F IOr n,, D e -t. o P e ,on e Y-r'i .
STATE OF FLORIDA
DEPARTMENT OF CONSERVATION 2r &
8. E. RICE, SUPERVISOR OF CONSERVATION
Herman Gunter, Director, Geological Survey
GEOLOGICAL BULLETIN No. 22
CONTRIBUTIONS TO FLORIDA
A FOSSIL SQUIRREL-FISH FROM
THE UPPER EOCENE OF FLORIDA
0. Miles Conrad
The American Museum of Natural History
New York, N. Y.
THE ROSTRUM OF
Joseph T. Gregory
Bureau of Economic Geology
The University of Texas
THE STATE GEOLOGICAL SURVEY
PUBLISHED SEPTEMBER 10, 1941
LETTER OF TRANSMITTAL
HONORABLE S. E. RICE
Supervisor of Conservation
I have the honor to transmit a report entitled Contributions to
Florida Vertebrate Paleontology for publication as Geological
This bulletin contains two papers, one by G. Miles Conrad of
the American Museum of Natural History entitled, "A Fossil
Squirrel-Fish From The Upper Eocene Of Florida," and one by
Joseph T. Gregory of the Bureau of Economic Geology of Texas
entitled, "The Rostrum Of Felsinotherium Ossivalense." These
are valuable contributions to the vertebrate paleontology of the
State and were generously contributed at no cost to the Geological
Survey by th? above mentioned institutions.
I want to take this opportunity of expressing my appreciation
of the interest you have shown in the Geological Survey and the
support you have given it.
HERMAN GUNTER, Director
June 17, 1941
A FOSSIL SQUIRREL-FISH FROM
THE UPPER EOCENE OF FLORIDA
G. MILES CONRAD
AMERICAN MUSEUM OF NATURAL HISTORY
Introduction ._----- .__..--
Generic and specific determination -_
Comparison with Holocentrus. _--------
Skull -_ -______--_ _
Pectoral girdle---- --- __________
Vertebral column -___----__
Fins -----. ....--_______
Anal fin -_ -
Phylogenetic significance of Holocentrites.
Literature cited --- _
--- -- ---. 9
--- -- ... 11
---- -- -- - 1 1
-- -- 14
Plate 1. Left side view of Holocentrites ovalis, showing
vertebral column and skull remains--___ -- 21
Plate II. Right side view of Holocentrites ovalis, show-
ing scales, anal fin spines and caudal fin rays ---- 22
Plate III. A. Comparison of fronto-occipital regions of
Holocentrus ascensionis, left; and Holocentrites
ovalis, right. B. Outer view of the right oper-
cula of Holocentrus, left; and Holocentrites,
right. C. Inner view of the right opercula of
Holocentrites, left; and Holocentrus, right ._-----__ 25
Figure I. A Phylogenetic History of the Holocentridae ------ 16
A FOSSIL SQUInnrt -rzoii rron r
THE UPPER EOCENE OF FLORIDA'
G. MILES CONRAD
Early in 1939 J. Clarence Simpson, Field and Museum Assist-
ant, Florida Geological Survey, found a fossil fish in the Florida
State Caverns situated two miles north of Marianna, in Jackson
County, Florida. The block of Ocala limestone in which the fos-
sil was embedded was carefully removed and forwarded by Her-
man Gunter, State Geologist of Florida, to the American Museum
of Natural History in New York for study and preparation.
When finally worked out of its fine-grained limestone matrix
the fossil proved to be almost complete and lacked only the an-
terior portion of the skull (Plates I and II). The vertebral col-
umn, which is almost complete, measures about 180 millimeters
in length. From the basioccipital to the most anterior portion of
the neurocranium which has been preserved the measurement is
about 40 millimeters. In total length, from the tip of the snout to
the most posterior part of the caudal fin the living fish probably
measured about 270 millimeters, or 10Y2 inches. In greatest depth
of body it is relatively shallow for a berycoid (of which group it
is a member), for it is only about 70 millimeters, or 25 % of the
length. In general body-form it is indeed fusiform and carries
little hint or the frequent berycoid tendency toward deep bodies
(cf. Hoplopteryx, Beryx).
The occipital region is the best preserved of all the skull ele-
ments represented (Plates I and III, Figure A). In dorsal aspect
the posterior half of the frontals, the parietals and the supraocci-
pital are present. The pterotic and sphenotic spines are much in
evidence as are the posttemporals and a large, oval scale bone on
the left side. Of the opercular series only a large fragment of the
left operculum is present; the right operculum has been removed
from the matrix, intact except for its spine (Plate III, Figure B
and C). The primary upper jaw is represented by the hyomandi-
bulars, the left metapterygoid, the right entopterygoid, and possi-
bly the left pterygoid (or palatine). On the left lateral aspect is
noted a portion of the maxilla and plastered upon it is the second
1Manuscript received December 29, 1939.
10 FLORIDA GEOLOGICAL SURVEY-BULLETIN 22
suborbital. Noticeable in front view and from the sides are the
alisphenoids, while on the right side the parasphenoid is in evi-
In the preparation of the fossil the squamation has been left
on the right flank (Plate II) while parts of the left side have been
removed in order to expose the almost perfect vertebral column.
Of the twenty-six vertebrae originally present, only two are com-
For permission to describe this fossil I am indebted to Mr. Gun-
ter and to Dr. William K. Gregory of the American Museum, to
whom it was originally sent. I also wish to extend thanks to Mr.
Edward Hawkins, Jr., who so carefully cleared the matrix from
GENERIC AND SPECIFIC DETERMINATION
HOLOCENTRITES,1 NEW GENUS
Genoholotype-Holocentrites ovalis, new species.
