Content uploaded by Christian de Muizon
Author content
All content in this area was uploaded by Christian de Muizon on Feb 15, 2021
Content may be subject to copyright.
387
Journal of Vertebrate Paleontology 24(2):387–397, June 2004
q
2004 by the Society of Vertebrate Paleontology
THE YOUNGEST SPECIES OF THE AQUATIC SLOTH THALASSOCNUS AND A REASSESSMENT
OF THE RELATIONSHIPS OF THE NOTHROTHERE SLOTHS (MAMMALIA: XENARTHRA)
CHRISTIAN DE MUIZON
1
, H. GREGORY MCDONALD
2
, RODOLFO SALAS
3
, and MARIO URBINA
3
1
UMR 8569 CNRS, De´partement Histoire de la Terre, Muse´um national d’Histoire naturelle, 8, rue Buffon,
F-75005 Paris, France, muizon@mnhn.fr;
2
National Park Service, 7333 West Jefferson Ave, P.O. Box 25287, Denver, Colorado 80225, U.S.A.;
3
Departamento de Paleontologı´a de Vertebrados, Museo de Historia Natural (UNMSM), Av. Arenales 1256, Lima 14, Peru
ABSTRACT—Two new specimens of the aquatic sloth Thalassocnus from the Pliocene of Peru are described, one of
T. carolomartini McDonald and Muizon, 2002 and the other T. yaucensis, sp. nov. Comparisons with the type species
of Thalassocnus,T. natans, demonstrates that T. carolomartini and T. yaucensis are more similar morphologically to
each other than to other species of the genus and are more derived. For example, both have a more elongated rostrum
(premaxillae and mandibular spout) and more robust, quadrate to circular molariform teeth. The aquatic sloth Thalas-
socnus is sufficiently distinct morphologically that it is placed in a new subfamily, the Thalassocninae, and the Noth-
rotheriinae is raised to family rank. The morphological features shared by T. carolomartini and T. yaucensis indicate
these species were better adapted to grazing on marine vegetation than earlier species of Thalassocnus.
INTRODUCTION
Two specimens recently discovered in the Pisco Formation
of the Sacaco Region, Peru, provide additional insight into the
evolutionary history of the Thalassocnus lineage of nothrothere
sloths (Muizon and McDonald, 1995; McDonald and Muizon,
2002; Muizon et al., 2003). One is a partial skull and mandible
along with associated hyoid elements, foot bones, and a femur.
This specimen is from the locality of Sacaco and was discov-
ered in the SAO (Sacaco) vertebrate-bearing horizon. It is,
therefore, late early to early late Pliocene in age (Muizon and
DeVries, 1985). Although slightly smaller and more gracile, it
is otherwise very similar to the holotype of Thalassocnus car-
olomartini McDonald and Muizon, 2002, also from the SAO
Horizon, and is referred to this species. The second specimen
was discovered at Yauca, 10 km south of Sacaco, in a verte-
brate-bearing horizon that was initially regarded as equivalent
in age to the SAO Horizon. No geologic correlation has been
established between the two localities and, so far, no biostrati-
graphic evidence indicates that the Yauca beds are the sameage
as the SAO Horizon. The Yauca specimen is a partial skeleton
with an exceptionally well-preserved skull and mandible, much
better than that of the holotype of T. carolomartini and one of
the best-preserved specimens of the Thalassocnus lineage.
Comparison of the Yauca specimen and those of T. carolomar-
tini with the older species of Pisco sloths (T. antiquus,T. na-
tans,T. littoralis) indicate that T. carolomartini and the Yauca
specimen are more similar morphologically and are more de-
rived. Despite the presence of numerous shared features, the
morphology of the mandibular spout in the Yauca specimen
differs significantly from that of the two specimens of T. car-
olomartini and, because of this feature, is referred here to a
new species of Thalassocnus.
The purpose of this paper is to describe the skull and man-
dible of the new referred specimen of T. carolomartini and the
Yauca specimen. They are compared to the holotypes of T. car-
olomartini and the genotypic species T. natans, which is the
best preserved of the other three species in the genus. The Yau-
ca specimen was closely associated with several bones of a
second smaller individual indicating a fetal or newborn onto-
genetic stage. This is regarded as an indication that the skeleton
is probably that of a female.
GEOLOGY
The Pisco Formation is a Neogene marine unit present along
the southern coast of Peru. It has yielded an abundant fauna of
marine vertebrates including fishes, reptiles, birds and mam-
mals. Hoffstetter (1968), Muizon (1981), Muizon and DeVries
(1985), and Marocco and Muizon (1988) reported abundant re-
mains of sloths, which were regarded by Muizon and Mc-
Donald (1995) as having semi-aquatic habits. This interpreta-
tion was initially based on the taphonomy of the specimens.
The Peruvian coast was a desert during the late Neogene (Se´-
brier et al., 1984; Todsal et al., 1984; Alpers and Brimhall,
1988). These data, combined with the extreme abundance and
excellent preservation (articulated skeletons) of the sloth re-
mains and the fact that virtually no other land mammals have
been found in the Pisco Formation, led to the conclusion that
they regularly entered the ocean to feed on marine vegetation,
either seaweed or sea grass. Many of the anatomical features
of the Pisco nothrothere, Thalassocnus natans, are compatible
with an interpretation of an aquatic or semi-aquatic life (Muizon
and McDonald, 1995).
McDonald and Muizon (2002) and Muizon et al. (2003) de-
scribed three additional species of Thalassocnus that were re-
covered from horizons above and below that in which the ho-
lotype of T. natans was found. The oldest species, Thalassocnus
antiquus, is from the Aguada de Lomas Horizon (AGL), ca. 7
to 8 Ma; T. natans from the Montemar Horizon (MTM), ca. 6
Ma; Thalassocnus littoralis from the Sud-Sacaco (SAS) Hori-
zon, ca. 5 Ma; and T. carolomartini from the Sacaco (SAO)
Horizon, ca. 4 to 3 Ma. The stratigraphy of the Pisco Formation
used here follows Muizon and DeVries (1985). Based on the
above four species, McDonald and Muizon (2002) and Muizon
et al. (2003) raised the hypothesis of the existence of a Thal-
assocnus lineage spanning four different vertebrate horizons
and at least 4 Ma. However, Muizon et al. (2003) also admitted
that Thalassocnus might have had a more complex evolutionary
history.
