri.


Figs. 1 A and B). My interpretation of the boundaries of various cranial
elements differs from Simpson as follows:
1) In Fig. 1B (a dorsal view), the anterolateral extensions of the parie-
tals should be moved anteriorly to a point lateral to the middle of
the frontal boss. These dimensions would then be in agreement with
the specimen, as well as virtually all other Sirenia.
2) The boundaries of the nasal bones should appear much more clearly
in the illustration, particularly in view of the fact that the relative
size and configuration of these bones, in relation to the anterior
midline of the frontals, is of considerable taxonomic importance.
3) The juncture of the lacrymals, jugals, postorbital process of the
frontals, and posterolateral arms of the premaxillaries is not at all
in accord with the illustration of the holotype. In the illustration,
there are no undulating rugosities on the premaxillaries immediately
anterior to the postorbital processes, and the posterolateral configura-
tion of the premaxillary arms is completely incorrect. The anterior
border of the orbital ring is formed by what appears to be a lacry-
mal, identified by its position and characteristic lateral protuberance,
and a jugal whose relationship is not distinct. Whether the maxilla
separates the jugal from the premaxillary arm cannot be determined.
The illustration of these elements, as rendered by myself, would
have a configuration that would not digress strongly from the con-
dition present in Eocene to Recent forms.
4) In Figure 1A (a lateral view of Hesperosiren) the pterygoids un-
doubtedly extended below a point lateral to the occlusal surface of
the cheek teeth, as they do to some degree in all Sirenia.
5) The dorsoanterior end of the zygomatic arch of the squamosal has a
greater forward extension than the ventroanterior end, which is op-
posite the condition in Figure 1A.
6) The outstanding superficial criterion for distinguishing Hesperosiren
is the interpretation that this genus bore a straight rostrum, as op-
posed to all other Sirenia in which this region is acutely down-
turned. Because of the crushed condition of the premaxillary arms,
it is difficult to ascertain the degree of flexure. In my opinion there
is little to distinguish between the rostra of Thalattosiren petersi
Sickenberg (1928: pl. 1) and Hesperosiren. Whether Hesperosiren
bore a flexure as acute as this form is a matter of conjecture, but
no more questionable than the interpretation of a straight rostrum.
It appears unlikely that the rostrum was as acutely downturned as
the condition found in Halitherium schinzi, Caribosiren, or Dugong.


GENUS Halitherium KAUP 1838
The presence of Halitherium, long represented by a number of fine
specimens from Europe and a cranial cap from Madagascar, has now
been definitely established in North America on the basis of Florida
Oligocene material. The term "established" is important, for poor speci-
mens from the Atlantic coast of the United States have been attributed
to this genus previously. Most sirenians from the eastern coast of the
United States have come either from the Calvert Formation of Maryland
or the Ashley River Formation of South Carolina. Both formations con-
tain a mixture of reworked faunas, and the sirenians found usually have
been not only fragmentary but of uncertain age. In Leidy's (1856)
short description of an isolated upper molar from the Ashley River de-
posits he mentioned fragments of ribs. These specimens, of which the
tooth was the type, were referred to Manatus antiquus. The same tooth
was later figured by Leidy (1860) in Holmes' Post Pliocene Fossils of
South Carolina. A second name, Manatus inornatus, was proposed by
Leidy in 1873, based upon the crown of a relatively small isolated lower
cheek tooth from the same deposits. The specific names antiquus and
inornatus are therefore based on teeth of indeterminate genera and un-
known age.
In reviewing the fossil mammals from the Ashley River deposits near
Charleston, South Carolina, Allen (1926) noted that the beds contained a
mixture of fossils from several different epochs. He considered the land
mammals to be chiefly of Pleistocene age (although some range from
Miocene to Recent), but, on the whole, the marine forms are from the
Miocene. From these deposits Allen collected a number of sirenian
remains, with the more diagnostic remains being cranial roofing com-
posed usually of some combination of the frontals, parietals, and supra-
occipital. Some water-worn humeri were also described. He divided
the skulls and humeri into two groups, referring those with strongly
convergent temporal crests and the small humeri with proximal tubero-
sities (which diverge at an angle of 60) to Halitherium antiquus. Allen
also referred Manatus inornatus to H. antiquus, although the teeth are
generically indeterminate. Simpson (1932) proposed discarding Leidy's
names (M. antiques and M. inornatus) as indeterminate, proposed a
new name (Halitherium alleni) based on Allen's (1926: pl. 11, fig. 1)
figure of fused parietals and supraoccipital. Whether the cranial ma-
terial is referable to Halitherium or to another genus is not known, but
I agree with Simpson that Allen's specimens (cranial roofs and humeri)
cannot be referred to Leidy's indeterminate teeth.
Allen's second division of sirenian specimens was composed of larger,
flatter, wider cranial roofs, with the temporal crests not so well developed


as in the smaller crania. The proximal tuberosities of this second group
diverged at an angle of approximately 85 to 90 degrees. Allen referred
his second group to Metaxytherium (=Halianassa) manigaulti Cope.
Reference of this material to Metaxytherium is reasonable, although the
specimens are so waterworn as to make a generic allocation somewhat
questionable. The type of M. manigaulti is a fragmentary premaxillary
from the Wando River, South Carolina, and was described by Cope
(1883) as Dioplotherium manigaulti. The generic distinction, according
to Cope, was that each premaxillary contained two incisors. Kellogg
(1925) showed this supposed differentiating character to be incorrect
and subsequently referred Dioplotherium to Metaxytherium mani-
gaulti. Since well-preserved specimens of Metaxytherium found to date
in the western hemisphere lack incisors, this premaxillary with its well-
developed incisor may be referable either to a male individual of the
genus Metaxytherium or to another genus.
A similar generic proposal was made by Lydekker (1887), who postu-
lated that a vertebral centrum and two ribs from the Ashley River
deposits belonged to Halitherium. A taxonomic assignment based on
such clearly indeterminate material is premature.
Matthew's (1916) reference to this genus may possibly be correct.
The type (from Juana Diaz, Puerto Rico) is a fragmentary mandible
bearing three heavily worn molars and the alveoli from three premolars.
The sediments around Juana Diaz are largely of Oligocene age, and this,
plus the number of teeth, gives credence to the determination. H.
antillense, however, might equally well be synonymous with Caribosiren
turner Reinhart 1959, which was collected in the same general locality
in undoubted Middle Oligocene strata, but which is distinctly different
from Halitherium. Unfortunately no common elements involving these
specimens exist.
Subsequent information gained from specimens indicates that cranial
caps with narrowly separated parietal-temporal crests cannot automatic-
ally be referred to Halitherium. Similar bony elements from Hespero-
siren, from young individuals of the Pleistocene Trichechus, and to a
lesser extent Metaxytherium calvertense and Caribosiren turner are
virtually indistinguishable from those of Halitherium when parietals
alone are examined. To distinguish between Hesperosiren and Hali-
therium it is necessary to have adjoining bones where distinctive differ-
ences exist as follows:
1) A large prominent domal boss characterized the frontal midline of
Hesperosiren, but is lacking in Halitherium.
2) Postorbital processes of the frontals in Hesperosiren have a com-


paratively greater distal expansion and are not abruptly recurved
posteriorly, as in Halitherium.
3) Nasals join anterior to the frontals in Halitherium, but are separated
in the midline by a forward extension of the frontals in Hesperosiren.
4) The rostrum of Halitherium is abruptly downturned. The degree
of flexure in the rostral region of Hesperosiren cannot be properly
assessed because of the crushed condition of the holotype, but I
believe it is more deflected than shown in reconstruction.
5) Well developed tusk-like incisors characterize the snout of Hali-
therium, whereas Hesperosiren either lacked tusks or bore only
minute vestigial incisors, as is found in some Florida metaxytheres.
Such incisors would have been shed before maturity, were probably
nonfunctional, and the area covered by a horny pad in the young
6) In Hesperosiren the exoccipital sutures form an angle of about 135,
and the ventral midline of these elements almost touches or borders
the foramen magnum in a fashion typical of metaxytheres and re-
ferred felsinotheres. A similar point in Halitherium schinzi displays
an exoccipital angle of 1670 and a 21 mm separation from the fora-
men magnum.
Perhaps other distinctive differences could be determined between
the two genera, but the above can be readily evaluated with a cursory
examination. Variance between Halitherium and the genus Metaxy-
therium, plus related felsinotheres, is well known, so a review of differ-
ences would not add to the readily available information. Comparison of
Halitherium with Caribosiren was made by Reinhart (1959: 8-21).

Halitherium olseni new species
TYPE (FGS V6094).-Cranium and mandible, essentially complete;
postcranial elements present from anterior half of left side; cervical
vertebrae present; thoracic vertebrae partially present; fragments from
right side.
TYPE LOCALITY.-FGS Locality No. 82, E bank Suwannee River at
water level, approximately 1.6 mi. below White Springs, Sec. 11, T2N,
R15E, Hamilton Co., Florida, 11 December 1961. Specimen found in
sand, fine to medium quartz in a matrix of clay, containing hard nodules
of sandy limestone. Bed is approximately 5 ft. above Suwannee (Oligo-
cene) limestone and contains impressions of pecten-like shells, too frag-
mentary for an accurate determination to be made.
AGE.-Middle to Late Miocene.1
SAge assigned by Dr. Robert O. Vernon, Division Director and State Geologist, Florida Geologi-
cal Survey, 18 March 1965.




Fr. boss


FIGURE 17.-Hesperosiren sp. (FGS V4250): dorsal view of nasals, frontals and
anterior end of parietals.

DIAGNOSIS.-Acute ventral bend to rostrum; functional tusklike in-
cisors present; supraorbital processes of frontals abruptly recurved; swol-
len, rounded parietals with temporal crests narrowly separated; unusually
large lacrymal, with duct and well developed lateral tuberosity; supra-
occipital-exoccipital sutural contact 1320; supraoccipital separated from
foramen magnum by 16.5 mm; dental formula: 1-0- 1 3
2or 3- 0-1-3



FIGcRE 18.-Halitherium olseni (FGS V6094): dorsal view of cranium.


FIGURE 19.-Halitherium olseni (FGS V6094): lateral view of cranium.