Distribution-Upper Eocene, Florida.
Diagnosis-Relatively fusiform holocentrid, the greatest body
depth about one-fourth of the total length. A single heavy mesial
ridge on each frontal bone, separated by a groove from a lateral
series of smaller, less pronounced ridges. Robust, thickened supra-
occipital crest; not produced posteriorly. Large. oval scale-bone.
Vertebral formula: 11+14+1=26. Parapophyses start on seventh
vertebra; more delicate than those in Holocentrus. Anterior uro-
neural united to twenty-sixth vertebra by suture only. Centra are
shorter than they are deep. Scales pectinate and ridged.
Holocentrites ovalis,2 new species.
Holotype-Florida Geological Survey, No. V-5453, Collected by
J. Clarence Simpson, 1939.
Occurrence and type locality-Ocala limestone. Upper Eocene.
Two miles north of Marianna, Jackson County, Florida at the
Florida State Caverns.
Diagnosis-Sole known species of the genus.
'Holocentr-um+ sts like.
2ovalis, oval; in allusion to the large, oval scale-bone.
CONTRIBUTIONS TO FLORIDA VERTEBRATE PALEONTOLOGY 11
COMPARISON WITH HOLOCENTRUS
Neurocranium-Comparison with a rather large series of acan-
thopterygians revealed that the fossil, Holocentrites, most nearly
resemolea berycoids of the family Holocentridae. Unfortunately,
however, many ot those osteological characters which define both
the order berycoidea and the family Holocentridae are missing
rrom the fossil specimen. The distinctive feature of Holocentrites
which led to its identification was the appearance of the fronto-
occipital region. As seen from above latee III, Figure A) the
frontals of Holocentrites and Holocentrus are prominently ridged
and grooved. The mesially placed ridges are hypertrophied in
Hoiocentrites, although they are rather prominent in Holocentrus.
The frontals of the fossil shelve over the parietals and supraocci-
pital in such a manner as to leave a slight undercut. This "shelv-
ing" effect is typical of the holocentrids. Surmounting the pterotic
spine is a scale bone. In Holocentrites it is large, oval and with
serrated edges. A scale bone is present in Holocentrus and Myri-
pristis, but it is proportionately smaller in both than in the fossil.
The supraoccipital crest is similar in extent in both Holocentrites
and Holocentrus, but it is considerably thicker and more robust in
the former. In the extent of the supraoccipital both of these genera
differ from the Cretaceous Caproberyx in which the crest extends
almost as far posteriorly as does the operculum. In recent holocen-
trids when the skull is in the normal position the vertical plane of
the occiput is more or less perpendicular to a line drawn from the
tip of the snout to the pygidion, that is, the occiput does not ex-
tend behind the supraoccipital. Although the jaw remains have
been crushed to the left, the fronto-occipital region of the fossil
does not seem to have been much affected. In Holocentrites the oc-
ciput meets the "snout tip-pygid'on" line at an angle of about
75 or 80 degrees.
The circumorbital series of Holocentrites is represented by only
two fragments which are firmly plastered over the left maxilla.
Their ventral edges are serrate as in most berycoids, recent and fos-
sil, while the dorsal edge has been inflected to form, with the other
suborbitals, a shelf. There is a slight overhang on the lateral dorsal
side of this suborbital, presumably for the circumorbital canal.
From the expanded size and disposition of these two fragments
they may be defined as parts of the second suborbital.
Preserved in the fossil is the posterior portion of the parasphe-
noid with its ascending wings which contact the alisphenoids. As
in other holocentrids the lateral flanges of the parasphenoid are
quite wide. The basisphenoid has not been preserved in Holocen-
Branchiocranium-The operculum of the holocentrids is note-
12 FLORIDA GEOLOGICAL SURVEY-BULLETIN 22
worthy for its spiny character, but in Holocentrites the posterior
edge or this bone has not been preserved. In the Holocentridae the
anterior border of the operculum (Plate III, Figure B) has, run-
ning its entire length, a smooth band of bone about a quarter of
an inch wide. Posterior to this smooth border the bone is marked
by numerous horizontally-arranged striations. The mesial surface
of the right operculum of the fossil (Plate III, Figure C) has been
preserved and comparison with that of Holocentrus reveals inter-
esting similarities. The socket which receives the hyomandibular
process is Y-shaped in both. The lower arm provides a seat for the
hyomandibular tacet while the upper arm is braced more dorsally
against the hyomandibular pedicle. In both a foramen tunnels un-
der the leg of the Y. There are no striations on the mesial surface
of the operculum of either Holocentrites or Holocentrus.
The preserved remnants of the primary upper jaw are quite
fragmentary, but insofar as they may be compared they are not
unlike their homologues in other holocentrids.
The posttemporals of both sides are well preserved in the fos-
sil. Posteriorly they are expanded into an oval-shaped plate with
sharply serrated edges as in Holocentrus and Myripristis.
The remainder of the pectoral girdle which is preserved has
much of the holocentrid character.
