Journal of Vertebrate Paleontology 22(3):510–534, September 2002
2002 by the Society of Vertebrate Paleontology
THE OSTEOLOGY OF MASIAKASAURUS KNOPFLERI, A SMALL ABELISAUROID
(DINOSAURIA: THEROPODA) FROM THE LATE CRETACEOUS OF MADAGASCAR
MATTHEW T. CARRANO
*, SCOTT D. SAMPSON
, and CATHERINE A. FORSTER
Department of Anatomical Sciences, Health Sciences Center T-8, Stony Brook University, Stony Brook,
New York 11794-8081, email@example.com;
Utah Museum of Natural History and Department of Geology and Geophysics, University of Utah,
1390 East Presidents Circle, Salt Lake City, Utah 84112-0050
ABSTRACT—We describe the osteology of the new small theropod dinosaur Masiakasaurus knopﬂeri, from the Late
Cretaceous Maevarano Formation of northwestern Madagascar. Approximately 40% of the skeleton is known, including
parts of the jaws, axial column, forelimb, pelvic girdle, and hind limb. The jaws of Masiakasaurus are remarkably
derived, bearing a heterodont, procumbent dentition that is unknown elsewhere among dinosaurs. The vertebrae are
similar to those of abelisauroids in the reduction of the neural spine, lack of pleurocoelous fossae on the centrum, and
extensively pneumatized neural arch. The limb skeleton is relatively gracile and bears numerous abelisauroid synapo-
morphies, including a rounded humeral head, peg-and-socket iliac-pubic articulation, prominent femoral medial epi-
condyle, expanded tibial cnemial crest, and double-grooved pedal unguals. The femora and tibiae show evidence of
dimorphism. More speciﬁc features shared between Masiakasaurus, the Argentine Noasaurus, and the Indian Laevi-
suchus suggest that these taxa form a clade (Noasauridae) within Abelisauroidea. This is supported by a cladistic
phylogenetic analysis of 158 characters and 23 theropod taxa. Additionally, Ceratosauria is rendered paraphyletic in
favor of a sister-taxon relationship between Neoceratosauria and Tetanurae that is exclusive of Coelophysoidea. The
unique dental and jaw specializations of Masiakasaurus suggest deviation from the typical theropod diet. Finally, the
distribution of noasaurids further supports a shared biogeographic history between South America, Madagascar, and
India into the Late Cretaceous.
The Late Cretaceous vertebrate fauna from the Maevarano
Formation of Madagascar has been known since the end of the
nineteenth century (Depe´ret, 1896; Thevenin, 1907). Originally
two dinosaur taxa were described: the sauropod Titanosaurus
madgascariensis Depe´ret, 1896 and the theropod Megalosaurus
crenatissimus Depe´ret, 1896 (
Lavocat, 1955). Both were based on fragmentary materials and,
although more complete materials were subsequently allocated
(e.g., Lavocat, 1955), are now considered nomina dubia (Mc-
Intosh, 1990; Sampson et al., 1998). Recent expeditions to the
Maevarano Formation have produced more complete dinosaur
remains, demonstrating the presence of at least two titanosau-
rian sauropods (Curry and Forster, 2001) and a large abelisaurid
theropod (Majungatholus atopus, Sampson et al., 1998). This
work has also unearthed exceptional materials from a diverse
non-dinosaurian fauna that includes freshwater actinopterygi-
ans, frogs, turtles, snakes, crocodyliforms, birds and mammals
(e.g., Forster et al., 1996, 1998; Krause et al., 1997, 1998, 1999;
Asher and Krause, 1998; Gottfried and Krause, 1998).
Numerous similarities have been demonstrated between the
Maevarano fauna and those from approximately contempora-
neous formations in India and South America, involving both
terrestrial and freshwater vertebrate clades (Krause et al., 1998,
1999; Sampson et al., 1996, 1998). More unusually, the Mae-
varano fauna includes at least seven crocodyliforms (Buckley
and Brochu, 1999) and ﬁve birds (Forster and O’Connor, 2000)
along with a comparatively low-diversity dinosaurian fauna.
The rarity of small non-avian dinosaurs—despite concerted, and
otherwise successful, attempts to recover small-bodied taxa
through screenwashing—remains a persistent and quizzical
characteristic of this Late Cretaceous fauna.
*to whom correspondence should be addressed
Thus, the presence of a small theropod taxon in the Maevar-
ano Formation is signiﬁcant on several levels. First, it indicates
that small-bodied non-avian dinosaurs are in fact present, al-
though of low diversity. Second, remains from most regions of
the skeleton indicate that this is a new taxon (Sampson et al.,
2001), raising the number of known non-avian dinosaur taxa to
four. Third, this taxon appears to belong to the theropod clade
Abelisauroidea, signiﬁcantly broadening the morphological di-
versity of this group and further strengthening the faunal links
between the Late Cretaceous faunas of Madagascar, India, and
Although these remains were found as isolated materials, a
number of factors lead us to assign them to a single taxon. First,
several elements (dentary, femur, tibia, metatarsal II, pedalpha-
langes) are known from multiple specimens that span a consid-
erable size range, but the maximum size attained by each ele-
ment is consistent with a single adult taxon of approximately
1.8–2.0 m in length. Second, these multiple elements offer no
conﬂicting phylogenetic information that might suggest the
presence of a second, distinct small-bodied taxon. Finally, low
dinosaur diversity appears to be a genuine characteristic of the
Maevarano fauna. As a result, we have referred all the small
theropod material in the Maevarano Formation to Masiakasau-
rus knopﬂeri (Sampson et al., 2001), described in detail below.
We look to future discoveries of articulated materials for con-
ﬁrmation of these associations.
DINOSAURIA Owen 1842
SAURISCHIA Seeley 1888
THEROPODA Marsh 1881
ABELISAUROIDEA (Bonaparte et Novas, 1985)
and Forster, 2001
Holotype UA 8680, a nearly complete right dentary with
511CARRANO ET AL.—OSTEOLOGY OF MASIAKASAURUS KNOPFLERI
FIGURE 1. Geographic data. A, Map of the Mahajanga Basin show-
ing outcrops of the Maevarano Formation (hatched area) and the ﬁeld
study area (box), near the village of Berivotra. B, Map of Madagascar
showing location of the Mahajanga Basin (black).
Hypodigm The holotype and referred specimens: dentaries
(FMNH PR 2177-79, 2222; UA 8682); splenial (FMNH PR
2124); angular (?) (FMNH PR 2166); maxilla (FMNH PR
2183); premaxillary/anterior dentary teeth (FMNH PR 2165,
2180, 2226); posterior maxillary/dentary teeth (FMNH PR
2164, 2170, 2181–82, 2198–2201, 2220–21, 2228); cervical
(FMNH PR 2139–41), dorsal (FMNH PR 2111, 2113–14,
2137–38, 2144–45, 2171, 2207, 2229; UA 8701), and sacral
vertebrae (FMNH PR 2142); proximal (FMNH PR 2133, 2204,
2230), medial (FMNH PR 2110, 2125–26; UA 8688, 8692),
and distal (FMNH PR 2127–28, 2156–57, 2162–63, 2168,
2202–03; UA 8689–91, 8695–96, 8702–03) caudal vertebrae;
humeri (FMNH PR 2143; UA 8693–94); manual phalanges
(FMNH PR 2132, 2205, 2224–25, 2227) and unguals (FMNH
PR 2136, 2169); pubes (FMNH PR 2108–09); femora (FMNH
PR 2115, 2117, 2120, 2123, 2148–50, 2153, 2208, 2215; UA
8681, 8684, 8712); tibiae (FMNH PR 2112, 2118–19, 2121,
2152, 2214; UA 8685, 8687, 8710–11); tibia with partial ﬁbula
and astragalocalcaneum (FMNH PR 2116, 2122); calcaneum
(FMNH PR 2235); metatarsals II (FMNH PR 2147, 2151, 2154,
2175, 2206; UA 8683), III (FMNH PR 2146, 2155), and IV
(FMNH PR 2214, 2234); pedal phalanges (FMNH PR 2129–
31, 2136, 2158–61, 2167, 2172–74, 2176, 2216–19, 2223; UA
8686, 8700, 8713–14) and unguals (FMNH PR 2134–35, 2236).
Locality and Horizon The holotype and nearly all referred
elements were recovered as isolated specimens from a single
locality, MAD 93-18. This locality lies in the Anembalemba
Member of the Maevarano Formation (Maastrichtian, Late Cre-
taceous), Mahajanga Basin, near the village of Berivotra in
northwestern Madagascar (Krause et al., 1998; Rogers and
Hartman, 1998; Rogers et al., 2000) (Fig. 1). Eight additional
localities (MAD 93-30, 93-35, 95-05, 99-26, 99-30, 99-38, 99-
50, and 99-51) in the Anembalemba member have produced
limited materials of Masiakasaurus.
Diagnosis (based on holotype only) Abelisauroid theropod
with anterior four dentary teeth procumbent, the ﬁrst inclined
above the horizontal and lying in an alveolus that is slung
below the ventral margin of the dentary. First alveolus large
and ventrally expanded, lying lateral to an anteroposteriorly
long dentary symphysis. Lower dentition markedly heterodont:
the ﬁrst four teeth are weakly spoon-shaped, elongate, and ter-
minate in a posteriorly hooked, pointed apex. Anterior dentary
teeth bear two weakly serrated posterior carinae and have faint
posterior ridges; more posterior teeth are transversely com-
pressed, recurved, and have a serrated posterior carina.
Skull and Lower Jaw
The skull and lower jaws of Masiakasaurus are represented
by a single right maxilla, six dentaries, a splenial, an angular
(?), and numerous isolated teeth.
Maxilla The maxilla (FMNH PR 2183; Fig. 2) is incom-
plete dorsally and posteriorly, but preserves the premaxillary
and nasal contacts as well as most of the antorbital fossa and
ventral (alveolar) margin. No erupted teeth are preserved, but
unerupted examples are present in the third and ﬁfth alveoli.
The anterolateral surface is slightly concave, a feature best ob-
served in anterior view. The premaxillary contact is concave,
relatively narrow, and terminates dorsally at the level of the
palatal process. An elongate, shallow fossa is dorsolaterally ad-
jacent to this and likely represents the nasal contact, suggesting
that the maxilla did not contribute to the external naris. Ma-
siakasaurus resembles the abelisaurid Majungatholus (see Ap-
pendix 1 for list of comparative theropod materials) in the over-
all arrangement of these features. The anterior margin of the
maxilla is angled 40
anteroventrally but has no distinct anterior
The antorbital fossa is elongate and bounded anteroventrally
by a distinct, narrow, raised rim. The fossa is broad anteriorly,
extending to the dorsal limit of the ascending ramus as pre-
served, but tapers posteriorly along with the maxilla itself.
Within the fossa, a subcircular depression is present at the base
of the ascending ramus, bordered by thin lateral and medial
laminae. Similar fossae are present in Ceratosaurus,Torvosau-
rus, and Megalosaurus. The antorbital fossa is not perforated
by additional foramina, but does invaginate some distance into
the body of the anterior maxillary ramus. This portion of the
fossa may be homologous to the promaxillary fenestra (Witmer,
The medial surface of the maxilla bears a prominent, short
triangular palatal process that protrudes medially and horizon-
tally (the anteriormost portion may be missing) immediately
below the maxillary antrum. It is not clear whether the palatal
process contacted that from the opposite maxilla, and there are
no traces of contact surfaces or ridges for the palatine or vomer
(unlike Dilophosaurus,Ceratosaurus and tetanurans). Further
ventrally, the paradental plates are smooth, continuous, and sep-
arated from the remainder of the medial maxilla by a prominent
horizontal groove that runs the length of the element. A small,
dorsomedially placed fossa at the posterior end probably rep-
resents part of the palatine contact.
Seven alveoli are preserved in FMNH PR 2183 and, although
the posteriormost portion is incomplete, it is unlikely that more
than two or three are missing. The ﬁrst alveolus has the most
peculiar orientation, being directed anteroventrally at about 40
to the horizontal (roughly parallel with the premaxillary con-
tact), indicating that the anterior teeth were somewhat procum-
bent. Successive maxillary alveoli are increasingly vertical until
becoming fully so by the ﬁfth. A row of small neurovascular
foramina, one per tooth position, is visible on the lateral surface
dorsal to the alveoli. The cross-section of the ﬁrst alveolus dif-
fers from those of subsequent alveoli in being almost circular
rather than anteroposteriorly elliptical. Thus, although the max-
illary dentition is not preserved, these features suggest that it
512 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 22, NO. 3, 2002
FIGURE 2. Right maxillae of Masiakasaurus knopﬂeri (A–C; FMNH PR 2124) and Noasaurus leali (D,E; PVL 4061) in lateral (A,D), anterior
(C) and medial (B,E) views. Alveoli are numbered sequentially from anterior to posterior; dashed lines indicate reconstructed outline of bone.
Abbreviations:aof, antorbital fossa; eaof, rim of external antorbital fenestra; iaof, internal antorbital fenestra; msf, maxillary sinus fossa; nc,
contact with nasal; nf, nutrient foramina; ng, nutrient groove; pdp, paradental plates; plc, contact with palatine; pmc, contact with premaxilla;
pp, palatal process. Scale equals 5 mm. D and E modiﬁed from Bonaparte and Powell (1980).
exhibited both procumbency and heterodonty, with teeth grad-
ing from conical to more blade-like posteriorly.
The maxilla of Masiakasaurus is quite primitive in certain
respects. For example, it bears only a possible homolog to the
promaxillary fenestra, a condition similar to that of Syntarsus
(Tykoski, 1998), Dilophosaurus (Welles, 1984), Ceratosaurus,
and abelisaurids (e.g., Carnotaurus; Bonaparte et al., 1991) but
not most tetanurans (e.g., Allosaurus,Sinraptor). In addition,
the anterior ramus is poorly developed, lacking even the narrow
anterior extension seen in coelophysoids. The simple palatal
process, together with the unusually low tooth count, the pro-
nounced rim on the external antorbital fenestra, and the prom-
inent medial longitudinal groove, are also found on the maxilla
of Noasaurus leali, from the Late Cretaceous of Argentina
(Bonaparte and Powell, 1980). A pronounced antorbital rim is
also present in coelophysoids, but the maxilla in these taxa has
a higher tooth count (
20 in coelophysids) and an angled an-
teroventral corner that forms part of the ‘‘subnarial gap’’
(Welles, 1984). The rim is also present in the coelurosaurs Or-
nitholestes and Compsognathus, and may be associated with
small size. The maxilla differs from those of abelisaurids in
having a less steeply inclined anterior margin, a much more
extensive antorbital fossa, no external texturing, fewer alveoli,
a slightly concave anterolateral surface, and poorly exposed
Dentary The dentary is the most unusual element pre-
served from Masiakasaurus. The anteriormost portion is absent
in the holotype (UA 8680), but it is well preserved in FMNH
PR 2179. Although the dorsal margin is nearly straight posterior
to the fourth alveolus, anteriorly it descends abruptly towards
the ventral border (Fig. 3). Most of the ventral margin is straight
posterior to the ﬁfth alveolus and slowly diverges from the dor-
sal margin. Anteriorly the ventral margin bulges downward to
accommodate the enlarged and nearly horizontal ﬁrst alveolus,
giving the ventral edge an arched shape overall. In total, the
dentary has 10–12 alveoli (12 in UA 8680). A series of small
neurovascular foramina, two per tooth position anteriorly and
one per position posteriorly, form a row along the lateral den-
tary ventral to the alveoli.
In lateral view, the ﬁrst alveolus is clearly demarcated from
the remainder of the dentary. It is anteroposteriorly elongate,
slung below the ventral margin of the dentary, and oriented
nearly directly forward (approximately 10
from the horizontal).
The second and third alveoli are less procumbent (approxi-
, respectively) but are also angled laterally
(Fig. 3E). Taken together, these features create a
scoop-like anterior dentary that opens both anteriorly and lat-
erally. This morphology diminishes posteriorly: the fourth al-
veolus is angled only slightly laterally, the ﬁfth not at all, and
both alveoli are more vertical. The eighth and all successive
alveoli are nearly vertical (e.g., FMNH PR 2178). Alveolar
cross-sections are greatest anteriorly and become progressively
smaller posteriorly. Cross-sections of the dentary indicate that
3–4 anterior alveoli are stacked dorsoventrally. Nothing com-
parable has been observed in any other theropod or dinosaur.
The posterior margin bears several posteriorly-projecting
processes. Dorsally, three prongs—two lateral and one more
medial—border a rounded notch that received the anterior pro-
cess of the surangular. A signiﬁcantly larger process (partially
absent in the holotype) forms the posteroventral portion of the
dentary and contacted the anterior part of the angular. Between
these four processes is a large notch marks the anterior margin
of the hypertrophied external mandibular fenestra. This mor-
phology is very similar to that of the abelisaurids Carnotaurus
and Majungatholus, both of which possess enlarged external
mandibular fenestrae. The only notable difference relates to the
relative size of the three dorsal processes: in Masiakasaurus
(UA 8680), the medial process is the longest of the three,
513CARRANO ET AL.—OSTEOLOGY OF MASIAKASAURUS KNOPFLERI
FIGURE 3. Holotypic right dentary of Masiakasaurus knopﬂeri (UA 8680) in lateral (A), medial (B), dorsal (C), ventral (D), and anterior (E;
stereopair) views. Alveoli are numbered sequentially from anterior to posterior. Abbreviations:anc, contact with angular; emf, external mandibular
fenestra; pdp, paradental plates; mg, Meckelian groove; sac, contact with surangular; ss, symphyseal surface. Scale equals 1 cm.