PREMAXILLARY.-The premaxillaries have a strong ventral deflection
similar to that of Halitherium schinzi and Metaxytherium ( = Felsi-
notherium) serresi. The downward deflection is intermediate between
that found in the larger metaxytheres and referred felsinotheres and the
more acute angle present in Caribosiren and Dugong. From a dorsal
view the overall construction of the rostrum most closely resembles that
of M. serresi and Miosiren kocki. As in all Sirenia, the rostrum narrows
ventrally, but the large alveoli and prominent incisors preclude a slightly
expanded rostral area necessary to house such tusks. This is in contrast
to such genera as Caribosiren, Hesperosiren, and North American metaxy-
theres in which the lack of incisors results in a narrow wedge-shaped
rostrum. A moderately swollen dorsal boss is present on either side of
the rostrum, a feature found to some degree in all Sirenia, even in Tri-
chechus, which bears the least developed rostrum of living or fossil
members of the order. The ventral surface of the premaxillaries is of a
general sirenian construction. It is concave, centered with a narrow
continuation of the palatal gutter, and bordered by thin, sharp, overhang-
ing sides. Its rugose surface is pitted with many small foramina, and
no doubt it bore a horny pad when the animal was living.
The mesorostral fossa is elongate, similar to that in Halitherium
schinzi (Lepsius 1882: fig. 92), but apparently differs from the specimem
illustrated in having a slightly developed anterior arch. Development
of this small anterior arch ranges from strongly pronounced in Dugong
to virtually absent in Trichechus. An incipient development of the arch
may be noted in Eotheroides (Reinhart 1959: 54, fig. 8-B), but it is
absent in Prorastomus (Owen 1875), Protosiren fraasi (Andrews 1906,
[=Eotherium aegyptiacum]), and in Sickenberg's (1934) stylized or
generalized illustrations of various Eocene forms. It is quite likely that


TABLE 6.-MEASUREMENTS OF Halitherium olseni.

UF 3965 H. olseni

Length of cranium 367.5
Maximum width mesorostral fossa 50.5
Posterocentral border of frontals to
anterior border of mesorostral fossa
Length of zygomatic arch 99.0

Length, cheek tooth row 62.0
Length M3 17.9
Width M3, anterior loph 13.5
Width M8, middle loph 11.5

Maximum width at posterior end, above squamosal wings 53.5e 53.7
Minimum width between temporal crests 12.0 12.0
Maximum width at anterior end of
frontal-parietal suture 39.0
Maximum length 86.0e

Maximum width 67.7e 71.0
Maximum height 48.2e 48.2
Thickness of ventral medial border 17.0 -
Width between lateral borders of supraorbital processes 122.5
Posterocentral border of frontals to
anterior end of supraorbital processes 136.0
Posterocentral border of frontals to
posterior end of supraorbital processes 102.5

Width between lateral extremities of
transverse processes 110.0
Height of transverse processes 32.7
Ventral surface of atlas to dorsal border of anterior cotyle 47.0
Width between lateral extremities of posterior cotyles 58.0
Estimated height if uncrushed 64.5

Maximum height 74.0
Anterior margin of odontoid process to posterior margin
of centrum 42.0
Length of facet on ventral side of odontoid process 24.3
Maximum width of facet on ventral side of
odontoid process 22.8
Maximum width between lateral extremities of cotyles 62.7
Height of anterior cotyle 23.5
Maximum width anterior cotyle (right) 21.6
Maximum width, neural canal 28.6
Height of neural canal, anterior border 24.5



UF 3965 H. olseni

Width between lateral extremities of
postzygapophyses 48.5
Width between lateral extremities of
dorsal transverse processes 74.7
Width between lateral extremities of
ventral transverse processes 52.1

Height 60.8
Width, estimated 77.0
Width between lateral borders prezygapophyses 56.1
Height, neural canal 23.6
Maximum width neural canal 29.7

Height, estimated 60.3
Width between lateral borders prezygapophyses 65.3
Height, neural canal 23.3
Width, neural canal 31.0

Minimum distance between ventral protuberances 46.0

Border of coracoid process to ventral edge of
glenoid cavity 52.3
Ventral border of glenoid cavity to projection for
teres major 128.6

Dorsal border of greater tubercle to ventral border of
trochoid 136.3
Lateral border of dorsal head to lateral border of
greater tubercle 48.7
Anterior border of dorsal head to posterior border of
lesser tubercle 46.7
Maximum width between lateral extremities of entocondyle
and ectocondyle 45.5
Maximum width of trochoid on medial side 20.0
Maximum width of trochoid on lateral side 27.7
Minimum distance between border of greater and lesser
tubercles 7.7

Length of radius 111.5
Maximum height of radial epiphysis 5.2
Length of ulna 134.0
Maximum height of ulna epiphysis 5.9
Length of interosseous space 50.0
Width across semilunar and radial notch 28.0
e = estimated


the apparent absence of the incipient arch in most of these early animals
results from lack of detail in illustration.
Insertion and posterior termination of the premaxillary rami are
located at a point lateral to the posterior end of the supraorbital process.
Although sutural contacts are not prominent in this area ,those present
lead one to interpret that the frontals and nasals covered the posterior
ends of the rami, and therefore dorsal exposure of these arms was close
to the forward edge of the supraorbital process.
NASAL CHAMBER.-A well preserved perpendicular plate, the lamina
perpendicularis, merges posteriorly with an indistinct crista galli and
cribiform plate. Dorsally the rugose surface of this plate bears evidence
of cartilaginous extensions. The ventral border is a thin, sharp edge,
clasped on either side by thin ascending plates of the vomer. From
either side of the posterior terminus of the perpendicular plate, promi-
nent nasoturbinals rise dorsad, and if these elements are in their correct
position and not crushed upward the following two nasal structure pos-
sibilities exist: 1) the nasals would probably be short in an anterior-
posterior direction, and 2) the anterior border of the nasals would rise
noticeably to override part of the nasoturbinals. Contacts between the
nasals and the adjacent frontals cannot be determined; therefore, the
configuration of these bones is unknown. Based on remains of the bones
in the area, however, I believe that if nasals were present they would
not have so great a forward extension as in the illustration of H. schinzi
Lepsius 1882: pl. 9, fig. 92), which shows the anterior nasal border well
in advance of the supraorbital process. In view of the actual specimen
shown by Lepsius (1882: pl. 2, fig. 5), it is quite possible that his inter-
pretation of this extension is incorrect. Extension of the anterior nasal
border slightly in front of, or even with, the forward edge of the supra-
orbital process is found in Eocene Sirenia because of the relatively large,
well-developed nasals. A reduction in size of sirenian nasal bones occurs
through time, ranging from the large Eocene forms, which have well
developed nasals that meet in the midline, to Mio-Pliocene metaxytheres
with reduced nasals lying behind the premaxillary rami and separated in
the midline by the frontals, to the virtually non-existent nasals of the
recent Dugong, Trichechus, and Hydrodamalis. A reduction in size of
nasals by telescoping occurs in all land mammals that have become ac-
climated to marine water and is clearly seen in the Cetacea.
The posterior half of the nasoturbinals is closely appressed to the
lateral borders of the nasal region, with the anterior half terminating in
a pronounced hooked processus hamatus lying well out into the meso-
rostral fossa. The nasoturbinals are thin, laterally compressed, concave
medially, convex laterally, and in general shape are similar to those of


Caribosiren. These are in contrast to the more stoutly built elements
present in the later Mio-Pliocene metaxytheres.
FRONTALS.-In overall configuration the cranial roof of H. olseni is
thin and elongate, and most closely resembles the European Halitherium
christoli (Spillman 1959: fig. 10), although it differs in certain details.
Elongation of the cranial cap is also characteristic of the lesser known
European Upper Oligocene Halitherium pergense, but the Florida speci-
men differs from this species by the presence of more strongly developed
and convergent temporal crests and by a flat, vertical temporal region,
rather than a gentle lateral widening in this area. Both H. olseni and
H. christoli differ from H. schinzi, which has foreshortened frontals and
less recurvature of the supraorbital processes.
The unique shape of the supraorbital processes of H. olseni serves to
distinguish it readily from all other sirenian genera and species. Re-
curved processes identify the specimen as Halitherium (Fig. 18), but
no other published specimen has been noted with such prominent re-
curved posterolateral borders. This feature is further accentuated by
a thin sharp crest that, as a continuation of the temporal crests, extends
along the lateral borders of the frontals to the middle of the supra-
orbital processes. At this point the crests diminish in height and recurve
to the posterolateral borders of the processes. The lateral borders of
the processes are strongly directed ventrally and are divided by a small
central indentation. These two features are present to some degree in
all Sirenia but usually are not as accentuated as in this specimen.
Unfortunately the anterior central portion of the frontals is missing,
and no contacts between frontals and nasals are available. From ad-
jacent fragments, a gentle rise in the midline of the frontals is suggested.
In the posterior portion of the midline, the frontal-parietal contact
forms the typical V-shaped juncture, with the parietals overriding the
frontals and extending well forward along the temporal borders.
LACRYMAL.-The lacrymal is a short thick semilunar shaped bone,
whose main configuration can be attributed to a prominent anterolateral
knob. Ventrally it is in contact with the jugal, dorsally with the post-
orbital process of the frontal and the premaxillary arm, and anteriorly
and lingually with the maxillary. These lacrymal contacts are essentially
the same from the Eocene Eotheroides sp. through the Recent genera,
although the element is reduced, nonfunctional, and essentially lost in
Dugong, Trichechus, and Hydrodamalis. Behind the anterolateral knob,
a small lacrymal duct passes toward the orbital region, a state that in-
dicates a functional duct. The presence of a functional duct suggests
that Halitherium olseni is not far removed from a terrestrial ancestor
in which an efficient tear duct was a necessity. Though a small sirenian,


H. olseni has a much larger lacrymal than do either the European hali-
theres or Caribosiren, and despite different proportions this bone is closer
in size to those of Metaxytherium cuvieri and Metaxytherium jordani
(=H. vanderhoofi). The latter two metaxytheres probably reached
about twice the body size of the Florida Halitherium, thus emphasizing
again the exceptional size of the lacrymals in this relatively small animal.
Evolution of the lacrymal through time was reviewed by Reinhart (1959:
57-58), so only the main points will be mentioned here.
1) There was an irregular progressive loss of function in the sirenian
lacrymal. It had already reached this stage in the Middle Oligo-
cene Caribosiren, but was still functional in the later Pontian Mio-
siren. Neither Trichechus, Dugong, nor the recently extinct Hydro-
damalis retained a true functional lacrymal duct.
2) The lacrymal bone has become progressively smaller and shows
much variation in size and shape in various phyletic lines.
3) As the lacrymal has lost some contact with the maxillary, it has
correspondingly gained contact with the jugal. The position of the
lacrymal has remained relatively constant, with consistent mainte-
nance of contact to the postorbital process of the frontal.
Measurements of the lacrymal are: maximum height, 33.5 mm;
maximum anteroposterior length, 21.5 mm; and maximum lateral width
at anterolateral tuberosity, 24.4 mm.
PARIETALS.-A planum parietal is almost non-existent, due to the
well developed lyriform temporal crests tending to meet in the midline.
As mentioned earlier in the study, cranial caps with narrowly separated
parietal temporal crests cannot automatically be referred to Halitherium.
If parietals alone are examined, such specimens could be referable to
Hesperosiren and, to a lesser extent, certain metaxytheres, (Caribosiren
and Trichechus). Because of this, Simpson's (1932: 445) referring the
two parietal-supraoccipital elements to Halitherium alleni cannot be com-
pletely accepted, although the generic designation is quite plausible and
probable. Validity of the specific name alleni is questioned, since it is
virtually impossible to distinguish the parietal from those of various
European halitherines.
The lyriform crests of H. olseni are broadly rounded, but bear a thin
sharp crest at the dorsolateral edge. A deep groove present at the dorso-
posterior border follows the medial side of the lyriform crests forward
and is the cause for accentuation of the crests in this area. In general
appearance the parietals are intermediate between the cranial caps of
Halitherium antiquum (Leidy) which were illustrated by Allen (1926:
pl. 11, figs. 1-2) and later referred by Simpson (1932) to H. alleni.
Internally the parietals form the greater part of the superior, lateral,