Abbreviations MNHN, Muse´um national d’Histoire natu-
relle, Paris, France; MUSM, Museo de Historia Natural de la
Universidad Nacional Mayor de San Marcos, Lima, Peru;
SMNK, Staatliche Museum fu¨r Naturkunde, Karlsruhe, Ger-
388 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 24, NO. 2, 2004
many; UPCH, Universidad Peruana Cayeteno Heredia, Lima,
Peru.
SYSTEMATIC PALEONTOLOGY
Order XENARTHRA Cope, 1889
Suborder PHYLLOPHAGA Owen, 1842
Superfamily MEGALONYCHOIDEA Simpson, 1931
Family NOTHROTHERIIDAE Ameghino, 1920, new rank
Included subfamilies Nothrotheriinae Ameghino, 1920;
Thalassocninae, subfam. nov.
Revised Diagnosis Skull elongated anterior to orbits and
muzzle smaller than posterior part of skull; supraorbital fora-
men present; zygomatic arch incomplete with jugal separated
from zygomatic process of squamosal; anteriorly directed zy-
gomatic root; enormous vomerine keel in nasopharynx; man-
dibular symphysis elongated into a spout. Dentition either 5/4
or 4/3; when dentition is 5/4, anteriormost tooth in skull and
mandible modified into a caniniform positioned midway be-
tween anterior of skull and molariform cheek teeth. Molari-
forms square in cross section with a well-developed transverse
loph on the mesial and distal edges of occlusal surface; molar-
iforms have a longitudinal groove on labial and lingual sides.
Ungual on digit two of manus semicircular in cross section
while all other unguals on the manus and pes are an isosceles
triangle; third trochanter present on femur; patellar trochlea of
femur separated from articular surfaces of distal condyles; tuber
calcis of calcaneum expanded equally both anteroposteriorly
and mediolaterally; astragalus with medial trochlea modified
into an odontoid process and lateral trochlea anteroposteriorly
curved; proximal and second phalanx of digit three of pes co-
ossified in later members of family; fifth metatarsal with me-
diolaterally expanded insertion for the fibularis longus; unguals
on digits 1–4 of pes.
History The earliest nothrothere to be described was Coe-
lodon maguinense Lund, 1838 (Nothrotherium Lydekker, 1889
is a replacement for Coelodon Lund, 1838, a preoccupied
name). Nothropus was described by Burmeister in 1882, and
Pronothrotherium by Ameghino in 1907. The North American
nothrothere now referred to as Nothrotheriops was originally
referred to Nothrotherium by Sinclair (1905). Nothrotheriops
was initially described by Hoffstetter (1954) as a subgenus of
Nothrotherium, but Paula Couto (1971) raised it to the level of
genus.
Kraglievich (1923) placed the Nothrotheriinae within the Me-
galonychidae as one of four subfamilies. Other than Nothroth-
erium, the type genus of the subfamily, he did not refer any
other genera to the subfamily. However, the provided geological
and geographic distribution included the Oligocene to Pliocene
of Argentina, Pliocene to Pleistocene of Brazil and Bolivia, and
Pleistocene of North America (California and Texas), which
suggests that Kraglievich (1923) was also including Nothropus,
Pronothrotherium, and the North American nothrothere now
known as Nothrotheriops. His diagnosis for the subfamily in-
cluded the following characters: nasal region low and elongat-
ed; preorbital fossa of maxilla shallow; first upper molar absent;
crown of the molars with grooves on the lateral face; pterygoid
inflated, with air sinuses; femur wide and flattened anteropos-
teriorly; fovea captis of femur totally included within the artic-
ular surface of the head; and the three distal articular facets of
the femur are independent.
Kraglievich’s (1923) placement of the subfamily Nothroth-
eriinae within the Megalonychidae was followed by Simpson
(1945). Patterson and Pascual (1968, 1972) were the first to
suggest that the nothrotheres were more closely related to the
megatheres. Paula Couto (1971) also placed the Nothrotheriinae
within the Megatheriidae. Engelmann (1985) listed characters
he felt united the Nothrotherinae and Megatheriinae within the
Megatheriidae, but the validity of these characters was ques-
tioned by De Iuliis (1994). Patterson et al. (1992) included the
Nothrotheriinae within the Megatheriidae using characters of
the ear region and noted that the Megatheriinae and Nothroth-
eriinae shared a number of derived characters. McKenna and
Bell (1997) altered this classification by creating the tribe Noth-
rotheriini, within the subfamily Megatheriinae, family Megath-
eriidae, thus eliminating the subfamily Nothrotheriinae alto-
gether. Gaudin and De Iuliis (1999) used the designation Noth-
rotheriidae in their discussion of the relationship of Nothropus
priscus and considered the group to be monophyletic based on
three shared characters: separation of the patellar trochlea from
the medial and lateral femoral condyles; an enormous vomerine
keel in the nasopharynx; and an anteriorly directed zygomatic
root.
It should be noted that the family Nothrotheriidae as used
here does not include the Miocene genera Hapalops,Synha-
palops and Diheterocnus, which have sometimes been included
within the nothrotheres. These genera may be considered equiv-
alent to the Schismotherium group identified by Patterson et al.
(1992) and formally identified as the subfamily Schismotheri-
inae by McKenna and Bell (1997). While there may be a close
relationship between these genera and the Nothrotheriidae (as
defined here), and although McKenna and Bell (1997) included
the genera Synhapalops and Diheterocnus in their Nothrother-
iini, as of yet there is insufficient evidence to establish exactly
what this relationship is. While further work will certainly elu-
cidate these relationships, such an analysis is beyond the scope
of this paper.