FIGURE 4. Left splenial (FMNH PR 2183) of Masiakasaurus knop-
ﬂeri in lateral (A) and medial (B) views. Abbreviations:anc, contact
with angular; f, fossa; imf, internal mandibular fenestra. Scale equals 1
whereas in Majungatholus the ventral process is the longest.
This is distinct from the condition in other theropods, in which
the ventral portion of the posterior dentary (below the suran-
gular notch) extends much further posteriorly and the external
mandibular fenestra is relatively smaller. Furthermore, in Ma-
siakasaurus the anterior margin of the surangular notch sits just
behind the last alveolus, rather than far posterior to the last
In medial view, the Meckelian groove tapers anteriorly to a
point below the ﬁfth alveolus (FMNH PR 2177), demarcating
the splenial contact. A single foramen is located midway along
the length of this groove (UA 8680). The symphysis extends
along the medial wall of the ﬁrst alveolus and has a smooth,
planar surface that occupies the ventral half of the anterior den-
tary. The long axis of the symphysis extends anteroposteriorly
back to the ﬁfth alveolus; it is considerably longer than in most
other theropods. When articulated, the left and right dentary
rami would have formed a V-shape in dorsal view, instead of
the U-shape characteristic of abelisaurids. The paradental plates
are shorter than those of the maxilla; in fact, the medial wall
of the dentary nearly obscures them from view. This pattern is
particularly marked posterior to the ﬁfth alveolus, where suc-
cessive, triangular paradental plates appear distinct from one
another. One well-preserved specimen (FMNH PR 2179) ex-
hibits a series of vertical ridges and grooves on the paradental
plates similar to (though less pronounced than) those present in
Splenial This is a thin, roughly triangular bone with a
rounded dorsal apex (Fig. 4). The roughened anterior border
bears a small, oblong fossa at the anteroventral corner of the
medial surface. The surface of this fossa is roughened, but it
seems to be too posterior to have contributed to the symphysis;
its associations remain unclear. The preserved ventral border is
slightly concave and lacks a splenial foramen, in contrast to
most other theropods. The ventral edge is thin anteriorly but
widens posteriorly to form a dorsally concave shelf for articu-
lation with the angular. The posterior edge is gently concave
posteriorly where it forms the border for the internal mandibular
fenestra, comparable to that of most non-tetanuran theropods.
514 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 22, NO. 3, 2002
FIGURE 5. Teeth of Masiakasaurus knopﬂeri. Anterior dentary or
premaxillary tooth (FMNH PR 2180) in medial/lateral (A), posterolat-
eral (B), and posterior (C) views. Scanning electron micrographs of
anterior dentary or premaxillary tooth (D; FMNH PR 2200) in postero-
lateral view, and a slightly more posterior tooth (E, FMNH PR 2182)
in posterior view. Abbreviations:c, carina; s, striations. Scale equals
The edge is broken ventral to this point. Both the medial and
lateral surfaces are smooth.
Angular(?) FMNH PR 2166 is a thin, curved fragment that
may pertain to the angular. On the apparent medial surface, the
element is convex and smooth, whereas it is concave laterally
and bears a narrow, longitudinal facet ventrally that is bounded
by a thin ridge. It is thinner anteriorly than posteriorly, and the
ventral edge is rounded and thickened.
Teeth As noted above, the anteriormost dentary teeth are
unlike those of any other known theropod in being extremely
procumbent, with the ﬁrst tooth lying nearly horizontal in the
jaw (Fig. 3). Successive teeth splay out both laterally and an-
teriorly from the jaw, as evidenced by the orientations of alveoli
2–5. The ﬁrst four dentary teeth are also unusual in being weak-
ly spatulate rather than mediolaterally compressed (coelophy-
soids, abelisaurids, basal tetanurans) or incisiform (allosauroids,
tyrannosaurids). The faint serrations on these teeth are arranged
along a pair of carinae, but these carinae are located on the
medial and lateral sides of the tooth rather than on the anterior
and posterior edges (Fig. 5). The carinae converge toward the
posteriorly-hooked apex, and numerous longitudinal striations
are present between them on the posterior surface. The apex of
the tooth is rounded rather than acuminate.
More posteriorly, the dentary teeth are increasingly similar
to those of other theropods, becoming laterally compressed and
recurved and bearing more pronounced serrations. The two ca-
rinae migrate in position to the anterior and posterior edges of
the tooth. The anterior carinae in more posterior teeth extend
only halfway down the crown, whereas the posterior carinae
extend along its entire length. The serrations are pronounced
on the posterior carina and weak to absent on the anterior ca-
rina. Representative teeth are not known for all tooth positions,
but those posterior to the ﬁfth position possess this more typical
theropod morphology. Although they remain more inﬂated lat-
erally than medially, these teeth are not procumbent or splayed
outward to any appreciable degree.
The maxillary dentition is represented by two in situ teeth,
located in the third and ﬁfth alveoli, but only that in the ﬁfth
can be described. It appears to be typical of most theropods and
resembles the morphology of posterior dentary teeth in being
mediolaterally compressed, recurved, short, and bearing pro-
nounced serrations. As described above, the morphology of the
anterior maxillary alveoli indicate that these teeth were increas-
ingly procumbent and, in the case of the ﬁrst tooth position,
circular in cross-section. Although the premaxilla of Masiaka-
saurus is unknown, based on the maxillary morphology we pos-
tulate that the premaxillary dentition was at least as procumbent
as that of the anterior maxilla.
The total number of vertebrae is not known in Masiakasau-
rus, and accurate vertebral counts are difﬁcult to obtain for most
theropods, particularly when the caudal series is considered.
However, the primitive dinosauromorph condition includes 24
presacral vertebrae (partitioned into 10 cervicals and 14 dorsals
or 9 cervicals and 15 dorsals) and two sacrals. These 24 pre-
sacrals are retained in theropods, but a variable number of dor-
sals are usually incorporated into the sacral series. The esti-
mated positions of isolated vertebrae of Masiakasaurus are
based on the assumption that it retained this basic vertebral
All of the pre-caudal vertebrae are characterized by a rela-
tively spacious and ﬂat-bottomed neural canal whose diameter
is more than 50% the diameter of the centrum. As in most other
abelisauroids, the neural arches of the pre-caudal vertebrate are
highly pneumatized, whereas the centra are not. Although it is
possible that neural arch pneumaticity extended into the cen-
trum via the pedicles, as it does in Majungatholus, there is no
direct evidence of this on any of the preserved vertebrae. Dam-
aged dorsal and caudal vertebral centra display an internal ca-
merate structure that is more consistent with spongy, trabecular
bone than pneumaticity.
Cervical Vertebrae The three cervical vertebrae recovered
are weakly amphicoelous, with anteriorly-placed diapophyses
and relatively elongate centra. The more anterior cervical
(FMNH PR 2140) is relatively shorter than more posterior ones
(FMNH PR 2139, 2141; Fig. 6), but the morphology among
them is quite consistent. The neural spines are highly abbrevi-
ated in both length and height and are located over the anterior
half of the vertebra. In dorsal view, the postzygapophyses and
epipophyses sweep posteriorly from the neural spine almost
twice as far as the prezygapophyses extend anteriorly. The epi-
pophyses themselves are quite small and do not extend poste-
riorly beyond the postzygapophyses. The four principal saur-
ischian laminae (prezygodiapophyseal lamina/prdl, postzygo-
diapophyseal lamina/podl, anterior centrodiapophyseal lamina/
acdl, posterior centrodiapophyseal lamina/pcdl) are present, in
addition to centroprezygapophyseal (cprl), centropostzygapo-
515CARRANO ET AL.—OSTEOLOGY OF MASIAKASAURUS KNOPFLERI
FIGURE 6. Posterior cervical vertebra (FMNH PR 2141) of Masiak-
asaurus knopﬂeri in right lateral (A), anterior (B), posterior (C), dorsal
(D), and ventral (E) views. Abbreviations:dp, diapophysis; ep, epi-
pophysis; f, foramina; nc, neural canal; ns, neural spine; poz, postzy-
gapophysis; pp, parapophysis; prz, prezygapophysis. Scale equals 1 cm.
physeal (cpol), and both spinal laminae (spinoprezygapophy-
seal/sprl, spinopostzygapophyseal/spol) (Wilson, 1999).
The parapophyses are situated at the anteroventral corner of
the centrum. The left parapophysis is broken in FMNH PR
2140, revealing a space that seems to communicate with the
centrum interior. In FMNH PR 2141, a single pedicle rises
steeply from the centrum and diverges into the cprl and the
acdl. A dorsally-directed foramen between the two opens into
a larger chamber beneath the prezygapophysis, which itself
opens anteriorly between the two laminae. The prezygapophy-
seal facet faces about 45
dorsomedially with little anteropos-
terior deviation. Its contact surface is ﬂat and rectangular, with
the exception of a modest triangular ﬂange that extends poste-
riorly. A wide infrapostzygapophyseal fossa is present between
the pcdl and cpol, housing a variable number of foramina that
open internally or into the infradiapophyseal fossa.
In the more posterior cervical (FMNH PR 2139), the pre-
zygapophyseal facets face slightly posteriorly but mainly dor-
somedially. The postzygapophyseal facets face ventrolaterally
and slightly anteriorly, and bear a small, rectangular posterior
projection. The neural spine remains mediolaterally narrow and
lacks obvious scars for interspinous ligaments. The epipophysis
increases slightly in size but still does not extend posteriorly
past the postzygapophysis. Changes also occur within the fos-
sae: the left and right sprl delimit a large midline fossa con-
taining a foramen that communicates with the hollow interior
of the postzygapophysis, and the large infrapostzygapophyseal
fossa becomes twice as large as the infraprezygapophyseal fos-
sa. The more complete diapophysis in FMNH PR 2139 appears
to have been ventrally pendant. Its dorsolateral surface is ﬂat
and continuous with the lateral surface of the prezygapophysis.
Although the cervical centra may be perforated by variable
numbers (0–2) of small foramina, potentially pneumatic fossae
are entirely lacking; therefore these foramina are interpreted as
vascular and non-pneumatic. In contrast, the cervical neural
arches bear the marks of extensive pnematization, both exter-
nally and internally, including a highly variable set of fossae
and foramina. For example, in FMNH PR 2139 there are two
foramina within the infrapostzygapophyseal fossa, one below
the postzygapophysis, and another more ventral that faces pos-
terolaterally. An additional foramen enters the base of the left
pcdl. Bilateral foramina enter the base of the postzygapophysis
from within the postspinal fossa, and small foramina pierce the
triangular fossa between the cpol and intrapostzygapophsyeal
lamina (tpol). A foramen is present dorsomedial to the right
prezygapophysis, but exists only as a fossa on the left side.
The posterior cervical vertebrae of Masiakasaurus lack any
elevation of the anterior face over the posterior face, suggesting
that the base of the neck was relatively straight. They differ
from those of coelophysoids in being taller and wider relative
to their length, bearing larger epipophyses, and having antero-
posteriorly shorter neural spines. In contrast, these cervicals are
particularly similar to those of abelisauroids (e.g., Majungath-
olus,Carnotaurus,Ilokelesia) in the size and shape of the neu-
ral spine and the disposition of the zygapophyses (Novas,
1992). However, the neural spine and diapophyses are more
anteriorly placed, the prezygapophyses and epipophyses are
more posteriorly placed, the posterior articular surface is not
concave, and the epipophyses are considerably less developed
in Masiakasaurus. Like abelisaurids and most other basal the-
ropods (but unlike coelurosaurs), the cervical centra zygapo-
physeal facets of Masiakasaurus are not ‘‘ﬂexed’’ and exposed
anteriorly. The three cervicals from the Late Cretaceous Lameta
Group of India ascribed to Laevisuchus indicus (Huene and
Matley, 1933) are similar to those of Masiakasaurus in most
respects (Fig. 7), as is the cervical neural arch of Noasaurus.
All three display short, anteriorly-placed neural spines, and
postzygapophyses that are swept back strongly posteriorly.
Dorsal Vertebrae Several isolated dorsal vertebral centra
are known, identiﬁed primarily by their proportions and lack of
a parapophysis on the centrum. All are incomplete and presum-
ably from juveniles because of their relatively small size (com-
pared to the cervicals and caudals) and unfused neurocentral
sutures (Brochu, 1996). They are spool-shaped, weakly amphi-
coelous, and lack foramina (Fig. 8). Unlike the dorsals of Ma-
jungatholus, the centra are not shortened but remain anteropos-
teriorly elongate (approximately twice as long as either wide or
tall), as in most coelophysoids, smaller theropods, and Ela-
phrosaurus. The centra are rounded or rounded-rectangular in
cross-section. Broken specimens reveal a layer of spongy bone
beneath the outer cortical shell, and a hollow central chamber
without signiﬁcant partitions.
A single posterior dorsal neural arch (FMNH PR 2144) is
preserved, associated with (but retaining no contact with) a par-
tial dorsal centrum (FMNH PR 2145). The arch is from a sub-
adult, having separated from its centrum along the neurocentral
suture. As in most theropod dorsals, the prezygapophyses are
positioned at the end of prominent, parallel pedicles that project
nearly directly anteriorly. Their articular surfaces are convex,
with a dorsal prezygapophyseal component and a medial hy-
posphene component. These complement the concave postzy-
gapophyses, which face ventrally and have laterally-facing hy-
pantra. The postzygapophyses are wider than long and project
only slightly beyond the posterior edge of the pedicles. The
eight laminae observed in the cervical neural arches are present
516 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 22, NO. 3, 2002
FIGURE 7. Cervical vertebrae of Masiakasaurus knopﬂeri (A,B;
FMNH PR 2140), Laevisuchus indicus (C,D; ISI K27/696), and Noas-
aurus leali (E,F; PVL 4061) in right lateral (A,C,E) and dorsal (B,
D,F) views. Dashed lines indicate reconstructed element outlines. Note
that the neural arch of ISI K27/696 was broken and reattached incor-
rectly, and has been digitally reversed for this illustration. Photographs
of Noasaurus courtesy of F. E. Novas. Scale equals 1 cm.
FIGURE 8. Dorsal vertebral centrum (FMNH PR 2113) and neural
arch (FMNH PR 2144) of Masiakasaurus knopﬂeri in left lateral (A),
dorsal (B), anterior (C), and ventral (D) views. Abbreviations:dp,
diapophysis; ha, hypantrum; hs, hyposphene; isl, interspinous ligament
scar; nc, neural canal; ncs, neurocentral suture; posf, postspinal fossa;
poz, postzygapophysis; pp, parapophysis; prz, prezygapophysis. Scale
equals 1 cm.
here as well. The arch is highly pneumatized, with foramina
piercing the infraprezygapophyseal, infradiapophyseal, and in-
frapostzygapophyseal fossae. These fossa may communicate
with the deep postspinal fossa (situated between the postspinal
laminae) through the arch interior. The relatively tall neural
spine is oriented nearly vertically, and occupies approximately
the posterior half of the arch. Interspinous ligament scars reach
nearly to the rounded top of the spine.
Sacral Vertebrae A single sacral series is known (FMNH
PR 2142), comprising three fused centra with partial neural
arches (Fig. 9). This series is incomplete, so the total sacral
count is unknown. Each vertebra is somewhat shorter antero-
posteriorly than tall dorsoventrally and, as a unit, the three co-
alesced sacrals are approximately twice as long as tall. The
centra are almost D-shaped, being somewhat ﬂattened dorsally,
and their exposed faces are ﬂat rather than concave. The cir-
cular neural canal is nearly two-thirds the area of the centrum.
A large intervertebral foramen opens into the canal within the
anterior half of the ﬁrst two vertebrae, and a much smaller
foramen is present in the third. As in most theropods, the ‘‘in-
tervertebral’’ foramina are actually enclosed within the neural
arch of each sacral vertebra, just anterior to the transverse pro-
cesses (which are preserved only at their bases). An additional
posterior foramen is present on the last sacral.
Each of the transverse processes is connected to a vertical
spinodiapophyseal (spdl) lamina that ascends the lateral face of
the neural spine. A second spinal lamina descends more ante-
riorly, terminating at the posterior border of the vertebral fo-
ramen. The three preserved neural spines are coalesced into a
single, sheet-like structure. In contrast to most other theropods,
the neural spines have been extensively pneumatized. In each
neural spine, the central strut (spdl) is ﬂanked by anterior and
posterior chambers, the former accessed by a large lateral fo-
ramen. Another foramen connects the anterior chamber of one
neural spine to the posterior chamber of the preceding element.