and posterior walls of the braincase. Extending forward from a pro-
nounced triangular internal occipital protuberance is the wedge-shaped
falx cerebri, which bifurcates immediately in front of the protuberance,
thins to a crest, and ends on either side in contact with the dorsal part
of the crista galli. In most adults (except Hydrodamalis), regardless
of genera, the parietal protuberance is separated from the supraoccipital
by a moderately deep, thin, transverse sulcus or groove, which marks
the contact between the bones. In H. olseni the groove is wide and not
noticeably incised, a factor believed to result from the youth of the in-
MAxLLA.-Ventrally, with the exception of the anterior end of the
maxillary-jugal contact, sutural relationship of the maxillaries with sur-
rounding elements is in all respects like that of Halitherium schinzi
(Lepsius 1882: pl. 10, fig. 96). In contrast to H. schinzi, the forward
edge of the orbital ring has a pronounced posterior slant on the lateral
border, as compared to the prominent anteriorly directed orbital ring
of the European form. In front of P4 the anterior continuation of the
palatal gutter is narrowly constricted almost to a point of closure, then
expands with a gradually widening taper that culminates in a maxillary
ridge swinging abruptly outward to the ventrolateral margins of the
rostrum. A very similar palatal condition is also present in Metaxy-
therium serresi (Dep6ret and Roman 1920: pl. 2, fig. Ic), in contrast to
other members of the metaxythere-felsinothere line (Caribosiren and
Dugong), in which the palatal gutter is less lyriform, more open, with
lateral termination of this groove formed by overhanging parallel borders.
A broad, flat, horizontal zygomatic-orbital bridge extends laterally,
its posterior border beginning at a point lateral from the rear end of M2.
The anterior border of the bridge cannot be determined exactly be-
cause of missing fragments. Laterally the maxillary meets the jugal
in a broadly arched and expanded vertical plate. The exact nature of
the posterior termination of the maxillary-jugal suture is believed to
be shown by a circular rugose surface on the inner surface of the jugal
at a point immediately posterior to the beginning of the vertical expan-
sion of the jugal. The bony relationship in this region is quite similar
to that in Caribosiren (Reinhart 1959: p. 10, figs. A-C).
VoMER.-There is a general similarity of this element to that of
Metaxytherium and Dugong, but it differs markedly from the condition
in Trichechus, which bears a flat anterior processus maxillaris, in con-
trast to the thin vertical parallel plates present in the Florida form. Be-
cause of the delicate construction of this bone and its general loose
sutures with surrounding elements, it is not usually preserved. As a re-
sult of its rarity in the fossil record, there is a lack of specimens avail-


able for comparison, and most illustrations either fail to show the vomer
or it is displayed in a very generalized manner unusable for comparison.
The lower anterior half of the elongate U-shaped vomer lies in a deep
indentation of the maxillary, and it is bordered on either side of the base
by a small canal leading from the anterior incisive foramen to its dis-
appearance on the sides of the processus maxillaris. The dorsal anterior
half of this element extends upward into two thin parallel bones, with
the greater accentuation posterior. If the thin vertical plates are in place,
as they appear to be, then the dorsal extension is unusually high. The
posterior half of the vomer is indistinct, except for the thin ascending
plates that clasp the perpendicular plate.
SUPRAOCCPITAL.-As is true even in advanced fetal stages of Dugong
and Trichechus, the supraoccipital parietal complex in H. olseni is firmly
ankylosed. Juncture of the dorsal surface of the parietal and the pos-
terior surface of the supraoccipital is 1150. The anterior border of the
superior nuchal line is upraised and arched forward. On the medial
nuchal line the triangular external occipital protuberance diminishes to
a thin crest, which extends two-thirds of the way down this element
and merges with a broad-grooved, indented ventral sulcus. The medial
crest is bounded on either side by an oblique rugose fossa. Sutural con-
tact with the exoccipitals forms an angle of 132, and the ventromedial
portion of the supraoccipital is separated from the foramen magnum by
an estimated 16.5 mm. These measurements are similar to those of the
large European Metaxytherium krahuletzi, which has complementary
figures of 1300 and 18 mm, but is otherwise structurally different and
readily distinguishable. The irregular progressive tendency for the
supraoccipital to touch the foramen magnum and for the supraoccipital-
exoccipitals to meet at a more closed or small angle is considered fully
on page --. From this discussion it will be noted that the measure-
ments given for the Florida specimen are about intermediate between
those of Eocene forms and the living Dugong.
EXOCCIPITAL.-Articular surfaces of the condyles are convex (as in
all Sirenia), have a semi-lunar outline, and, due to the excellent state of
preservation, bear sharp borders as in living forms. The lateral border
of the condyles is paralleled by the outermost borders of the exoccipitals,
with a deep supracondylar fossa separating the two parts. Borders of the
foramen magnum form a transverse flattened oval, with a slight rise in
the midline. Such a description fits the shape of the foramen magnum in
the bulk of fossil Sirenia, but differs from the broader, flatter oval form
of Trichechus, which also lacks a dorsal indentation and can be differenti-
ated from Dugong, which bears a triangular foramen. In this same
feature, Halitherium schinzi Lepsius (1882: pl. 5, fig. 52; pl. 10, fig. 97)


differs in the presence of an abrupt small dorsal arch. Another point of
contrast between the Florida specimen and H. schinzi is the paroccipital
process, which in the latter, as in metaxytheres and referred felsinotheres,
lies below the ventral surface of the condyles, as compared to a higher
placement in H. olseni.
The condyloid foramen, a short, thick canal that provides an exit
for the hypoglossal nerve, is present on the lateral side of either condyle,
at which point it ascends dorsally and medially, bifurcating into two
branches on the inner surface of the cranium. As would be expected,
this foramen maintains a relatively constant size; that is, it enlarges as
the animal grows and reaches greatest size in the largest genera.
BASIOCCIPITAL.-A review of this element in Eocene to Recent Sirenia
reveals no basic change in shape and but minor superficial differences.
Fusion has obliterated the basioccipital-basisphenoid contact, a mod-
erately unusual condition. Sickenberg (1931: 428) notes the general
occurrence of a nonfusion of this suture, discusses its relationship to
other neotenic conditions in the Sirenia, and attributes these to hypo-
thyroid metabolism. Nonfusion of this suture occurs in Caribosiren,
Halitherium schinzi, H. christoli, Metaxytherium jordani, Hesperosiren,
and many other forms. From examination of living and fossil specimens,
it is suggested that, as with most other sutures, the basioccipital-
basisphenoid contact tends to remain open in young animals and to fuse
in old age.
The sella turcica is a shallow depression that is moderately well de-
veloped considering the small size of the individual. It was reported
to be present by Lepsius (1882) in H. schinzi. It is absent in the adult
specimens of Dugong I have examined, an observation in keeping with
those of Freund (1908). Sickenberg, however, reported a gentle de-
pression in this area in very old individuals of Dugong. Adult trichechids
show this minor feature well developed; it is relatively less conspicuous
in Hydrodamalis. A small ridge, the dorsum sellae, is present, but it is
not known if the lateral clinoid processes were ever present. As in
Metaxytherium jordani ( = H. vanderhoofi), the typical upraised rugosity
found on the anteroventral surface reveals some details not generally
preserved. A narrow median ridge in the posterior half divides the
rugosity into two oval rugose areas, which in turn are split by an anter-
oposterior directed sulcus.
PTERYGom.-Posteriorly the pterygoid processes are deeply incised
by a broad groove, which widens ventrally, terminates laterally in thin
edges, and ends dorsally in a small lateral spine. Rugose surfaces on the
lateral and anterior surfaces of the processes indicate that sutural con-
nections with the alisphenoid and palatines are missing. Although the


pterygoid processes are not exactly like those of other sirenians, neither
are they distinctive from other halitheres or the metaxythere-felsinothere
JUGAL.-The anterior-dorsal part of the maxillary separates the lacry-
mal and jugal from the premaxillary arm, a condition also found in the
Eocene Eotheroides sp. (Reinhart 1959: fig. 8). Abel (1912: 312)
illustrated the maxillary of Eotheroides aegyptiacum as forming the entire
lateral ring of the orbit. Weber (using Abel's reconstruction [1928:
fig. 330]) illustrated, and Sickenberg (1934: 196) stated, that in Eother-
oides libyca the forward end of the orbit is composed of the lacrymal
and jugal. H. olseni, like Eotheroides sp. above, is intermediate between
the position interpreted by Abel and that of Weber and Sickenberg. It
has the lateral surface of the orbital ring composed of the jugal instead
of the maxillary, which differentiates it from E. aegyptiacum as described
and illustrated by Abel. Furthermore the forward border of the orbital
ring is composed of the maxillary, jugal, and lacrymal, rather than ex-
clusively of the two latter bones, as Sickenberg and Abel have interpreted
for E. libyca.
The maxillary-jugal contact at the anterior end of the orbital ring, as
seen in the illustration of H. schinzi by Lepsius (1882: pl. 8, fig. 87), is
basically the same as in the Florida halithere, except that a lacrymal is
not present in the former. The same series of illustrations by Lepsius
is used by Dechaseaux (1958: 350, fig. 25), but here the maxilla is
labeled as the jugal, and the jugal bears no designation. Because the
lacrymal is loosely sutured to the surrounding bones, it is rarely present,
and for this reason I place strong emphasis on the interpretation of the
bony relationships displayed in this area in both H. olseni and Eother-
oides sp. This is in contrast to many illustrations that I feel are incor-
rectly interpreted and illustrated, due to lack of clarity in the specimens.
Dorsoanteriorly the jugal abuts abruptly into the ventral end of the
lacrymal. The orbital rim is laterally widened and flat, as is true of
Sirenia in general. The ventral surface beneath the orbit is broadly
rounded; dorsally the orbital ring ends in a thick rugose surface, and it
appears that this area abutted abruptly into the zygomatic arch. As a
result of crushing during preservation, a strong skew exists in the orbital
ring. Nevertheless, it is interpreted that the posterior end of the orbit
was more closed than in H. schinzi, but less so than in Trichechus. Be-
hind the orbit there is a marked vertical expansion, which thins rapidly
posteriorly and terminates as a thin vertical plate. An indentation and
general rugose ventral surface of the zygomatic arch indicates that the
malar process ended at a point lateral to the middle of the glenoid plate.
SQUAMOSAL.-The overall configuration of this element is similar to