Recent research on the Thalassocnus lineage and its relation-
ship to other nothrothere sloths has demonstrated that nothroth-
eres constitute a monophyletic group that has had an evolution-
ary history distinct from that of megalonychids and megatheres.
As such, we propose the formal recognition of a family Noth-
rotheriidae. We also propose that they are more closely related
to the megalonychids than to the megatheres, because they do
not share any of the unique derived characters present in the
Megatheriidae (McDonald and Muizon, 1995 and in prep.). As
such they are placed in the superfamily Megalonychoidea. The
closer relationship between nothrotheres and megalonychids
was discussed briefly in Muizon and McDonald (1995). There-
fore, the classification presented here differs from that of Mc-
Kenna and Bell (1997).
Subfamily NOTHROTHERIINAE Ameghino, 1920
Revised Diagnosis Premaxillae primitive and triangular,
shorter than length of molariform tooth row; post-palatine notch
extends anteriorly to level of last molariform; caniniform may
or may not be present; third trochanter of femur either posi-
tioned on midshaft or continuous with lateral epicondyle; length
of tibia shorter than femur.
Type Genus Nothrotherium Lyddekker, 1889.
Included Genera Nothropus Burmeister, 1882; Xylophorus
Ameghino, 1887; Pronothrotherium Ameghino, 1907; Noth-
rotheriops Hoffstetter, 1954; Chasicobradys Scillato-Yane´, Car-
lini, and Vizcaino, 1987; Huilabradys Villarroel, 1998; Amphi-
bradys Scillato-Yane´ and Carlini, 1998.
Age Range Middle Miocene to late Pleistocene (Collon-
curan to Lujanian in South America, Irvingtonian and Ran-
cholabrean in North America).
Subfamily THALASSOCNINAE, subfam. nov.
Diagnosis Premaxilla large, robust, and greatly elongated
with anterior edge expanded and spatulate with a correlative
extension of mandibular symphysis in some taxa; anteroventral
processes of maxillae much longer than in other nothrotheres;
dentition 4/3, caniniform absent; in lateral view, dorsal profile
389MUIZON ET AL.—NEW AQUATIC SLOTHS FROM PERU
of frontal relatively straight as in Nothrotherium and Pronoth-
otherium, not dorsally convex as in Nothropus and Nothroth-
eriops; sagittal crest not continuous with dorsal edge of zygo-
matic process of squamosal; infraorbital foramen short antero-
posteriorly and located above M3. Ectotympanic wider than in
Pronothrotherium and Nothrotherium, expanded medially, and
partially covering entotympanic; lateral edge of ectotympanic
thickened at inferior edge of external auditiory meatus; mastoid
process of squamosal relatively stouter than in other nothroth-
eres; zygomatic process of squamosal anteroposteriorly short;
proximal phalanx of digit 1 on manus absent; proximal and
second phalanx of digit 3 of manus fused; third trochanter of
femur continuous with greater trochanter, with distal end of
femur narrower than proximal end; tibia relatively longer com-
pared to length of the femur than in other nothrotheres; distal
caudal vertebrae with bifurcate transverse processes or process
perforated by a foramen.
Type and Only Genus Thalassocnus Muizon and Mc-
Donald, 1995.
Age Range Late Miocene to late Pliocene, Huayquerian to
Chapadmalalan, possibly Uquian.
THALASSOCNUS Muizon and McDonald, 1995
Diagnosis As for subfamily (see also McDonald and Mui-
zon, 2002:350).
Type species Thalassocnus natans Muizon and McDonald,
1995.
Included Species Thalassocnus littoralis McDonald and
Muizon, 2002; Thalassocnus carolomartini McDonald and
Muizon, 2002; Thalassocnus antiquus, Muizon et al., 2003;
Thalassocnus yaucensis, sp. nov.
THALASSOCNUS CAROLOMARTINI McDonald and Muizon,
2002
Holotype SMNK PAL 3814, a nearly complete skull and
mandible lacking part of the pterygoids, jugal, zygomatic pro-
cesses of the squamosals, exoccipitals, condyloid and angular
processes of the dentaries. Complete right and left manus
(SMNK PAL 3815) are associated with the skull.
Referred Specimens MNHN SAO 203, a partial skull
(lacking basioccipital, right auditory region and most of the
occipital) and mandible (lacking left condyle and angular pro-
cesses), both stylohyals (lacking proximal extremity), right hy-
pohyal and ceratohyal, basihyal, partial basihyal and fused right
thyrohyal, the right femur, partial right calcaneum, right navic-
ular, partial right cuboid, and partial right metatarsal 5; MNHN
SAO 201, a partial postcranial skeleton including strongly
weathered limbs and vertebral column; MNHN SAO 13, prox-
imal right humerus; MNHN SAO 76, distal left humerus; SAO
204, distal left radius; MNHN SAO 77, right metacarpal 5;
MNHN SAO 12, distal left femur; MNHN SAO 75, proximal
right tibia; MNHN SAO 16, right calcaneum; MNHN SAO 14
right astragalus; MNHN SAO 51 co-ossified first and second
phalanx of digit 3 of right foot.
Type Locality The holotype and referred specimens of T.
carolomartini are all from the locality of Sacaco, southern coast
of Peru (McDonald and Muizon, 2002:fig. 1).
Horizon and Age SAO Horizon, Pisco Formation (Mui-
zon, 1981; Muizon and DeVries, 1985), Chapadmalalan, ca. 3
to 4 Ma (late early to early late Pliocene).
Description Both skulls of T. carolomartini (holotype and
MNHN SAO 203) are very similar. However, MNHN SAO 203
differs from SMNK PAL 3814 in its smaller size, the greater
slenderness of its teeth and the slightly shorter premaxillae (Fig.
1A–D; Table 1). These differences could be related to a sexual
dimorphism. For instance, the lesser development of the pre-
maxillae could indicate a smaller upper lip in MNHN SAO 203.
In some recent mammals with a well-developed upper lip or a
small proboscis such as elephant seals (Ling and Bryden, 1981)
and saiga antelope (Sclater and Thomas, 1897), this feature is
always more pronounced in the male. Therefore, this condition
could be extrapolated to T. carolomartini. McDonald and Mui-
zon (2002) already mentioned that the great individual variation
observed in T. littoralis could be related to sexual dimorphism.