A short horizontal shelf extends laterally near the top of the
spines, adjacent to a continuous, low central ridge. This ‘‘spine
table’’ is also hollow, with a dorsal foramen that communicates
with the interior of the central strut.
Caudal Vertebrae Twenty-four isolated caudals are pre-
sent from all regions of the tail. The caudals have relatively
smaller neural canals than the more anterior vertebrae. There
are no foramina or fossae invading the centra; in fact, there is
no evidence of pnematization within the caudal series. All cau-
dal centra are amphicoelous and have protruding ventral chev-
ron facets. The ventral midline is concave and ﬂanked by weak
keels except in the central region of the tail.
Only one anterior caudal is known (FMNH PR 2133), prob-
517CARRANO ET AL.—OSTEOLOGY OF MASIAKASAURUS KNOPFLERI
FIGURE 9. Sacral vertebrae (FMNH PR 2142) of Masiakasaurus knopﬂeri in left lateral (A; stereopair), ventral (B), and dorsal (C) views.
Sacrals numbered as preserved from anterior to posterior. Abbreviations:cs, central strut; ivf, intervertebral foramen; ns, neural spine; st, spine
table; tvp, transverse process. Scale equals 1 cm.
ably one of the ﬁrst ﬁve. Both faces are elliptical and tall, with
a slightly ﬂattened ﬂoor of the neural canal. The transverse
processes are directed posterodorsally and extend to the pos-
terior centrum edge, but its true lateral extent remains uncertain
due to existing damage. The zygapophyses are rounded quad-
rilaterals oriented at about 30
from vertical. The L-shaped neu-
ral spine rises nearly vertically over the posterior half of the
vertebra, with the anterior portion rising half as high as the
posterior portion. The anterior edge ends abruptly at a shallow
prespinal fossa; the postspinal fossa is deeper and mediolater-
ally narrow. The spine has been broken and may have been
taller, comparable to the condition in most other theropods.
With regard to laminae, the acdl, prdl, podl, sprl, and spol are
present, but the pcdl is absent. This anterior caudal is similar
to those of Majungatholus and Carnotaurus in having very
steeply dorsally angled transverse processes. The centrum is tall
and long relative to its width, as in Majungatholus. Most non-
abelisauroid theropod caudals are considerably shorter relative
to their width and have more horizontally-oriented transverse
The mid-caudals have elliptical centrum faces that are slight-
ly broader than tall (Fig. 10A). The neural canal is round rather
than ﬂat-bottomed. The prezygapophyses extend anteriorly past
the anterior border of the centrum for nearly half their length,
with nearly vertical, medially-oriented facets. The prezygapo-
physes do not extend laterally beyond the edge of the centrum.
The transverse process is located on the posterior half of the
centrum, with the distal half of the process swept approximately
posteriorly. The neural spine is low and anteroposteriorly
long, situated directly above the transverse processes, and con-
nected to a faint, anteriorly-directed ridge. The postzygapophy-
seal facets are nearly vertical and somewhat posteriorly orient-
ed. A low, triangular bump extending posteriorly from the pos-
terior margin of the postzygapophysis may be the epipophysis.
All laminae are lacking from this and successively more distal
Several specimens represent more distal mid-caudals. The
transverse process of FMNH PR 2163 is actually a small pro-
trusive lateral ridge located near the posterior end of the cen-
trum, indicating that it is a more anterior vertebra than the re-
maining elements. The neural spine is weakly evident along the
posterior dorsal midline, but the epipophyses are absent. The
zygapophyses are oriented similarly to those of FMNH PR 2126
and, like them, extend anteroposteriorly beyond the bounds of
the centrum. As in most theropods, a double ventral keel is
present on these centra (e.g., FMNH PR 2127) but is usually
The distal caudals are elongated relative to the more proximal
elements, and ridges are present in place of the transverse pro-
cesses and neural spine (Fig. 10B). The postzygapophyses are
relatively shorter than in more anterior caudals, eventually ex-
tending only as far posteriorly as the centrum. In contrast, the
prezygapophyses are quite long and thin, with up to two-thirds
of the process extending anteriorly in front of the centrum. A
weak double ventral keel is present on most of these elements.
More distally, the neural canal becomes smaller and the cen-
trum more box-like.
No cervical or dorsal ribs, haemal arches, or gastralia are
known for Masiakasaurus.
All preserved limb bones (humerus, femur, tibia, metatarsals)
are hollow and relatively slender in dimensions, comparable to
those of coelophysids.
Humerus The proximal two-thirds of the slender, straight
518 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 22, NO. 3, 2002
FIGURE 10. Medial (FMNH PR 2126; A,C,E) and distal (FMNH
PR 2202; B,D,F) caudal vertebrae of Masiakasaurus knopﬂeri in left
lateral (A,B), dorsal (C,D), and ventral (E,F) views. Abbreviations:
cf, chevron facet; epi?, epipophysis?; ns, neural spine; nsr, neural spine
ridge; poz, postzygapophysis; prz, prezygapophysis; tvp, transverse
process. Scale equals 1 cm.
FIGURE 11. Right humerus (FMNH PR 2143) of Masiakasaurus
knopﬂeri in anterior (A), medial (B), posterior (C), lateral (D), and
proximal (E) views. Abbreviations:dp, deltopectoral crest; gt, greater
tubercle; hh, humeral head; it, internal tuberosity. Scale equals 1 cm.
humerus is known (FMNH PR 2143; Fig. 11), the only example
among small-bodied abelisauroids. The only perceivable cur-
vature on this element is a slight medial concavity, although
this may be exaggerated by the protruding internal tuberosity.
The rounded humeral head occupies most of the proximal end,
ﬂanked laterally and medially by the small greater tubercle and
internal tuberosity, respectively. The head slightly overhangs
the shaft anteriorly. In proximal view, the head is somewhat
oblong with the long axis oriented mediolaterally. The internal
tuberosity is contiguous with two sizable rugose regions, one
on the posterior surface and another medially. Two small fo-
ramina sit within a lateral sulcus just distal to the humeral head.
The deltopectoral crest is rugose and elevated from the shaft in
mediolateral view, projecting mostly anteriorly. It is approxi-
mately as tall as long, and is distinct from the humeral head
and internal tuberosity. The shaft of the humerus narrows to a
minimum near the base of the deltopectoral crest, and ﬂares
again towards the (missing) distal end.
The humerus of abelisaurids (Carnotaurus,Majungatholus)
bears a similarly rounded head, but in these taxa the bone is
much shorter and more robust than in Masiakasaurus.Inad-
dition, the deltopectoral crest in abelisaurids (and in more basal
theropods) is less distinct and tends to be conﬂuent with the
humeral head. That of Elaphrosaurus is somewhat intermediate
in dimensions and crest development between these two forms.
The deltopectoral crest of Masiakasaurus somewhat resembles
that of Deltadromeus in its placement but differs in its relatively
Manus Portions of the manus represent the only other fore-
limb elements of Masiakasaurus known. These include several
manual phalanges of uncertain identiﬁcation. Two (FMNH PR
2132, 2224) may represent I-1, as they are extremely short and
lack collateral ligament pits. The proximal articular surface is
asymmetrically biconvex and bears two prongs, the ventral be-
ing far more prominent than the dorsal. In contrast, the distal
articular surface is extensive and laps onto the dorsal surface
of the bone, suggesting that the successive phalanx enjoyed
considerable hyperextension. Another (FMNH PR 2225) is rel-
atively longer, and lacks the dorsal extension of the distal ar-
ticular surface. However, it is similar to the preceding in the
morphology of the proximal prongs and asymmetrical proximal
concavities. Several other theropod phalanges are referred to
this taxon but cannot be assigned further until more complete
materials are available. The two fragmentary manual unguals
(FMNH PR 2136, 2169) are mediolaterally compressed and re-
519CARRANO ET AL.—OSTEOLOGY OF MASIAKASAURUS KNOPFLERI
FIGURE 12. Manual ungual (FMNH PR 2136) of Masiakasaurus
knopﬂeri in lateral (A) and medial (B) views. Abbreviations:bg,
‘‘blood’’ groove; dbg, dorsal ‘‘blood’’ groove; dp, dorsal process; ft,
ﬂexor tubercle; vbg, ventral ‘‘blood’’ groove. Scale equals 5 mm.
FIGURE 13. Left pubis (FMNH PR 2108) of Masiakasaurus knopﬂeri in lateral (A), medial (B), posterior (C), anterior (D), and proximal (E;
stereopair) views. Abbreviations:ac, acetabular region; amb, origin of M. ambiens; ilc, contact with ilium; isc, contact with ischium; li, lateral
inset of boot; of, obturator foramen; pa; pubic apron; pb, pubic boot; ps, pubic symphysis. Scale equals 2 cm.
curved, with two vascular (‘‘blood’’) grooves on one side and
a small ﬂexor tubercle (Fig. 12).
Pelvic Girdle Only the pubis is known, represented by two
specimens (FMNH PR 2108, 2109), one of which (FMNH PR
2108; Fig. 13) is virtually complete. The pubis has a long, slen-
der shaft with a modest distal expansion. The iliac peduncle is
unusual in containing a deep, ﬂat-bottomed socket with steep,
straight sides that received the pubic peduncle of the ilium.
Presumably the pubic peduncle of the ilium was complemen-
tarily peg-shaped, as it is in Majungatholus,Ceratosaurus (Britt
et al., 1999, 2000) and probably Carnotaurus. Overall, the iliac
peduncle is slightly longer anteroposteriorly than it is wide me-
diolaterally. The acetabular region is nearly as large as the iliac
contact but more rectangular and only gently concave. The tri-
angular ischial contact is vertical and angled slightly poster-
oventrally, at about 90
to the acetabular margin.
A relatively large, oval obturator foramen pierces the pubis
below the acetabulum, set into a thin ventral ﬂange. No other
openings are evident. The lower margin of the ventral ﬂange is
continuous with a low ridge that in turn becomes conﬂuent with
the pubic apron. A shallow fossa is present just distolateral to
the iliac peduncle, probably marking the origination of M. am-
biens (Romer, 1923; Hutchinson, 2001a). In lateral aspect, the
pubis is gently curved, being slightly concave along the pos-
teroventral margin. In medial view the pubis is smooth and
nearly featureless, except for a shallow fossa along the dorsal
rim of the obturator foramen. The medial surface develops a
laminar ﬂange distally, giving the bone a teardrop-shaped cross-
section. This medial lamina develops a contact surface for the
lamina from the opposing pubis. Together, these laminae form
the pubic apron, which ran continuously to the symphysis at
the distal end. The exact contact point between left and right
pubes is difﬁcult to determine, but the pubic apron appears to
have occupied between one-half and three-ﬁfths of the total
length of the pubic shaft. Longitudinal scars on the anterior
surface of the apron (FMNH PR 2108) probably represent the
origination of M. puboischiofemoralis medialis (Hutchinson,
This distal end is enlarged into a relatively small, rounded
‘‘boot’’ that projects posteriorly from the main shaft axis. A
shallow, wide furrow runs between the pubic apron and a weak-
er ridge that runs proximally from the posterior border of the
boot. The posterior half of the boot is inset medially from the
lateral shaft, forming a second shallow furrow that extends onto
the shaft. The distal surface of the boot is ventrally convex.
This boot generally resembles that of Carnotaurus (Bonaparte
et al., 1990) in being laterally inset and posteriorly-projecting.
In most other theropods, the boot is either small, lobular and
unremarkable (coelophysoids), enlarged anteroposteriorly (al-
losauroids), or lacks an inset (most coelurosaurs).
Femur The femur is the most abundantly preserved ele-
ment, represented by 13 specimens of varying size (Table 1).
The femoral shaft is bowed strongly anteriorly and more subtly
medially (Fig. 14). The midshaft cross-section is triangular with
the apex pointing anteriorly, becoming rectangular distally. The
midshaft cortices are also much thicker than those distally. A
pronounced epicondylar crest runs along the distal one-fourth
of the anteromedial edge of the bone, bearing a series of parallel
ridges along its posteromedial surface. As a result, the femur
is much wider distally than at midshaft and bears a broad, de-
pressed anterior intercondylar groove for the origin of M. fe-
520 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 22, NO. 3, 2002
TABLE 1. Measurements of femora and tibiae of Masiakasaurus
knopﬂeri. Categorization of a bone as ‘‘gracile’’ or ‘‘robust’’ is based
on prominence of muscle scars and/or presence of fusion between ele-
anteroposterior diameter, c
mediolateral diameter, r
robust scars, w
fusion absent, ND
Specimen l ap ml c scars fusion
FMNH PR 2115
FMNH PR 2117
FMNH PR 2120
FMNH PR 2123
FMNH PR 2148
FMNH PR 2149
FMNH PR 2150
FMNH PR 2153
FMNH PR 2208
FMNH PR 2215
FMNH PR 2112
FMNH PR 2116
FMNH PR 2118
FMNH PR 2119
FMNH PR 2121
FMNH PR 2122
FMNH PR 2152
FMNH PR 2214
morotibialis (Hutchinson, 2001b; Carrano and Hutchinson,
2002). The medial epicondylar crest is similarly well-developed
in Carnotaurus,Genusaurus,Elaphrosaurus, and a femur from
the Tendaguru formation of Tanzania (HMN 68; Janensch,
As in all non-tetanuran theropods, the femoral head is ori-
anteromedially. A shallow groove runs anteroposte-
riorly across the middle of the proximal articular surface, and
likely articulated beneath the iliac supraacetabular shelf as in
many basal theropods (Padian and Olsen, 1989). In anteropos-
terior view, the dorsal margin of the femoral head is angled
slightly ventrally rather than horizontally. On the posterior sur-
face, a deep, non-articular sulcus runs dorsoventrally and dis-
tinguishes a medial lobular projection that contributes to the
ventral overhang of the head. The large, low bulge that forms
the greater trochanter is conﬂuent with the femoral head and
probably served as the insertion of Mm. puboischiofemorales
externi (Hutchinson, 2001a).
The lesser (
anterior) trochanter is elevated above the shaft
and rises to approximately the lower third of the femoral head,
as in abelisauroids and basal tetanurans. It has a complex sur-
face topography that probably reﬂects the attachments of two
deep dorsal muscles, Mm. iliotrochanterici caudalis (proximal-
ly) and medialis/cranialis (distolaterally) (Carrano, 2000;
Hutchinson, 2001b; Carrano and Hutchinson, 2002). A single
foramen opens distomedial to the base of the anterior trochanter.
As in tetanurans, the lateral edge of this base is connected, at
least in some specimens (e.g., FMNH PR 2208), via a thin ridge
with a more lateral rugose bump that represents the insertion
for M. iliofemoralis externus (remainder of the trochanteric
shelf; Hutchinson, 2001b). Thus a distinct trochanteric shelf is
The fourth trochanter represents the insertion for Mm. cau-
dofemorales brevis (laterally) and longus (medially) (Dollo,
1888; Romer, 1923; Hutchinson, 2001b). Here it is a low ridge
that runs approximately one-ﬁfth the length of the bone, with
its distal end somewhat proximal to midshaft. It shows no ev-
idence of having been pendant. Similarly reduced fourth tro-
chanters are found in other abelisauroids as well as coelophys-
ids and coelurosaurs, contrasting the prominent condition typ-
ical of most other theropods. As in most theropods, a single
foramen is present about midway along the posterolateral edge.
A large, circular fossa—more pronounced in some specimens
(e.g., FMNH PR 2208)—is present anteromedial to the fourth
trochanter, probably for additional attachment of M. caudofe-
On the distal articular surface, the ﬁbular condyle is approx-
imately twice the breadth of the tibial condyle. The posterolat-
eral tuberous process (
crista tibioﬁbularis) is ﬂattened and
protrudes about the same distance posteriorly as the tibial con-
dyle. The intercondylar groove disappears into a shallow pop-
liteal fossa immediately proximal to the tuberous process. The
tuberous process is oblong in posterior view, with a marked
sulcus between its lateral face and the ﬁbular condyle. A similar
sulcus is observed in many other theropods and is less promi-
nent than that of coelophysids (Rowe, 1989). The two distal
condyles diverge posteriorly in distal view.
Two relatively distinct morphs are present within the total
sample of 13 femora, one with more pronounced muscle scars
(e.g., FMNH PR 2208) and the other in which these features
are muted or absent (e.g., FMNH PR 2120). This variation does
not appear to be size-related—similar sized femora may exhibit
either condition (Table 1)—but statistical comparisons reveal
that pronounced scars are associated with femora of more ro-
bust dimensions (though not always signiﬁcantly so; Table 2A).
A similar pattern has been observed in the femora of Coelo-
physis (Colbert, 1990), Syntarsus (Raath, 1990), and the
‘‘Shake-N-Bake’’ taxon (Tykoski, pers. comm.) (see Discus-
Tibia The tibia is long, relatively slender, and straight-
shafted (Fig. 15). At the proximal end, the long axis of the shaft
is directed anteriorly along a pronounced cnemial crest. This
crest arises out of the medial shaft surface and curves laterally
as it extends far anterior to the shaft. In mediolateral view, the
cnemial crest rises above the proximal articular surface, bearing
dorsal and ventral expansions at its distal end. A longitudinal
groove for the knee extensor tendon is present along the lateral
surface of the anteriormost cnemial crest. A similarly developed
cnemial crest is also present in abelisauroids (e.g., Xenotarso-
saurus,Majungatholus,Carnotaurus,Lametasaurus), two the-
ropod tibiae from the Tendaguru formation of Tanzania (HMN
37 and 69; Janensch, 1925), and several tibiae from the Lameta
formation of India (Huene and Matley, 1933). The cnemial crest
is usually less pronounced in most other theropods, and in par-
ticular lacks the lobular expansion.
The two proximal tibial condyles project equally far poste-
riorly, but mirror the corresponding femoral condyles in that
the lateral (ﬁbular) condyle is the larger of the two. The artic-
ular surface of the lateral condyle is nearly planar, whereas that
of the medial condyle slopes upward as it approaches the pos-
terolateral margin. The two condyles are separated posteriorly
by a short rectangular notch, and several small channels are
apparent along the posterior surface of the medial condyle. A
small, elliptical scar on the proximal medial surface is probably
for insertion of M. ﬂexor tibialis internus 3 (e.g., UA 8685;
Carrano and Hutchinson, 2002). The proximal posterior surface
bears a slight rugosity medial to the ﬁbular crest.
The narrow ﬁbular crest sits on the lateral surface and runs
parallel to the main shaft axis, terminating just proximal to a
small gap (perhaps for passage of interosseous neurovascula-
ture) that contains a single foramen. Further distally, the lateral
tibia bears an extensive ﬂattened contact surface for the ﬁbula
521CARRANO ET AL.—OSTEOLOGY OF MASIAKASAURUS KNOPFLERI
FIGURE 14. Left femur (FMNH PR 2123) of Masiakasaurus knopﬂeri in anterior (A), lateral (B), posterior (C), medial (D), proximal (E), and
distal (F) views. Distal left femur (FMNH PR 2115) in posterior (G) view. Abbreviations:aig, anterior intercondylar groove; cﬂf, fossa for M.
caudofemoralis longus; f, foramen; fc, ﬁbular condyle; fcg, ﬁbular condyle groove; 4t, fourth trochanter; gt, greater trochanter; ife, origin of M.
iliofemoralis externus; lt, lesser trochanter; mec; medial epicondylar crest; pag, proximal articular groove; pig, posterior intercondylar groove;
ps, proximal sulcus; tc, tibial condyle; tfc, tibioﬁbular crest. Scale equals 2 cm.
that runs continuously along the shaft and tapers toward the
distal end. This resembles the condition in all theropods more
derived than Herrerasaurus and Eoraptor. The posterior sur-
face of the distal tibia has a rugose, ﬁbrous texture that ends at
the articulation with the astragalus.
The anterior tibia shaft is relatively ﬂat, with two longitudinal
intermuscular lines (lateral and anterior) that delineate the
bounds of the M. tibialis anterior origin (UA 8685; Carrano and
Hutchinson, 2002). The distal anterior surface is somewhat in-
ﬂated and rugose proximal to the astragalar articulation, and the
ascending process sits within a large concave area on this part
of the shaft. The distal end of the tibia is bilobate, with a larger,
ﬂatter, and more distally-projecting lateral malleolus. The distal
tibia is fused to the astragalocalcaneum (although not with the
ﬁbula) in several specimens, a feature associated with more
pronounced muscle scars.
As with the femora, there appears to be no correlation be-
tween the presence of fusion or prominent scars and the overall
size of the animal: several large tibiae (e.g., FMNH PR 2121,
UA 8685) show no signs of fusion although one smaller spec-
imen (FMNH PR 2122) does (Table 1). Nevertheless, there is
a tendency for tibiae with more pronounced muscle scars to be
more robust (again, not always signiﬁcantly so; Table 2A). In
addition, tibiae exhibiting fusion also tend to exhibit more pro-
nounced muscle scars (Table 2B). Thus there appears to be
some basis for segregating these elements (and the femora) into
‘‘robust’’ and ‘‘gracile’’ morphs.
Fibula Only the distal end is preserved (Fig. 16), but the
ﬁbula was clearly slender and only slightly shorter than the tibia
(FMNH PR 2116). The extensive articular facets for the ﬁbula
on the tibial shaft indicate that these two bones were closely
appressed for most of their lengths. The distal ﬁbula ﬂares be-
fore contacting the proximal tarsals, and is approximately as
wide mediolaterally as anteroposteriorly. There is a broad distal
contact surface with the calcaneum, and a small portion of the
distal ﬁbula articulates with the body of the astragalus, abutting
and fusing to the lateral edge of the ascending process.
Tarsus The preserved astragali of Masiakasaurus are part-
ly or entirely fused to the tibia, ﬁbula, and calcaneum (Fig. 16);
as a result, the lateral and proximal astragalar surfaces are not
visible. The astragalus is much larger than the calcaneum, and
caps the entire distal tibia as well as the medial half of the
522 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 22, NO. 3, 2002
TABLE 2. Statistical comparisons between femoral and tibial morphs of Masiakasaurus knopﬂeri.A. Results of student’s unpaired two-group
t-test and Mann-Whitney U-test comparing robusticity measures for elements with pronounced and weak muscle scars. B. Results of chi-square
test on the co-occurrence of tibial fusion and pronounced rugosities. ap
anteroposterior diameter, c
mediolateral diameter, N(w)
number of elements with weak scars, N(p)
number of elements with pronounced scars, t
t-test value, p
signiﬁcance level, x(p)
mean of elements with pronounced scars, x(w)
mean of elements with weak scars,
chi-square value, Z
A. t-test and U-test
Variable x(w) x(p) N(w) N(p) t p Z p
B. Chi-square test
Weak Pronounced Totals
tibia fusion by rugosity 8 4.444 0.0350*
distal ﬁbula. Its hourglass-shaped distal articular surface is ori-
ented anteroventrally when articulated, and bears a single small
foramen located towards the anterior edge of the distal surface.
A weak groove crosses the anterior face of the astragalus dis-
tolaterally, probably marking the transit of a vascular structure.
Although the proximal astragalar surface is not visible, it is
clear from FMNH PR 2116 that the tibial articulation extends
posterior to the ﬁbula, similar to the condition in tetanurans but
differing from that in coelophysoids.
The base of the ascending process is concave anteriorly, and
bears the small foramen common in archosaurs. The ascending
process is difﬁcult to distinguish from the tibial shaft, but it
appears to be relatively tall and rectangular in shape. The
oblique proximal edge is highest at its contact with the ﬁbular
shaft. More distally, the process is fused to the anterior tibia,
but at its base it is distinct from both the tibia and ﬁbula. It is
similar in shape and extent to that of Xenotarsosaurus. In con-
trast, the ascending process is triangular in Herrerasaurus and
coelophysoids (e.g., Coelophysis,Syntarsus,Dilophosaurus,
Liliensternus), is taller and tapered apically in allosauroids (e.g.,
Allosaurus,Sinraptor), and is considerably taller and more lam-
inar in coelurosaurs (e.g., Tyrannosaurus,Struthiomimus).
In fused specimens (e.g., FMNH PR 2116), it is not possible
to identify the anterior suture between the astragalus and cal-
caneum with certainty, but the suture is clearer on the posterior
surface. The calcaneum appears to form approximately the lat-
eral one-ﬁfth of the proximal tarsal block (Fig. 15). An isolated
calcaneum (FMNH PR 2235) reveals that the contact with the
astragalus is roughened and bears two small knobs (Fig. 17).
The concave ﬁbular facet is rhomboidal and faces dorsally. The
tibial facet is relatively large, rectangular, and opens medially.
It faces posterodorsally and is separated from the ﬁbular facet
by a thin, oblique ridge that runs posteromedial–anterolateral.
Ventrally, the articular surface is smooth and has a similar pro-
ﬁle to that of Sinraptor,Torvosaurus, and Allosaurus. The lat-
eral calcaneal surface has a central depression that is surround-
ed by raised edges, and bears several small foramina. The distal
tarsals are not known.
Pes The pes of Masiakasaurus is generally rather slender,
a feature particularly evident in the metatarsals. It shows none
of the pedal specializations evident in coelurosaurs (e.g., Holtz,
1994a), and follows a generalized theropod design. Although
metatarsals I and V are lacking, much of the remainder of the
pes has been recovered.
Metatarsal II has unusual proportions (Fig. 18). The proximal
shaft is narrow, only one-third the thickness of the distal end,
but the most proximal part is slightly expanded and bears a ﬂat,
elliptical surface for contact with the distal tarsals. The ﬂat me-
dial surface extends to the distal end, suggesting that it abutted
against metatarsal III for nearly its entire length. The distal
articular surface is more typically theropod in size and shape,
bearing a double-ﬂanged condyle of which the lateral ﬂange is
substantially larger than the medial. The ginglymus and collat-
eral ligament pits are well developed, with the lateral pit par-
ticularly pronounced. A small, ﬂat facet on the posterior prox-
imal end may represent the articular surface for (unpreserved)
metatarsal I (Tarsitano, 1983). Further distally along the sharp
posterior edge, there is a small facet for insertion of M. gas-
trocnemius medialis (Carrano and Hutchinson, 2002).
A mediolaterally reduced metatarsal II is also present in the
enigmatic Argentine theropod Velocisaurus unicus (Bonaparte,
1991a), but the shaft is more rounded than in Masiakasaurus.
Although the full length of the bone is not preserved in the
holotype and only specimen, the proximal and distal ends were
found in articulation with metatarsal III, revealing that the sec-
ond metatarsal of Velocisaurus was considerably shorter and
more slender than metatarsal III. Metatarsal II of Noasaurus is
nearly identical in proportions and morphology (Bonaparte and
Powell, 1980) to that of Masiakasaurus: the distal end retains
a more typical size and shape, whereas the proximal half of the
shaft is signiﬁcantly reduced and ﬂattened. In addition, Ela-
phrosaurus has a proximally reduced metatarsal II that, like
Masiakasaurus, retains typical theropod proportions at the dis-
tal articular surface.
Metatarsal III is straight and relatively slender, lacking the
mediolateral curvature present in most theropods (e.g., Allosau-
rus,Sinraptor,Dilophosaurus) (Fig. 19). The proximal articular
surface is rectangular but not hourglass-shaped, although there
is some suggestion of invaginations along the medial and lateral
edges. It resembles the condition in coelophysoids and Herrer-
asaurus more than that in tetanurans such as Allosaurus.A
shallow hyperextensor pit is present, as well as deep collateral
ligament pits. As in most theropods, the proximal contact sur-
face for metatarsal II is larger and deeper anteroposteriorly than
that for metatarsal IV. The contact surface for metatarsal II is
patent and extends approximately two-thirds of the way down
the length of the shaft; that for metatarsal IV is less clear.
523CARRANO ET AL.—OSTEOLOGY OF MASIAKASAURUS KNOPFLERI
FIGURE 15. Left tibia, ﬁbula, and proximal tarsals (FMNH PR 2122) of Masiakasaurus knopﬂeri in anterior (A), lateral (B), posterior (C), and
medial (D) views. Proximal left tibia (FMNH PR 2118) in lateral (E) and proximal (G) views. Unfused distal left tibia in anterior view (F; FMNH
PR 2119), and unfused distal right tibia in distal view (H; FMNH PR 2112). Abbreviations:as, astragalus; ca, calcaneum; cn, cnemial crest; f,
foramen; fap, facet for ascending process; fc, ﬁbular crest; ff, ﬁbular facet; ﬁ, ﬁbula; g, gap; icg, intercondylar groove; keg, knee extensor groove;
lc, lateral condyle; lf, lateral fossa; lm, lateral malleolus; mc, medial condyle; mm, medial malleolus; ms, muscle scar. Scale equals 2 cm.
Metatarsal IV is also long and relatively slender, but not me-
diolaterally constricted (Fig. 20). The proximal half (FMNH PR
2234) was apparently appressed against the lateral surface of
metatarsal III; in this it resembles the metatarsal IV of most
other theropods. The articular surface is D-shaped and lacks the
postermedial process present in most tetanurans, instead resem-
bling the condition in coelophysoids, Ceratosaurus, and Ela-
phrosaurus. As in most theropods, metatarsal IV diverges
markedly laterally along its distal half. Although this diver-
gence is abrupt, it does not appear to be due to deformation.
The mediolaterally compressed distal end (FMNH PR 2214)
bears two distinct articular condyles. The medial condyle is
larger, and its collateral ligament pit more marked, than the
The pedal phalangeal formula cannot be explicitly deter-
mined for Masiakasaurus, but it is 2-3-4-5-x in most other non-
avian theropods. Several phalanges from digits II, III, and IV
are known, including unguals, but no elements from digits I or
V are preserved. The non-ungual phalanges are generally very
similar to those of other small theropods in both shape and
dimensions. Most bear a dorsal hyperextension pit as well as
two collateral ligament pits (one medial and one lateral), of
which the lateral tends to be the deeper of the two.
Phalanx II-1 is long and slightly medially curved, with a
prominent hyperextensor pit for insertion of M. extensor digi-
torum (Carrano and Hutchinson, 2002). In lateral view, the pit
is evident as a sulcus in the dorsal proﬁle of the shaft. The
proximal articular surface bears a single shallowly concave sur-
face for metatarsal II, whereas the distal ginglymus is well-
developed. Proximally, the ventral surface bears two pro-
nounced heels, the medial being the larger of the two. II-2 is
generally similar but relatively shorter and less medially
curved. The proximal articular surface bears two distinct facets
for the distal ginglymus of II-1. A modest dorsal projection and
two ventral heels are present at the proximal end.
Phalanx III-1 is somewhat more stout and symmetrical than
II-1. It also lacks the ventral ‘‘notching’’ of the distal articular
surface, although the proximal articular surface is similarly
weakly concave. Both surfaces are subrectangular. The hyper-
extensor pit is pronounced and triangular, and both collateral
pits are approximately equally excavated. III-2 is similar to III-
1 but relatively shorter for its width. Both the articular surfaces
are subrectangular, although the proximal surface has a slightly
rounded apex, and the hyperextensor pit is triangular but shal-
524 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 22, NO. 3, 2002
FIGURE 16. Drawing (A) and stereopair (B) of left tibiotarsus (FMNH PR 2122) of Masiakasaurus knopﬂeri in anterior view. Abbreviations:
ahg, anterior horizontal groove; ap, ascending process; as, astragalus; ca, calcaneum; ﬁ, ﬁbula; mb, medial buttress; ti, tibia. Scale equals 1 cm.
FIGURE 17. Right calcaneum of Masiakasaurus knopﬂeri (FMNH PR
2235) in dorsal (A), lateral (B), ventral (C), and medial (D) views.
Abbreviations:asc, astragalar contact; dc, distal condyle; fc, ﬁbular
cup; plr, posteromedial–anterolateral ridge; tc, tibial cup. Scale equals
Phalanx IV-1 is short and markedly curved laterally. The dis-
tal articular groove is deep, and the lateral condyle is nearly
twice as large as the medial. The proximal articular surface is
concave, elliptical (with the long axis oriented dorsoventrally),
and bears a small projection at the ventrolateral corner. There
is no clear hyperextensor pit. IV-2 is extremely short but more
symmetrical than IV-1. The proximal articular surface shows
two clear concavities for contact with the proximal part of IV-3.
A hyperextensor pit and both collateral pits are visible. IV-3 is
the shortest phalanx of all those preserved; nearly the entire
bone is composed of the two well-developed articular surfaces,
with little shaft between them. It is otherwise very similar to
The pedal unguals cannot be clearly assigned to particular
digits, but two types are preserved in the collection (Fig. 21).
The ﬁrst (FMNH PR 2129) is relatively mediolaterally com-
pressed (but less so than the manual unguals), whereas the sec-
ond (FMNH PR 2134, 2135) is more ﬂattened dorsoventrally.
Given the morphology of the pedal digits, FMNH PR 2129
probably represents II-3, but the remaining unguals could be-
long to either digit III or IV. In both, the two concavities in the
proximal articular surface are visible. The unguals are only
slightly ventrally curved, but both the dorsal and ventral sur-
faces terminate as edges rather than being ﬂattened. They also
bear an unusual pattern of vascular (‘‘blood’’) grooves along
the medial and lateral surfaces. Instead of the single groove
typical of most theropod unguals, each side exhibits a pair of
grooves that converge distally and meet near the dorsal edge
(Sampson et al., 2001). A weaker groove also runs perpendic-
ular to these two (dorsoventrally) near the proximal end, form-
ing a triangular network. A similar pattern of grooves is also
seen in several abelisaurid pedal unguals from the Lameta For-
mation, India (Huene and Matley, 1933; Novas and Bandyo-
padhyay, 2001), as well as those of Majungatholus, although
these are considerably larger than those of Masiakasaurus.
The skeleton of Masiakasaurus (Fig. 22) exhibits several
characters that clearly diagnose it as a theropod, including the
presence of: (1) recurved, serrated, and laterally compressed
teeth in the maxilla and dentary; (2) curved, claw-like unguals;
(3) markedly hollow vertebral centra and long bones; and (4)
paradental plates (Gauthier, 1986; Sereno et al., 1993; Holtz,
1994b). In order to place this taxon more speciﬁcally within
Theropoda, we performed a phylogenetic analysis of 158 char-
acters distributed across 21 ingroup taxa, using Herrerasaurus
525CARRANO ET AL.—OSTEOLOGY OF MASIAKASAURUS KNOPFLERI
FIGURE 18. Right metatarsal II of Masiakasaurus knopﬂeri (FMNH
PR 2147) and Noasaurus leali (PVL 4061) in dorsal (A), medial (B,
C), proximal (D), lateral (E), ventral (F,G), and distal (H) views. Ab-
breviations:cp, collateral ligament pit; fm1, possible facet for meta-
tarsal I; fm3, facet for metatarsal III; gm, origin of M. gastrocnemius
medialis; hp, ‘‘hyperextensor’’ pit; lf, lateral ﬂange; mf, medial ﬂange;
vs, ventral sulcus. Scale equals 1 cm. C and G modiﬁed from Bonaparte
and Powell (1980).
FIGURE 19. Left metatarsal III (FMNH PR 2147) of Masiakasaurus
knopﬂeri in dorsal (A), lateral (B), ventral (C), medial (D), and distal
(E) views. Abbreviations:cp, collateral ligament pit; hp, ‘‘hyperexten-
sor’’ pit; lf, lateral ﬂange; mf, medial ﬂange; vs, ventral sulcus. Scale
equals 1 cm.
FIGURE 20. Right metatarsal IV (FMNH PR 2214) of Masiakasaurus
knopﬂeri in dorsal (A), lateral (B), ventral (C), medial (D), and distal
(E) views. Abbreviations:fm3, facet for metatarsal III; lf, lateral
ﬂange; mf, medial ﬂange; vs, ventral sulcus. Scale equals 1 cm (A–D)
and Eoraptor as successively more distant outgroups (see Ap-
pendices 2, 3). Although the speciﬁc placement of these taxa
has been disputed, this arrangement is common to all competing
hypothesis relative to our ingroup taxa. 142 characters were
binary and 16 were multistate (unordered); all were equally
weighted. This matrix was analyzed using PAUP 4.0*b10
(Swofford, 2002) under both heuristic and branch-and-bound
options, and character-state transformations were evaluated un-
der both ACCTRAN and DELTRAN optimizations.
This analysis produced 85 most parsimonious trees, each of
255 steps with a CI of 0.68 and a RI of 0.83 (Fig. 23). Strict
consensus analysis of these trees resolves the ingroup taxa into
three main clades: a basal Coelophysoidea ((Coelophysis
Syntarsus)Liliensternus) that is the sister-taxon to Tetanurae
Neotetanurae) plus Neoceratosauria (most
other taxa). Resolution within Neoceratosauria is poor, but a
monophyletic Abelisauridae (Abelisaurus,Xenotarsosaurus,
Carnotaurus and Majungatholus) is recovered, while Elaphro-
saurus and Ceratosaurus are placed basal to all other neocer-
atosaurs. Noasaurids (Noasaurus,Laevisuchus, and Masiaka-
saurus) are united in 88% of these trees, while weaker evidence
suggests that Ilokelesia and Genusaurus are successive sister-
taxa to Abelisauridae.
Our results agree with recent analyses (Rauhut, 1998, 2000;
Carrano and Sampson, 1999; Forster, 1999) in arranging Coe-
526 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 22, NO. 3, 2002
FIGURE 21. Pedal ungual (FMNH PR 2135) of Masiakasaurus knop-
ﬂeri in lateral (A), distal (B), medial (C), and proximal (D) views.
Abbreviations:dbg, dorsal ‘‘blood’’ groove; dp, dorsal process; laf,
lateral articular facet; maf, medial articular facet; pbg, proximal
‘‘blood’’ groove; vbg, ventral ‘‘blood’’ groove; vk, ventral keel. Scale
equals 1 cm.
FIGURE 22. Composite skeletal reconstruction of Masiakasaurus knopﬂeri in left lateral view, based on the holotype and referred ma-
terials. The materials recovered to date represent approximately 40% of the skeleton. Approximate length of adult skeleton is estimated at
lophysoidea and Neoceratosauria as successive sister-taxa toTe-
tanurae, rather than as a monophyletic Ceratosauria (e.g., Gau-
thier, 1986; Rowe, 1989; Holtz, 1994b, 2000; Sereno et al.,
1996); 12 extra steps are needed to acquire the latter arrange-
ment. They also agree generally with previous studies (e.g.,
Rowe, 1989; Holtz, 1994b, 2000) in the relationships of coe-
lophysid taxa. Interestingly, we recover the placement of Dil-
ophosaurus suggested by Rauhut (2000), whereas three addi-
tional steps are required to retain it within a monophyletic Coe-
lophysoidea. Finally, these results further support the hypothe-
sis that Noasauridae and Abelisauridae are united as a clade
(Novas, 1989, 1991, 1992, 1997a; Bonaparte, 1991b).
We performed a heuristic search on 1,000 bootstrap repli-
cates. Neoceratosauria and Neoceratosauria
present in 69% and 78% of these replicates, respectively, while
a traditional Ceratosauria was present in fewer than 5%. Decay
analysis revealed poor support for most nodes within the larger
clades, but moderate decay indices for Abelisauridae (3), Neo-
ceratosauria (3), Neoceratosauria
Tetanurae (3), and Tetanu-
rae (9). The lower support of several nodes within Neocerato-
sauria is largely due to the incomplete nature of Xenotarsosau-
rus,Genusaurus, and Ilokelesia; this also causes many of the
characters supporting neoceratosaurian nodes to be resolved as
ambiguous. The majority of characters supporting Abelisauri-
dae, Abelisauria, or Abelisauroidea are correspondingly ambig-
uous because many of the taxa between these nodes are too
incomplete to score for many of these characters.
Within Theropoda, Masiakasaurus shares a number of fea-
tures with neoceratosaurs (76, 87, 100, 123, 136, 138) and more
inclusive clades within Neoceratosauria (49, 50, 54, 55, 72, 78,
135, 142, 155, 156). In addition, several of the skeletal elements
of Masiakasaurus (as well as those of other neoceratosaurs)
exhibit synapomorphies usually associated with Tetanurae (56,
58, 63, 121, 122, 133, 134, 139, 143–147). These (and many
other) characters strongly support a Neoceratosauria
urae clade in the current analysis. However, along with other
abelisauroids, Masiakasaurus retains symplesiomorphies such
as an anteromedially oriented femoral head, a lobate and ven-
trally directed femoral head, a pubic obturator foramen, a dis-
tally imperforate pubic apron, anteroposteriorly elongate dorsal
vertebral centra, and vertebral centra that lack pneumatic fos-
sae. Such a combination of features supports the placement of
Neoceratosauria as sister-taxon to Tetanurae sensu stricto but
derived relative to Coelophysoidea. This clade (Neoceratosauria
Tetanurae) may or may not currently bear a name; recent
revisions of theropod nomenclature (e.g., Sereno, 1998; Padian
et al., 1999) have produced contradictory results in this regard.
Rather than create a new name at this time (or untangle its
history and priority), we adhere to a moratorium on theropod
nomenclature pending a more detailed phylogenetic revision.
The overall morphological similarities between Masiakasau-
rus,Noasaurus, and Laevisuchus are difﬁcult to evaluate be-
casue the latter two taxa are extremely incomplete. Nonetheless,
Noasaurus and Masiakasaurus share several synapomorphies
(cervical neural spine placed on the anterior half of the centrum,
reduced shaft of metatarsal II, simple maxillary palatal process,
and prominent rim on the antorbital fossa) that suggest these
two taxa form a clade, Noasauridae. Laevisuchus shares fea-
tures with abelisauroids (Novas and Bandyopadhyay, 1999) and
noasaurids (anterior placement of cervical neural spine, cervical
epipophyses small posteriorly). Additional specimens will per-
mit a more detailed analysis of the relationships of these three
527CARRANO ET AL.—OSTEOLOGY OF MASIAKASAURUS KNOPFLERI
FIGURE 23. Cladograms showing phylogenetic relationships of Masiakasaurus knopﬂeri and other theropods. A, Strict consensus results.
Numbers represent decay indices for individual nodes. B, One representative most parsimonious tree. Numbers show unambiguous character
support (left of node) and percentage of bootstrap replicates containing that clade (right of node). Values below 50% are not reported. Numbered
nodes are as follows: 1, Coelophysinae; 2, Coelophysidae; 3, Abelisauridae; 4, Abelisauroidea; 5, Neoceratosauria. The placement of several of
these names is temporary pending ongoing phylogenetic revision and a full review of nomenclatural history and priority. No formal redeﬁnitions
(in terms of content or otherwise) are offered.
As might be expected with any sample including multiple
individuals, marked variation exists among the numerous spec-
imens of Masiakasaurus. Speciﬁcally, two discrete morphs are
present in the sample and are demonstrable on comparably
sized elements. These ‘‘gracile’’ and ‘‘robust’’ morphs are most
easily discerned on the appendicular elements, probably be-
cause these have the largest sample sizes. The ‘‘robust’’ morph
is characterized by pronounced muscular and ligamentous at-
tachments, whereas these sites are less distinct in the ‘‘gracile’’
morph. In addition, ‘‘robust’’ tibiae show fusion with the tarsal
elements, whereas this fusion is lacking in ‘‘gracile’’ tibiae. Six
528 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 22, NO. 3, 2002
femora (FMNH PR 2117, 2123, 2149, 2208; UA 8684, 8712)
and four tibiae (FMNH PR 2116, 2119, 2122) demonstrate the
‘‘robust’’ morphology, while seven femora (FMNH PR 2115,
2120, 2215, 2148, 2150, 2153; UA 8681) and eight tibiae
(FMNH PR 2112, 2118, 2121, 2152, 2214; UA 8687, 8710-11)
are ‘‘gracile’’ (Table 1). No other elements are sufﬁciently well
sampled to permit distinguishing these two morphs. Indeed, our
sample sizes of even these elements are insufﬁcient to dem-
onstrate statistically signiﬁcant differences for several measured
parameters, although consistent trends are apparent (Table 2).
Furthermore, although both morphs are present in both femora
and tibiae, without articulated specimens we cannot empirically
demonstrate that ‘‘robust’’ femora correspond to ‘‘robust’’ tib-
Such variation is reminiscent of that described as possible
sexual dimorphism for the coelophysoids Coelophysis (Colbert,
1990) and Syntarsus (Raath, 1990). However, it is not clear how
sexual variation (i.e., two genders within the sample) might be
satisfactorily distinguished from taxonomic variation (e.g., two
populations or subspecies within the sample) when a bimodal
distribution is apparent. This is particularly problematic when
the variation in question does not involve actual primary or
secondary sexual features. The presence of two morphs mixed
in the same locality might favor sexual dimorphism, as it is
perhaps unlikely that two closely-related taxa would have lived
in such proximity. Nevertheless, at this point we describe these
as individual variations and note the possibility that they may
represent sexual dimorphism. The isolated nature of most spec-
imens, and the small sample sizes associated with most skeletal
elements, do not permit exploring this idea further in Masiak-
asaurus at this time.
Dentition and Diet
The most striking characteristics of Masiakasaurus are the
unusually heterodont, procumbent dentition and associated an-
terior jaw morphology. In addition to the extent of variation
within the tooth row, the speciﬁc morphologies of the anterior
dentary teeth are unknown elsewhere among theropods. Indeed,
although other theropods, such as spinosaurids and therizino-
saurids, also deviate from the typical theropod tooth design, the
teeth of neither clade resemble those of Masiakasaurus.Inad-
dition, neither group displays either procumbency or hetero-
donty to the degree seen in the new Malagasy taxon. The pro-
cumbency in the small coelurosaurs Ornitholestes and Procer-
atosaurus (Paul, 1988) is considerably less marked than that of
Masiakasaurus, and is not accompanied by heterodonty.
Somewhat analogous anterior tooth morphologies can be ob-
served in certain mammals. Carpolestid primates (Plesiadapi-
formes), for example, possess an elongate, acuminate, broad-
based lower medial incisor (Rose, 1975; Biknevicius, 1986).
Like the anterior dentary teeth of Masiakasaurus, this incisor
lacks prominent wear facets and is strongly procumbent. Other
plesiadapiforms, including paromomyids, saxonellids, and ple-
siadapids, also show procumbency of the incisors, although in
the former these teeth tend to be more slender and narrow (Flea-
gle, 1999). Kay and Cartmill (1977) drew a comparison be-
tween the anterior dentition of plesiadapiforms and that of ex-
tant caenolestid marsupials. They noted that in these living taxa,
robust, broadly pointed incisors are used for cutting food into
smaller pieces, whereas narrower, more elongate incisors are
used for grasping and puncturing prey items.
The dentition of Masiakasaurus is difﬁcult to interpret in the
light of functional and feeding considerations, particularly giv-
en that most of the skull remains unknown. Nonetheless, the
orientation and morphology of the anterior dentary teeth sug-
gest use in prehension—grasping small, whole prey items (such
as invertebrates, small vertebrates, or even fruits)—rather than
tearing or clipping parts of larger prey. The posteriorly posi-
tioned carinae on the anterior teeth are also better positioned
for grasping than tearing, and the rounded apex argues against
its use in incising. In living taxa, the upper anterior dentition
obviously plays a signiﬁcant role in food procurement as well,
and often varies in a manner consistent with lower tooth use.
Until these upper teeth are known for Masiakasaurus, we avoid
more speciﬁc hypotheses of dental function.
The ﬂat dentary symphysis of Masiakasaurus does not sug-
gest rotation or another unusual motion between the two halves
of the lower jaw, but it lacks strong ridges or other structures
that might indicate a strong ligamentous connection. In these
qualities the dentary symphysis of Masiakasaurus is similar to
those of other theropods. The surface area of the symphysis is
unusually large, but this could reﬂect a ﬁrmer attachment be-
tween the two dentaries or merely be a consequence of the
expanded anterior dentition. The posterior dentary greatly re-
sembles that of other abelisauroids, particularly the loose peg-
and-socket articulation between the surangular and dentary.
This may have imparted a greater degree of motion or shock
absorption between the anterior and posterior mandible, but it
is not unique to Masiakasaurus and is therefore not likely to
have been speciﬁcally related to its bizarre dental morphology.
More puzzling still is the fact that the posterior dentary and
maxillary teeth of Masiakasaurus retain the more generalized
carnivorous morphology seen in most other theropods. Thus,
unlike therizinosaurs, the entire dentition has not been modiﬁed
from the basic theropod plan; changes are localized to the an-
terior region. The posterior teeth are recurved, laterally com-
pressed, and ﬁnely serrated. In addition, the serrated carinae are
anteriorly and posteriorly positioned, suggesting that the teeth
were used for holding, cutting, and/or slicing animal tissues.
The combination of this morphology with that of the anterior
dentition is not yet explicable. One possibility is that Masiak-
asaurus was insectivorous or piscivorous, using its anterior
teeth for acquiring small, whole prey items and its posterior
teeth for maceration. More complete materials of the skull and
forelimb will undoubtedly shed light on the feeding and pred-
atory habits of Masiakasaurus.
The recognition of Masiakasaurus as a small-bodied abeli-
sauroid broadens our perspective on the evolutionary radiation
of this poorly-understood group of theropods. Currently, nearly
all known abelisauroids—Abelisaurus,Carnotaurus,Indosau-
rus, and Ilokelesia—are large-bodied animals, with estimated
masses in the single-ton range. Other, less complete abelisau-
roid remains (Martı´nez et al., 1993; Bertini, 1996) also fall
within this size range. Other than Noasaurus,Ligabueino (Bon-
aparte, 1996), and Laevisuchus, there are no deﬁnitive abeli-
sauroid remains from animals of signiﬁcantly smaller body siz-
es. However, unlike the Mesozoic terrestrial localities of Laur-
asia, Gondwanan formations have thus far produced few re-
mains of any small-bodied theropods. As a result, these animals
are poorly known and either very fragmentary (e.g., Compso-
suchus, Huene and Matley, 1933; Velocisaurus, Bonaparte,
1991a) or only distantly related to abelisauroids (e.g., Alvar-
ezsaurus, Bonaparte, 1991a; Patagonykus, Novas, 1997b).
With Masiakasaurus and Laevisuchus, it is now possible to
identify a radiation of small-bodied abelisauroids and to hy-
pothesize that Noasauridae (Bonaparte and Powell, 1980) is a
more diverse clade (Bonaparte, 1991b). Like coelophysoids and
coelurosaurs, abelisauroids include a diversity of taxa spanning
several orders of magnitude in body size. The dental morphol-
ogy of Masiakasaurus, at least, also hints at a diversity of feed-
ing habits presumed elsewhere only among coelurosaurs. The
529CARRANO ET AL.—OSTEOLOGY OF MASIAKASAURUS KNOPFLERI
long unsampled lineages of most known (Late Cretaceous)
abelisauroids, coupled with the less well-exploited Gondwanan
fossil record, suggests that currently known taxa far underre-
present the true diversity of abelisauroids.
Abelisauroid remains have been described from Morocco
(Russell, 1996) and France (Buffetaut et al., 1988; Le Loeuff
and Buffetaut, 1991), but these specimens are extremely incom-
plete and their assignment to Abelisauroidea is equivocal. Nev-
ertheless, abelisauroids were certainly more widely dispersed
than has generally been recognized. Fragmentary material re-
covered from the Cenomanian of Egypt (Stromer and Weiler
1930:pl. III, ﬁg. 47a, b) may represent an abelisauroid, but un-
fortunately this material was destroyed in World War II along
with nearly all of Stromer’s Egyptian collections. In addition to
Elaphrosaurus, several specimens collected from the Tithonian
Tendaguru Formation of Tanzania probably pertain to at least
one other abelisauroid (HMN 37, 68, and 69). According to the
present analysis, abelisauroids were also represented in the Al-
bian of Europe (Genusaurus; Accarie et al., 1995).
However, in the two decades since their diagnosis(Bonaparte
and Novas, 1985), abelisaurids have been unequivocally iden-
tiﬁed only in South America, India, and Madagascar. The par-
ticularly close phylogenetic relationships between Carnotaurus
(South America), Majungatholus (Madagascar), and Indosuchus
(India) have been used to support the hypothesis that these three
regions share a more recent tectonic connection (and therefore
a closer biogeographic afﬁnity) with each other than they do
with other Gondwanan landmasses (Krause et al., 1998; Samp-
son et al., 1998). A recent tectonic reconstruction also supports
this interpretation (Hay et al., 1999). Like carnotaurine abeli-
saurids, the distribution of noasaurids (which may also include
tiny Ligabueino andesi, from the Lower Cretaceous of Argen-
tina; Bonaparte, 1996) provides additional support of a South
America–India–Madagascar biogeographic connection into the
Nevertheless, it should be noted that Late Cretaceous faunas
are simply unknown from most other Gondwanan landmasses,
most importantly Africa and Antarctica. Without any faunal in-
formation from such regions, reputed similarities (or differenc-
es) to those of South America, India, and Madagascar must be
viewed with caution (Forster, 1999). At this time we cannot
determine whether the distributions of carnotaurine abelisaurids
and noasaurids is evidence of a biogeographic connection be-
tween these latter continents exclusive of other Gondwanan
landmasses, or represents a truly Gondwanan distribution af-
fected by uneven preservation and sampling. Further collection
in the Upper Cretaceous deposits of Africa (e.g., Gemmellaro,
1921; Rauhut and Werner, 1997) and Antarctica remains the
best opportunity to illuminate these important components of
Late Cretaceous Gondwanan faunal history.
The remains of a new theropod, Masiakasaurus knopﬂeri
from the Late Cretaceous Maevarano formation of Madagascar,
are described in detail. Despite their incomplete nature, these
specimens represent approximately 40% of the skeleton, in-
cluding portions of the skull, vertebral column, limbs, and pel-
vic girdle. They reveal the presence of a small-bodied theropod
with hitherto unknown characteristics of the dentition and jaws.
The anterior dentary teeth are procumbent and spoon-shaped,
with weak serrations, whereas the posterior dentary and max-
illary teeth are more typically theropod in form. Dimorphic var-
iation is evident in the appendicular skeleton. Features through-
out the skull and skeleton reveal Masiakasaurus to be an abel-
isauroid theropod. More speciﬁcally, it may belong to the clade
Noasauridae, containing the Late Cretaceous Noasaurus, from
South America, and Laevisuchus, from India. If so, the geo-
graphic distribution of this clade parallels that of carnotaurine
abelisaurids, further supporting a hypothesis of shared biogeo-
graphic history for South America, Madagascar, and India. Fi-
nally, the presence of Masiakasaurus on Madagascar docu-
ments a more global distribution of small-bodied abelisauroids
than has been previously appreciated.
The authors would like to thank D. Krause and all the par-
ticipants of the SUNY-Stony Brook/Universite´ d’Antananarivo
1993–1999 ﬁeld expeditions to the Mahajanga Basin, Mada-
gascar: R. Asher, G. Buckley, M. Gottfried, J. Hartman, C.
Lockwood, J. Miller, C. Norovelo, P. O’Connor, Prosper, J. A.
Rabarison, L. Rahantarisoa, L. L. Randriamiaramanana, N. H.
S. Ravelomanantsoa, F. Ravoavy, K. Curry Rogers, R. Rogers,
K. Samonds, N. Stevens, E. Roberts, R. Terry, C. Wall, N.
Wells, and R. Wunderlich. Particular thanks are due to A. Ra-
soamiaramanana, G. Ravololonarivo and B. Rakotosamimanana
(Universite´ d’Antananarivo), B. Andriamihaja and P. Wright
(Institute for the Conservation of Tropical Environments), and
the villagers of Berivotra for invaluable logistical support
throughout these expeditions.
This paper was greatly improved thanks to discussions with
O. Rauhut, P. Sereno, and H. Larsson, as well as reviews by F.
Novas, and R. Tykoski. We also acknowledge P. Holroyd, A.
Milner, P. Powell, A. Prieur, P. Sereno, C. Schaff, P. Taquet, and
D. Unwin for access to specimens in their care, F. Novas for
photographs of Noasaurus, F. Grine for assistance in obtaining
the SEM photographs, and L. Betti-Nash for artistic advice.
English translations of Accarie et al. (1995), Bonaparte and
Novas (1985), Depe´ret (1896), Dollo (1888), Lavocat (1955),
Le Loeuff and Buffetaut (1991), Martı´nez et al. (1993), Novas
(1991, 1992, 1993), and Thevenin (1907) are available from the
Polyglot Paleontologist website (http://www.uhmc.sunysb.edu/
Accarie, H., B. Beaudoin, J. Dejaz, G. Frie`s, J.-G. Michard, and P.
Taquet. 1995. De´couverte d’un Dinosaure The´ropode nouveau
(Genusaurus sisteronis n. g., n. sp.) dans l’Albien marin de Sisteron
(Alpes de Haute-Provence, France) et extension au Cre´tace´ infe´r-
ieur de la ligne´e ce´ratosaurienne. Compte Rendus de l’Academie
des Sciences, Paris, se´rie IIa 320:327–334.
Asher, R. J., and D. W. Krause. 1998. First pre-Holocene (Cretaceous)
record of Anura from Madagascar. Journal of Vertebrate Paleon-
Bakker, R. T., M. Williams, and P. J. Currie. 1988. Nanotyrannus, a new
genus of pygmy tyrannosaur, from the latest Cretaceous of Mon-
tana. Hunteria 1:1–30.
Bertini, R. J. 1996. Evide´ncias de Abelisauridae (Carnosauria: Sauris-
chia) do Neocreta´ceo da Bacia do Parana´. Boletim do 4o Simposio
sobre o Creta´ceo do Brasil 1996:267–271.
Biknevicius, A. R. 1986. Dental function and diet in the Carpolestidae
(Primates, Plesiadapiformes). American Journal of Physical An-
Bonaparte, J. F. 1991a. Los vertebrados fo´siles de la Formacio´n Rio
Colorado, de la Ciudad de Neuque´n y Cercanı´as, Creta´cico Supe-
rior, Argentina. Revista del Museo Argentino deCiencias Naturales
‘‘Bernardino Rivadavia’’ e Instituto Nacional de Investigacio´n de
las Ciencias Naturales: Paleontologı´a 4:17–123.
——— 1991b. The Gondwanian theropod families Abelisauridae and
Noasauridae. Historical Biology 5:1–25.
——— 1996. Cretaceous tetrapods of Argentina. Mu¨nchner Geowis-
senschaften Abhandlungen (A) 30:73–130.
———, and F. E. Novas. 1985. Abelisaurus comahuensis, n. g., n. sp.,
Carnosauria del Cre´tacico Tardio de Patagonia. Ameghiniana 21:
———, and J. E. Powell. 1980. A continental assemblage of tetrapods
from the Upper Cretaceous beds of El Brete, northwestern Argen-
530 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 22, NO. 3, 2002
tina (Sauropoda-Coelurosauria-Carnosauria-Aves). Me´moires de la
Socie´te´Ge´ologique de France, Nouvelle Se´rie 139:19–28.
———, F. E. Novas, and R. A. Coria. 1990. Carnotaurus sastrei Bon-
aparte, the horned, lightly built carnosaur from the Middle Creta-
ceous of Patagonia. Contributions in Science, Natural History Mu-
seum of Los Angeles County 416:1–41.
Britt, B. B., C. A. Miles, K. C. Cloward, and J. H. Madsen. 1999. A
juvenile Ceratosaurus (Theropoda, Dinosauria) from Bone Cabin
Quarry West (Upper Jurassic, Morrison Formation), Wyoming.
Journal of Vertebrate Paleontology 19(3, suppl.):33A.
———, D. J. Chure, T. R. Holtz, Jr., C. A. Miles, and K. L. Stadtman.
2000. A reanalysis of the phylogenetic afﬁnities of Ceratosaurus
(Theropoda, Dinosauria) based on new specimens from Utah, Col-
orado, and Wyoming. Journal of Vertebrate Paleontology 20(3,
Brochu, C. A. 1996. Closure of neurocentral sutures during crocodilian
ontogeny: implications for maturity assessment in fossil archosaurs.
Journal of Vertebrate Paleontology 16:49–62.
Buckley, G. A., and C. A. Brochu. 1999. An enigmatic new crocodile
from the Upper Cretaceous of Madagascar; pp. 149–175 in D. M.
Unwin (ed.), Cretaceous Fossil Vertebrates. Special Papers in Pa-
Buffetaut, E., P. Mechin, and A. Mechin-Salessy. 1988. Un dinosaure
the´ropode d’afﬁnite´s gondwaniennes dans le Cre´tace´ supe´rieur de
Provence. Comptes Rendus de l’Academie des Sciences Paris, Se´r-
ie II 306:153–158.
Carrano, M. T. 2000. Homoplasy and the evolution of dinosaur loco-
motion. Paleobiology 26:489–512.
———, and J. R. Hutchinson. 2002. The pelvic and hind limb mus-
culature of Tyrannosaurus rex (Dinosauria: Theropoda). Journal of
———, and S. D. Sampson. 1999. Evidence for a paraphyletic ‘Cera-
tosauria’ and its implications for theropod dinosaur evolution. Jour-
nal of Vertebrate Paleontology 19(3, suppl.):36A.
Colbert, E. H. 1990. Variation in Coelophysis bauri; pp. 81–90 in K.
Carpenter and P. J. Currie (eds.), Dinosaur Systematics: Perspec-
tives and Approaches. Cambridge University Press, Cambridge.
Coria, R. A., and L. Salgado. 2000. A basal Abelisauria Novas 1992
(Theropoda-Ceratosauria) from the Cretaceous of Patagonia, Ar-
gentina. GAIA 15:89–102.
Currie, P. J., and X.-J. Zhao. 1993. A new carnosaur (Dinosauria, Ther-
opoda) from the Jurassic of Xinjiang, People’s Republic of China.
Canadian Journal of Earth Sciences 30:2,037–2,081.
Curry Rogers, K. A., and C. A. Forster. 2001. The last of the dinosaur
titans: a new sauropod from Madagascar. Nature 412:530–534.
Depe´ret, C. 1896. Sur l’existence de dinosauriens sauropodes et the´ro-
podes dans le Cre´tace´ supe´rieur de Madagascar. Comptes Rendus
de l’Academie des Sciences, Paris 122:483–485.
Dollo, L. 1888. Sur la signiﬁcation du ‘‘trochanter pendant’’ des dino-
sauriens. Bulletin Scientiﬁque de la France et de la Belgique 19:
Fleagle, J. G. 1999. Primate Adaptation and Evolution. 2nd ed. Aca-
demic Press, New York, pp.
Forster, C. A. 1999. Gondwanan dinosaur evolution and biogeographic
analysis. Journal of African Earth Sciences 28:169–185.
———, L. M. Chiappe, D. W. Krause, and S. D. Sampson. 1996. The
ﬁrst Cretaceous bird from Madagascar. Nature 382:532–534.
———, and P. M. O’Connor. 2000. The avifauna of the Upper Creta-
ceous Maevarano Formation, Madagascar. Journal of Vertebrate Pa-
leontology 20(3, suppl.):41–42A.
———, S. D. Sampson, L. M. Chiappe, and D. W. Krause. 1998. The
theropod ancestry of birds: new evidence from the Late Cretaceous
of Madagascar. Science 279:1915–1919.
Gauthier, J. 1986. Saurischian monophyly and the origin of birds; pp.
1–47 in K. Padian (ed.), The Origin of Birds and the Evolution of
Flight. Memoirs of the California Academy of Sciences, San Fran-
Gemmellaro, M. 1921. Rettili mae¨strichtiani di Egitto. Giornale di
Scienze Naturali ed Economiche 32:340–351.
Gottfried, M. D., and D. W. Krause. 1998. First record of gars (Lepi-
sosteidae, Actinopterygii) on Madagascar: Late Cretaceous remains
from the Mahajunga Basin. Journal of Vertebrate Paleontology 18:
Harris, J. D. 1998. A reanalysis of Acrocanthosaurus atokensis, its phy-
logenetic status, and paleobiogeographic implications, based on a
new specimen from Texas. New Mexico Museum of Natural His-
tory and Science Bulletin 13:1–75.
Hay, W. W., R. M. Deconto, C. N. Wold, K. M. Wilson, S. Voigt, M.
Schulz, A. Wold-Rossby, W.-C. Dullo, A. B. Ronov, A. N. Balu-
khovsky, and E. Soeding. 1999. An alternative global Cretaceous
paleogeography; in E. Barrera and C. Johnson (eds.), Evolution of
the Cretaceous ocean-climate system. Geological Society of Amer-
ica Special Paper 332:1–48.
Holtz, T. R., Jr. 1994a. The arctometatarsalian pes, an unusual structure
of the metatarsus of Cretaceous Theropoda (Dinosauria: Sauris-
chia). Journal of Vertebrate Paleontology 14:480–519.
——— 1994b. The phylogenetic position of the Tyrannosauridae: im-
plications for theropod systematics. Journal of Paleontology 68:
——— 2000. A new phylogeny of the carnivorous dinosaurs. GAIA
Huene, F. v., and C. A. Matley. 1933. The Cretaceous Saurischia and
Ornithischia of the Central Provinces of India. Memoirs of the Geo-
logical Survey of India: Palaeontologica Indica 21:1–72.
Hutchinson, J. R. 2001a. The evolution of pelvic osteology and soft
tissues on the line to extant birds (Neornithes). Zoological Journal
of the Linnean Society 131:123–168.
——— 2001b. The evolution of femoral osteology and soft tissues on
the line to extant birds (Neornithes). Zoological Journal of the Lin-
nean Society 131:169–197.
Janensch, W. 1925. Die Coelurosaurier und Theropoden der Tendaguru-
Schichten Deutsch-Ostafrikas. Palaeontographica Supplement VII
(1) (teil 1, leif 1):1–100.
Kay, R. F., and M. Cartmill. 1977. Cranial morphology and adaptation
of Palaechthon nacimienti and other Paromomyidae (Plesiadapo-
idea, ?Primates), with a description of a new genus and species.
Journal of Human Evolution 6:19–35.
Krause, D. W., G. V. R. Prasad, W. von Koenigswald, A. Sahni, and F.
E. Grine. 1997. Cosmopolitanism among Gondwanan Late Creta-
ceous mammals. Nature 390:504–507.
———, C. A. Forster, S. D. Sampson, G. A. Buckley,and M. Gottfried.
1998. Vertebrate fossils from the Late Cretaceous of Madagascar:
implications for Gondwanan plate tectonics and biogeography; pp.
128–129 in J. Almond, J. Anderson, P. Booth, A. Chinsamy-Turan,
D. Cole, M. J. De Wit, B. Rubridge, R. Smith, J. van BeverDonker,
and B. C. Storey (eds.), Gondwana 10: Event Stratigraphy of Gond-
wana, Journal of African Earth Sciences 27. Cape Town, South
———, R. R. Rogers, C. A. Forster, J. H. Hartman, G. A. Buckley,
and S. D. Sampson. 1999. The Late Cretaceous vertebrate fauna of
Madagascar: implications for Gondwanan paleobiogeography. GSA
Lavocat, R. 1955. Sur une portion de mandibule de the´ropode proven-
nant du Cre´tace´ supe´rieur de Madagascar. Bulletin du Muse´um Na-
tional d’Histoire Naturelle, se´rie 2, 27:3.
Le Loeuff, J., and E. Buffetaut. 1991. Tarascosaurus salluvicus nov.
gen., nov. sp., dinosaure the´ropode du Cre´tace´ Supe´rieur du sud de
la France. Geobios 24:585–594.
Makovicky, P. J. 1997. A new small theropod from the Morrison For-
mation of Como Bluff, Wyoming. Journal of Vertebrate Paleontol-
Martı´nez, R., A. Maure, M. Oliva, and M. Luna. 1993. Un maxilar de
Theropoda (Abelisauria) de la Formacion Bajo Barreal, Cretacico
Tardio, Chubut, Argentina. Ameghiniana 30:109–110.
McIntosh, J. S. 1990. Sauropoda; pp. 345–401 in D. B. Weishampel, P.
Dodson, and H. Osmo´lska (eds.), The Dinosauria. University of
California Press, Berkeley.
Molnar, R. E., S. M. Kurzanov, and Z. Dong. 1990. Carnosauria; pp.
169–209 in D. B. Weishampel, P. Dodson, and H. Osmo´lska (eds.),
The Dinosauria. University of California Press, Berkeley.
———, A. L. Angriman, and Z. Gasparini. 1996. An Antarctic Creta-
ceous theropod. Memoirs of the Queensland Museum 39:669–674.
Novas, F. E. 1989. Los dinosaurios carnivorous de la Argentina. Ph.D.
dissertation, Universidad Nacional de La Plata, La Plata, 510 pp.
——— 1991. Phylogenetic relationships of ceratosaurian theropod di-
nosaurs. Ameghiniana 28:401.
——— 1992. La evolucion de los dinosaurios carnivoros; pp. 126–163
in J. L. Sanz and A. D. Buscalioni (eds.), Los Dinosaurios y Su
Entorno Biotico: Actas del Segundo Curso de Paleontologia in
Cuenca. Instituto ‘‘Juan Valdez’’, Cuenca, Argentina.
531CARRANO ET AL.—OSTEOLOGY OF MASIAKASAURUS KNOPFLERI
——— 1993. Diagnosis y ﬁlogenia de los Dinosauria. Ameghiniana
——— 1997a. Abelisauridae; pp. 1–2 in P. J. Currie and K. Padian
(eds.), Encyclopedia of Dinosaurs. Academic Press, New York.
——— 1997b. Anatomy of Patagonykus puertai (Theropoda, Avialae,
Alvarezsauridae), from the Late Cretaceous of Patagonia. Journal
of Vertebrate Paleontology 17:137–166.
———, and S. Bandyopadhyay. 1999. New approaches on the Creta-
ceous theropods from India; pp. 47 in H. A. Leanza (ed.), VII
International Symposium on Mesozoic Terrestrial Ecosystems, Ab-
stracts, Ameghiniana 36. Buenos Aires, Argentina.
——— 2001. Abelisaurid pedal unguals from the Late Cretaceous of
India. VII International Symposium on Mesozoic Terrestrial Eco-
systems, Asociacio´n Paleontolo´ gica Argentina Publicacio´ n Especial
Padian, K., and P. E. Olsen. 1989. Ratite footprints and the stance and
gait of Mesozoic theropods; pp. 231–241 in D. D. Gillette and M.
G. Lockley (eds.), Dinosaur Tracks and Traces. Cambridge Uni-
versity Press, Cambridge.
———, J. R. Hutchinson, and T. R. Holtz, Jr. 1999. Phylogenetic def-
initions and nomenclature of the major taxonomic categories of the
carnivorous Dinosauria (Theropoda). Journal of Vertebrate Pale-
Paul, G. S. 1984. The archosaurs: a phylogenetic study; pp. 175–180
in W.-E. Reif and F. Westphal (eds.), Third Symposium on Meso-
zoic Terrestrial Ecosystems, Short Papers. Attempto Verlag, Tu¨-
——— 1988. Predatory Dinosaurs of the World: A Complete Illustrated
Guide. Simon & Schuster, New York, 464 pp.
Raath, M. A. 1990. Morphological variation in small theropods and its
meaning in systematics: evidence from Syntarsus rhodesiensis; pp.
91–105 in K. Carpenter and P. J. Currie (eds.), Dinosaur System-
atics: Perspectives and Approaches. Cambridge University Press,
Rauhut, O. W. M. 1995. Zur systematischen Stellung der afrikanischen
Theropoden Carcharodontosaurus Stromer 1931 und Baharisaurus
Stromer 1934. Berliner Geowissenschaften Abhandlungen 16:357–
——— 1998. Elaphrosaurus bambergi and the early evolution of the-
ropod dinosaurs. Journal of Vertebrate Paleontology 18(3, suppl.):
——— 2000. The interrelationships and evolution of basal theropods
(Dinosauria, Saurischia). Ph.D. dissertation, University of Bristol,
Bristol, 583 pp.
———, and C. Werner. 1997. First record of a Maastrichtian sauropod
dinosaur from Egypt. Palaeontographica Africana 34:63–67.
Rogers, R. R., and J. H. Hartman. 1998. Revised age of the dinosaur-
bearing Maevarano Formation (Upper Cretaceous), Mahajanga Ba-
sin, Madagascar; pp. 160–162 in J. Almond, J. Anderson, P. Booth,
A. Chinsamy-Turan, D. Cole, M. J. De Wit, B. Rubridge, R. Smith,
J. van Bever Donker, and B. C. Storey (eds.), Gondwana 10: Event
Stratigraphy of Gondwana, Journal of African Earth Sciences 27.
Cape Town, South Africa.
———, ———, and D. W. Krause. 2000. Stratigraphic analysis of Up-
per Cretaceous strata in the Mahajanga Basin, northwestern Mad-
agascar: implications for ancient and modern faunas. Journal of
Romer, A. S. 1923. The pelvic musculature of saurischian dinosaurs.
Bulletin of the American Museum of Natural History 48:605–617.
Rose, K. D. 1975. The Carpolestidae: Early Tertiary primates from
North America. Bulletin of the Museum of Comparative Zoology
Rowe, T. 1989. The early history of theropods; pp. 100–112 in K.
Padian and D. J. Chure (eds.), The Age of Dinosaurs, 12th Annual
Short Course of the Paleontological Society. University of Tennes-
———, and J. Gauthier. 1990.Ceratosauria; pp. 151–168 in D. B. Weis-
hampel, P. Dodson, and H. Osmo´lska (eds.), The Dinosauria. Uni-
versity of California Press, Berkeley.
Russell, D. A. 1996. Isolated dinosaur bones from the Middle Creta-
ceous of the Taﬁlalt, Morocco. Bulletin du Muse´um National
d’Histoire Naturelle, Paris, 4e se´rie 18, Section C:349–402.
———, and Z. Dong. 1993. The afﬁnities of a new theropod from the
Alxa Desert, Inner Mongolia, People’s Republic of China. Cana-
dian Journal of Earth Sciences 30:2,107–2,127.
Sampson, S. D., M. T. Carrano, and C. A. Forster. 2001. A bizarre new
predatory dinosaur from Madagascar. Nature 409:504–506.
———, D. W. Krause, P. Dodson, and C. A. Forster. 1996. The pre-
maxilla of Majungasaurus (Dinosauria: Theropoda), with implica-
tions for Gondwanan paleobiogeography. Journal of Vertebrate Pa-
———, L. M. Witmer, C. A. Forster, D. W. Krause, P. M. O’Connor,
P. Dodson, and F. Ravoavy. 1998. Predatory dinosaur remains from
Madagascar: implications for the Cretaceous biogeography of
Gondwana. Science 280:1,048–1,051.
Sereno, P. C. 1998. A rationale for phylogenetic deﬁnitions, with ap-
plication to the higher-level taxonomy of Dinosauria. Neues Jahr-
buch fu¨r Geologie und Pala¨ontologie Abhandlungen 210:41–83.
——— 1999. The evolution of dinosaurs. Science 284:2,137–2,147.
———, A. L. Beck, D. B. Dutheil, B. Gado, H. C. E. Larsson, O. W.
M. Rauhut, R. W. Sadleir, C. A. Sidor, D. J. Varricchio, G. P. Wil-
son, and J. A. Wilson. 1998. A long-snouted predatory dinosaur
from Africa and the evolution of spinosaurids. Science 282:1,298–
———, D. B. Dutheil, M. Iarochene, H. C. E. Larsson, G. H. Lyon, P.
M. Magwene, C. A. Sidor, D. J. Varricchio, and J. A. Wilson.1996.
Predatory dinosaurs from the Sahara and Late Cretaceous faunal
differentiation. Science 272:986–991.
———, C. A. Forster, R. R. Rogers, and A. M. Monetta. 1993. Prim-
itive dinosaur skeleton from Argentina and the early evolution of
Dinosauria. Nature 361:64–66.
———, J. A. Wilson, H. C. E. Larsson, D. B. Dutheil, and H.-D. Sues.
1994. Early Cretaceous dinosaurs from the Sahara. Science 266:
Stromer, E., and W. Weiler. 1930. Ergebnisseder Forschungsreisen Prof.
E. Stromers in den Wu¨sten A
¨gyptens. VI. Beschreibung von Wir-
beltier-Resten aus dem nubischen Sandsteine Obera¨gyptens und aus
a¨gyptischen Phosphaten nebst Bemerkungen u¨ber die Geologie der
Umgegend von Mahamıˆd in Obera¨gypten. Abhandlungen der Bay-
erischen Akademie der Wissenschaften Mathematisch-naturwissen-
schaftliche Abteilung, Neue Folge 7:1–42.
Swofford, D. L. 1998. PAUP*. Phylogenetic Analysis Using Parsimony
(* and other methods). Version 4. Sinauer Associates, Sunderland,
Tarsitano, S. 1983. Stance and gait in theropod dinosaurs. ActaPalaeon-
tologica Polonica 28:251–264.
Thevenin, A. 1907. Pale´ontologie de Madagascar. Annales de Pale´on-
Tykoski, R. 1998. The osteology of Syntarsus kayentakatae and its im-
plications for ceratosaurid phylogeny. M.S. thesis, University of
Texas at Austin, pp.
Welles, S. P. 1984. Dilophosaurus wetherilli (Dinosauria, Theropoda):
osteology and comparisons. Palaeontographica Abteilung A, 185:
———, and R. A. Long. 1974. The tarsus of theropod dinosaurs. An-
nals of the South African Museum 64:191–218.
Wilson, J. A. 1999. A nomenclature for vertebral laminae in sauropods
and other saurischian dinosaurs. Journal of Vertebrate Paleontology
Witmer, L. M. 1997. The evolution of the antorbital cavity of archo-
saurs: a study in soft-tissue reconstruction in the fossil record with
an analysis of the function of pneumaticity. Journal of Vertebrate
Paleontology Memoir 3, 17(1, suppl.):1–73.
Zhao, X., and P. J. Currie. 1993. A large crested theropod from the
Jurassic of Xinjiang, People’s Republic of China. Canadian Journal
of Earth Sciences 30:2,027–2,036.
Received 13 March 2001; accepted 3 November 2001.
Data available from SVP website: http://www.vertpaleo.org/jvp/
List of characters and character-states used in this phylogenetic anal-
ysis. Original citations are provided along with signiﬁcant subsequent
(1) Sculpturing on craniofacial elements: absent (0), present (1) (No-
532 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 22, NO. 3, 2002
(2) Diastema in dentition between premaxilla and maxilla: absent (0),
present (1) (Welles, 1984; Rowe, 1989).
(3) Premaxillary tooth row ends: ventral (0), anterior (1) to naris
(4) Maxillary/palatal process of premaxilla: large ﬂange (0), blunt
triangle (1) (Sampson et al., 1998).
(5) Height:length ratio of premaxilla below external naris: 0.5–1.25
1.25 (2) (Holtz, 1994b, 2000).
(6) Anterior maxillary ramus: absent/small (0), extends beneath half
of naris (1), extended anteriorly to end of naris (2) (Sereno et al., 1994;
(7) Maxillary fenestra: absent (0), present (1) (Gauthier, 1986; Wit-
(8) Promaxillary fenestra: absent (0), present (1) (Currie and Zhao,
1993; Witmer, 1997).
(9) Pneumatic excavation/antrum in maxillary anterior ramus: absent
(0), present (1) (Sereno et al., 1994).
(10) Pneumatic fossa in maxillary ascending process absent (0) or
present (1) (Witmer, 1997; Holtz, 2000).
(11) Palatal process of maxilla: ridged ﬂange (0), simple process (1)
(Sereno et al., 1998).
(12) Anteroventral border of antorbital fenestra: indistinct (0), de-
marcated by raised ridge (1) (Rowe, 1989).
(13) Ventral extent of antorbital fossa: substantial(0), absent (1) (No-
(14) Nasals: unfused (0), fused (1) in adults (Sereno, 1999).
(15) Pneumatic nasal foramina: absent (0), present (1) (Forster, 1999).
(16) Posterior narial margin: fossa (0), hood (1).
(17) Prefrontal-frontal articulation: butt joint (0), peg-in-socket (1)
(Sereno et al., 1994).
(18) Frontals and parietals: fused (0), unfused (1) in adults (Forster,
1999; Sereno, 1999).
(19) Median fossa in saddle-shaped depression overlapping frontal-
parietal contact: absent (0), present (1) (Sampson et al., 1998).
(20) Knob-like dorsal projection of parietals: absent (0), present (1)
(21) Nuchal wedge and parietal alae: small (0), hypertrophied and
elevated (1) (Forster, 1999; Sereno, 1999).
(22) Ventral process of postorbital: transversely narrow (0), wide and
U-shaped (1) (Sereno et al., 1994).
(23) Suborbital ﬂange of postorbital: absent (0), present (1) (Gauthier,
(24) Anterior postorbital prominence: absent/small (0), large (1), con-
tacts lacrimal/excludes frontal from orbit (2) (Sereno et al., 1994, 1996).
(25) Long axis of postorbital: dorsal-ventral (0), anteroventral-pos-
terodorsal (1) (Novas, 1989).
(26) Stepped-down ventrolateral fossa on postorbital: absent (0), pre-
sent (1) (Sampson et al., 1998).
(27) Dorsoventral thickness of lacrimal anterior process: moderate
(0), greatly reduced (1) (Sereno et al., 1994).
(28) Lacrimal anterior ramus length:
65% (1) ventral
ramus length (Sereno et al., 1998).
(29) Lacrimal foramen size: moderate/large (0), small (1) (Sereno et
(30) Lacrimal pneumatization and horn: absent/moderate (0), large
(1) (Novas, 1989; Currie and Zhao, 1993).
(31) Suborbital process of lacrimal: absent (0), present (1) (Sampson
et al., 1998).
(32) Jugal-maxilla overlap length:
50% (1) total length
of jugal (Sampson et al., 1998).
(33) Jugal participation in antorbital fossa: absent (0), present (1)
(Currie and Zhao, 1993).
(34) Facet on ventral ramus of lacrimal for jugal: absent (0), present
(1) (Sereno et al., 1994).
(35) Jugal pneumatic foramen: absent (0), present (1) (Zhao and Cur-
(36) Quadratojugal-squamosal contact: at tip (0), absent (1), broad
(2) (Rauhut, 2000; Holtz, 2000).
(37) Quadratojugal fused to quadrate in adults: no (0), yes (1) (Holtz,
(38) Squamosal contribution to broad, arching nuchal crest: absent
(0), present (1) (Novas, 1989; Sampson et al., 1998).
(39) Quadrate foramen size and position: large/at edge (0), small/
enclosed (1), absent (2) (Novas, 1989; Harris, 1998).
(40) Interorbital region: unossiﬁed (0), ossiﬁed (1) (Russell and
Dong, 1993; Novas, 1997a).
(41) Cranial nerve V exit foramen: single (0), partly split (1), fully
split (2) (Currie and Zhao, 1993).
(42) Ventral extent of distal paroccipital process: at (0), below (1)
occipital condyle (Paul, 1988; Bakker et al., 1988).
(43) Basal tubera width:
(1) occipital condyle width (Holtz,
(44) Depth of median supraoccipital ridge:
(1) depth of oc-
(45) Jugal process of palatine: tapered (0), expanded (1) (Sereno et
(46) Palatine pneumatic recess: absent (0), present (1) (Currie and
(47) Deeply excavated pocket on ectopterygoid ﬂange of pterygoid:
absent (0), present (1) (Gauthier, 1986).
(48) Lateral depth of ectopterygoid fossa: shallow (0), deeply incised
(1) (Sereno et al., 1994).
(49) External mandibular fenestra: moderate/small (0), hypertrophied
(1) (Gauthier, 1986; Sampson et al., 1998).
(50) Anterior end of external mandibular fenestra: posterior (0), ven-
tral (1) to last dentary tooth (Sereno, 1999).
(51) Posterior edge of splenial: straight/curved (0), notched (1) (Ser-
eno et al., 1994).
(52) Pendant medial process on articular: absent (0), present (1) (Ser-
eno et al., 1994).
(53) Retroarticular process faces: dorsally (0), posterodorsally (1)
(Sereno et al., 1994).
(54) Surangular articulation for dentary: small notch (0), large socket (1).
(55) Posteroventral dentary process: far posterior (0), ventral (1) to
posterodorsal process (Sereno, 1999).
(56) Premaxillary teeth: symmetrical (0), asymmetrical (1) (Bakker
et al., 1988).
(57) Posterior end of maxillary tooth row: beneath (0), anterior (1)
to orbit (Gauthier, 1986).
(58) Maxillary tooth count:
(59) Dentary tooth count:
18 (0), 20–29 (1) (Russell and Dong,
(60) Paradental plates: widely visible (0), obscured (1) in medial
(61) Medial surface of paradental plates: smooth (0), striated (1)
(Sampson et al., 1998).