that of Halitherium schinzi and Metaxytherium serresi. Should it differ
from any Eocene through Pliocene sirenian, it would be in the more
posterior extension of the malar process, which in this instance is opposite
the center of the tuberculum articulare. The articular plate is roughly
oval, with the long axis transverse, and (with the exception of a slightly
upraised border) it is flush with the ventral surface of the zygomatic
root. A deep postarticular fossa, of the same size and shape as that in
Dugong, is also present.
Above the posttympanic process, the squamosal swings in a small
irregular circle around the mastoid part of the periotic, thus leaving
this bone exposed externally. Internally the petrosal part of the periotic
is loosely enclosed in a large groove of the squamosal.
EAR BoNEs.-The ear bones are dense and compact, as is true of most
marine mammals. An irregular ovoid to reniform periotic is set in a
deep groove of the squamosal. The petrosal is separated from the mal-
leus by a wide fossa incudus, which is broadly rounded dorsally, as
compared to the condition in Dugong and Trichechus in which the dorsal
termination is a narrow V. The posteroexternal face bears a flat rugose
surface, which is exposed externally and partly enclosed by the squamosal
and exoccipital. Because of the skewed condition of the cranium in this
area, the relationship of the supraoccipital to the exposed mastoid is not
known although the general condition appears to have been very similar
to that in Dugong. Both pars labyrinthiae are missing, as well as the
various foramina located at the juncture with the main body of the
In H. olseni both tympanic rings are detached. These are of typical
horseshoe shape, with the anterointernal arm bearing several small, weak
points for attachment to the petrosal, and the larger posteroexternal
attachment is missing. The ventral margin terminates in a thin crest,
similar to that in Trichechus, with a ventroanterior edge produced into
a prominent spine. In general outline there is a close resemblance to
Halitherium schinzi (Lepsius 1882, pl. 2, fig. 12).

LEFT MALLEUS.-A vertical, transversely thin, helmet-shaped manu-
brium extends ventral from the main body of the malleus, and is uniquely
different in details from that in Dugong and Trichechus. It is noticeably
thinner than in Dugong, more like that of Trichechus, and although
slightly broken it clearly lacks the prominent ventroposterior extension
of the manubrium present in the two living genera. The main body is
twisted and bears a conical, internally directed prominence termed the
lateral process. On the outer side, the spur of the malleus is slightly


twisted and directed inward as in Dugong. Anteriorly the general con-
figuration also resembles Dugong, but posteriorly H. olseni has a simpler
construction (i.e. fewer and less accentuated undulations).
RIGHT INCUS.-The incus in its entirety is almost an exact replica
of that present in Dugong. The crus breve and crus longum diverge
from one another nearly at right angles. The crus breve is short and
conical, and extends dorsally into the fossa incudis. On the anterior face
of the body is a concave oval facet, which articulates with the head of
the malleus. Lying immediately dorsal to this is the V-shaped facet,
into which the spur of the malleus is inserted. As an extension of the
body the crus longum is twisted medially and terminates in a small oval
facet against the stapes.
RIGHT STAPES.-Although the stapes is separated from its articulation
with the incus, a small oval articular facet (the lenticular process) is
present on the incus, and a corresponding articular surface is undoubtedly
present on the head of the stapes. The base is a flat oval plate, and
the general proportions of the stapes are similar to those of all Dugongi-
dae. The stapes is 8.2 mm long, with a distance of 3 mm separating
the base from the center of the stapedial foramen. As near as can be
ascertained from living and fossil specimens, no taxon is distinct be-
cause of unusually large, or small auditory apparatus, but rather the
sizes of ear bones and ossicles are in proportion to the size of the in-
The value gained from a description of the auditory ossicles is
limited, because of the difficulties involved in interpreting distinctions
among such intricately constructed elements, based on descriptions
and/or illustrations in the literature. A detailed study of actual speci-
mens covering the time sequence from Middle Eocene to Recent might
reveal features characteristic to various taxa that are not readily per-
ceivable except by direct examination.
UPPER INCISOR.-As with all tusked Sirenia, the upper tusk-like in-
cisors appear almost too large for the cranium, and seem grotesquely
inconsistent with other cranial proportions. They are similar to incisors
of the European Halitherium, Metaxytherium, and referred felsinotheres
in having a laterally compressed oval outline and a heavily-striated,
rough surface. Extending down from the base for approximately 21 mm
and encircling either incisor is a series of rings formed by small rugose
round knobs. A small, laterally compressed dental cavity is present in
the base of the tusk. In cross-section the occlusal end is roughly tri-
angular, and the lateral face sharply beveled with a chisel-like termina-
tion. Throughout the length of the tusks there is a gently lateral diver-


sion, the maximum separation of the incisors occurring at the occlusal
end. Maximum length of the incisor is 93.7 mm.
DENTITION.-An oval alveolar region is present in H. olseni. Very
similar counterparts are found in H. schinzi, Caribosiren, Metaxytherium
serresi, and M. forest, as compared to some of the anteroposterior
straight alveolar rows present in Trichechus and occasionally in some of
the larger Sirenia, such as Metaxytherium calvertense (FGS V5454).
Four heavily worn cheek teeth, interpreted as being M1-3 and P4, reveal
little detail, due to obliteration of cusp structure. Although H. olseni is
a young individual, the worn condition of the teeth suggests an animal of
old age. It suggests either an inability to efficiently masticate vegetation,
or the consistent presence of hard, abrasive sediments mixed with the
food. The worn molars could well have been a factor contributing to
the apparent premature death of the individual.
M3 is subquadrangular, with the three anterior teeth quadrangular.
In M2-1 minor remains of enamel on the occlusal surface show that an
anterior cingular valley was present on the buccal side, and a transverse
median valley present in M1-3. p4 also appears to have been lophodont.
The right M3 has two roots that recurve anteriorly and one or possibly
two vertically straight roots. M2 through P4 bear one root on the lingual
side and two on the buccal. The condition of recurved or hooked roots
is typical in fossil Sirenia, as opposed to straight roots in Trichechus,
vertical pegs in Dugong, and an absence of teeth in Hydrodamalis.
MANDIBLE.-TOrsiOn during preservation has warped the mandibles,
but dental differences between the North American and European
Halitherium are readily recognizable. A long diastema separates four
cheek teeth from 2-3 alveoli in the symphyseal plate. Broad, shallow
alveoli reveal the former presence of three nonfunctional incisors in the
left mandible and two in the right. The small number of symphyseal
teeth is reminiscent of the metaxythere-felsinothere Mio-Pliocene line
rather than of Halitherium, which normally bears alveoli for three in-
cisors and one canine. Whether this is an individual anomaly or a
trend similar to the loss of incisors, as in Trichechus and Hydrodamalis,
is not known. Innervation of the symphyseal face is accomplished by a
medial foramen, which enters the dorsoposterior margin, then becomes a
moderate furrow from which minor grooves branch throughout the
symphysis. Lateral to the incisor alveoli the lateral borders of the
symphyseal region flare out to a thin edge, the degree and prominence
of flare corresponding to the development of the incisors. A thin-edged
diastema separates the P4 from the symphysis. It can be clearly ascer-
tained that Ms is trilophodont, the size of the fragmented M2 suggests
a similar construction, and two lophs are discernible for M,. All cheek


FIGURE 20.-Halitherium olseni (FGS V6094): posterior view of cranium.

teeth are anchored by two roots: those of MI,- are thin anteroposteriorly
and transversely broad; the M, roots are more rounded; and the alveoli
of P4 suggest small round roots. No comparison of cusp details is pos-
sible because of the heavy wear. Except for their small size, the cheek
teeth of H. olseni cannot be distinguished from those of the European
H. abeli or H. christoli. The ventral border of the horizontal ramus
forms a broad symmetrical arc, similar to that in Dugong, H. abeli, and
H. christoli. The comparable region in Halitherium schinzi is less arcu-
ate; that is, it is more elongate and straight as the angle is approached.
Remains of the angle appear to be typically dugongid and not arched
anteriorly, as in Trichechus.
ATLAs.-There is, in general, a dearth of fossil atlases with which to
compare H. olseni, but the atlas of Halitherium abeli is essentially an
exact duplicate, and that of the Eocene Protosiren fraasi has the same
type of transverse processes and general proportions. The major dif-
ference existing between FGS V6094 and forms such as M. jordani, M.
cuvieri, M. forest, and M. serresi is the much greater height and promi-
nence of the transverse processes in the former. This condition results
from the development of the base of the transverse processes lateral
to the ventral arch, rather than midway up the atlas as in other fossil
specimens and recent genera. Unlike the distinct vertebrarterial canal


FIGiUE 21.-Halitherium olseni (FGS V6094): (A) anterior view of atlas; (B) pos-
terior view of atlas.

foramina, which pierce the transverse processes in H. abeli, those of the
Florida specimen are incipient and hardly functional. Presence of these
foramina is highly variable, often being well developed on one side and
closed on the other, sometimes not developed on either side, or incised
at the base of the process by a deep groove as in M. jordani (= H.
vanderhoofi). The double arch for the odontoid process and the neural


canal is in the shape of a figure 8, which is the state found in Oligocene-
Pliocene sirenians, in contrast to the laterally expanded, open neural
canal in Trichechus. The neural arch is crushed, so the correct height
of the element can only be approximated.
Axis.-Except for details, the axis of H. olseni does not differ from
that of M. serresi. The base of the odontoid process is bounded dorsally
and laterally by a deeper groove than any illustration or specimen
known to the writer, but this condition is undoubtedly due to the juve-
nile stage and the excellent degree of preservation. Beneath the process
the outline of the articular epiphysis is similar to an exaggerated egg,
with the anterior half pinched and the posterior half expanded suddenly
into a bulbous posterior border. This articular facet, well separated
from the anterior cotyles and lacking a prominent articular platform, is
in contrast to the condition present in adult metaxytheres and Trichechus,
in which the facet is elevated on a short platform, and a proximity of
cotyles to the facet is the rule. In general outline the epiphysis not only
differs from the more oval border of various metaxythere axes on hand
and from the round epiphysis of Trichechus but, most interestingly, it
is as large as those in axes 50-75% larger in overall proportions. An-
terior cotyles are quadrangular and more like those of Florida metaxy-
theres than those of M. serresi or Metaxytherium cuvieri, both of which
are vertically elongate. The border of the neural canal has a rounded
arch, as in Trichechus, rather than the condition typical of Oligocene-
Pliocene sirenians in which the dorsal half of the neural canal is more
quadrate. A rugose surface marks the dorsal surface of the neural arch
upon which a central weakly developed spinous process is present. The
anterior tubercle of the neural canal is roughly oval in outline, and
rugose where a pad of cartilage lays between this face and the atlas.
Postzygapophyses are as large as in the larger Trichechus, a feature in
keeping with the large articular facet on the ventral side of the odontoid
process. The base of the centrum is broadly concave, as opposed to the
usual flattened condition of both fossil and recent forms. Laterally the
centrum is flanked by two pairs of transverse processes; a thin, promi-
nent dorsal extension directed lateroposteriorly; and a small ventral pair.
These are present to a minor degree on a number of well preserved
fossil specimens and in Trichechus, but in no instance do any display
the development of H. olseni. Whether this is a character of taxonomic
value or the result of excellent preservation is not known.
CERVICAL VERTEBRAE.-The cervical vertebrae from 3 through 7 bear
slender neural arches that are nested within one another, and have the
reduced thickness characteristic of all mammals that have become
adapted to an aquatic environment. A triangular spinal canal is similar