It is noteworthy that the femur associated to MNHN SAO 203
is very small (22%) when compared to the femur of MNHN
SAO 201 from the same locality and horizon. It resembles the
size of the femur of the holotype of Thalassocnus yaucensis,
sp. nov., which is interpreted as a female (see below).
The ascending ramus of the mandible is incomplete in the
holotype, but is almost complete in MNHN SAO 203 (Fig. 1E;
Table 2). The condylar process is small and its neck is short. It
is located above the tooth row and the articular surface faces
essentially dorsally. In fact, the orientation has a small posterior
component, although much less developed than the dorsal. It is
triangular (equilateral) in shape and its medial side is almost
parallel to the axis of the tooth row. It is not transversely ex-
panded as is observed in the earlier species of Thalassocnus (T.
antiquus,T. natans, and T. littoralis) and in Choloepus. The
masseteric fossa is relatively small when compared to the ear-
lier species of Thalassocnus and the apex of the coronoid pro-
cess is distinctly bent medially. On the lateral side of the pro-
cess, a distinct posteroventrally curved ridge joins the poster-
oventral angle of the apex to the anterior base of the process.
This ridge delimitates a medially bent anterodorsal surface,
which received the anterior insertion of the temporalis muscle
and could correspond to the M. temporalis superficialis pars
zygomatica observed in Choloepus by Naples (1985). The pro-
cess is long with a deeply concave posterior edge and appears
strongly recurved posteriorly.
THALASSOCNUS YAUCENSIS, sp. nov.
Holotype MUSM 37, partial skeleton including an almost
complete skull (only lacking right jugal), mandible (lacking
right angular process); stylohyals; cervical, dorsal, lumbar, and
3 proximal caudal vertebrae; 2 chevrons; ribs; both scapulae;
right humerus, radius, and ulna; right trapezoid; left scaphoid;
right metacarpals 1, 2, and proximal 3; first and second phalan-
ges of right digit 2; distal left metacarpals 3, 4 and 5; phalanges
1–3 of left digit 2; right or left pisiform; right femur; and prox-
imal left femur.
Referred Specimens MUSM 434, isolated femur; and
MUSM 347, partial skeleton including fragments of premaxilla,
maxilla, and mandible, astragalus, partial humerus, radius, ulna,
and tibia.
Etymology From Yauca, a small village on the Peruvian
coast, which is located close to the type locality.
Type Locality The holotype was discovered on the south-
ern side of the Panamerican Highway, approximately 3 km east
of Yauca. This site is here called East Yauca. It is located ap-
proximately 30 km southeast of Sacaco.
Horizon and Age Pisco Formation. Because the rostrum of
T. yaucensis is distinctly more elongated (and therefore more
derived) than in T. carolomartini, the horizon in which the new
species was discovered is probably slightly younger than the
SAO Horizon. The locality East Yauca is probably late Pliocene
or early Uquian (1.5 to 3 Ma, Flynn and Swisher, 1995).
Diagnosis Nasals longer than in T. carolomartini, extend-
ing posteriorly behind the maxilla-frontal suture and the trans-
verse plane at the level of the lachrymal foramina; longer pre-
dental portion of the maxillae; larger buccinator fossa; ventral
edge of the choanae more posterior than in T. carolomartini;
ventral process of the jugal larger and longer than dorsal pro-
cess; mandibular symphysis longer than in T. carolomartini;
390 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 24, NO. 2, 2004
FIGURE 1. Skull of Thalassocnus carolomartini, referred specimen MNHN SAO 203, in ventral (A), dorsal (B), lateral left (C), and lateral
right (D) views; mandible in lateral right (E) and dorsal (F) views.
391MUIZON ET AL.—NEW AQUATIC SLOTHS FROM PERU
TABLE 1. Comparative measurements of the new skulls of Thalassocnus. All measurements are in mm. Measurements followed by one decimal
measured using digital calipers. Palatal measurements do not include the premaxillae. Abbreviations: AP, anteroposterior; e, estimate; L, length;
MD, mesiodistal; W, width.
Thalassocnus
carolomartini
MNHN SAO 203
Thalassocnus
yaucensis
MUSM 37 holotype
Maximum skull L (in dorsal view) 288e 299.4
Basicondylar L from anteroventral angle of ascending process of the maxilla to posterior edge of
occipital condyles — 215.8
L from anterior edge of M1 to posterior of occipital condyle — 180.3
L along midline from anterior edge of postpalatine notch to anterior edge of foramen magnum — 93.3
L along midline of palate from medial base of anteroventral process of the maxilla to posterior
edge 90.1 96.5
L from anterior of M1 to posterior of M4 58.5 60.7
L from anterior of M1 to anterior edge of anteroventral process of maxilla 45.3 53.5
Maximum W between labial edges of M1 39.9 35.3
Maximum W between labial edges of M4 38.0 41.3
L from posterior edge of M4 to posterior edge of occipital condyle — 119.5
Maximum W between postorbital processes of frontal 51.5 60.9
Least interorbital breadth 49.1 52.9
Maximum W between mastoid processes — 100.9
Maximum W between lateral margins of occipital condyles — 63.6
Maximum height of occipital condyle — 27.1
W of foramen magnum between ventral edges of occipital condyles — 16.0
Height of occiput in midline from dorsal edge of foramen magnum to nuchal crest — 47.3
Maximum W between lacrimal foramina 67.2 72.3
L from postorbital process of frontal to nuchal crest (measured horizontally) 138.6 136.5
Least distance from posterolingual edge of M4 to postpalatine notch 14.6 10.4
Minimum W of palate between M1 14.0 10.4
Minimum W of palate between M4 15.1 15.2
M1 AP L 10.9 11.7
M1 MD W 12.7 12.3
M2 AP L 13.6 13.6
M2 MD W 14.4 13.6
M3 AP L 12.5 12.2
M3 MD W 14.6 14.3
M4 AP L 8.2 6.2
M4 MD W 11.9 12.5
TABLE 2. Measurements of the mandibles of the new specimens of Thalassocnus. All measurements are in mm. Measurements followed by
one decimal measured using digital calipers. Abbreviations as in Table 1.