The vertebral formula of Holocentrus and Myripristis, accord-
ing to Starks (1904), is: abdominals 11+ caudals 15+ hypural
=27. However, Regan (1911) indicates that the Holocentridae
have either 26 or 2D vertebrae. The specimen of Holocentrus with
which Holocentrites was compared had 26 vertebrae. Holocentrites
has a vertebral formula of 11+14+1=26, and thus it falls within
the definition of the family. Although dissimilar in skull details
the fossil resembles Monocentr s japonicus in vertebral count. Ac-
cording to Regan the Trachichthyidae have 26 vertebrae, but he
makes no mention of their disposition into abdominals, caudals,
The vertebral column (Plate I) satisfies Regan's general defi-
nition for the Berycomorphi in that it is composed of soldd centra
which are co-ossified with the arches; anterior ribs sessile: posterior
ribs on parapophyses; hypurals more or less fused and expand-
In Holocentrus a parapophysis is first present on the sixth ver-
tebra, whereas in Holocentrites it appears first on the seventh. It is
long, thin and pointed in the latter, but broad and squat in Holo-
centrus. It carries a pleural rib in Holocentrus, but this has appar-
ently been lost in Holocentrites. In general the remaining para-
CONTRIBUTIONS TO FLORIDA VERTEBRATE PALEONTOLOGY 13
pophyses seem more robust in Holocentrus. The parapophyses are
first connected to form a haemal arch on the ninth vertebra of
Holocentrus, whereas on the fossil such a connection is first defi-
nitely preserved on the eleventh. In both forms, however, the
parapophyses are completely fused at their tips to form a haemal
spine on the thirteenth vertebra.
The posterior abdominals and the caudals are marked by two
lateral fossae which seem to be more deeply incised in Holocentrites
than in Holocentrus. The bar which separates these two pits is
thicker in the former than in Holocentrus. In the Holocentrus
skeleton with which this comparison was made the antero-posteri-
or dimension of the centra was greater than the dorso-ventral,
while the centra of Holocentrites seem to be shorter than they are
deep. However, the figure of the Holocentrus skeleton given by
Agassiz (1831) has vertebrae whose centra are proportionately
much as those of Holocentrites. Such proportional differences are
probably only of specific rank.
Spread throughout the matrix in the region of the abdominal
vertebrae of Ho!ocentrites were numerous epipleural ribs. Unfor-
tunately all of them were very fragile and broke at the least touch,
so that nothing was learned of their insertion. Holocentrus is well
provided with epioleurals, some of which are attached to the cen-
tra and others to the neural ribs. Pleural ribs are preserved on the
third, fourth and fifth vertebrae of Holocentrites. In the lack of
pleurals on the first and second vertebrae Holocentrites agrees with
Holocentrus in which the first two vertebrae bear only epipleurals.
In Holocentrus the neural spine is expanded into a thin lamina
rear its proximal end which extends the length of the centrum.
The neural spines of Holocentrites are much less laminate in the
anterior abdominals and much more delicate in the posterior ab-
The anterior uroneural of the hypural complex in Holocentrus
is fused to the last centrum, whereas in Holocentrites this element
is united with the centrum by suture only. The tail segment (those
vertebrae which enter into the support of the caudal fin) of both
Holocentrites and Holocentrus apparently includes the last two ver-
tebrae plus the hypural. The neural and haemal spines of Holocen-
trus are expanded on vertebrae 23, 24. and 25, but the spines of
23 stand free of the caudal fin rays. This is also the condition in
Holocentrites, although the neural spine of the vertebra 23 is not
Except for a few isolated rays neither the dorsal nor the pectoral
fins are present. The pelvic fins are thoracic in position a-d are
represented by one spine each with fragments of a number of
14 FLORIDA GEOLOGICAL SURVEY-BULLETIN 22
Anal Fin-One of the family characters of the Holocentridae
is the presence of four anal spines, of which series the third is the
most prominent. Holocentrites, too, possesses four anal spines. Al-
though the photograph (Plate II) shows only the second and
third of the series, the others are present. These spines in Holocen-
trus are ornamented by very fine striations which anastomose much
as do the friction ridges of the human finger-tips. In Holocentrites,
however, the striations are very marked and do not anastomose,
but retain their identity in an easy spiral around the body of the
According to Cockerell (1912) the scales of the Holocentridae
"resemble in shape those of the Berycidae, being broad, with prom-
inent laterobasal angles . . the apical margin is armed with
strong, straight, comb-like teeth .... The basal margin is straight
or nearly so, except that in the middle it is thrown into one or
more folds or lobes, indicative of the rudiments of a basal radial
system. The scales are broader than long . ."
In a very general way the scales of Holocentrites and other fos-
sil genera fall into the family scale definition. However, their most
obvious similarity among themselves and with the modern genera
is in the condition of their apical margins which are always "armed
with strong, straight, comb-like teeth." However, the basal mar-
gin in Holocentrites is more curved than straight and shows no
evidence of the "rudiments of a basal radial system" noted by
Cockerell among recent genera. Hoplopterqx, a berycid, shows
these basal ridges, but Homonotus, assigned by Regan to the Holo-
centridae, seems to lack them. In all the fossil genera, including
the Miocene Myripristis meliteensis the apical teeth are merely the
free termini of strong longitudinally-placed ridges. This is also
true of the recent species Holocentrus ascensionis and Myripristis
multiradiatus, although the ridges are not as pronounced.
The scales of Holocentrites are considerably larger in absolute
size than those of a Holocentrus of comparable dimensions.
PHYLOGENETIC SIGNIFICANCE OF HOLOCENTRITES
Throughout the Cretaceous of Europe and North America the
Berycoidea were well represented by the Bervcidae, such as Hov-
lopteryx and Berycorsis. Contemnorarv with this primitive family
was a genus found in England. Caoroberux. which Regan rightly
places in the Holocentridae. In point of time, Canroherlx can
claim no greater primitiveness than other Cretaceous holocentrids,
but in its skull structure it seems to be nearer to the bottom of the
direct line leading to the modern Penera than they are. Its occinito-
frontal region resembles tbht of Holocentr, or Murinristis, more
nearlv than it does that of Holocortri;ts. However, it departs from
all three of these latter genera in that its supraoccipital crest is
CONTRIBUTIONS TO FLORIDA VERTEBRATE PALEONTOLOGY 15
much extended posteriorly. The lack of a preopercular spine in
Caproberyx suggests affinity to Myripristis.