FIGUxm 22.-Halitherium olseni (FGS V6094): (A) anterior view of axis; (B) pos-
terior view of axis.


to that of virtually all living and fossil sirenians, but is different from the
more rounded arch of M. serresi. The cervical centra form a broad
transverse ellipse, are of equal thickness in numbers 3, 4, and 5, and
increase in thickness rapidly in 6 and 7. Broad, posteriorly directed
transverse processes, pierced by a vertebrarterial foramen, are present
on number 3 and are believed to have been present on numbers 4 and 5.
A similar situation is found in cervical vertebrae of European halitheres.
Peculiar to the sixth cervical is a powerfully constructed, ventrally di-
rected knob at either ventrolateral corner of the centrum. The base of
each of these tuberosities is pierced by a foramen. The seventh cervical
bears large thick transverse processes that begin laterally from the flat
base of the centrum, a feature also true of M. serresi and M. forest. A
similar knob, but of different shape, is present on the sixth cervicals of
the Egyptian Eocene Eotherium aegyptiacum and E. libycum, and on
the third cervical of M. forest. Generally such protuberances are
missing on fossil cervicals and are completely lacking in Trichechus. No
definite criteria that can be used for generic or specific distinction appear
to be among the reduced cervicals. Perhaps more information concern-
ing the size and shape of the transverse processes will prove to be of
value in such a diagnosis. The first thoracic, or eighth, vertebra bears
the typically wide transverse processes with distinct facets for articulation
with tuberculum and capitulum. No vertebrarterial foramen pierces
either the seventh or eighth vertebra.
SCAPULA.-The overall outline and proportions of the scapula agree
best with Halitherium abeli, Metaxytherium krahuletzi, and M. serresi,
except that in H. olseni this bone is smaller. The glenoid cavity is deeply
concave, with an egg-shaped articulate surface, and the coracoid process
is well developed for a juvenile individual. Beginning at the ventral
edge of the glenoid cavity, the curving axillary border becomes gradually
thinner, ending in a prominent projection for reception of the teres major
muscle. This projection, which is also pronounced in Dugong, is seldom
found preserved, but where it is not destroyed, as in scapula of M. serresi
and M. forest, the projection is poorly developed. The border remains
thin posterior to this projection. The only spine seen was fragmented,
but appeared to be short, as in Halitherium christoli.
HUMERUs.-The humerus is fore-shortened, as in all Sirenia, but it
nevertheless lacks the massiveness of the larger metaxythere humeri.
It is relatively slender, and in this regard resembles both the European
Halitherium schinzi f. delheidi (Sickenberg 1934: pl. 9, fig. 2a-b) and
M. serresi (Dep6ret and Roman 1920: pl. 2, fig. 3) in some detail. The
dorsal head is hemispherical, is joined to the body by the short, restricted
anatomical neck, and assumes a relatively greater size in the distal end


FIGURE 23.-Halitherium olsent (FGS V6094): (A) anterior view of third cervical
vertebrae; (B) posterior view of third cervical vertebrae.


of the humerus than in the large metaxytheres. The long axis of the
greater tubercle lies lateral and parallel to the dorsal head, the two
separated by a deep U-shaped sulcus. A continuation of this bicipital
groove between the greater and lesser tubercles becomes deeper, with
the accentuation greatly increased where the border of the greater tubero-
sity curves abruptly toward the lesser, and partially closes the U-shaped
sulcus or groove. Closure of this intertubercular groove is greater than
in any illustration or description known to me, although the more massive
Metaxytherium krahuletzi (Abel 1904: pl. 14, fig. 4b) and Metaxytherium
sp. (UF 3619) approach this condition. Juvenile humeri of Metaxyther-
ium cuvieri display a relatively open notch, with more closure occurring
as the lesser tubercle develops a larger size with age. The head of the
greater tubercle is a modified semilunar outline, the entire surface ap-
parently having served for muscle insertion.
A small circular area for muscle insertion lies at the base of the
greater tubercle, immediately above the sharply recurved deltoid process.
This thin process is usually worn or broken in fossil humeri, but the
sharp curvature is present to a similar degree in H. schinzi f. delheidi,
M. serresi, M. cuvieri, and Dugong. Below the distal head the triangular
shaft displays a slight degree of curvature, which is similar in the humeri
of all sirenians. The proximal third of the humerus is virtually the
same in Sirenia from the Oligocene through the Pliocene. A semicylindri-
cal, oblique trochoid is bounded laterally by a prominent entocondyle
and a small ectocondyle. Both the olecranon and coronoid fossae are
deep and well developed. In metaxytheres the olecranon fossa remains
a deep depression throughout the adult stage, but the coronoid fossa
tends to fill with bone, with the the result that only a shallow depression
marks the site.
RADIUS AND ULNA.-The coronoid process of the ulna is well fused
in the upper extremity, but unfused at the distal end, except for a
small internal area immediately above the epiphyses. The usual crossed
torsional position between these bones is less developed than in any
other sirenian specimen known to me, regardless of genus. This rela-
tively straight relationship could be a function of age with torsion and
a greater bony opposition occurring as it becomes necessary for the
animal to propel a larger body. A long, thin interosseous space is
present that is similar to that in Halitherium and Metaxytherium, more
closed than in Dugong, and lacks the outward bowed radius of Triche-
chus. Shafts of the radius and ulna are slightly bowed, with the only
major change in configuration being a deep depression running the
length of the internal face of the ulna. Semilunar and radial notches
form a concave, transversely broad semicircle. The olecranon process of


the ulna is the same as that in M. serresi. The distal epiphysis of the
ulna is larger than that of the radius, as is usual in sirenians. The
epiphyseal surface of the ulna is concave, rounded, and elongate, whereas
the corresponding articular surface of the radius is transversely broad,
somewhat flatter, and kidney-shaped. Lack of fusion of the epiphyses,
plus the cellular structure of these elements, indicates a young animal.
In general proportions it appears to show closest affinities to the Euro-
pean Halitherium schinzi f. delheidi and to M. serresi.

METACARPALS.-Remains of the proximal ends of the left metacarpals
3, 4, and 5, and the distal portion of metacarpal 2, reveal partially
broken articular surfaces, but provide no new information regarding the
genus or species. A partially combined element of the intermedium and
radiale is, as near as can be ascertained, the same as that of Halitherium
schinzi f. delheidi.
Rms.-Heads of ribs 1 through 3 are missing or fragmented, but
remaining portions strongly suggest the usual condition of the anterior
ribs (i.e. a tuberculum and capitulum separated by a broad groove).
As in Trichechus, the first rib is anteroposteriorly flattened, acutely
curved at the proximal end, and bears a small tuberosity on the distal
termination. Articular facets on the seventh cervical and first thoracic
vertebra indicate the usual angular, two-sided capitulum characteristic
of a number of interior ribs.
The second rib is much more robust than the first and is terminated
distally by a flattened thin end, the inner posterior face being formed
by a flat plane. The third rib is marked by an increasingly robust con-
struction, a noticeably larger size than the two anterior ribs, a typically
banana-like shape, and a distal termination that ends in a flat plane like
that of the second rib.
No difference in size of ribs 4 through 9 is evident. Separation of
tuberculum and capitulum is essentially unchanged, but the double
face of the capitulum tends to become a single facet in a posterior pro-
gression of the ribs. An obvious change is noted in the angle, which
is marked by a minor groove in the first rib and becomes a gentle round
protuberance bounded on either side by a broad shallow groove. Flat-
tening of the distal end is evident in ribs 1 through 7, but thereafter the
rib termini become more rounded. The distal portions of ribs 10-12 are
enlarged, bulbous, and were probably attached to a cartilaginous or bony
sternum, as suggested by the rugose medial termination. Scattered re-
mains of many rib fragments near the posterior end and possibly right
side of the specimen are heavily weathered and add no new information.


MATERIAL.-TWO ribs (UF 4053).
LoCALITY.-Abandoned quarry of New Lebanon dolomite pit, Sec. 12,
T 16 S, R 16 E, SW Lebanon Station, Levy Co., Florida.
FORMATION.-Avon Park limestone.
AGE.-Late Middle Eocene.
/ Sirenian ribs are found abundantly in a number of north African sites;
However, the ribs here reported from the Middle Eocene of Florida are
considered a rarity. Vernon (1951: 108-110) substantiated this record
in his brief mention of a "Manatee rib" from the above dolomite pit,
which is accompanied by a detailed stratigraphic section. Because of the
significance of the discovery and the importance of eventually establish-
ing a stratigraphic relationship among Eocene sirenians, the section by
Vernon is included here:

On November 15, 1947, and July 11, 1948, boulders of this lime-
stone, completely dolomitized, were mined in the north side of
the pit, and these contained the large Lucinia of L-123, a Mana-
tee rib, and numerous dolomite casts of "Cerithium" n.sp.