(62) Anterior face of anterior presacrals: amphiplatyan or amphicoe-
lous (0), convex (1) (Gauthier, 1986).
(63) Posterior face of anterior presacrals: amphiplatyan or amphicoe-
lous (0), concave (1) (Gauthier, 1986).
(64) Pleurocoel immediately posterior to parapophysis in anterior pre-
sacrals: absent (0), present (1) (Gauthier, 1986; Rowe, 1989).
(65) Pleurocoel at posterior edge of centrum in anterior presacrals:
absent (0), present (1) (Gauthier, 1986; Rowe, 1989).
(66) Vertebral centrum pneumaticity: absent (0), camerate (1), ca-
mellate (2) (Britt, 1993).
(67) Axial epipophyses: moderate (0), extend well past postzygapo-
physis (1) (Novas, 1989).
(68) Axial neural spine: broad (0), invaginated laterally (1) (Molnar
et al., 1990).
(69) Axial parapophyses: moderate/prominent (0), reduced/absent (1)
(70) Axial diapophyses: moderate (0), reduced/absent (1) (Rowe,
(71) Axial pleurocoels: absent (0), present (1) (Rowe, 1989).
(72) Cervical prezygapophyseal-epipophyseal lamina: absent/weak
(0), marked (1) (Coria and Salgado, 2000).
(73) Cervical prespinal fossa: narrow (0), broad (1) (Coria and Sal-
(74) Cervical neural spine location: posterior (0), anterior (1) half of
(75) Anterior prongs on cervical epipophyses: absent (0), present (1)
(76) Cervical zygapophyses: close to midline (0), displaced laterally
(1) (Makovicky, 1997).
(77) Anterior cervical epipophyses: low/blunt (0), long/thin (1), long/
robust (2) (Novas, 1993).
(78) Cervical neural spines: long (0), short (1) anteroposteriorly (No-
533CARRANO ET AL.—OSTEOLOGY OF MASIAKASAURUS KNOPFLERI
(79) Cervical centrum length:
(1) 3 times height (Russell and
Dong, 1993; Sereno, 1999).
(80) Dorsal transverse processes: rectangular (0), triangular (1) in
dorsal view (Rowe, 1989).
(81) Posterior dorsal parapophyses: close to centrum/arch (0), ex-
tended far laterally (1) (Currie and Zhao, 1993).
(82) Posterior dorsal parapophyses and transverse processes: separate
(0), connected by a web (1).
(83) Dorsal vertebral centrum length
(1) twice height (Sereno,
(84) Sacrum composition: primordial 2 sacrals (0),
1 dorsal and
3 dorsals and
1 caudal (2) (Gauthier, 1986; Novas,
(85) Mid-sacral centra dimensions: normal (0), strongly constricted
(1) (Sereno, 1999).
(86) Ventral sacral margin: horizontal (0), arched (1) (Sereno, 1999).
(87) Sacral neural spines: separate (0), fused (1) in adults (Rowe,
(88) Anterior caudal neural spines: long (0), narrow (1) anteropos-
teriorly (Rauhut, 2000).
(89) Transverse processes of caudal vertebrae: unexpanded (0), ex-
panded (1) anteroposteriorly at ends (Coria and Salgado, 2000).
(90) Cervical ribs and centra: separate (0), fused (1) in adults (Gau-
(91) Pneumatic excavations in cervical rib heads: absent (0), present
(1) (Harris, 1998).
(92) Posterior process of anterior cervical ribs: narrow (0), wide and
ﬂat (1) (Coria and Salgado, 2000).
(93) Mid and distal chevrons: curved (0), L-shaped (1) (Russell and
(94) Anterior process of chevron base: absent/weak (0), large (1)
(Molnar et al., 1990).
(95) Transition between acromial process and scapular blade: gradual
(0), abrupt (1) (Currie and Zhao, 1993).
(96) Distal expansion of scapular blade: marked (0), weak/absent (1)
(Currie and Zhao, 1993).
(97) Scapular blade anteroposterior width: broad (0), narrow and
strap–like (1) (Bakker et al., 1988).
(98) Posteroventral process of coracoid: moderate (0), expanded far
past glenoid (1) (Sereno et al., 1996).
(99) Humerus length:
(1) one-third of femur length (Novas,
(100) Humeral head shape: elongate (0), globular (1) (Rauhut, 2000).
(101) Distal humeral condyles: rounded (0), ﬂattened (1).
(102) Humeral shaft torsion: absent (0), present (1) (Holtz, 2000).
(103) Deltopectoral crest length:
(1) 45% of humeral length
(Sereno et al., 1998).
(104) Distal carpals 1
2: separate (0), fused without facets (1),
fused with facets (2) (Gauthier, 1986; Sereno et al., 1994).
(105) Basal half of metacarpal I: loosely (0), closely (1) appressed
to metacarpal II (Gauthier, 1986).
(106) Metacarpal I length:
(1) 50% of metacarpal II length
(107) Metacarpal III width:
(1) half metacarpal II width (Gau-
thier, 1986; Sereno et al., 1994).
(108) Metacarpal IV: present (0), absent (1) (Paul, 1984; Gauthier,
(109) Fusion between pelvic elements in adults: absent (0), present
(1) (Gauthier, 1986).
(110) Posterior width of iliac brevis fossa: narrow (0), broad (1)
(Molnar et al., 1990; Sereno et al., 1994).
(111) Lateral wall of iliac brevis fossa: deeper (0), shallower (1)
ventrally than medial wall.
(112) Iliac supraacetabular crest: pendant (0), shelf-like (1) (Gauthier,
(113) Iliac pubic peduncle orientation: mostly ventral (0), mostly an-
terior (1) (Sereno, 1999).
(114) Iliac pubic peduncle size: subequal (0), larger (1) than iliac
ischial peduncle (Sereno et al., 1994).
(115) Iliac pubic peduncle length:
(1) twice width (Sereno et
(116) Iliac preacetabular length: to edge of (0), well past (1) anter-
iormost pubic peduncle (Gauthier, 1986; Carrano, 2000).
(117) Shape of dorsal iliac margin: convex (0), straight (1).
(118) Iliac postacetabular length:
(1) acetabulum length (For-
ster, 1999; Carrano, 2000).
(119) Posterior margin of ilium: convex (0), indented (1) (Sereno et
(120) Pubic apices: contacting (0), separate (1), fused with proximal
foramen (2) (Rauhut, 2000).
(121) Pubic obturator opening: absent (0), closed (1), notch (2) (Ser-
eno et al., 1994).
(122) Distal pubis: rounded (0), expanded anteroposteriorly (1) (Gau-
thier, 1986; Holtz, 1994b).
(123) Iliac-pubic articulation: concavo-convex (0), peg-in-socket (1)
(Sampson et al., 2001).
(124) Pubic shaft: straight (0), ventrally curved (1) (Rowe, 1989).
(125) Ischial obturator opening: absent (0), present/notch (1) (Gau-
thier, 1986; Harris, 1998).
(126) Notch ventral to ischial obturator process: absent (0), present
(1) (Rauhut, 1995).
(127) Distal ischium: rounded (0), expanded anteroposteriorly/trian-
gular (1) (Novas, 1993).
(128) Ischial antitrochanter: large and notched (0), reduced (1) (Ser-
(129) Length of ischium: approximately equal to (0), less than (1)
length of pubis (Gauthier, 1986).
(130) Femoral head orientation: 45
anteromedial (0) 20–35
medial (1), 90
/medial (Bonaparte, 1991b; Carrano, 2000).
(131) Dimorphism in femoral morphology: absent (0), present (1)
(Rowe and Gauthier, 1990).
(132) Femoral head directed: ventrally (0), horizontally (1) (Molnar
et al., 1990; Harris, 1998).
(133) Anterior M. iliofemoralis insertion on proximal femur: shelf/
spike (0), ﬁnger-like trochanter below proximal end (1) aliform tro-
chanter near proximal end (Hutchinson, 2001b).
(134) Posterior M. iliofemoralis insertion on proximal femur: shelf
(0), mound (1) (Hutchinson, 2001b).
(135) Femoral medial epicondyle: rounded (0), ridge (1), hypertro-
phied and ﬂange-like (2) (Forster, 1999).
(136) Sulcus along distal femoral tibioﬁbularis crest: weak (0),
marked (1) (Rowe, 1989).
(137) Tibial lateral malleolus: lobular (0), tabular (1) (Sereno,1999).
(138) End of cnemial process: rounded (0), proximodistally expanded
(1) (Forster, 1999).
(139) Distal tibia shape: circular/elliptical (0), mediolaterally elongate
(1) (Gauthier, 1986; Sereno et al., 1994).
(140) Medial ﬁbular fossa: posterior groove (0), deep sulcus (1), shal-
low/absent (2) (Gauthier, 1986; Rowe, 1989; Sereno et al., 1996).
(141) Anterior protuberance on ﬁbula for M. ilioﬁbularis: moderate
(0), pronounced (1) (Rauhut, 2000).
(142) Fibula fused to astragalar ascending process in adults: absent
(0), present (1).
(143) Astragalar distal condyles oriented: ventrally (0), anteroven-
trally (1) (Sereno et al., 1994).
(144) Fossa at base of astragalar ascending process: absent (0), pre-
sent (1) (Welles and Long, 1974).
(145) Astragalar ascending process: wedge-shaped/blocky (0), plate-
like/laminar (1) (Sereno et al., 1994).
(146) Ascending process of astragalus height:
(1) depth of
body (Welles and Long, 1974).
(147) Horizontal groove across anterior face of astragalar condyles:
absent (0), present (1) (Welles and Long, 1974).
(148) Astragalar facet for ﬁbula: large/dorsal (0), reduced/lateral (1).
(149) Astragalus and calcaneum: separate (0), fused (1) in adults
(Welles and Long, 1974).
(150) MT I length:
(1) 50% MT II length (Gauthier, 1986).
(151) Metatarsal midshaft widths: II
IV and both
III (0), II
both IV and III (1).
(152) Proximal end of metatarsal III: rectangular (0), hourglass-
shaped (1) (Paul, 1984; Sereno et al., 1994).
(153) MT V distal end: articular (0), non-articular (1) (Gauthier,
1986; Rauhut, 2000).
(154) Metatarsal V length:
(1) 50% length of metatarsals
II–IV (Gauthier, 1986).
(155) Antarctometatarsus (posteriorly expanded metatarsal III): ab-
sent (0), present (1) (Holtz, pers. comm.).
(156) Single (0) or double (1) vascular grooves on pedal unguals
(Sampson et al., 2001; Novas and Bandyopadhyay, 2001).
(157) pedal digit II ungual: symmetrical (0), asymmetrical (1).
(158) pedal digit I phalanges 1
(1) length of III-
1 (Sereno et al., 1994).
534 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 22, NO. 3, 2002
Taxon-character state matrix used in this phylogenetic analysis. 0, 1, 2
character states; ?
not known; multistate codings
10 20 30 40 50 60 70 80
000?0 00000 ?1000 0?000 00000 00000 0100? 000?? ???0? ?0?00 0?000 00000 ?0000 0?0?? 00000 00000
00000 00000 00100 0?000 00000 00000 0010? 00000 00000 00?00 00000 00000 00000 00001 00000 00000
01101 00000 01000 0?000 00000 00000 00000 00000 00?00 ?0?00 0?000 00010 00011 10011 00000 01011
01101 00000 01000 00000 00000 00000 00000 00000 00000 00?00 0?000 00010 00011 10011 00000 01011
0???? 0?00? ?10?? ????0 00??0 0???? ???00 ?0?0? ???0? ????? 0???? ???10 00011 1???? ?0000 01011
01101 00100 00000 ?0000 00000 00001 00110 (01)0000 00000 0??00 00000 01101 00111 10011 00000 01001
????? ????? ????? ????? ????? ????? ????? ????? ????? ????? ????? ????? ?0011 ????? ?0000 0?011
00012 00101 00010 1?000 00000 00001 01000 11021 10010 ?1000 0?00? 10100 00110 21100 10000 12001
????? ????? ????? ????? ????? ????? ????? ????? ????? ????? ????? ????0 0???? ????? ????? ?????
????? ????? ????? ????? ?0120 0???? ????? ?0??? ????? ????? ????? ????? ?0110 ????? ?1100 12100
????? ????? ????? ????? ????? ????? ????? ????? ????? ????? ????? ????? ?0110 ????? ????? ?????
100?2 00100 ?0111 1?100 10121 ????0 1???? 11121 ?001? ????? ????? ??1?? 10110 ????? ????? ?????
100?2 00100 ?0111 1?111 10121 11000 11010 11121 ?001? ???11 0?011 10100 10110 21100 11101 12101
10012 00100 00111 1?111 10121 11000 11010 1?121 10010 01011 00011 10100 10110 21100 11100 12101
????? ????? ????? ????? ????? ????? ????? ????? ????? ????? ????? ????? ?0100 ????? ?1010 1?10?
00??? 00101 110?? ????? ????? ????? ????? ????? ????? ???11 0??11 1?101 00100 1???? ?1010 1?10?
00??? 0010? 110?? ????? ????? ????? ????? ?0?2? ????? ????? ????? ??1?1 00100 ????? ?1011 1?10?
00??? 20111 000?? ????? ?1000 00111 00110 ??02? ????? ????? ????? ?11?0 01110 10100 10000 00000
00000 20111 000?? ?1??0 ?1000 01110 0???? ?0?20 0?0?? ???00 ????? 1?100 01110 10100 00000 00000
000?2 20111 000?? ????? ?1000 01110 0?110 ?0?2? ????? ????? ????? 0?1?0 01110 10100 10000 02000
00000 11110 00001 01000 00010 00001 00111 20010 21101 11100 11100 11100 01110 11100 10000 02000
00000 11110 00001 01000 00110 00001 00111 20010 21101 11100 11100 11100 01110 11100 10000 02000
000?0 011?0 ?0000 0?000 00000 000?0 0011? 2001? ???0? ???00 ??100 ?1101 ?1110 ????? ?0000 00000
APPENDIX 3. (Extended)
90 100 110 120 130 140 150
00000 00000 ?0000 00000 00000 00000 00000 00000 10000 00000 ?0000 0000? 00000 00000 00000 000
00100 00000 00001 ?1?00 00000 00000 00000 00000 01000 00000 00000 00000 00000 00000 00000 000
00010 01001 00000 00000 00011 00011 00100 11101 10010 00010 10001 ?1000 00000 00011 00100 000
00010 01001 00000 00000 00011 00011 00100 11101 10010 00010 10002 11000 00000 00011 00100 00?
00010 0?000 00??0 00000 000?1 0?001 00100 10101 00010 00010 ?0000 01000 00000 0001? 00??0 0??
00010 00000 000?0 00000 000?1 00001 00100 1010(02) 00000 10000 ?0110 00002 00010 11011 00100 00?
00020 11001 ????? ?0001 101?? ???01 00010 1010(02) ??000 010?0 ?0(12)11 ?0012 1?0?? ?0?1? 10??1 ???
10120 11001 10000 10000 000?1 ??011 00010 101?(02) 11100 1?000 ?0111 10111 10111 1111? 00??1 0??
??0?? ????? ????? ????? ????? ???1? ?0?1? 111?? (12)???? ????(01) ?0112 1?1?1 1???? ????? ????? ???
110?? ???10 1100? ????? ????? ????? ????? ????? ????? ????? ????? ????? ????? ????? ????? 11?
??1?? ????? ????? ????? ????? ????? ????? ????? ????? ????0 ?0112 00111 11111 1111? ????? ???
????? ????? ????? ????? ????? ????? ????? ????? ????? ????? ????? ????? ????? ????? ????? ???
11121 11110 11?00 10111 101?? ???11 00010 11110 11?00 11000 ?0112 1?1?1 1???? ???1? ????? ???
111(12)1 11100 11000 10111 101?? ????1 00010 1111? ??1?? ????0 ??11? ?0111 11111 1111? 00??1 11?
??0?? ????0 ????? ????? ????? ????? ????? ????? ????? ????? ????? ????? ????? ????? ????? ???
??0(12)0 ?1000 ????0 ?0??1 ????? ???0? ????? ????0 1111? ????0 10112 1011? ?1111 1111? 10??? 11?
??0?? ????0 11??? ????? ????? ????? ????? ????? ????? ????? ????? ????? ????? ????? 1???? ???
001?0 ??00? ??0?? ???00 01021 11??0 11011 1010(02) (12)1001 10101 ?1111 00012 001?1 11101 01110 0?1
00110 0?00? ????? 10?00 011?? ????0 11010 10100 (12)100? ??101 ?1111 00012 0?111 111?? 01??0 0??
00110 01000 ??000 10000 011?1 10?00 11010 10100 11001 00101 ?1111 00012 00111 1110? 01??0 ???
00110 00000 10111 11000 01021 11100 01011 10102 21001 10102 01211 00011 00111 11101 01110 001
00110 00000 10??1 110?? ????? ?1000 01011 10102 11001 10102 ?1211 00011 00111 11101 01110 001
00010 00000 ??11? ???00 010?1 11100 01011 1010? 2?001 00112 ?1211 0???2 00??? ????? 01??0 ???