Trachichthyoides ornatus, another Cretaceous berycomorph,
seems to be referable to the Holocentridae. The mesial frontal
ridges are strengthened as in Holocentrites, but they actually meet
at their posterior ends. The lateral group of ridges fan out pos-
teriorly in Holocentrites, but in Trachichthyoides they come to
a point (Woodward, 1902, Plate VIII, Fig. 5 A). The operculum
of the latter seems more expanded antero-posteriorly and there
does not seem to be the same distinction between the striated and
smooth portions as in Holocentrites, or the recent genera. The
clavicle of Trachichthyoides seems also to be more expanded an-
tero-posteriorly than in Holocentrites. In its emphasis of the me-
sial frontal ridges Trachichthyoides resembles Holocentrites more
than it does either Caproberyx or Holocentrus.
Homonotus, an extremely small form from the Cretaceous of
England, has been assigned to the Holocentridae. The material, as
figured by Woodward, is rather imperfect and difficult to compare
with Holocentrites. One salient feature of its anatomy is its lack
of a preopercular spine.
Dinopteryx spinosus, provisionally assigned by Regan to the
Holocentridae, is also notable for its lack of a preopercular spine.
This consistent lack of preopercular spine among the Cretaceous
holocentrids seems to indicate that its presence is secondary. The
implication being that Holocentrus, with one such spine, is more
advanced phylogenetically than Myripristis. Unfortunately, how-
ever, Holocentrites lacks any indication of the condition of its
preoperculum, so that no clue is here given to its phylogenetic
The North American Cretaceous representative of the Holo-
centridae is Kansius. from the Niobrara of Kansas. Kansius is a
small Holocentrus-like fish. The enlarged anal spine is characterized
by an "ornamentation of low, anastomosing, longitudinal wrin-
kling or ridging" (Hussakof, 1929). In this character Kansius
resembles Holocentrus rather than Holocentrites in which the
ridging is prominent and does not anastomose. In discussing the
anal spines of Kansius, Hussakof wrote as follows:
"The greatly enlarged anal fin-spine calls for special com-
ment. This character occurs in only a few families of teleosts,
such as the Holocentridae (Sauirrel-fishes), Haemulidae
(Grunts), and the Llltianidae (Snappers). It seems to be
best developed in the Holocentridae. It is remarkable that a
seemingly unimportant character like this should persist for
such a great length of time-from the Cretaceous up to the
present day. It is even more remarkable that its ornamenta-
tion of very fine longitudinal wrinklings, which are hardly
visible to the unaided eye, and for which we can conceive no
16 FLORIDA GEOLOGICAL SURVEY-BULLETIN 22
function of service to the fish, should likewise have persisted
during this vast period of time, unaffected by evolutional
It seems evident from these observations that Holocentrites
early broke away from the main phylogenetic stem in this char-
acter and attempted a more novel anal spine. Fluting similar to
that on the anal spines of Holocentrites is to be noted on the dorsal
and pelvic spines of Myripristis melitensis as described by Bas-
Unfortunately the skulls of the two cotypes of Kansius are
much fragmented and were not studied by Hussakof, so that its
/HoOCENTrr IU /)s Mu fer/s
) -^ fOuTICHTHYy
H/L X^A HoT'o
I-/cOCSVT ITF HZ L Z-- "
Figure I. A Phylogenetic History of the Holocentridae.
CONTRIBUTIONS TO FLORIDA VERTEBRATE PALEONTOLOGY 17
phylogenetic position is determined largely by fin counts and posi-
tions. It seems rather close to the main holocentrid stem, although
the specific diagnosis ot five, instead of tour anal spines places it
outside of the family as defined. Of the other Cretaceous forms,
Caproberyx and Homonotus are definitely stated to have only four
anal spines. Hussakof points out that in one cotype three spines are
preserved and in the other there were four "with a strong proba-
bility that there were five, as in the existing Holocentrus." How-
ever, all species of H.olocentrus seem to have only four such spines,
so that a corrected definition of Kansius sternbergi might read:
"Anal IV, and about 8."
By the upper Eocene the modern genera Myripristis and Holo-
centrus had become established in Europe. From the Oligocene of
Rumania, Pauca has recently (1931) described a new berycid,
Hotocentroides, which may pertain to this family.
An analysis of comparable items in the various fossil genera
seems to indicate that tne cocene Holocentrites and the Cretaceous
Trachichthyoides are on an independent line of holocentrid devel-
opment which has stressed anisomeristic development of the
frontal ridges, while the main line leading toward the recent gen-
era is characterized by a polyisomerism or the frontal ridges. At
no stage, however, are any or the holocentrid genera very remote
from each other, for the family is extremely compact and shows
little adaptive radiation of body-form or essential anatomy. Nor
is this devotion to a basic pattern confined to this particular fam-
ily, for in writing of berycomorph scales, Cockerell says: "The
Berycoidea . . form a sufficiently compact group, quite isolated
from the groups standing on either side of it in the system."
The recent "genera" of Holocentridae are in most cases merely
sub-genera of either Holocentrus or Myripristis. The more primi-
tive Myripristis has given rise to Ostichthys and Holotrachys. The
first being characterized and distinguished from Myripristis by its
rough scales. Holotrachys differs from both in its very much
smaller scales; for example, whereas Ostichthys has 29 scales along
the lateral line; Holotrachys has from 40 to 45. These have been
treated by authors both as genera and sub-genera. The "sub-fam-
ily" represented by this group seems to be characterized by its lack
of a preopercular spine.
Holocentrus seems to have given rise to a number of recent
"genera," or, more probably, sub-genera which are characterized
in large part by the degree of spination of the preopercle. Among
this group of holocentrids may be numbered Flammeo, Sargocen-
trum, Plectrypops and Adioryx.
Figure I, while hardly representative of an exhaustive study,
suggests the phylogenetic history of the family Holocentridae.
18 FLORIDA GEOLOGICAL SURVEY-BULLETIN 22
Agassiz, J. L. R. 1833-43. "Recherches sur les Poissons Fossiles." Vol. 3 and
Bassani, F. 1914. "Sopra un Bericide del calcare Miocenico di Lecce, di Rosig-
nano Piemonte e di Malta (Myripristis melitensis A. S. Woodward sp.)."
Atti Acad. Sci. Napoli, 2 ser. no. 1, 1-14.
Cockerell, T. D. A. 1913. "Observations on fish scales." Bull. Bureau Fisheries,
1912 (1913), 32, 119-174.
Fowler, H. W. 1904. "New, little known and typical berycoid fishes." Proc.
Acad. Na Sci. Phila., 56, 222-238.
Hussakof, L. 1929. "A new teleostean fish from the Niobrara of Kansas." Amer.
Mus. Novitates, no. 357, 1-4.
Jordan, D. S. and Evermann, B. W. 1905. "The aquatic resources of the
Hawaiian Islands. I. The shore fishes of the Hawaiian Islands, with a
general account of the fish fauna." Bull. U. S. Fish Comm., 1903 (1905),
Pauca, M. 1931. "Neue fische aus dem Oligozan von Piatra-Neamt." Bull. Sec-
tion Scientifique de l'Acad. Roumaine, 14, 29-34.
Regan, C. T. 1911. "The anatomy and classification of the teleostean fishes
of the orders Berycomorphi and Xenoberyces." Ann. Mag. Nat. Hist., 8,
ser. 7, 1-9.
Starks, E. C. 1904. "The osteology of some berycoid fishes." Proc. U. S. Nat.
Mus., 27, 601-619.
Woodward, A. S. 1902. "The fossil fishes of the English Chalk." Part I,
Monograph Palaeontogr. Soc., 56, 1-56.
L 19 I
Left side view of Holocentrites ovalis, showing vertebral column and skull
[ 20 1
FLORIDA GEOLOGICAL SURVEY BULLETIN TWENTY-TWO, Plate 1
[ 21 ]
FLORIDA GEOLOGICAL SURVEY BULLETIN TWENTY-TWO, Plate 2
'~~~~~ ~~~ 22------- I ] -------
Right side view of Ho!ocentrites ovalis, showing scales, anal fn spines and
caudal fin rays.
[ 23 ]
A. Comparison of fronto-occipital regions of Holocentrus ascensionis, left;
and Holocentrites ovalis, right. B. Outer view of the right opercula of Holo-
centrus, left; and Holocentrites, right. C. Inner view of the right opercula of
Holocentrites, left; and Holocentrus, right.
[ 24 ]
FLORIDA GEOLOGICAL SURVEY BULLETIN TWENTY-TWO, Plate 3
[ 25 ]
THE ROSTRUM OF
JOSEPH T. GREGORY
BUREAU OF ECONOMIC GEOLOGY
UNIVERSITY OF TEXAS, AUSTIN, TEXAS
In July 1918, Mr. Anton Schneider, Superintendent of the
Amalgamated Phosphate Company, submitted to the writer at
Tallaiassee, Florida, three photographs of the skull of a sirenian
which he had obtained in connection with phosphate mining in
the Bone Valley formation of Polk County, Florida. Following
the receipt of photographs, Mr. Schneider was requested to send
the skull for description, which he did. A few weeks later, the
writer came to Texas, and in the ensuing years the skull, still un-
described, was lost sight of in the collections of the Bureau of
Economic Geology of the University of Texas. Recently in over-
hauling the Bureau collections, the long lost specimen was found.
In the meantime, Dr. G. G. Simpson (1932) described the fossil
Sirenia of Florida. As it was not possible at that time to locate the
skull, Simpson included in his memoir an illustration from photo-
graphs made by Schneider. Now that the skull has been found,
the description already given from the photographs can be some-
what supplemented. Accordingly, at the writer's request, Dr.
Joseph T. Gregory has described the skull as indicated in the ac-
companying manuscript. The skull is being returned to the Flor-
ida Geological Survey.
E. H. SELLARDS
[ 29 ]
Introduction __-- ___ -- ----- 33
Deftition --- ---- 33
Measurements of dentition _______-..36
Skull _--- 37
Measurements of Felsinotherium skull ______- 39
Discussion ___ -----___39
Literature cited __ _____________- 41
Plate I. Photograph of left side of front portion of
skull, x 2 ---- ----- ------ 45
Plate II. Photograph of top of rostral portion of skull, x Y2 46
Figure I. Upper molars of Felsinotherium ossivalense
Simpson, occlusal view, x 1 __ -----34
Figure II. Palatal aspect of anterior portion of skull of
Felsinotherium ossivalense, x Y2- --- 35
1 31 J
THE ROSTRUM OF
JOSEPH T. GREGORY
In the collections of the Florida Geological Survey is a well
preserved rostral portion of the skull of the little-known, extinct
dugongid, Felsinotherium ossivalense Simpson. This specimen,
number V-5454, lacks the downturned extremity of the premaxil-
laries and their included tusks, the posterior portion of the brain
case, and the zygoma, but is otherwise complete and little crushed.
It is of particular interest as it contains the most complete den-
titiop of this Pliocene sirenian thus far found in America, and
is the only known specimen of the anterior portion of the skull.
Photographs of this skull were published by Dr. G. G. Simpson
(1932, p. 452, figs. 12 A-C). Examination of the specimen it-
self reveals few features which he failed to observe on those illus-
Dr. Simpson has offered a number of valuable suggestions in
connection with this study, and has most kindly read the manu-
script. I am indebted to Dr. E. H. Sellards and to Mr. Herman
Gunter for permission to study this specimen. Assistance in prepa-
ration of illustrations has been rendered by Dr. H. B. Stenzel and
Mr. J. Wilson. This contribution has been made possible by co-
operation between the University of Texas. Bureau of Economic
Ceology, and the Federal Work Projects Admiristration, Texas
O. P. No. 665-66-3-233.
The tip of the rostrum is broken off opposite the end of the
maxillary bone. No trace of roots of the tusks nor of alveoli for
them occurs on the broken surface.
In addition to the two pairs of molars which Simoson described
from photographs, the third molar of the left side is present, not
yet fully erupted. In size and cusp arrangement it agrees fairly well
with the tvoe, though the proportions are somewhat closer to
those of F. floridanum. The length is 1.37 times the width across
the hypocone, much closer to 1.32, the proportion of F. flori-
darnm. than to 1.77, the ratio in F. ossivalense.
A large protocone connects the protoloph with the anterior
cingulum, which also connects at a lower level with the base of
1Manuscriot received May 2, 1940.
34 FLORIDA GEOLOGICAL SURVEY-BULLETIN 22
the paracone, leaving a lake in front of the protoconule. The para-
cone is slightly smaller than the protocone, and inclines inward
Figure 1. Upper molars of Felsinotherium ossioalense Simpson, occlusal view, x 1.
strongly from the outer margin of the tooth, so that it appears to
squeeze the small protoconule against the protocone. A moderate
sized metaconule lies nearly in front of the large hypocone, and,
together with the posterior salient of the protocone, partially
blocks the transverse median valley. The metacone is slightly for-
ward relative to the hypocone, as in the type of F. ossivalense, and
slopes inwardly even more than the paracone. One accessory cusp
lies behind the metacone, a second at the middle of the rear of the
tooth, lateral to the rear edge of the hypocone. A small postero-
internal pillar is present as in F. floridanum. The minor differ-
ences in cusp arrangement and proportions between these various
specimens suggest that these characters are of rather wide individ-
ual variation rather than specifically constant.
The two anterior molars have already been described by Simp-
son. On M2 a small enamel lake remains to indicate the separation
of protoloph and cingulum. There is a rather marked inflection of
the enamel at the posterior inner edge of the protocone marking
the position of a posterior salient which may have been more
strongly developed than on M3. The worn hypocone and meta-
conule are joined, as may be seen from the figures. On the right
side a slight evagination of the enamel at the posterointernal
corner of the hypocone suggests an accessory pillar like that of M3;
this is more faintly shown on the left side. A single hyDoconule
is present, with a ridge extending laterally from it to behind the
The only detail which may be observed on the greatly worn
CONTRIBUTIONS TO FLORIDA VERTEBRATE PALEONTOLOGY 35
first molars is a slight inflection of the lateral border, marking the
position of the median valley.
skull of Felsinotherium
Figure. 2. Palatal
ossivalense, x Y2
36 FLORIDA GEOLOGICAL SURVEY-BULLETIN 22
Anterior to the preserved teeth is an alveolar area 2 Y2 cm. long,
narrower anteriorly than behind. Traces of the 4 root-sockets of
Dp4 and an indistinct smaller anterior region which lodged Dp3
may be observed on the right side. Partial resorbtion of the alve-
olus suggests that the anterior of these teeth, at least, had been shed
before the death of the animal. The amount of wear on M1 prior
to the eruption of M3 is striking, whether Dp3 and Dp4 had been
shed or not. The time of eruption of the last molar appears to be
variable. The skull of F. forestii described by Capellini has M3 in
use and Dp4 and Dp3 still present, though worn. That of F. ser-
resi described t l Deperet and Roman has lost all trace of Dp3 and
has Dp4 greatly worn but present. M3 is moderately worn. Sicken-
berg (1931, p. 428) has emphasized the correlation between loss
of tooth replacement with non-union of epiphyses and other
neotenic characteristics which he attributes to a hypothyroid con-
dition in the Sirenia. Possibly this retardation of growth also re-
sults in irregular time of tooth eruption.
Simpson (1932, p. 449) mentioned the possible alternative
interpretation of the worn teeth in this skull as Dp4 and M1. The
amount of wear on the anterior tooth seems great for M1 in an
individual in which M3 has not yet erupted, whereas one would
expect Dp4 to be worn before the posterior molars were implaced.
According to this view the tooth described above as M8 would be
the second molar, and the third would still be contained in its
crypt. However, the absence of any space in which a more poste-
rior molar could form behind the last tooth present in the maxil-
lary definitely indicates that this tooth is M3. On the right side
of the skull the pterygoid process is broken away so that the short,
solid end of the maxillary can be seen behind the alveolus of M3,
leaving no doubt on this point. Moreover the anterior alveolar
area is far too large for Dop of Felsinotherium. and must also in-
clude space for Dp4. Lastly, the position of M1 opposite the base
of the zvgomatic process is in accord with the conditions in other
halitheriines, as pointed out by Simpson.
Deperet and Roman describe the molars of F. forestii as some-
what more complex than those of F. serresi. Those of F. ossiva-
lense seem close to the latter species in this respect, though they
differ in detail, and are considerably larger.
Measurements of dentition, left side.
Length M1 20+
Width M1 22+
Length M2 28.0 Length M3
Width M2 26.2 Width heel M3
Length M3 30.1
Width M3 across nrotocone 24.2 Length M3 1.24
Width M3 across heel 22.0 Width M3
CONTRIBUTIONS TO FLORIDA VERTEBRATE PALEONTOLOGY 37
'Ihe premaxiliary rostrum is broken off at the base of the ex-
panced region or spongy Done at the distal end or the palatal sur-
race. nhe maxilio-premaxillary suture passes trom the lateral to
the paiatal surfaces at the base or ridges which bound the concave
portion ot the rostrum laterally. behind this expanded region the
rostrum is markedly elongate. i'or aoout IU cm. anterior to the
alveolus of Dp6, its palatal surface is subcylindrical, convex. Com-
pared to P. torestii and P. serresi this region is much longer and
more slender in F. ossivalense, approaching Metaxytherium cuvi-
ern more closely in this respect though apparently even more
elongate than in that species. In the European specimens of Fel-
snotnenrum strong ridges arise anterior to the cheeK teeth, approach
somewhat in this diastema region, and flare out again rather ab-
ruptly to join the base or the lateral ridges of the premaxillary
rostrum. 'I these ridges are scarcely distinguishable in F. ossivalense,
in which they approach each other only slightly anterior to their
origin at the alveolus for Dp3, and then extend forward, parallel,
15 mm. apart, bounding the median palatal grooves for about
75 mm. before turning outward to merge with the edges of the
rostrum. So nearly as i can determine from published illustrations,
this specimen is rather similar to Metaxytherium cuvieri in this
respect, and it is certainly closer to it than to other Felsinotherium
The tendency toward elongation of the skull is reflected in the
structure of the palate, which is exceedingly narrow. The width
between the two first molars is 0.26 the length Dp3-M3 in F. ossi-
valense and 0.33 in a comparable specimen of F. serresi.1 The
tooth rows are less strongly arched than in either European species,
and are most widely separated at M1 instead of M2. This may be
a character of immaturity, for the second palate figured by Deperet
and Roman shows M1 most widely separated. The narrow
V-shaped inner narial notch reaches the rear of M2. In this it is
intermediate between F. serresi, in which the notch reaches the
anterior part of M2, and F. forestii in which it only reaches M3.
The palatine bones send extremely narrow processes forward to
the anterior palatine foramina between Dp3. Posteriorly the pala-
tines extend backward along the sides of the narial notch and turn
down to form the anterointernal portion of the heavy pterygoid
processes. The main portion of these processes is fused with the
alisphenoid, and bears a shallow groove on its posterior edge.
These processes overlap the posterior end of the alveolar portion
of the maxillary about 3 cm., to the edge of the alveolus for M3.
1Measurements taken from illustrations by Deperet and Roman. The specimen
may be crushed and thus give a low value. An adult skull gives a ratio of 0.45,
but Dp3 is missing.
38 FLORIDA GEOLOGICAL SURVEY-BULLETIN 22
The posterior portion of the skull has been broken off on a line
through the occipitosphenoidal suture. Simpson has described a
skull from the same beds (Fla. Geol. Surv. V-3211) in which
the basioccipital and basisphenoid are fused; this is also the condi-
tion in the type of Hesperosiren crataegensis. In the present speci-
men the end surface of the basisphenoid shows the irregular and
porous structure characteristic of open sutures. Sickenberg (p.
428) has stressed the general occurrence of a widely open suture
at this point in the Sirenia, and shown that it may be related to
other neotenic conditions in the skeleton which he attributes to
hypothyroid metabolism. Simpson's record therefore appears un-
usual; perhaps these specimens were from extremely old or ab-
Because of abrasion of the anterior edge of the premaxillaries,
one cannot determine the extent of the median boss. There is defi-
nite indication of its presence, but it may have been smaller than
in F. serresi. The angle of deflection of the rostrum appears about
the same as in that species.
The sides of the naso-frontal region have been crushed inward
somewhat, so that the narial opening appears narrower than it
was in life. The aperture was elongate-oval, probably narrowing
most gradually posteriorly. Deperet and Koman state that the
external nares of F. forestii are narrower than those of F. serresi,
but suspect that the difference may be due to crushing. F. ossiva-
lense likewise appears to be narrower than F. serresi in this region,
with some allowance for distortion. The nasal bones are exposed
on the dorsal surface. They are widely separated, and, posteriorly
at least, were separated from the nasal cavity by thin lamina of
the frontal which are broken in the specimen but not improbably
covered their medial surface.
The supraorbital processes are well preserved. Their surface is
rugose, possibly due to abrasion of the surface bone. The temporal
ridges extend forward onto the dorsal surface of these processes,
which thus appear like horn-like extensions of the deeply cleft
frontal. From the supraorbital processes the temporal ridges con-
verge posteriorly, and nearly parallel one another along the
angle between the top and side of the braincase. On the parietals
they suddenly thicken opposite the extreme posterior point of the
frontoparietal suture, the medial boundaries converging inward,
the lateral turning out. At the narrowest point the outside width
is 67 mm. A slight median boss is present on the frontal behind
the nasal opening. Slight differences in detail between this speci-
men and that figured by Simpson (p. 450, fig. 13) are probably
not significant. There is general agreement with the outline of
the ridges in F. forestii.
The upper edge of the vomer bears a slight projecting ridge on
either side which reaches the frontal and separates the respiratory
CONTRIBUTIONS TO FLORIDA VERTEBRATE PALEONTOLOGY 39
passage from the olfactory space. Above this ridge, on the sides of
the mesethmoid, are traces ot two or three pairs or turbinals. Their
details could not be determined. Evidently the entire ossified por-
tion of the mesethmoid is preserved, ror its abruptly truncate an-
terior end bears evidence of cartilaginous extension, and is not
broken. The vomer may have extended farther anteriorly. No trace
of a separate orbitosphenoid bone could be observed.