A visit to the site of the abandoned quarry produced a number of
rib fragments from scattered boulders composed of massive tan sugary
dolomite, as well as a dolomite conglomerate. In what appeared to be
rocks of comparable altitude and lithology 10 mi. north of the New
Lebanon dolomite pit, more sirenian rib fragments were found. The
second collecting site (SE 1 Sec. 21, T14S, R16E, 1 mi. SE Gulf Ham-
mock, Levy Co.) is being actively worked as a dolomite pit. Since
Owen's record of the Jamaican Prorastomus in 1875, two recent reports
have been made on Middle Eocene sirenians in the western hemisphere.
Siler (1964) provided the first information of Eocene Sirenia in the
United States. This is a rib of typical dense structure, found in place in
the Gosport sandstone, Monroe Co., Alabama (assigned to Middle Eo-
cene, Auversian age, by Stenzel et al. 1957). The second record (Arata
and Jackson 1965) is a distal half of a rib from Little Stave Creek, base
of the Gosport sandstone, Clarke Co., Alabama (see Russell 1955: 454-457,
469). In neither case was the described rib figured. The Gosport sand-
stone of Alabama and the Avon Park limestone of Florida are equivalent
in age.
Upper-Middle Eocene beds of the Fayum bear moderate numbers of
sirenian specimens and, largely because of the relative abundance of
fossils from this region of Egypt, some students, such as Kurten (1969),


FIGURE 24.-Halitherium olseni (FGS V6094): articulation of humerus with radius
and ulna, left side.


r (restricted)




the world, this probably results from insufficient investigation of Middle
Eocene beds rather than from a lack of fossils. The earliest sirenians
with representatives in Jamaica, Florida, Alabama, Egypt, and possibly
Java would obviously have had a wider geographic range than these
isolated regions.
From the two illustrated Florida specimens, it can be seen that the
head and particular facets are missing in the more complete rib, but the
remainder is present from a point immediately anterior to the angle to
the ventral termination. Curvature of the shaft is moderate throughout,
with a slightly more acute curve at the angle. The other rib consists of
the lower half of the shaft, from which the distal terminus is missing.
Both te density of the bony structure and plumpness of the ribs im-
mediately suggest sirenian affinity, but beyond this no specific reference


FIGURE 26.-(UF 4053) Middle Eocene sirenian ribs from Avon Park limestone,

can be given. Ribs from the uppermost Middle Eocene sirenians of the
Egyptian Fayum also demonstrate the typical sirenian massiveness at an
early geological stage. Andrews (1906: 217) stated: "This specimen
shows that in Eosiren the ribs had attained the great thickness and gen-
erally massive structure characteristic of some of the later Sirenians."
A cylindrical marine boring and a number of circular, nickel-size indenta-
tions (probably echinoderm or limpet depredation) indicate that the
more complete rib either was not immediately covered by sediment after
death or was later uncovered after fossilization occurred. Similar bor-
ings or markings by marine invertebrates on sirenian ribs are moderately
common throughout the fossil record of this order. The greatest di-
ameter of the larger rib fragment is 30 mm at a point just ventral to the
Other than the ordinal designation, no taxonomic relationship to
the Middle Eocene Prorastomus from Jamaica is proposed or even ,
contemplated. The West Indian specimen was collected in the 1400 ft.
thick Richmond Formation, from which invertebrate studies (which
could be used to correlate between this formation and the Florida Avon
Park limestone) are lacking. Absence of comparable bony elements
further complicates any suggestion of relationship between the sirenians
from these two sites. The rib fragments from Florida can be readily
identified as sirenian, but to provide them with a taxonomic designation
below that of order is unwarranted. A concentrated collecting effort in
Florida and Alabama Eocene formations would undoubtedly produce
material that would permit a taxonomic diagnosis.


Inasmuch as the Recent Dugong ranges throughout large areas of the
Indian and southwestern Pacifiic oceans, one would expect to find re-
mains of fossil specimens from this region. Continental shores and
islands in this area, however, have not been consistently investigated
geologically or paleontologically, except in isolated sites, and knowledge
of fossil Sirenia is most fragmentary. Fossil Dugongidae are well known
from North Africa, Europe, and North America. From a study of the
distribution of this family through various Cenozoic epochs, and from a
consideration of the genera common to Europe and North America, one
might assume that these forms crossed the Atlantic, thus accounting for
the trans-Atlantic distribution of the genera and what appears to be
parallel development. An east-west distribution of the sirenians by way
of a Pacific route is seldom, if ever, mentioned as a possible dispersal
route, perhaps due to a lack of knowledge of trans-Pacific forms. Lack
of such information has made the work of Koenigswald (1952) on the
fossil sirenians of Java of unusual importance.
Koenigswald (1952: 610) listed the earliest sirenian remains as a
thick pachyostotic rib from Upper Eocene beds. To this he added, "too
fragmentary for description, this find must be mentioned here, as it is
the oldest indication for the presence of sirenians in the East Indies."
The age of the rib fragment is of marked significance in providing a
possible link between the North African Eocene sirenians and the Middle
Eocene Prorastomus from Jamaica. The importance, however, is lessened
by the fact that the rib is too fragmentary for description or illustration
and no catalogue number is given.
Javanese sirenian remains from the Lower Miocene include a "frag-
ment of a heavy bone in a marine limestone" and two tusks referred by
Riggs (1839) to the genus Sus. In reference to the tusks, Koenigswald
states, "As the formations in that neighborhood are marine, and also
'tusks' are typical for Halicore, these teeth most probably indicate a
Sirenian. The original specimen seems to be lost." Doubt was sub-
sequently expressed as to the proper reference of the tusks (Riggs 1839:
612) ". . the 'tusks' from the Lower Miocene of Djasinga-if they really
belong to a Sirenian . ." The assignment of the Lower Miocene speci-
mens to a sirenian appears vague and indefinite to me. The "heavy
bone" is not illustrated or described in any manner and the "tusks" are
lost. Identification of marine mammal tusks is a special problem and
little has been accomplished toward differentiating the myriad of sizes
and shapes with their varied triturating surfaces. Tusks of marine mam-
mals are found unidentified in many private collections and in all large





CHRONOZOON (indet. sirenian)

FIGURE 27.-Fossil sirenian localities in Java and Australia.

A most important sirenian is Indosiren javanense (von Koenigswald
1952) from the Upper Miocene (probably Sarmatian) of western Java.
This is a small, unknown, bilophodont, upper milk molar that completely
lacks development of roots. The anterior cingulum is typical in M1 of
most sirenians from Eocene forms up to Trichechus. Some variation,
however, may block an opening or the anterior loph and cingulum may
be worn low, thus giving the impression of an open cingular valley. In
most fossil Dugongidae, M3 is more likely to bear an open cingulum on
the buccal side, with M2 intermediate in this character between the
closed M1 and the open M3. The anterior loph consists of a transverse
arrangement of the paracone, protoconule, and protocone, all of which
are similar in size. A very straight transverse median valley divides a
slightly larger and wider anterior part of the tooth from the posterior.
An unusual feature for a tooth of this age is the absence of any major
or accessory cusp blocking the median valley. An open median valley
is characteristic of the Eocene Eotheroides, Prototherium, and the living
Trichechus. Miocene and Pliocene metaxytheres of a similar age as
Indosiren are characterized by a forward displacement of the metaconule.
The posterior loph is composed of a closely appressed and equal-sized
hypocone and metaconule, which are separated from a shorter stout


metacone. The posterior cingular valley is open on the buccal side, a
molariform trait common to most fossil sirenians.
In general appearance the occlusal surface bears a resemblance to
Trichechus and to the Upper Eocene Prototherium. A lateral view
again suggests Trichechus and also the tapir Tapirus. The tooth, as
Koenigswald (1952) states, differs from other genera of similar age
(Halianassa, Miosiren, and Felsinotherium), and reference to Eocene
forms is unlikely. In addition, the tooth, which lacks any sign of abra-
sion, cannot be logically reworked from older strata. In the analysis of
this tooth, certain developmental changes in sirenian teeth should be
remembered. Fossil sirenian milk molars are rare, the usual finds being
adult cheek teeth that usually differ from their lacteal predecessors. Be-
cause there are few milk teeth of comparable development with which
to compare Indosiren, its significance remains unknown. With a single
tooth a number of questions remain unanswered: Would the perma-
nent adult tooth differ from the lacteal counterpart, or would it bear the
same cusp occlusal pattern? In a number of fossil Dugongidae a greater
number of accessory cusps appear in successive molars in the same
individual as it grows older. The question may be asked whether an
adult molar of Indosiren would be comparable to other genera of similar
geologic age or a distinct entity, as the milk molar appears to be? More
specimens are needed before the correct taxonomic assessment of Indo-
siren can be made.
Measurements of the upper left molar of Indosiren javanense are (in
mm): length, 14.6; maximum width, 13.5 and height of entire loph, 10.6.

The only other figured sirenian from this general sector of the world
is a parietal-supraoccipital complex (De Vis 1883), which bears the pro-
posed name of Chronozoon australe. This specimen is from the Chin-
chilla Drift, Darling Downs, New South Wales, in the southeast sector of
Australia. The age of the element is not known, but it may be Pliocene.
Associated land and freshwater vertebrates indicate a freshwater deposit.
Based on the illustration, there is some suggestion of similarity of this
specimen to comparable elements of Metaxytherium serresi and Hali-
therium antiquum (Leidy), the latter of which was later referred by
Simpson (1932: 445) to Halitherium alleni and assigned possibly to
Metaxytherium or Felsinotherium by Kellogg (1966: 91). Other than
this general observation, there is no indication of the relationship of this
form to other genera. Because diagnostic qualities of a new genus are
lacking in this specimen, the name is of little value other than to serve


PL A.c.
M L P .

E 28.-ndosiren aanese: occlusal view of upper left milk molar.
i ,.


FxcmiR 28.-Indosiren javanense: occiusal view of upper left milk molar.

as a reminder of the presence of a sirenian from this isolated geographic

The recent specimen of a young Dugong (FMNH 68781) used for
purposes of comparison reveals an interesting anomaly in the symphysial
portion of the mandible. Four alveoli for three incisors and a canine
are present on the left side, but five alveoli are found on the right. If


"_ _. MILK


FIGURE 29.-Dugong (FMNH 68781): premaxillary incisor tusks in young Dugong.

the three ventral alveoli on the right are ascribed to the incisors, the
upper two would then have contained a canine and a premolar. To
attribute the dorsal alveolus to the premolar would be odd indeed, for
this would necessitate separation of an anterior premolar from the cheek
teeth by an elongate diastema. Separation of canine and premolar is a
condition found in many animals and this is possibly the situation in
FMNH 68781. If this postulate should be accepted, however, some
explanation must be given to account for an extra incisor. Holmes and
Harrison (1892: 790) may possibly have found a Dugong with five
alveoli on either side of the mandibular symphysis, but their explanation
is brief, vague, confusing, and lacks illustrations. Since all mandibular
alveoli, except 12 on the right side, are shallow and essentially filled with a
cancellous bony growth, presumably there was but one vestigial incisor
that in life was undoubtedly covered by a horny pad. The alveolus of 12
probably held a milk tooth that was absorbed during early adulthood.
The alveolus for 12 is elongate (14.5 mm), cylindrical (7.5 mm diameter),
bears a round posterior termination, and is slightly expanded anteriorly.
Several other accounts mention vestigial incisors in Dugong. Home
(1820: 153, pl. 14) stated that in a young female the third alveolus on
each side of the jaw bore a small tooth. Owen (1840-45: 366) and later
Lyman (1939: 229) noted an adult male Dugong in which the third
alveolus in the left mandible bore a vestigial tooth. Kiikenthal (1897:
68) reported four teeth-bearing alveoli in each mandible of an embryo.



FIGURE 30.-Dugong (FMNH 68781): mandibular incisor anomaly.