Thalassocnus
carolomartini
MNHN SAO 203
Thalassocnus
yaucensis, sp.
nov., MUSM 37
L from anterior of spout to posterior edge of angular process
L from anterior of spout to posterior edge of condyle
L from anterior edge of spout to anterior edge of m1
L from anterior edge of spout to posterior edge of m3
L of spout from anterior edge to posterior edge of symphysis
—
221.4
95.4
144.8
90.8
250.8
232.7
114.2
166.8
105
Alveolar L of cheek teeth
Maximum depth of horizontal ramus
W of spout at anterior edge
Angle of ventral edge of spout relative to cheek tooth row
m1 AP L
51.9
45.9
38.6
24
8
14.1
52.9
46.4
33e
12
8
14.0
m1 MD W
m2 AP L
m2 MD W
m3APL
m3MDW
13.9
14.7
14.4
14.4
13.9
13.6
14.3
13.6
16.0
13.7
mandibular spout horizontal with apex located below the al-
veolar border.
COMPARATIVE DESCRIPTION
The description below will focus on the major anatomical
differences between Thalassocnus carolomartini,T. yaucensis,
and the genotypic species, T. natans. In order to avoid redun-
dancy with McDonald and Muizon (2002), the description will
be comparative. The new specimen of T. carolomartini, which
is extremely similar to the holotype, allows a better understand-
ing of the individual variation in this species and, therefore, a
better definition of the new species.
The size of the skull of T. yaucensis is similar to that of the
holotype of T. carolomartini in total length and mastoid width.
The predental palate (i.e., distance between the anterior edge of
M1 and the apex of the premaxillae) is slightly longer in T.
yaucensis (107 mm) than in T. carolomartini (101.5 mm and
87.5 mm) (Fig. 2).
As in T. carolomartini, the premaxillae of T. yaucensis are
distinctly longer than in T. natans and much wider at the apex.
As a consequence, the lateral edges of the premaxillae are sub-
392 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 24, NO. 2, 2004
FIGURE 2. Skull of Thalassocnus yaucensis, sp. nov., holotype, MUSM 37 in lateral (A), ventral (B), dorsal (C), views; mandible in lateral
left (D) and dorsal (E) views.
parallel in T. carolomartini and T. yaucensis, while they form
a distinct ‘V’ in T. natans (Fig. 3). Furthermore, the premaxillae
of T. yaucensis are also longer than in T. carolomartini.
The upper molariforms (essentially M2 and M3) of T. yau-
censis are slightly more quadrate than in the T. carolomartini
and the anterior and posterior crests more pronounced and more
transverse. However, they are conspicuously less compressed
anteroposteriorly than in T. natans (Fig. 4).
The left jugal of the holotype of T. yaucensis is completely
preserved (Fig. 2A). The dorsal process is stout and thick. Half-
393MUIZON ET AL.—NEW AQUATIC SLOTHS FROM PERU
FIGURE 3. Outlines of the premaxillae in thalassocnines. Thalassocnus antiquus (holotype, MUSM 228) (A); T. natans (holotype, MNHN SAS
734) (B); T. littoralis (holotype, MNHN SAS 1615) (C); T. carolomartini (MNHN SAO 203) (D); T. yaucensis (holotype, MUSM 37, lateral
sides of the apex reconstructed) (E). Not to scale.
FIGURE 4. Dental morphology of advanced Thalassocnus in posterolabial view. T. carolomartini (MNHN SAO 203) left M2 and posterior
edge of M1 (A) and m2 and anterior half of m3 (B). T. yaucensis (holotype, MUSM 37) left M2–M3 (C) and m2–m3 (D). Abbreviations: cf,
convex wear facet; pf, planar wear facet; tf, facet indicating some transverse movement of the mandible.
394 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 24, NO. 2, 2004
FIGURE 5. Comparative lateral views of the mandible in thalassoc-
nines. Thalassocnus natans (holotype, MNHN SAS 734) (A); Thalas-
socnus carolomartini (MNHN SAO 203) (B); Thalassocnus yaucensis
(MNHN SAO 203) (C). Not to scale.
way between its apex and the ventral border of the orbit its
anterior edge bears a salient short crest (more a knob in the
holotype), which probably corresponds to the ventral extension
of the postorbital ligament. This ligament attached anteriorly on
supraorbital process of the frontal, which is reduced to a small
bump located posteroventral to the temporal crest at approxi-
mately 5 cm from the maxillary-jugal articulation. The medial
face of the dorsal process of the jugal (its distal half) received
part of the origin of the zygomaticomandibularis muscle. The
posterior edge of the dorsal process of the jugal is distinctly
concave as in T. carolomartini. However, it is not as notched
as in Nothrotheriops and was not almost contacting the zygo-
matic process of the squamosal as in this genus. The jugal and
squamosal processes are separated by 1.5 cm and the two bones
are not imbricated as in Nothrotheriops (although never in con-
tact, Naples, 1989). This condition is similar to that in T. car-
olomartini. The ventral process of the jugal is a long and wide
plate, the apex of which is deflected posteriorly. On its lateral
side attached the origin of the five portions of the masseter
superficialis (Naples, 1989). Although the preservation of the
specimen does not allow identification of the five heads of the
masseter superficialis as in Nothrotherium (Naples, 1987), the
size of the ventral process of the jugal indicates that this muscle
was very powerful in both T. carolomartini and T. yaucensis.
The posterior edge of the ventral process is also deeply con-
cave. It is separated from the concave posterior edge of the
dorsal process by a posteriorly expanded convex area, which is
ca. 3 cm long (ca. 4 cm in the holotype of T. carolomartini).
This condition however presents some variation because in the
new specimen of T. carolomartini a second salient posterior
spine is also present on the posterior border of the ventral pro-
cess. This region of the jugal is similar to that of Thalassocnus
natans, but does not form a true posterior process, ventral to
the apex of the zygomatic process of the squamosal as in Noth-
rotherium. This convex part of the posterior edge of the jugal
is approximately at the level of the inferior border of the orbit,
while the posterior process of Nothrotherium is distinctly more
dorsal than in the thalassocnine sloths.
Assuming that the condition of T. yaucensis was similar to
that reconstructed in Nothrotherium (Naples, 1989), the mas-
seter profundus originated on the medial side of the jugal, in a
relatively central position (i.e., the ventral part of the dorsal
process, the posterodorsal part of the ventral process and the
apex of the zygomatic process of the squamosal). In T. yaucen-
sis, the dorsal process of the jugal is distinctly shorter than the
ventral process, contrary to the condition in T. natans. The ratio
ventral process/dorsal process is 1.17 (81.7/69.5) in T. yaucen-
sis (MUSM 37), 0.83 (65e/78) in T. carolomartini (MNHN
SAO 203), and 0.76 (59.7/77.8) in T. natans (MNHN SAS 734)
(the length of the processes is measured from the ventralmost
point of the orbit to the apex of the process). In T. yaucensis,
the dorsal edge of the jugal is distinctly lower than the sagittal
crest.
The pterygoids of the holotype of T. yaucensis are remark-
ably well preserved. The pterygoid flanges are relatively small
when compared to those of Thalassocnus natans and even
smaller relatively to those of Nothrotheriops and Glossother-
ium. However, they are thickened and stout at their posterior
base, as in T. carolomartini (Fig. 2B). The anteroventral edge
of the flange is a thick crest, ca. 25 mm long and 5 mm wide.
Dorsal to this crest is a conspicuous, shallow fossa for origin
of the pterygoideus medialis. Dorsal to this fossa and posterior
to the sphenoidal fissure and optical foramen is another, slightly
larger fossa for the origin of the pterygoideus lateralis. It is not
possible to identify separate origins for the upper and lower
head of this muscle, although they could be distinguished on
the right side of the skull by a very subtle division into two
shallow fossae.
On the lateral wall of the braincase of MUSM 37, the orbi-
tosphenoid bears a distinct sphenorbital fissure and optical fo-
ramen. Separation of these foramina has not been observed in
any of the other members of Thalassocnus and could represent
individual variation. The remainder of the skull of the holotype
of T. yaucensis does not present significant differences with the
two known specimens of T. carolomartini.
The singular most obvious character of T. yaucensis is ob-
served on the mandible. The mandibular spout is distinctly lon-
ger than in T. carolomartini (Figs. 5–6). The ratio of the length
of the mandibular spout (from apex of the mandible to anterior
edge of m1) to the foraminal length (from apex to anterior edge
of the medial mandibular foramen) is 0.60 in T. yaucensis, 0.52
in T. carolomartini (average of SMNK 3814 and MNHN SAO
203), 0.49 in T. natans (one specimen), and 0.50 in T. antiquus
(one specimen). The difference between T. yaucensis and T.
carolomartini is almost 15% and is regarded here as too great
to be the result of simple individual variation.
Another important difference of the mandible of T. yaucensis
is the orientation of the mandibular spout (Fig. 5). In the ho-
lotype and only specimen for which a complete skull is known,
it is oriented almost horizontally. Its dorsal and ventral edges
form a similar angle with the tooth row and the apex is located
distinctly below the alveolar border. The dorsal edge of the
spout is concave and forms an angle of approximately 165
8
to
170
8
below the alveolar border. In T. carolomartini, the dorsal
edge of the spout is straight and aligned with the alveolar bor-
der. This condition is also present in T. natans. Therefore, the
mandibular spout of T. yaucensis is unique as far as its length
and orientation are concerned.
The ascending ramus is short anteroposteriorly; the distance
between the anterior base of the coronoid process and the pos-
395MUIZON ET AL.—NEW AQUATIC SLOTHS FROM PERU
FIGURE 6. Outlines of the mandibular spout in Thalassocnus.T. antiquus (holotype, MUSM 228) (A); T. natans (holotype, MNHN SAS 734)
(B); T. littoralis (MNHN SAS 736, juvenile) (C); T. carolomartini (MNHN SAO 203) (D); T. yaucensis (holotype, MUSM 37, restored) (E). Not
to scale.
TABLE 3. Measurements of bones of the stylohyal in Thalassocnus yaucensis, sp. nov. (MUSM 37). All measurements are inmm. Measurements
followed by one decimal measured using digital calipers. Abbreviations as in Table 1.
Left Right
L from articular facet for tympanohyal to articular facet for ceratohyal
AP L of proximal end —
24.5 56.2
25.4
terior base of the condyloid process is shorter than in T. natans
and T. carolomartini (Fig. 5). The condyloid process is small,
its neck is short and it faces more dorsally than posteriorly. The
condyle is triangular and its anterolateral edge is distinctly
shorter than the anteromedial and the posterior, which is the
longest. Therefore, the condyle is transversely expanded. Its
medial portion is more slender than the lateral and projects
medially. The condyloid process is located above the level of
the tooth row.
The coronoid process is high and its posterior edge is deeply
concave. It is recurved posteriorly and medially at its apex. Its
anterior edge is strongly convex and reaches the horizontal ra-
mus at an angle close to 90 degrees.
A portion of the hyoid apparatus is preserved in MUSM 37,
but only the right stylohyal is complete (Table 3). It is short
and massive relative to that of T. natans. Its proximal extremity
is triangular-shaped and articulates in the tympanohyal fossa of
the skull. The articular process is a rugose condyle. From this
condyle the stylohyal extends anteroventrally in a long rod-like
process. This process is proportionally shorter and thicker than
in T. natans. Another small process extends posteroventally and
is approximately as developed as the condyle. The triangular
proximal portion of the bone is gently concave laterally. Me-
dially the bone is slightly rugose and do not present any fossa
as in T. natans. The apex of the anterior process is expanded
and bears the articulation for the epihyal.
CONCLUSIONS
As indicated by the features described above, both T. caro-
lomartini and T. yaucensis have more advanced adaptations to
an aquatic lifestyle relative to the other species of Thalassoc-
nus. The newest species of aquatic sloth described here enlarges
the breadth of intrageneric variation of the genus Thalassocnus,
currently represented by five species. The two youngest species
of Thalassocnus differ significantly in their cranial and post-
cranial anatomy from the three earlier species, which could jus-
tify their placement in a separate genus. However, this would
make Thalassocnus a paraphyletic taxon (see Muizon et al.,
2003). Considering the fact that Thalassocnus is apparently an
endemic taxon from the southern coast of Peru (and possibly
northern Chile), we retain the five species in a single genus
reflecting our argument that Thalassocnus represents a single
evolutionary lineage, extending from the late Miocene (T. an-
tiquus) to the late Pliocene (T. yaucensis) and illustrates the
multiple stages of the adaptation of a terrestrial mammal to an
aquatic environment.
On the skull, the most spectacular differences between the
youngest two species of Thalassocnus and earlier species are
in the anterior extension and widening of the premaxillae and
in the splayed apex of the mandibular spout. These modifica-
tions are clearly related to feeding adaptations, including the
development of strong lips. Muizon et al. (2004) provide a
functional analysis of the evolution of feeding adaptations in
Thalassocnus.
The aquatic specializations of the thalassocnines relative to
other nothrotheres are also present on the postcranial skeleton,
although generally less spectacular than those of the skull. The
most notable change is on the radius, which is distinctly trans-
versely widened in T. yaucensis and approaches the condition
of pinnipeds (the radius figured by Muizon et al., 2003:fig. 3E
belongs to the holotype of T. yaucensis). This is an indication
that, as mentioned by Muizon and McDonald (1995), the thal-
assocnines had some locomotor modifications adapted to facil-
itate feeding and/or locomotion in an aquatic environment. In
396 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 24, NO. 2, 2004
this respect, T. yaucensis is the most obviously derived relative
to all other species of Thalassocnus. A study of the postcranial
skeleton of the thalassocnines is currently in progress by the
authors.
Despite the many aspects of the anatomy of Thalassocnus
that are unique, the genus shares many features in common with
other nothrotheres that distinguish the group as both monophy-
letic and distinct from other traditionally recognized families of
sloths. The presence of both derived features unique to the noth-
rotheres and the group’staxonomic diversity including the high-
ly derived thalassocnines justifies the recognition of the family
Nothrotheriidae with two subfamilies, Nothrotheriinae and
Thalassocniinae.
ACKNOWLEDGMENTS
Funds for field work were provided by the IFEA (Institut
Franc¸ais d’E
´tudes Andines) and the CNRS (Centre National de
la Recherche Scientifique). The type and all of the referred
specimens of Thalassocnus carolomartini were discovered by
Carlos Martı´n. The holotype of T. yaucensis was discovered
and collected by Martı´n Ostrowiki. Photographs are by P. Loub-
ry and D. Serrette.
LITERATURE CITED
Alpers, C., and G. Brimhall. 1988. Middle Miocene climatic change in
the Atacama Desert, northern Chile: evidence from supergene min-
eralization at La Escondida. Bulletin of the Geological Society of
America 100:1640–1656.
Ameghino, F. 1907. Notas sobre una pequen˜a collecio´n de huesos de
mamı´feros procedentes de las grutas calca´reas de Iporanga del Es-
tado de Sa˜o Paulo-Brazil. Revista del Museo Paulista 7:59–124.
Ameghino, F. 1920. Sur les e´dente´s fossiles de l’Argentine. Examen
critique, re´vision et correction de l’ouvrage de la M. R. Lydekker.
Obras Completas y Correspondencia Cientifica 11:447–909.
Burmeister, H. 1882. Nothropus priscus, ein bisher unbekanntes fossils.
Faulthier Sitzungberichte der (Ko¨ niglische-preussischen) Akademie
der Wissenschaften, Berlin 1882(2):613–620.
Cartelle, C., and G. A. Boho´rquez. 1986. Descric¸a˜o das pre-maxilas de
Nothrotherium maquinenese (Lund) Lydekker, 1889. (Edentata,
Megalonychidae) e de Eremotherium laurillardi (Lund) Cartelle &
Boho´rquez, 1982. (Edentata, Megatheriidae). Iheringia Se´rie Geo-
logia 11:9–14.
De Iuliis, G. 1994. Relationships of the Megatheriinae, Nothrotheriinae,
and Planopsinae: some skeletal characteristics and their importance
for phylogeny. Journal of Vertebrate Paleontology 14:577–591.
Engelmann, G. F. 1985. The phylogeny of the Xenarthra; pp. 51–64 in
G. G. Montgomery (ed.), The Evolution and Ecology of Armadil-
los, Sloths, and Vermilinguas. Smithsonian Institution Press,Wash-
ington, D.C.
Flynn, J. J., and C. C. Swisher. 1995. Cenozoic South American land
mammal ages: correlation to global geochronologies; pp. 317–333
in Time Scales and Global Stratigraphic Correlation. SEPM Special
Publication 54, Tulsa.
Gaudin, T. J., and G. De Iuliis. 1999. The Late Miocene slothNothropus
priscus, with comments on the phylogeny of the Nothrotheriidae.
Journal of Vertebrate Paleontology 19 (3,supplement):46A.
Hoffstetter, R. 1954. Les gravigrades (Edente´s Xe´narthres) des cavernes
de Lagoa Santa (Minas Gerais, Bre´sil). Annales des Sciences Na-
turelle, Zoologie, 11
e`me
Se´rie, 16:741–764.
Hoffstetter, R. 1968. Un gisement de verte´bre´s tertiaires a` Sacaco (Sud-
Pe´rou), te´moin ne´oge`ne d’une migration de faunes australes au long
de la coˆte occidentale sudame´ricaine. Comptes Rendus de
l’Acade´mie des Sciences, Se´rie D, 267:1273–1276.
Kraglievich, L. 1923. Descripcio´n de dos cra´neos y otros restos del
ge´nero Pliomorphus Ameghino procedentes de la formatio´n En-
trerriana de las barrancas del Rı´o Parana´. Anales del Museo Na-
cional de Historia natural de Buenos Aires 33:1–56.
Ling, J. K., and M. M. Bryden. 1981. Southern elephant seal—Miroun-
ga leonina; pp. 297–327 in S. H. Ridgway and R. Harrison (eds.),
Handbook of Marine Mammals, Vol. 2. Seals. Academic Press,
London, 359 pp.
Lund, P. W. 1838. Coup d’oeil sur les espe`ces e´teintes de mammife`res
du Bre´sil. Annales des Sciences Naturelles, se´rie 2, (zoologie) 11:
214–234.
McDonald, H. G., and C. de Muizon. 2002. The cranial anatomy of
Thalassocnus (Xenarthra, Mammalia), a derived nothrothere from
the Neogene of the Pisco Formation (Peru). Journal of Vertebrate
Paleontology 22:349–365.
McKenna, M. C., and S. K. Bell. 1997. Classification of Mammals
Above the Species Level. Columbia University Press, New York,
631 pp.
Marocco, R., and C. de Muizon. 1988. Los vertebrados del Neogeno
de la costa sur del Peru´: ambiente sedimentario y condiciones de
fosilizacio´n. Bulletin de l’Institut Franc¸ais d’Etudes Andines 17:
105–117.
Muizon, C. de. 1981. Les ve´rte´bre´s fossiles de la Formation Pisco (Pe´r-
ou). Premie`re partie. Deux nouveaux Monachinae (Phocidae, Mam-
malia) du Plioce`ne de Sud-Sacaco. Travaux de l’Institut Franc¸ais
d’E
´tudes Andines, 22:1–160.
Muizon, C. de, and T. J. DeVries. 1985. Geology and paleontology of
Late Cenozoic marine deposit in the Sacaco area (Peru). Geologis-
che Rundschau 74:547–563.
Muizon, C. de, and H. G. McDonald. 1995. An aquatic sloth from the
Pliocene of Peru. Nature 375:224–227.
Muizon, C. de, H. G. McDonald, R. Salas, and M. Urbina. 2003. An
early species of the aquatic sloth Thalassocnus (Mammalia, Xe-
narthra) from the late Miocene of Peru. Journal of Vertebrate Pa-
leontology 23:886–894.
Muizon, C. de, H. G. McDonald, R. Salas, and M. Urbina. 2004. The
evolution of feeding adaptations of the aquatic sloth Thalassocnus.
Journal of Vertebrate Paleontology. 24:398–410.
Naples, V. L. 1985. Form and function of the masticatory musculature
in the tree sloths, Bradypus and Choloepus. Journal of Morphology
183:25–50.
Naples, V. L. 1987. Reconstruction of cranial morphology and analysis
of function in the Pleistocene ground sloth Nothrotheriops shasten-
sis (Mammalia, Megatheriidae). Natural History Museum of Los
Angeles County, Contributions to Science 389:1–21.
Naples, V. L. 1989. The feeding mechanism in the Pleistocene ground
sloth, Glossotherium. Natural History Museum of Los Angeles
County, Contributions to Science 415:1–23.
Patterson, B., and R. Pascual. 1968. Evolution of mammals on southern
continents. Quarterly Review of Biology 43:409–451.
Patterson, B., and R. Pascual. 1972. The fossil mammal fauna of South
America; pp. 247–309 in A. Keast, F. C. Erk, and B. Glass (eds.),
Evolution, Mammals, and Southern Continents. State University of
New York Press, Albany.
Patterson, B., W. Segall, W. D. Turnbull, and T. J. Gaudin. 1992. The
ear region in xenarthrans (
5
Edentata: Mammalia) Part II. Pilosa
(Sloths, Anteaters), Palaeanodonts, and a miscellany. Fieldiana Ge-
ology new series 24:1–79.
Paula Couto, C. de. 1971. On two small Pleistocene ground-sloths. An-
ais de Academia Brasileira de Cie´ncias 43:499–513.
Paula Couto, C. de. 1976. The manus of Nothrotheriops shastense (Sin-
clair, 1905). Anais do XXVIII Congresso Brasileiro de Geologia2:
166–176.
Phillips, R. C., S. Wyllie-Escheverria, and T. J. DeVries. 1996. Fossil
Heterozostera tasmanica from south-central Peru; pp. 369 in J.
Kuo, R. C. Phillips, D. J. Walker, and H. Kirkman (eds.), Seagrass
Biology. Proceedings of an International Workshop, Rottnest, Is-
land, Western Australia, 25–29 January 1996.
Sclater, P. L., and O. Thomas. 1897. The Book of Antelopes, Volume
III. R. H. Porter, London, 245 pp.
Se´brier, M., R. Marocco, and J. Machare´. 1984. Evolution tectonique
du piedmont pacifique et ses relations avec la cordillie`re des Andes
du Pe´rou central et me´ridional; pp. 49–70 in Montagnes et Pied-
monts, Revue ge´ ographique des pyre´nne´es et du Sud-Ouest. Special
Issue, Toulouse.
Simpson, G. G. 1945. The principles of classification and a classification
of mammals. Bulletin of the American Museum of Natural History
85:1–350.
Sinclair, W. J. 1905. New mammalian from the Quaternary caves of
California. University of California Publications, Bulletin of the
Department of Geology 4:145–161.
Stock, C. 1925. Cenozoic gravigrade edentates of western North Amer-
397MUIZON ET AL.—NEW AQUATIC SLOTHS FROM PERU
ica with special reference to the Pleistocene Megalonychinae and
Mylodontidae of Rancho La Brea. Carnegie Institution of Wash-
ington Publication 331:1–206.
Todsal, R., A. Clark, and E. Farrar. 1984. Cenozoic polyphaselandscape
and tectonic evolution of the Cordillera Occidental, Southernmost
Peru. Bulletin of the Geological Society of America 95:1318–1332.
Received 14 January 2002; accepted 30 October 2003.