There are two foramina in the upper side of the groove which
extends forward from the large sphenoidal fissure to the orbit.
The more anterior and smaller of these is situated deep in the
groove 3 cm. anterior to the base of the pterygoid process. The
internal openings of these foramina are on the floor of the brain-
case about 1 cm. from the base of the crista galli, directly behind
the narrow fossae for the olfactory lobes. Shallow canals from
them converge posteriorly. Apparently they carried the optic nerve
and tract. A somewhat larger pair of foramina opens about 1 cm.
behind the optic, near the outer edge of the groove. The internal
openings lie at the edge of the floor of the braincase slightly an-
terior and lateral to the optic foramina.
Pachyostosis and osteosclerosis are shown by the various skull
elements, particularly the parietals, as in other dugongids.
Measurements of Felsinotherium skull.
Length, occipitosphenoidal suture-premaxillary boss 27
Length, anterior border of posterior nares to base of
flared portion of rostrum 14
Length anterior narial opening 19
Length maxillary 20
Length supraorbital process, from back end narial
Minimum width, diastema region 6.2
Width between opposite M2s 1.8
Depth, parietal to basisphenoid 11
It has been pointed out that the specimen de~ribed above com-
pares more closely with Metaxytherium cuvieri than with the
European species of Felsinotherium in the elongation of the proxi-
mal portion of the rostrum, and in the weak development of
ridges on the anterior part of the palate. The ratio of length M3/
width M3, 1.24, agrees far better with various species of Metaxy-
therium than with Felsinotherium. The relatively narrow posterior
portion of M3, compared to the anterior portion, also resembles
M. cuvieri. Other resemblances, of doubtful importance, are the
narrowness of the external nares compared to F. serresi, and possi-
bly the presence of somewhat larger nasals than in that species.
The size is closer to that of M. cuvieri than to F. serresi, but F.
forestii is also of large size. No importance can be attached to size
40 FLORIDA GEOLOGICAL SURVEY-BULLETIN 22
differences of the magnitude involved in determining generic rela-
Un the other hand the greater deflection of the rostrum in the
Florida specimen, ana tne cusp development of the molar teeth
approach F. serrest more closely than M. cuvien.
Simpson assigned the elnocene sirenians from Florida to Feisino-
therium rather than to Mletaxytherium because "in most or all of
the principal characters permitting or close comparison they seem
rather more advanced anc more on a par with Pelsinothertum
(p. 469). Among the characters mentioned are certain doubttui
characters or the teetn (oiie or which does not hold tor this speci-
pen), the rorm or the supraoccipital and its relationship to the
foramen magnum, the form or the temporal ridges, the similar
deflection of the rostrum in this specimen and t'. serresi, and a
number of postcramal characters. He recognized also that there
were close resemblances to Metaxytherium in other characters, and
stressed the difficulty of defining these genera, which are separated
on degree of phyletic evolution alone.
The presence or two species of dugongids in the Bone Valley
Pliocene, in neither of which is the dentition associated with the
posterior portion of the cranium or postcranial skeleton, adds to
the difficulty of determining the generic affinities of these animals.
It is possible that the occiput and postcranial elements which
Simpson considered more like Felsinotherium than Metaxytherium
all pertain to F. fioridanum, and that F. ossivalense is more primi-
tive and comparable to M. cuvteri in those parts, as well as in the
anterior portion of the skull. If further discoveries show this to be
true, it will be necessary to refer the species to Metaxytherium.
But if the advanced features of the supraoccipital and other por-
tions of the skeleton are combined with the Metaxytherium-like
features mentioned above in one species, its generic allocation be-
comes a matter of uncertainty and individual choice. It is probable
that the form of th? supraoccipital is of greater significance as an
indication of advancement than the elongation or angle of the
rostrum. But the proportions of the teeth likewise merit serious
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 Felsinother-
ium in Metaxytherium, recognizing that some species of that genus
showed certain evidence of advancement over their earlier relatives.
Until more material is available, a formal proposal of this nature
would be premature. At present, then, the specimen described
above may be referred to Simpson's species, F. ossivalense, but the
generic reference remains open to doubt until better specimens are
CONTRIBUTIONS TO FLORIDA VERTEBRATE PALEONTOLOGY 41
Cottreau, J. Le Metaxytherium cuvieri du Golfe de la Loire. Annales de Pal6on-
tologie, T. XVII, pp. 3-20, 1928.
Depiret, C., and Roman, F. Le Felsinotherium serresi des sables pliocenes de
Montpelier et les rameaux phyl4tiques des sireniens fossiles de l'ancien
Monde. Archives du Musium d'Histoire Naturelle de Lyon, T. XII, Mem.
4, pp. 1-56, 1920.
Sickenberg, O. Morphologie und Stammesgeschichte der Sirenen. Palaeobio-
logica, Bd. IV, pp. 404-444, 1931.
Simpson, G. G. Fossil Sirenia of Florida and the Evolution of the Sirenia.
Bulletin of the American Museum of Natural History, vol. 59, pp. 419-
(Contains extensive bibliography on Sirenia).
E 43 1
Photograph of left side of anterior portion of skull, x /2.
[ 44 1
FLORIDA GEOLOGICAL SURVEY BULLETIN TWENTY-TWO, Plate 1
[ 45 ]
FLORIDA GEOLOGICAL SURVEY BULLETIN TWENTY-TWO, Plate 2
[ 46 ]
'o 'W 7.Fr
Photograph of top of rostral portion of skull, x /2.
L 47 J