From these few cases one would surmise that in Dugong there is in
progress an irregular reduction of incisors in which I is generally the
most stable and the last to disappear. This condition of I, stability was
pointed out by Heuvelmans (1941: pl. 4), who stated that the last rudi-
ments of the labial teeth are always found in the third alveolus. FMNH
68781 differs from what might be the usual condition by the presence of


the right 12 alveolus that is believed to have contained a milk incisor.
The primitive condition of a vestigial incisor can occasionally be expected
in the young of an animal species in an evolutionary stage intermediate
from one in which the adults had functional incisors and one in which
the incisors are replaced by a horny pad for mastication. The uneven
number of five alveoli, however, is an oddity not attributable to this
general evolutionary trend but rather to an individual anomaly.
In the cranium of the same individual the maxillaries reveal two sets
of incisors, a condition described by a number of authors in some detail.
The milk incisors are smaller, anterior, and are firmly attached in their
sockets. The larger posterior permanent incisors are short (27 mm),
stout, roughly triangular, longitudinally striated, incipiently cuspid, and
lie loose in their alveoli (sockets). The layer of enamel and dentine is
very thin, and the teeth are hollow, with the vacuity extending through-
out the tooth. This condition differs from that reported by Heuvelmans
(1941: 2, fig. 1), in which the internal vacuity is short and the dentine
well developed. It is possible that the specimen at hand was a young
female, in which the permanent incisors would not have become well
developed. The posterior termination of the permanent incisors lies in
a broadly swollen recess of the premaxillaries. The swollen sockets are
thin and open, thus revealing internally the cancellous bony tissue
through which the permanent incisors were innervated. A thin trans-
verse bony plate separates the anterior and posterior incisors. In con-
trast to the larger permanent incisors, the lacteal counterparts are strongly
rooted, cylindrical, more elongate, lack striations and enamel, and bear
no signs of cusps. The etched condition of the anterior incisors appears
not to be the result of use, but rather of physiological resorption. The
usual number of five cheek teeth are present in both upper and lower
jaws of this specimen.

REFERRED SPECIMEN.-Upper left M3. NUC 4301, Department of Paleontology,
Physical and Natural Sciences of the National University of Cordoba, Argentina.
AGE.-Collected immediately below strata of Mesopotamian age. Upper Para-
namian age, equivalent to Middle or Upper Miocene.
LOCALITY.-Cliffs along Parana River, Villa Urquiza, Province of Entre Rios,
Argentina. Stratigraphic marine sequence provided by Dr. Florencio G. Acefiolaza.
Parana Formation ( =Paranense+Entrerriense.1 Frenguelli 1920-1947).
The Tertiary marine section outcropping in the Villa Urquiza region is composed
of two members.
Overlying: "Mesopotamiense-Rionegrense2 fluvial" (Frenguelli 1920-1947).
-2 See Pascual, R., et al. 1965, p. 174, table 11.


FIGURE 31.-Metaxytherium sp. indet. (NUC 4301): occlusal view, left M3.

FiGURF 32.-Metaxytherium sp. indet. (NUC 4301): (A) Buccal view; (B) Lingual









SCALE APPROX. 1:20,000

FIGURE 33.-Argentina, collecting site of Metaxytherium sp.

UPPER MEMBERS: Thickness 1%-3 m. Yellowish and whitish sand, clayey; mas-
sive in lower level; a fossiliferous bed in the upper level, contains: Ostrea patag6nica
D'Orb, Ostrea alvarezil D'Orb, Chlamys paranensis (D'Orb), and Amussium darwini-
anus (D'Orb).
The best outcrop of these members is situated in the Port Villa Urquiza area.
In the sand and clay there are abundant vertebra, fish-scales, and teeth of sela-
chians (Carcharodon, Carcharias raja, etc.).
BASAL MEMBERS: Thickness 8-10 m. Green and green-yellowish clay, with
thinly bedded, fine white-yellowish sand in the upper part.
Clay with an intercalation of whitish fossiliferous limestone (thickness 25 cm)



I I i I




FIGUBE 34.-Fossil sirenian localities in South America and the Caribbean.

outcrops in El Brete area; containing: Ostrea parassitica Gm., Morophoraster darwini
(Desor), and Sirenia indet. (tooth).
In the Las Palmas area an oyster bed 30 cm. thick, intercalated between dark
green clay, contains: Ostrea parassitica Gm. and Placunanomya (Pododesmus) papi-
racea (Phil)
BASE: "Not seen."

Although fossil sirenians achieved their greatest range and probable
zenith in Late Miocene time, it is significant that a dugongid has been


identified as far south (latitude 30 38' south) on the east coast of South
America as Argentina. A report of another dugongid (Kellogg 1966),
Metaxytherium ortegense from Colombia, further corroborates the co-
existence of dugongids and trichechids in the Caribbean, and probably
the entire coast of South America, south to the Parana River Basin during
Late Miocene-Early Pliocene time. What factors caused the demise of
the more widespread dugongids, but permitted the continued existence
of the trichechids? Except for Hydrodamalis, the most obvious limiting
factor affecting recent Sirenia is the narrow range of water temperature
that the animals can withstand. Increased death of Florida Trichechus
is notable during short periods of unusual cold weather. This fact ap-
plied to cooling marine currents of Pliocene and Pleistocene time may
have accounted for a drastic reduction in numbers, as well as an asso-
ciated change of flora due to cooler temperatures. Cooler temperatures
may also have caused a crowding of herbivorous marine mammals in a
narrow range of latitude, causing increased competition for food, as well
as encouraging predators to feed on this concentrated food supply. Both
living genera of Sirenia are basically shoreline and river-dwelling in-
habitants, but reports on the sightings of recent dugongids suggest that
they are more likely to be found in the open ocean than are trichechids.
This gives rise to the thought that dugongids would be affected more by
cooling marine currents than would their relatives. The above premises
ar interesting, but offer no explanation for the presence of the living
Dugong from the Solomon Islands in the east, to the gulfs of Suez and
Aquaba and the north coast of Kenya in the west.

sp. indet. calvertense ossivalense
NUC 4301 USNM 23409 UF 3281
Length 25.0 28.5 29.3
Width across protocone 17.7 18.5 22.9
Width across hypocone 13.9 18.5 19.0
Length M3/width M3
across protocone 1.41 1.54 1.28
Length M3/width M3
across hypocone 1.80 1.54 1.54

Stage of development of the M3 from Argentina is similar to that of
metaxytheres and the assigned felsinotheres from Middle Miocene
through Lower Pliocene of both Europe and North America. The lopho-
dont molar is an elongate oval, subequally divided by a moderately-open,
transverse median valley into a large anterior and a small posterior. The


prominent anterior cingulum is open on the buccal side, rather than
closed by an accessory cuspule, and but for this difference most closely
resembles Metaxytherium ossivalense (UF 3281) from Florida and Metax-
ytherium calvertense (USNM 23409) from Maryland. As in typical
metaxytheres, the second loph is formed from a large protocone that is
fused with the lingual portion of the anterior cingulum and the ap-
proximately equal-sized protoconule and paracone. The transverse
median valley is deep, and essentially straight transversely, being inter-
rupted by the forward displacement of the metaconule, a feature char-
acteristic of the Mio-Pliocene metaxytheres. In most specimens attri-
buted to this lineage, the anterior end of the metaconule has a more
central location than that present in the Argentina specimen; however,
the holotype of M. ossivalense, except for its larger size appears almost
identical in this character. The hypocone is confluent with the meta-
conule, and the two cusps form a single unit about twice the size of the
metacone. The posterior cingulum is formed by two small cusps, of
which the larger is lingual and unattached to the hypocone. The pos-
terior series of cusps, metaconule, hypocone, metacone, and two posterior
cusps form a small circle that encloses an open lake in this section of the
Despite its small size, the South American specimen is well developed,
and the occlusal surface shows much wear. Tooth size indicates an
inmature individual, but this cannot be substantiated without more evi-
dence. Certain specimens of Metaxytherium ( = Felsinotherium) serresi
bear a cusp and size resemblance to NUC 4301, and these are considered
to have belonged to a group in which the mature adult was small in
contrast to its larger contemporaneous relatives.
A determination that the tooth is an M3 is made on the length/width
ratio, the M3 being elongate, the M2 more quadrate.
Typically the M3 through the P4, inclusive, of North American and
European metaxytheres bear two roots on the buccal side and one on
the lingual. In the South American M3 the anterobuccal quadrant is
underlain by a root that spirals and turns in a posterolingual direction.
The posterior and smaller portion of the molar is completely underlain
by a large root that curves abruptly in an anterolingual direction near
the tip or base. In similar teeth from North America another root would
underlie the protocone and curve in a posterobuccal direction. The small
tooth size may account for the lack of a third root.
The question to be resolved is whether or not the M3 from Argentina,
which closely resemble similar teeth from North America, can be judged
a new species. Basis of determination must be reached on two criteria:


(1) minor morphological variations, and (2) spatial relationships (i.e.
geographical separation of the teeth in question).

In terms of morphological variation, the sample under consideration
obviously is inadequate. In neither case (North or South America) can
the specimens be classed as representative of the former population or
populations that existed in these present widespread geographic com-
munities. A comparison of the M3 from Argentina with molars of similar
age from the south Atlantic coast of the United States indicates minor
differences in occlusal cusp pattern, length/width ratio, and size, but
not more than what would be expected within members of a species.
This, of course, is my subjective concept inferred from the specimens at
hand, a knowledge of the order, plus information available from litera-

Based largely upon the distance separating Florida and Maryland
from the Parana River Basin, can one with any certainty assume that
two different species are involved, or are we dealing with a species hav-
ing a north-south range on the east coast of North and South America
extending from approximately 400 N to 310 S? This is not an unusual
range for certain living marine mammals, such as the larger whales, but
it is a moderately broad range for a near-shore form. Among living
Sirenia in the Atlantic region, Trichechus inunguis occurs from the Ama-
zon River and its tributaries and Trichechus manatus from the Antilles,
the Gulf Coast, and Mexico. The latter species is divided into the fol-
lowing subspecies by various authors: T. manatus manatus from the
Antilles, and T. manatus latirostris and T. manatus americanus from the
continental gulf coast. I do not know if the latter names are now re-
garded as synonymous nor the basis for their separation into subspecies.
Trichechus senegalensis occurs along the West African coast and asso-
ciated major rivers. Though the supposed diversity of living trichechids
in a relatively small geographic area suggests the possibility that the
Late Miocene dugongids in question were separate species, this cannot
be proved from the specimens. Finally, one must consider continental
drift and the possibility that South American and North American
Sirenia are now separated by greater distances than in the past.

My conclusion is to withhold description of a new species until other
specimens are found that will permit a clearer determination to be made
than is now possible. Should the small tooth represent an adult rather
than an immature individual, a new species would be justified. Because
the M3 is in the general metaxythere lineage it is referable to Metaxy-
therium sp.


In 1883 Florentino Ameghino established a new genus, Ribodon,
based on several trichechid teeth from the cliffs along the Parana River,
Argentina, and referred the specimens to the Order Perissodactyla.
Since that time the taxonomic reference to these fossils, as well as the
sedimentary strata from which they were collected, has been the subject
of much discussion.
The earliest investigators of the fossils in the cliffs along the Parana
River were D'Orbigny (1842) and Darwin, neither of whom did much to
differentiate between the sedimentary units composing the cliffs. Later,
Ameghino, in his 1883 study of the Pampean Formation, attributed the
major portion of this sedimentary unit to a Pliocene age and suggested
that the lower Pampean sediments that formed the cliffs along the Parana
River would be of an older age and therefore at least Miocene. Doering
(1881), however, established three stratigraphic horizons within the
formation, a lower marine that he correlated with the Eocene, a middle
unit of fluvial terrestrial origin placed in the Lower Oligocene, and an
upper marine that corresponded with the upper Mesopotamian strata
from which the molars of Ribodon were collected. Ameghino (1906)
placed the Mesopotamian in the Upper Oligocene. Here the subject of
stratigraphic equivalency rested until Frenguelli (1920a,b), on the basis
of (a) ichthyological and (b) invertebrate evidence, established the
area from which the fossils were collected as belonging to the Upper Mio-
cene, and he further corroborated the age (Frenguelli 1922) on the basis
of structural evidence, which suggested an Upper Miocene-Lower Plio-
cene age. Later, Simpson (1932), in his review of the Sirenia, synony-
mized Ribodon with the genus Trichechus, and the sediments in which it
was found to be Pleistocene. However, Pascual (1953), in a well docu-
mented review of the genus Ribodon, revalidated the genus and placed
it in a Mio-Pliocene age.
The genus Ribodon has had a history of varied taxonomic assign-
ments, which may well account for its former general lack of acceptance
as the important pre-Pleistocene trichechid it has proved to be. On the
basis of an upper molar collected in the cliffs along the Parana River,
Ribodon limbatus was proposed in 1883 by Ameghino, who placed it in
the Order Perissodactyla, Superfamily Tapiroidea. In 1885 Ameghino
also referred a lower molar and three upper molars to this taxon, and
continued to attribute it to the tapirs. In the following year Burmeister


(1886) mistakenly assigned Hyrachyus agrarius Leidy (1873), to Ribo-
don, an excusable error since the teeth of the small running rhinoceros,
Hyrachyus, bear a close resemblance to the teeth of the South American
tapirs. Ameghino (1886: 147), although accepting the taxonomic assign-
ment of Hyrachyus to Ribodon, emphasized throughout the paper the
many differences existing between the two genera, rather than the close
relationship suggested by Burmeister. By 1891 reservations as to the
relationship of Ribodon begin to appear in the writings of Ameghino.
In a letter to H. V. Ihering he stated (translation) ". . The Ribodon
known only by isolated molars could have completely different affinities
from those that I previously supposed." A short time later he stated
(Ameghino 1893: 448), "While I am speaking of the fossils of that region,
I must add that the doubts which I had concerning the real nature of
Ribodon are now at rest. The new material at hand shows Ribodon
limbatus belongs to the Sirenia, and is one of the family of Halitheri-
idae."1 The same year Lydekker (1893) also referred Ribodon to the
Sirenia, suggesting that it had close relationship with the European
Halitherium or with Prorastomus. Later, Ameghino (1906: 481) cor-
rectly removed Ribodon from the family Halitheriidae and placed it in
the Trichechidae. Little more was written on this genus until Simpson
(1932) determined that the teeth belonged to the Pleistocene Trichechus.
I (Reinhart 1951) followed Simpson in this matter, and it was not until
better specimens (a mandible and more lower and upper molars) were
described by Pascual (1953) that the generic validity of Ribodon could
be categorically substantiated.

BRAZILIAN Trichechus
Another interesting South American trichechid was reported upon
by Couto (1956). This consists of the typical cranial remains, a fused
parietal-supraoccipital element, from the Jurua River, Acre Territory, in
northwest Brazil. Though referred to Trichechus manatus, a Pleistocene
and Recent species of North and South America, it was collected from
river sediments that may possibly have been reworked from older beds.
Couto (1956: 110) had indicated this possibility:

"The fossil mammals from the upper Jurua and Purus rivers re-
gions are, in part, of Tertiary (Upper Miocene, Pliocene), and
in part, of Pleistocene aspect. Their stratigraphical origins are,
for the most part, doubtful (most of them were collected on the
edges or banks of the rivers, after having been transported from

'Dugongidae Gray 1821 (=Halitheridae Gill 1872).


their original sediments and rolled down river). Only the ex-
ploration of those regions by competent paleontologists can give
us safe information concerning the stratigraphical origins of the
fossils, and their relative ages, as well as the correlation of the
respective strata with other South American Cenozoic deposits
of which the geological age is more or less known."

With present information it is not possible to correlate the Cenozoic
deposits of the upper Jurua with those of the Mesopotamian of Argentina,
but it is an interesting speculation in view of the fact that Ribodon is
known to be from the Mesopotamian and the specimen referred to T.
manatus may possibly have been reworked from beds of similar age.
Potamosiren from the La Venta fauna, Upper Miocene, of Colombia,
is of a similar age range as Ribodon, and Pascual (1965) and Couto
(1956) believe it may be referable to that genus. Location of the
Brazilian specimen, approximately 1200 miles from the present ocean,
is not unusual when one considers the present dispersal of Trichechus
manatus1 in the Amazon River valley. The river and its tributaries are
readily navigated by such water dwellers today, and it is believed that
physiographic conditions for this region were similar in late Cenozoic
An earlier occurrence of a fossil sirenian was reported by Dilg (1909)
to have been found in Pliocene beds near Para (Belem), at the mouth
of the Amazon. This form was named Trachypleurotherium, but failure
to describe or figure the material has rendered the name a nomen nudum.

Because of the restriction of fossil and living Trichechidae to the
West Indies, southeastern United States, the east coast of South America,
and the west coast of Africa, it is of unusual interest to note that Macaro-
vici and Oescu (1941) have referred remains of sirenians from Bessarabia
(southern Russia) to Manatus maeoticus Eichwald 1853. The specimens
consist of 15 ribs of adults and 8 ribs attributed to young individuals.
From the characteristic bulbous shape of the ribs, one would readily
agree to the sirenian assignment, but whether they belong to the Tri-
chechidae or to the Dugongidae I cannot ascertain. Identifying the
ribs to genus and species is impossible on the basis of morphologic
features. The authors gave specific assignment to the ribs because they
were found in the limestones of Chisinau, where Eichwald (1853) found
the bones he referred to Manatus maeoticus. This reference, while

1 Often referred to as T. inunguis.


specifically and probably generically incorrect, has the advantage of
keeping the specific name consistent for all indeterminate sirenian speci-
mens found at a given locality. Following Eichwald, Nordmann (1860:
330-333) described and referred to M. maeoticus ribs, two scapulae, and
26 vertebrae from the stone quarries of Kischinew (now Kishinev [Chi-
sinau when under Roumanian rule], capital of the Republic of Moldavia,
Russia, at the northwest end of the Black Sea). Unfortunately Nord-
mann classified the sirenians as belonging to the whales. Since 1771,
when Brunnich proposed the genus Manatus (which was preceded by
Trichechus Linnaeus 1758 nee 1766), the generic name has continued to
be widely but incorrectly used in reference to both recent and fossil
forms. Hemming (1952) recently presented a well documented report
on the nomenclatorial usage of Manatus, both as a generic and trivial
name. His recommendations are as follows:
(1) place the generic name Manatus Brunnich, 1771, an objec-
tive junior synonym of the name Trichechus Linnaeus, 1758
(already placed on the Official List of Generic Names in
Zoology by the decision taken in Opinion 112), on the Offi-
cial Index of Rejected and Invalid Generic Names in Zo-
(2) place the trivial name manatus Linnaeus, 1758 (as pub-
lished in the binominal combination Trichechus manatus)
(trivial name of the type species of Trichechus Linnaeus,
1758) on the Official List of Specific Trivial Names in Zo-
I have been unable to find any reference to the geologic age of the
referred collecting sites. In the general nearby region of Europe the
most common genera have been Metaxytherium and Halitherium, and
the postcranial materials from this area should be compared to these
genera. The geographic area strongly suggests that the specimens be-
long to the family Dugongidae rather than to the Trichechidae, and quite
possibly to one of the two genera mentioned above.

Cryptomastodon FROM JAVA

Vertebrate paleontologists have discussed on numerous occasions the
possible relationship of Cryptomastodon and Desmostylus. I believe
Cryptomastodon is a proboscidean, the category in which it is classified
at the present time, and thus Desmostylus would have no closer relation-
ship to this genus than to any other proboscidean. The rare remains of


Cryptomastodon consist of teeth fragments, of which one tooth is almost
complete, and a questionable distal end of a humerus. A series of short
columns inclined toward each other at the occlusal surface give the tooth
a superficial resemblance to Desmostylus; this characteristic being shared
with many proboscidean teeth, particularly fragmented columns. The
individual columns differ from those of the desmostylids in (1) being
broader at the base, (2) having a less circular outline, (3) being ap-
pressed less closely, (4) having more cement between the columns, and
(5) lacking a small upraised boss in the center of the occlusal surface in
unworn columns. The thickness of the enamel may offer some compari-
son, as would the outline of a perfect tooth, but the former aspect of the
Java specimens is unknown to me and the latter remains to be found.
Comparisons between Cryptomastodon and desomostylids can be made
using the works of Stehlin (1925), Es (1931), Koenigswald (1933), and
Dietrich (1934).

A 1961 report by Drewes et al. on Unalaska Island, in the Aleutian
chain, contains the interesting report on the presence of Cornwallius'
from the Unalaska Formation, Lower Miocene. Inasmuch as this area is
at the geographic center of the desmostylid range from Japan to the west
coast of North America, this find is not unexpected. The known range
of latitude for this genus during the Lower Miocene is therefore from
530 40' north (at Unalaska) to 250 20' north (Baja California site).
Using this broad geographic range as a base, a number of points of in-
formation on the life habits and taxonomy of this genus can be surmised.
If one assumes that a single species of Cornwallius lived approximately
concurrently at both localities, and climatic conditions were similar to
those at present, a eurythermal physiological condition might have been
a necessity. General climatic conditions in the Miocene are believed to
have been more cosmopolitan than at present, however; a fact that is
indicated by widespread seas and many far-ranging marine vertebrates
and invertebrates. If this were the case, ocean currents may have
caused similar climatic conditions in the broad swampy bays and estuar-
ies in the widely separated regions. A single species would have a wide
range of latitude and even a similar diet. It is equally possible that
within the two regions a southern and a northern species are represented,
although scarce dental remains and a complete lack of cranial and post-
cranial bones give no such clue.

'Identified by Dr. G. E. Lewis (Drewes et al. 1961: 606). Lewis further states that identification
as Desmostylus is possible, however, the manuscript indicates a preference for Cornwallius.

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs