The appendicular skeleton of Majungasaurus crenatissimus (Theropoda: Abelisauridae) from the Late Cretaceous of Madagascar

Abstract

The appendicular skeleton of the abelisaurid theropod Majungasaurus crenatissimus (Depéret, 1896) Lavocat, 1955 is described for the first time. The available materials include an incomplete pectoral girdle and forelimb, along with the ilium and a nearly complete hind limb. These materials display a number of ceratosaur, abelisauroid, and abelisaurid synapomorphies, supporting the phylogenetic placement of Majungasaurus based previously on cranial anatomy. As in Ceratosaurus and Carnotaurus, the scapular blade is relatively wide and has a pronounced dorsal lip over the glenoid. The humerus is short and bears a globular head, but is more slender than in Carnotaurus. The ilium has a preacetabular hook, a strong supraacetabular crest, a notched posterior margin, and peg-and-socket articulations with both the pubis and ischium. Hind limb elements are proportionally stocky, as in some other abelisaurids. The femur lacks a trochanteric shelf, the tibia has a greatly enlarged cnemial crest, and the fibula bears a deep, posteriorly facing medial fossa. The abelisaurid astragalocalcaneum is described here in detail for the first time, and is more similar to that of tetanurans than to those of coelophysoids. Taken together, these materials illustrate that the appendicular skeleton of abelisaurids was specialized over the typical condition in basal theropods, particularly through the development of enlarged muscle attachment processes.

Full text

THE APPENDICULAR SKELETON OF MAJUNGASAURUS CRENATISSIMUS (THEROPODA:
ABELISAURIDAE) FROM THE LATE CRETACEOUS OF MADAGASCAR
MATTHEW T. CARRANO
Department of Paleobiology, Smithsonian Institution, P.O. Box 37012, MRC 121, Washington, DC 20013-7012;
carranom@si.edu
ABSTRACT—The appendicular skeleton of the abelisaurid theropod Majungasaurus crenatissimus (Depéret, 1896)
Lavocat, 1955 is described for the first time. The available materials include an incomplete pectoral girdle and forelimb,
along with the ilium and a nearly complete hind limb. These materials display a number of ceratosaur, abelisauroid, and
abelisaurid synapomorphies, supporting the phylogenetic placement of Majungasaurus based previously on cranial
anatomy. As in Ceratosaurus and Carnotaurus, the scapular blade is relatively wide and has a pronounced dorsal lip over
the glenoid. The humerus is short and bears a globular head, but is more slender than in Carnotaurus. The ilium has a
preacetabular hook, a strong supraacetabular crest, a notched posterior margin, and peg-and-socket articulations with
both the pubis and ischium. Hind limb elements are proportionally stocky, as in some other abelisaurids. The femur lacks
a trochanteric shelf, the tibia has a greatly enlarged cnemial crest, and the fibula bears a deep, posteriorly facing medial
fossa. The abelisaurid astragalocalcaneum is described here in detail for the first time, and is more similar to that of
tetanurans than to those of coelophysoids. Taken together, these materials illustrate that the appendicular skeleton of
abelisaurids was specialized over the typical condition in basal theropods, particularly through the development of
enlarged muscle attachment processes.
MALAGASY ABSTRACT (FAMINTINANA)—Sambany izao no namelabelabelarin’i Lavocat tamin’ny 1955 ny
momban’ny taolan-drambon’ny abelisaurid theropod Majungasaurus crenatissimus (Depéret, 1896). Ireo karazan-taolana
nisy tamin’ireo dia maro ny ceratosaur sy abelisauroid ary abelisaurid synapomorphies, izay nanamarina ny toerana
phylogenetic –n’ny Majungasaurus izay tamin’ny bikan’ny karan-doha ny nametrahana azy. Toy ny an’ny Ceratosaurus
sy Carnotaurus dia azo ambara fa mivelatra ny taolan-tsoroka ary iny faritra ambonin’ny glenoid iny dia misy molony
aoriana mivoitra mazava tsara. Fohy ny taolan-tsandry ary borobory ny lohany, saingy marotsadrotsaka raha ampitahaina
ny an’i Carnotaurus. Ny ilium dia ahitana faingoka alohan’ny acetabular, ny tampony supraacetabular matanjaka, misy
faingoka aoriana amin’ny sisiny, ary lavaka fitoerana miaro vohitra mahatazona ny fifanohizan’ny taolana izay miaraka
amin’ny pubis sy ischium. Ny taolan-tongotra dia mitovitovy ny fahafohezany, toy ireo sasany amin’ny abelisaurids. Tsy
ahitana trochanteric self ny taolam-pe, ny tibia (taolan-dranjo iray) dia misy vohitra cnemial mivelatra be, ary ny fibula
(taolan-dranjo iray hafa) dia mitondra lavaka lalina manatrika aoriana. Ny abesilaurid astragalocalcaneum dia novela-
belarina volalohany tamin’ny antsipirihiny eto, izay toy ny natao tamin’ny tetanurans ka mihoatra ny natao tamin’ny
coelophysoids. Rehefa nojerena miaraka dia tsapa fa ireo taolana ireo dia manazava fa ny taolan-damosin’ny abelisaurids
dia voatokana manokana tamina fisehoan-javatra izay tsy mahazatra teo amin’ny faritra ambany amin’ny vatan’ny
theropods, indrindra indrindra ny fisian’ny fivelaran’ny toerana fipetrahan’ny hozatra.
INTRODUCTION
The first theropod materials in Madagascar were discovered
by French military personnel in the 1890s and subsequently
described by Charles Depéret (1896a, b). As was common in
19
th
-century dinosaur paleontology, Depéret allocated these
fragmentary theropod elements from the Upper Cretaceous
Maevarano Formation to a species of Megalosaurus (M. cren-
atissimus), and subsequently to the genus Dryptosaurus
(Depéret and Savornin, 1928). Although Lavocat (1955) later
referred a dentary to this form, separating it from Megalosaurus
as Majungasaurus crenatissimus, its anatomy and broader phy-
logenetic relationships remained obscure.
Two decades after Depéret described the Malagasy remains,
Matley (1921) reported the presence of at least two ‘megalosau-
rians’ from the Upper Cretaceous Lameta Formation of India.
He later removed some of these materials—including ilia, tibiae,
a sacrum, and a number of scutes—and placed them into a
new stegosaur taxon, Lametasaurus indicus (Matley, 1924).
Lametasaurus was later recognized as a theropod by Chakra-
varti (1934, 1935) and Walker (1964), thus ‘re-associating’ it with
the original Lameta theropod materials. Huene and Matley
(1933) named two additional theropods from the same deposit
(Indosaurus matleyi and Indosuchus raptorius), along with Dryp-
tosauroides grandis. Like Depéret and Lavocat, however, Huene
and Matley did not appreciate the distinctive nature of their taxa
at the suprageneric level, and referred them to the Allosauridae.
Chatterjee (1978) later explicitly supported the notion of two
distinct large theropods in the Lameta Beds by suggesting that
Indosuchus was a tyrannosaurid.
The peculiar morphological specializations of the Indian and
Malagasy theropods went unappreciated until the discovery and
description of the more complete South American forms Abeli-
saurus comahuensis (Bonaparte and Novas, 1985) and Carnotau-
rus sastrei (Bonaparte, 1985; Bonaparte et al., 1990). These taxa
were recognized as distinct from allosaurids, ‘megalosaurs,’ and
tyrannosaurids, instead having shared a closer phylogenetic his-
tory with the unusual North American theropod Ceratosaurus
nasicornis (and possibly with Indosaurus and Indosuchus;
Bonaparte and Novas, 1985). In particular, the abelisaurid skull
and vertebral column were noted as being highly derived, differ-
ing markedly from those of tetanurans. The appendicular skel-
eton also appeared to be very specialized, but was comparatively
incompletely known.
With these South American examples in hand, several authors
(Bonaparte et al., 1990; Molnar, 1990) suggested that the Mala-
gasy and Indian theropods were probably also members of the
Abelisauridae. Unfortunately, these remained fragmentary and
Society of Vertebrate Paleontology Memoir 8
Journal of Vertebrate Paleontology
Volume 27, Supplement to Number 2: 163–179, June 2007
© 2007 by the Society of Vertebrate Paleontology
163

therefore poorly understood. More recently, however, Sampson
and colleagues (1996, 1998) described new, more complete
theropod materials from the Maevarano Formation of Madagas-
car. They regarded M. crenatissimus as a nomen dubium because
it was not distinguishable on morphological grounds from other
abelisaurids. Instead they referred the newly discovered materi-
als to Majungatholus atopus, which originally had been described
as a pachycephalosaurid (Sues and Taquet, 1979). However, as
detailed elsewhere (Krause et al., this volume), new materials
have made it clear that: (1) Majungasaurus can be distinguished
from other abelisaurids; (2) only one large abelisaurid is present
in the Maevarano Formation; and (3) this taxon should be re-
ferred to as Majungasaurus crenatissimus, with M. atopus as a
junior synonym.
Majungasaurus is now known from numerous specimens that
preserve nearly the entire skull and vertebral column, as well as
most of the appendicular skeleton, making it one of the best-
known abelisaurids (Fig. 1). Although only portions of the fore-
limb have been recovered, much of the hind limb skeleton is
preserved. This is in contrast to nearly all other abelisaurids,
where this region of the skeleton is either incomplete (e.g., Car-
notaurus,Xenotarsosaurus,Lametasaurus,Ekrixinatosaurus)or
entirely unknown (e.g., Abelisaurus,Indosaurus,Indosuchus).
Although the holotype of Aucasaurus garridoi includes nearly
complete hind limb materials, these have been only preliminarily
described (Coria et al., 2002).
In this article, I describe the appendicular materials of Majun-
gasaurus and discuss their relevance to phylogenetic and func-
tional interpretations of this taxon and other abelisaurids. Be-
cause many major theropod clades are diagnosed by features of
the hind limb, understanding these structures in abelisaurids sig-
nificantly affects assessments of their phylogenetic position. In a
broader sense, these features also bear on the placement of abe-
lisauroids and neoceratosaurs relative to coelophysoids and tet-
anurans (i.e., the monophyly or paraphyly of Ceratosauria sensu
Gauthier, 1986). Finally, the apomorphic nature of the abelisau-
rid appendicular skeleton highlights several potential functional
specializations in this unusual theropod clade.
Institutional Abbreviations—DGM, Museu de Ciências da
Terra, Rio de Janeiro, Brazil; FMNH, Field Museum of Natural
History, Chicago, IL; FSL, Facultédes Sciences, Universitéde
Lyon, France; GSI, Geological Society of India, Kolkata, India;
HMN, Humboldt Museum für Naturkunde, Berlin, Germany;
ISI, Indian Statistical Institute, Kolkata, India; MACN, Museo
Argentino de Ciencias Naturales “Bernardino Rivadavia,”
Buenos Aires, Argentina; MCF-PVPH, Museo “Carmen
Funes,”Plaza Huincul, Argentina; MNHN, Muséum National
d’Histoire Naturelle, Paris, France; MPCA-PV, Museo Provin-
cial “Carlos Ameghino,”Cipoletti, Argentina; MWC, Museum
of Western Colorado, Fruita, CO; UA, Universitéd’Antanana-
rivo, Antananarivo, Madagascar; UMNH VP, Utah Museum of
Natural History, Salt Lake City, UT; UNPSJB-PV, Universidad
Nacional de Patagonia “San Juan Bosco,”Comodoro Rivadavia,
Argentina; USNM, National Museum of Natural History, Smith-
sonian Institution, Washington, DC.
Comparative Taxa and Specimens—The following specimens
were examined for the comparisons mentioned in this paper.
When literature illustrations were used, the appropriate refer-
ence is given below. Abelisauridae indet. (GSI 296, K27/525, 539,
558, 560, 568-570, 620, 653-654, 658-659, 671, Huene and Matley,
1933; ISI R91/1); Aucasaurus garridoi (MCF-PVPH 236); Car-
notaurus sastrei (MACN-CH 895); Ceratosauria indet. (HMN 37,
69); Ceratosaurus nasicornis (MWC 1.1; UMNH VP 5278,
USNM 4735); Elaphrosaurus bambergi (HMN Gr. S. 38-44); Ge-
nusaurus sisteronis (MNHN Bev-1); Ilokelesia aguadagrandensis
(MCF-PVPH 35); Lametasaurus indicus (GSI uncat., Matley,
1924); Masiakasaurus knopfleri (FMNH PR 2112, 2114-2119,
2129-2132, 2120-2123, 2134-2136, 2143, 2146-2155, 2158-2161,
2167, 2169, 2171-2176, 2205-2106, 2208, 2214-2119, 2223-2225,
2227, 2234, 2236; UA 8681, 8683-8684, 8686, 8693-8694, 8700,
8710-8711, 8713-8714); Pycnonemosaurus nevesi (DGM 859-R,
Kellner and Campos, 2002); Quilmesaurus curriei (MPCA-PV
100); Rajasaurus narmadensis (GSI 21141/1-33, Wilson et al.,
2003); Tarascosaurus salluvicus (FSL 330 201-3); Xenotarsosau-
rus bonapartei (MACN 1468, cast of UNPSJB-PV 184/PVL 612).
SYSTEMATIC PALEONTOLOGY
DINOSAURIA Owen, 1842
SAURISCHIA Seeley, 1888
THEROPODA Marsh, 1881
CERATOSAURIA Marsh, 1884
ABELISAUROIDEA (Bonaparte and Novas, 1985)
ABELISAURIDAE Bonaparte and Novas, 1985
MAJUNGASAURUS Lavocat, 1955
MAJUNGASAURUS CRENATISSIMUS (Depéret, 1896)
Lavocat, 1955
Type Specimen—MNHN MAJ-1, nearly complete right den-
tary of subadult individual (Lavocat, 1955).
Referred Specimens—See complete listing in Krause et al.
(this volume).
FIGURE 1. Skeletal reconstruction of Majungasaurus crenatissimus in left lateral view, showing known skeletal elements in white, unknown
elements in gray.
SOCIETY OF VERTEBRATE PALEONTOLOGY, MEMOIR 8164

Revised Diagnosis—See Krause et al. (this volume).
Age and Distribution—Appendicular materials of Majungas-
aurus were found at several localities (MAD93-01, 93-18, 93-19,
93-20, 93-32, 93-33, 93-35, 93-73, 95-14, 95-16, 96-01, 96-07, 96-18,
96-21, 99-26, 99-32, 99-33, and 01-05) near the village of Beriv-
otra, in the Mahajanga Basin of northwestern Madagascar
(Krause et al., 1997; Sampson et al., 1998). Most materials of
Majungasaurus derived from the Anembalemba Member, the
uppermost white sandstone stratum of the Maevarano Forma-
tion (Maastrichtian, Upper Cretaceous) (Rogers and Hartman,
1998; Rogers et al., 2000, this volume).
Described Material—FMNH PR 2423 (right humerus), UA
9031 (left humerus), UA 8678 (left and right ilia), UA 9032 (left
tibia), FMNH PR 2424 (left tibia), UA 9077 (left tibia and
fibula), UA 9078 (right fibula), FMNH PR 2425 (left astragalo-
calcaneum), UA 9033 (right astragalocalcaneum), UA 9082 (as-
tragalus), UA 9034 (left metatarsal II), UA 9079 (left metatarsal
III), UA 9035 (left metatarsal IV), UA Bv 532 (left pedal pha-
lanx I-2), UA Bv 1658 (left pedal phalanx I-2), UA 9036 (left
pedal phalanx II-1), FMNH PR 2426 (right pedal phalanx II-1),
UA Bv 1260 (right pedal phalanx II-1), FMNH PR 2428 (right
pedal phalanx II-1, left pedal phalanx II-3 and III-2), FMNH PR
2427 (right pedal phalanx II-2), UA 9037 (right pedal phalanx
II-2), UA 9038 (left pedal phalanx II-3), UA Bv 1265 (left pedal
phalanx III-1), FMNH PR 2429 (left pedal phalanx III-1), UA
9039 (right pedal phalanx III-1), UA 9042 (left pedal phalanx
III-2), UA 9081 (right pedal phalanx III-1 or III-2), FMNH PR
2430 (right pedal phalanx IV-1), UA 9040 (right pedal phalanx
IV-1), FMNH PR 2431 (left pedal phalanx IV-3), UA 9041 (right
pedal phalanx IV-2), FMNH PR 2432 (left pedal phalanx IV-4),
FMNH PR 2433 (right pedal phalanx IV-4), FMNH PR 2434
(left pedal phalanx IV-5), UA 9043 (pedal phalanx IV-5), and
FSL 92.290 (pedal ungual phalanx).
FMNH PR 2278 is an associated skeleton that includes cranial,
axial, and appendicular elements from site MAD99-26. Among
the latter have been identified a left scapulocoracoid, partial left
ilium, left femur, left and fragmentary right tibiae, left and partial
right fibulae, left astragalocalcaneum, left metatarsals II-IV,
right pedal phalanges II-1 and IV-2, and left pedal phalanges
IV-2 and IV-3.
Specimens UA 9031, 9033-9037, 9040-9041, 9077-9079, 9081-
9082, and FMNH PR 2430 come from a single quarry horizon
(site MAD99-33), along with a nearly complete skull. Although
there is some duplication of elements (notably two right pedal
IV-2 phalanges and two right premaxillae), most of the materials
are consistent with having been derived from a single subadult
individual.
DESCRIPTION AND COMPARISONS
Pectoral Girdle
The pectoral girdle of Majungasaurus is represented only by a
single, incomplete scapulocoracoid (FMNH PR 2278). The
clavicles and sternum are not known.
Scapulocoracoid—FMNH PR 2278 includes an incomplete,
coossified scapulocoracoid (FMNH PR 2278; Fig. 2). The scapu-
lar blade is long, with nearly parallel anterodorsal and postero-
ventral edges where they are complete (approximately the ven-
tral one-third). Although the anterodorsal edge of the distal end
is missing, the posteroventral edge does not flare appreciably,
thus resembling the condition in tetanurans and Herrerasaurus
more than that in coelophysoids and Eoraptor. Unlike the con-
dition in tetanurans, however, the blade is relatively wide an-
teroposteriorly, as in abelisauroids (Masiakasaurus,Carnotau-
rus,Aucasaurus), Ceratosaurus, and more primitive taxa. Only
the base of the acromial process is preserved. The posteroven-
trally-facing glenoid has prominent lips both dorsally and ven-
trally, with the former being the most pronounced. The postero-
ventral process appears to have been relatively deep dorsoven-
trally, with a moderately concave notch between it and the
glenoid. In general shape and proportions, this element is most
similar to the scapulocoracoids of Masiakasaurus,Carnotaurus,
and Aucasaurus.
On the lateral surface, the scapulocoracoid suture is faintly
visible as a roughened ridge that passes anteriorly from the glen-
oid along a slightly undulating line. A large, elliptical rugosity sits
ventral to this, just anterior to the glenoid and dorsal to the
coracoid foramen (Fig. 2). The coracoid foramen faces ventrally
in lateral view, and opens into a dorsally directed channel that
passes through the bone to the medial side. A small, faint tu-
FIGURE 2. Left scapulocoracoid of Majungasaurus crenatissimus
(FMNH PR 2278) in lateral view. Abbreviations:cf, coracoid foramen;
dl, dorsal lip; g, glenoid; pvp, posteroventral process; r, rugosity. Dashed
lines indicate reconstructed outlines of element. Scale bar equals 10 cm.
CARRANO—APPENDICULAR SKELETON OF MAJUNGASAURUS 165

bercle is visible ventral to the glenoid, between it and the pos-
teroventral process; this may have been associated with the origi-
nation for M. biceps (Ostrom, 1974; Welles, 1984, Carpenter and
Smith, 2001; Jasinoski et al., 2006) and/or M. coracobrachialis
brevis (Osmólska et al., 1972; Walker, 1977; Jasinoski et al.,
2006). The lateral surface of the scapular blade is smooth.
The medial surface of the scapular blade is more poorly pre-
served and illuminates few additional details. In posterior view,
the blade can be seen to curve broadly, with its concave face
directed medially (against the ribcage). Its thickness decreases
from proximal to distal, as well as from posteroventral to an-
terodorsal. The glenoid is approximately reniform, with a slightly
longer dorsal (scapular) component. The scapulocoracoid suture
between these two components is unclear. Additionally, there is
a deep fossa just dorsal to the glenoid that exaggerates the
prominence of the dorsal lip, as in other ceratosaurs.
Forelimb
The forelimb of Majungasaurus is represented only by the
humerus. All remaining elements (radius, ulna, carpus, and ma-
nus) are unknown and cannot be compared to the highly derived
corresponding elements of the abelisaurids Carnotaurus and Au-
casaurus. These elements are also largely unknown in other abe-
lisauroids (with the exception of individual manual phalanges in
Noasaurus and Masiakasaurus; Carrano et al., 2004), and thus
are currently of limited use in phylogenetic assessments of these
taxa. The forelimb materials known for Ceratosaurus are rela-
tively short compared to those of other theropods, but lack the
peculiar derived morphology seen in abelisaurids.
Humerus—The humerus is short but not particularly stocky
(Fig. 3; Table 1), similar in proportions to the same bone in adult
tyrannosaurids (Carpenter and Smith, 2001; Currie, 2003). The
long axes of the proximal and distal ends are slightly twisted
relative to each other in Majungasaurus and other abelisaurids,
more than in coelophysoids but less than in tetanurans. The
humeral shaft is nearly straight in anteroposterior view (Fig. 3B,
D), bowed anteroposteriorly, and flares as it approaches the dis-
tal end. The proximal end is dominated by a large, rounded head,
and the accompanying internal tuberosity and greater tubercle
are comparatively small. The humeral head and internal tuber-
osity both expand anteroposteriorly from the shaft.
The deltopectoral crest is a low, rugose ridge that curves from
the lateral surface (proximally) onto the anterior midshaft. It
FIGURE 3. Right humerus of Majungasaurus crenatissimus (FMNH PR 2423). A, anteromedial view; B, anterolateral view; C, posterolateral view;
D, posteromedial view; E, proximal view; F, distal view. Abbreviations:dpc, deltopectoral crest; ecc, ectepicondyle; enc, entepicondyle; gt, greater
tubercle; hh, humeral head; it, internal tuberosity; lds, scar for M. latissimus dorsi; rc, radial condyle; uc, ulnar condyle. Scale bar equals 10 cm.
SOCIETY OF VERTEBRATE PALEONTOLOGY, MEMOIR 8166

likely served as the insertion point for Mm. pectorales (Dilkes,
2000, 2001; Carpenter and Smith, 2001; Jasinoski et al., 2006).
The crest is widest at its distal end, and approximately half again
as high as the shaft is wide. A large fossa located just distal to the
humeral head on the anterior surface may mark the insertion of
M. coracobrachialis (Carpenter and Smith, 2001; Jasinoski et al.,
2006).
The cross-section of the proximal shaft is elliptical, whereas
that of the distal half is nearly circular. At the distal end, a faint
ridge runs obliquely proximomedially away from the radial con-
dyle, delineating the lateral border of the shallow intercondylar
depression. The medial border is more pronounced, and runs
more directly proximally from the weakly convex ulnar condyle.
The convexity of the radial condyle is particularly evident in
posteromedial view (Fig. 3D).
In anteroposterior view, the humeral head appears more
asymmetrically rounded than in mediolateral view. A large, oval,
rugose ridge extends roughly parallel to the distal deltopectoral
crest on the posterolateral side of the bone, with a small fossa
between them proximally. An additional fossa is positioned ad-
jacent to this second ridge, separating it from a rounded, rugose
bump more posteriorly; this probably represents the insertions of
M. latissimus dorsi and part of M. deltoideus (Fig. 3C; Dilkes,
2000, 2001; Carpenter and Smith, 2001; Jasinoski et al., 2006).
In proximal view (Fig. 3E), the humeral head is rounded and
bulbous, with no clear long axis. The internal tuberosity is visible
as a triangular medial projection, but the greater tubercle is in-
distinct from the deltopectoral crest, which descends from the
greater tubercle as a ridge. The distal end is unusual in lacking
prominent, convex condyles for the radius and ulna, instead
bearing nearly flat articular surfaces. In distal view (Fig. 3F), this
end is irregularly shaped, with two slightly concave swellings that
represent the articular surfaces for the radius and ulna. A small
central depression separates these indistinct condyles.
The humerus of Majungasaurus is strikingly similar to those of
Carnotaurus and Aucasaurus, the only other abelisaurids for
which this bone is known. In all three taxa, this bone is propor-
tionally short, lacks significant mediolateral curvature, and bears
a globular head and flattened distal condyles. Of the two, the
humerus of Majungasaurus more closely resembles that of Au-
casaurus than the bulkier element of Carnotaurus. The humeri of
Masiakasaurus and Elaphrosaurus also share several features
with that of Majungasaurus (Carrano et al., 2002), but are rela-
tively longer and more slender.
Pelvic Girdle
The pelvic girdle of Majungasaurus is represented only by the
ilium; the pubis and ischium are not known.
Ilium—The ilium is considerably longer than tall (Fig. 4; Table
1), and greatly resembles the same element in Carnotaurus,Au-
casaurus,Lametasaurus,Rajasaurus, and Genusaurus.Asin
other neotheropods, the ilium is elongated anteroposteriorly
past the pubic and ischial peduncles, and bears an extensive,
largely concave lateral surface (Fig. 4). In lateral view, the an-
terior blade is lobate, curving ventrally below its point of origin
on the pubic peduncle. This flat surface bears several weak lon-
gitudinal crenulations and is angled out ventrolaterally. The ven-
tralmost point of the postacetabular blade (Fig. 4A) forms a
“hook”that is located midway between the pubic peduncle and
the anteriormost edge, demarcating the anterior edge of the gap
for passage of M. puboischiofemoralis internus 2. The origina-
tion fossa for M. puboischiofemoralis 1, on the anterior surface
of the pubic peduncle, is faint. The central portion of the lateral
face is slightly concave with a straight dorsal margin. Along the
dorsal edge, a faint band of parallel, dorsoventrally-oriented
lines marks the origination of Mm. iliotibiales. As in most thero-
pods, there is no clear division between the origins of M. ilio-
femoralis and M. iliofibularis.
More posteriorly, the dorsal edge curves ventrally, meeting
the posterior border at a distinct corner. As in other abeli-
sauroids, the posterior border itself is unusual in having an un-
dulating margin, concave in its dorsal third and convex below.
Ventrally, the posterior iliac blade is concave, and the thick,
rounded edge is rugose along its posterior third. The blade is
concave in this region as well, a feature exaggerated by the lat-
eral position of the ventral border, which marks the lateral extent
of the brevis fossa. Medial to this edge, the brevis fossa both
widens and deepens posteriorly as in other ceratosaurs. The ac-
etabulum is tall, open medially, and has a parabolic outline.
Dorsolaterally, it is roofed by a well-developed and somewhat
pendant supraacetabular shelf. The shelf extends ventrolaterally
to form a lobate extension over the posterior half of the acetabu-
lum.
As in tetanurans, the ischial peduncle is smaller than the pubic
peduncle. In Majungasaurus (as in Masiakasaurus and Gigano-
tosaurus), it is developed into a ventrally extending projection
that presumably lodged into a corresponding fossa in the iliac
peduncle of the ischium. The peduncle is oriented posteroven-
trally, and the process occupies only the anterior half of its ven-
tral contact surface. The pubic peduncle is more unusual; it is
larger than the ischial peduncle (as in tetanurans), but also bears
a distinct peg that is extended ventrally below the remainder of
the peduncle. This peg is presumed to have lodged into a corre-
sponding socket in the pubis, although this bone is unknown in
Majungasaurus. This is similar to the condition in Masiakasaurus
(Sampson et al., 2001; Carrano et al., 2002), Lametasaurus (Mat-
ley, 1924), Rajasaurus (Wilson et al., 2003), and Ceratosaurus
(Britt et al., 1999, 2000).
Medially, the contact points for at least four sacral vertebrae
are evident as paired dorsal and ventral rugosities; only a single
TABLE 1. Measurements (in cm) of appendicular elements of Majungasaurus crenatissimus.Abbreviations:4t, distance from proximal end to
fourth trochanter; AP, anteroposterior diameter; C, circumference; e, estimated; GH, glenoid height; GW, glenoid width; L, left; ML, mediolateral
diameter; TL, total length. Dashes indicate where measurements could not be taken; plus signs (+) indicate that some length is missing from the
element as preserved.
Specimen Element TL AP ML C 4t GW GH
FMNH PR 2278 L scapulocoracoid 645 111 21.7 39.7 64.3
R femur 568+ —80.5 —205+
L tibia 455++ 55.2 70.2 218
L fibula (406)e 32.8 18.9 84
R fibula (406)e 31.8 19.5 84
MT II 198.9 32.4 24.3 105
MT III 250.0 35.2 39.5 132
MT IV 207.7 24.8 35.1 93
UA 9077 L fibula 406.4 33.4
UA 9036 L II-1 88.0 30.0 26.0
UA 9037 R II-2 45.0 19.0 23.0
CARRANO—APPENDICULAR SKELETON OF MAJUNGASAURUS 167

contact is present for the first sacral (Fig. 4B). Along the dorsal
margin, long, parallel dorsoventral ridges run along nearly the
entire length of the bone. The brevis fossa is inset medially to a
considerable degree, forming a distinct medial shelf whose ven-
tral border turns from convex to strongly concave as it ap-
proaches the ischial peduncle. The pegs of both peduncles are
more prominent in medial view, appearing to extend directly
from the medial wall of the bone. The anterior blade diverges
laterally from the midline, beginning at a point just anterior to
the base of the pubic peduncle. The medial edge of the dorsal
acetabulum is thin.
In posterior view (Fig. 4F), the brevis fossa can be seen to
expand both medially and laterally beneath the main iliac blade.
Longitudinal striations are visible on most of the internal surface
of the fossa. Its lateral wall extends farther ventrally than medi-
ally. In anterior view (Fig. 4E), the lateral deviation of the an-
terior iliac blade is quite evident, which creates a distinct passage
for M. puboischiofemoralis internus 2. A well-defined fossa at
the junction of the blade and pubic peduncle marks the begin-
ning of this passage. The anteriormost end of the iliac blade is
thinner and rougher than the other edges.
Hind Limb
Most of our knowledge of the hind limb of Majungasaurus is
derived from one specimen, FMNH PR 2278, a partial skeleton
FIGURE 4. Left ilium of Majungasaurus crenatissimus (UA 8678). A, lateral view; B, medial view; C, dorsal view; D, ventral view; E, anterior view;
F, posterior view. Abbreviations:bf, brevis fossa; ifis, scar for M. iliofibularis; ip, ischial peduncle; its, scar for Mm. iliotibiales; lbf, lateral wall of
brevis fossa; mbf, medial wall of brevis fossa; pp, pubic peduncle; sac, supraacetabular crest; sr2-5, attachment surfaces for sacral ribs 2-5; st1-5,
attachment surfaces for transverse processes of sacral vertebrae 1-5; vph, ventral preacetabular hook. Scale bar equals 10 cm.
SOCIETY OF VERTEBRATE PALEONTOLOGY, MEMOIR 8168

that includes most of the hind limb elements. This specimen
indicates that the hind limb is unusually short relative to other
skeletal elements (Fig. 1; Table 1). This stockiness is especially
apparent in the tibia and metatarsals, which are robust for their
size, but even the femur is relatively short (although not particu-
larly robust).
Nearly the entire hind limb of Majungasaurus is known, with
the exception of metatarsals I and V, and pedal phalanges I-1
and III-4. Among abelisaurids only Aucasaurus has a more com-
pletely preserved hind limb (Coria et al., 2002).
Femur—The femur is known only from parts of two damaged
elements, both pertaining to FMNH PR 2278. The head, neck,
and much of the anterolateral surface have been weathered away
in even the most complete specimen (Fig. 5). The exposed inter-
nal structure reveals a large central cavity that permeated the
entire femoral shaft, as is typical for theropods. Proximally, the
base of the lesser trochanter is preserved, as well as the deepest
portion of the sulcus between it and the main femoral shaft. This
sulcus indicates that the lesser trochanter was elevated, probably
comparable to the condition in Carnotaurus,Xenotarsosaurus,
Tarascosaurus,Genusaurus,andCeratosaurus. The preserved
portion of the lesser trochanter is extremely rugose. Near its
base, but more distolaterally positioned, a rounded bump marks
the insertion of M. iliofemoralis externus. These two structures
result from complete separation of the primitive trochanteric
shelf (Hutchinson, 2001), as in tetanurans but not coelophysoids.
A flat, longitudinally striated facet marks the lateral edge of the
greater trochanter, and represents the insertion of Mm. pubois-
chiofemorales externi (Hutchinson, 2001).
Still farther distally, the posteromedial surface shows a well-
preserved, ridge-like fourth trochanter (for Mm. caudofemo-
rales) situated approximately two-fifths of the way down the
shaft. A rugose region extending proximally onto the posterior
shaft surface represents the additional insertion area for M. cau-
dofemoralis brevis. Compared to the other prominent limb
muscle attachment sites, the fourth trochanter appears to be
surprisingly small, although the relevance of this to actual muscle
size is questionable (e.g., Bryant and Seymour, 1990). The inser-
tion of M. adductor femoris 1 cannot be identified with certainty
because the posteromedial side of the distal shaft is damaged.
However, M. adductor femoris 2 insertion is complete and lo-
cated about two-thirds down the shaft, along its lateral edge; the
insertion is flat and longitudinally elliptical.
The shaft exhibits bowing in both the anteroposterior and me-
diolateral planes, being slightly concave posteriorly and medi-
ally. Both distal condyles are present but poorly preserved. The
fibular condyle is approximately twice as wide mediolaterally as
the tibial, but the two extend the same distance distally (Fig. 5A,
C). In distal view, the fibular condyle is expanded and bulbous,
as in Masiakasaurus,Carnotaurus, and Xenotarsosaurus. The
deep, broad posterior intercondylar sulcus contains no marked
longitudinal ridges for attachment of the knee flexors. A medio-
laterally narrow tibiofibular crest is separated from the main
fibular condyle by a broad, shallow groove. The medial edge of
the bone is battered, so the presence of a prominent entepicon-
dylar crest (as in Masiakasaurus and Genusaurus; Carrano et al.,
2002) cannot be ascertained.
Tibia—The tibia of Majungasaurus is known from five speci-
mens, two isolated (UA 9032, FMNH PR 2424), one provision-
ally associated with a skull and partial skeleton at MAD93-33
(UA 9077), and a pair belonging to another partial skull and
skeleton (FMNH PR 2278). As in other abelisauroids (including
Genusaurus and Quilmesaurus), the well-developed cnemial
crest of Majungasaurus extends far anteriorly and bears a promi-
nent extensor groove on its lateral face (Fig. 6). In proximal view,
the cnemial crest extends in a broad arch from the main body of
FIGURE 5. Right femur of Majungasaurus crenatissimus (FMNH PR 2278). A, posterolateral view; B, posterolateral view of proximal end; C,
posterior view of distal end; D, distal view. Abbreviations:4t, fourth trochanter; add2s, scar for M. adductor femoris 2; fc, fibular condyle; ifes, scar
for M. iliofemoralis externus; lt, lesser trochanter; pifes, scar for M. puboischofemoralis externus; pig, posterior intercondylar groove; tc, tibial
condyle; tfc, tibiofibularis crest; tfcs, tibiofibularis sulcus. Dashed lines are reconstructed outlines. Scale bars equals 5 cm.
CARRANO—APPENDICULAR SKELETON OF MAJUNGASAURUS 169

the bone, with a wide lateral concavity that does not resemble
the distinct notch of forms such as Allosaurus,Sinraptor, and
many coelurosaurs. In lateral view (Fig. 6A), the cnemial crest
exhibits a strong dorsal curvature such that a substantial portion
of this structure occurs dorsal to the lateral and medial condyles.
The anteriormost portion of the cnemial crest is slightly ex-
panded mediolaterally, with a distinct sulcus (presumably for the
knee extensor tendons) visible along the lateral edge. The two
proximal condyles are similarly sized and separated by a weak
notch. In addition, the posterior edge of the tibia is angled
obliquely relative to the mediolateral axis, as evident in proximal
view (Fig. 6B).
The lateral fossa between the cnemial crest and the tibial shaft
is also correspondingly well-developed. In addition, the terminus
of the cnemial crest is expanded both proximally and distally,
and does not taper to a rounded point as in most theropods. This
structure is thin and delicate, and is completely preserved (or
nearly so) only in UA 9077. More posteriorly, the lateral tibia
bears a stout ridge, the fibular crest, which runs for approxi-
mately one-fourth of the bone’s length and articulates with the
fibula. A short gap sits distal to the crest, and a large foramen is
located posterior to it. Farther distally, the long, flat fibular facet
runs to a point just proximal to the expansion of the lateral
malleolus. This facet is exceptionally rugose and wide, as in other
abelisaurids, contrasting with the thinner, fainter structure of
most other theropods.
The anterior aspect of the tibia is relatively smooth and flat-
tened, with a rounded anteromedial edge. At the distal end (Fig.
6C), a triangular facet for the astragalar ascending process can be
observed, bounded proximally by an obliquely oriented buttress.
In this view, the lateral malleolus is clearly larger and descends
farther distally than the medial. The medial malleolus bears a
short proximal ridge. Medially, the tibial surface is smooth and
largely featureless, with a broadly rounded cross-section
throughout most of its length. The distal end shows the same
‘torsion’relative to the proximal end as do most theropod tibiae;
that is, the long axis of the distal end is oriented perpendicular to
that of the proximal end. The posterior shaft is also rounded and
fairly smooth, with a modest distal ridge that terminates at the
sulcus between the two malleoli.
In distal view (Fig. 6D), the tibia of Majungasaurus more
closely resembles those of tetanurans than those of coelophy-
soids or Herrerasaurus. The expanded lateral malleolus and dis-
tal ridge lend an elongate shape to the distal tibia, in contrast to
the rectangular or circular form of more primitive taxa. The
astragalar facet invaginates the anterior profile opposite the pos-
terior distal ridge. The outer ends of both malleoli are tapered.
The tibia of Majungasaurus is relatively short and stocky
(Fig. 7; Table 1), as are those of Lametasaurus (Matley, 1924),
other Indian theropod tibiae (GSI K19/579, K27/568; Huene and
Matley, 1933), Quilmesaurus, and a large ceratosaurian from the
Tendaguru Formation of Tanzania (HMN 37, 69). In contrast,
the tibiae of Rajasaurus,Xenotarsosaurus and Aucasaurus are
more slender than that of Majungasaurus. The tibia of Carno-
taurus is incomplete, but given its general similarity to those of
the latter two taxa, it was probably proportionally longer than
that of Majungasaurus but still shorter than originally recon-
structed (Bonaparte et al., 1990).
Fibula—The fibula is correspondingly short but not propor-
tionally stocky (Fig. 8; Table 1). Its lateral face is generally
smooth and rounded. The anterolateral edge below the proximal
end bears a roughened, bulbous expansion that is elevated above
the rest of the concave proximal surface. This tubercle, the in-
sertion of M. iliofibularis (Romer, 1923; Carrano and Hutchin-
son, 2002), is more prominent than in most other theropods, but
similar to those of Ceratosaurus,Xenotarsosaurus,Genusaurus,
Carnotaurus,Rajasaurus, and Aucasaurus. Farther distally, the
posterior edge is expanded into a prominent, obliquely oriented
trochanter whose thin, ridge-like proximal half yields to a
broader, slightly concave distal surface. Both anterior and pos-
terior edges of the fibular shaft are acute but rounded. At the
distal end, the shaft flares out laterally and anteroposteriorly,
terminating in a rugose calcaneal articulation. The anterior edge
of the shaft at this point bears a roughened flange that (based on
FMNH 2278) abutted against the lateral edge of the astragalar
ascending process. The distal articular surface is convexly
rounded in mediolateral view.
The medial face is shallowly concave near the proximal end
(Fig. 8D) in a manner that approximately corresponds to the
curvature of the lateral tibia. A raised ridge extends distally
down the proximal one-third of the anterior shaft edge, probably
marking the attachment of the interosseus membrane. Just pos-
terior to this ridge, the fibula is excavated into a wide, deep
medial fossa that occupies most of the proximal third of the
bone. The fossa is more open anteriorly than posteriorly (giving
it an L-shape in cross-section), and the posterior portion of the
bone is expanded medially to create a thin but prominent wall.
This medial fossa is unlike the much deeper excavation of de-
rived coelurosaurs, the narrow incisure of coelophysoids (Rowe
and Gauthier, 1990), or the broadly shallow surface of spinosau-
roids, but instead closely resembles the condition in Ceratosau-
rus,Rajasaurus,Xenotarsosaurus, and Genusaurus. Nonetheless,
all these structures are likely homologous, representing the in-
FIGURE 6. Left tibia of Majungasaurus crenatissimus. Proximal end of
FMNH PR 2424 in A, lateral view and B, proximal view. Distal end of
FMNH PR 2278 in C, anterior view and D, distal view. Abbreviations:
af, astragalar facet; cn, cnemial crest; das, distal articular surface; fc,
fibular crest; ff, fibular facet; lc, lateral condyle; lf, lateral fossa; lm,
lateral malleolus; mc, medial condyle; mm, medial malleolus; pas, proxi-
mal articular surface. Dashed line in Bindicates reconstructed outline of
element. Scale bars equals 1 cm.
SOCIETY OF VERTEBRATE PALEONTOLOGY, MEMOIR 8170

sertion of M. popliteus (Carrano and Hutchinson, 2002). Distal
to the fossa, the fibula is flattened medially near midshaft and
becomes increasingly concave farther distally.
As in all neotheropods, the fibula was aligned closely with the
tibia for much of its length, contacting directly at the fibular crest
and via an interosseous membrane along the lengthy fibular
facet. The distal fibula lodges into a concave fibular facet that
spans both proximal tarsals. The distalmost one-fifth of the me-
dial surface would have contacted the broad lateral malleolus of
the tibia.
Astragalus—Representative elements of the tarsus of Majun-
gasaurus are limited to the astragalus and calcaneum; the distal
tarsals are not known. In anterior view, the astragalus bears
two distinct and asymmetrical condyles separated by a shallow
central sulcus (Fig. 9). On the anterior face, a (presumably vas-
cular) groove runs horizontally, parallel to the ventral condylar
edge. A large, oval fossa sits at the base of the ascending process
and extends somewhat onto it. It contains two foramina, the
larger of which passes through the base of the ascending process
to the tibial facet, whereas the smaller disappears into the inte-
rior of the bone. Laterally, the astragalus bulges anteriorly
near its calcaneal contact, marking the beginning of the fibular
facet.
The most complete ascending process is present in UA 9033. It
is taller and more laminar than those of coelophysoids, Herre-
rasaurus, and Ceratosaurus. Although incomplete, the medial
edge is oblique and lodged beneath a similarly oriented buttress
on the anterior distal tibia. Rising perpendicularly to about one-
eighth the length of the tibia, the ascending process tapers but
remains wide proximally. This conformation more closely re-
sembles that of Xenotarsosaurus (Martínez et al., 1986) and Ma-
siakasaurus (Carrano et al., 2002) than it does the acuminate
process of Dilophosaurus and most tetanurans. The lateral edge
of the process runs nearly vertically down to the calcaneal con-
tact, where it connects to two distinct ridges. The first runs a
short distance to merge with the anterolateral edge of the fibular
facet, and the second crosses to the posterolateral corner of the
calcaneum and divides its fibular and tibial facets.
The calcaneal contact is largely intact in UA 9033, preserving
a complex facet with concave and convex portions that suggest
an interlocking articulation. The tibial facet is deep and acute in
lateral view, facing mostly posteriorly. In this view, it is apparent
that the astragalar condyles face anteroventrally (as in tet-
anurans) rather than ventrally (as in more primitive theropods).
Posteriorly, the astragalus rises obliquely from lateral to medial,
reaching a peak near the medial border; it simultaneously ex-
FIGURE 7. Tibial proportions within non-avian Theropoda. Log-log plot of tibial anteroposterior midshaft diameter versus tibial length (both
measured in millimeters), showing relatively robust proportions of some ceratosaur taxa.
CARRANO—APPENDICULAR SKELETON OF MAJUNGASAURUS 171

pands anteroposteriorly, forming a rounded bulge at this peak. A
small fossa is visible near the calcaneal contact. In medial view,
the astragalar surface is broad, reniform, and faintly concave.
The medial malleolar facet faces proximomedially and flanks the
medial ascending process. The anteromedial corner is blunt.
Dorsally, the astragalus bears two distinct facets for the tibial
malleoli (Fig. 9C). The medial facet is divided from the lateral by
a low ridge that runs from the ascending process to the postero-
medial bulge. The lateral edge of this ridge is sharper than the
medial, and borders a discrete fossa. The lateral facet is also
deeper and narrower than the medial, and contains several fos-
sae as well as the internal counterpart to the medial foramen on
the external ascending process. This lateral facet is continuous
with the calcaneal facet for the tibia, and includes two large
foramina that enter into mediolateral channels, which them-
selves are confluent with two large calcaneal foramina. The an-
terolateral corner of the astragalus extends anteriorly beyond the
ascending process, marking the beginning of the fibular articu-
lation. Thus the tibia articulates partly posterior to the fibula, as
in tetanurans but not more primitive theropods. The fibular facet
contains several small foramina, as well as an additional small
fossa lodged within its medial corner. Ventrally, the astragalus is
obliquely hourglass-shaped, due to the enlargement of the an-
teromedial corner relative to the anterolateral corner.
Several astragali are known: one isolated (UA 9082), two con-
nected to their respective calcanea (FMNH PR 2425, UA 9033),
and a fourth articulated with both the tibia and calcaneum
(FMNH PR 2278). None of these specimens show evidence of
having been directly fused to the tibia or fibula, although all of
the ascending processes show some breakage along the potential
fibular contact surface.
Calcaneum—The three known calcanea (FMNH PR 2278,
2425, UA 9033) are fused to their astragali (Fig. 9), obscuring the
medial surface in these specimens. Anteriorly, the fibular facet is
slightly visible where it rises at its posterolateral corner, and
contains numerous small foramina. The anterior wall is raised
and obscures the tibial facet from anterior view. Laterally the
calcaneum has a broad, slightly concave face that is rounded
except for two triangular projections, one marking the antero-
lateral edge of the fibular cup, and the second its posterolateral
edge.
The bone is smooth and convex in posterior view, with a low,
concave proximal ridge that bounds the tibial facet posteriorly.
This ridge descends to a low point at the (faintly visible) astraga-
localcaneal suture. Much of the tibial facet can be viewed pos-
teriorly. Ventrally, the calcaneum shows a slightly pitted, convex
articular condyle that is continuous with that of the astragalus.
The contact between the two runs nearly directly anteroposteri-
orly, and the horizontal groove on the anterior surface of the
astragalus is continued onto the anteroventral part of the calca-
neum. The anterior bulge beneath the fibular cup of the astraga-
lus also has a counterpart on the calcaneum, just above the hori-
zontal groove.
In dorsal view, the calcaneum shows two distinct articular fac-
ets. The concave fibular facet is a rounded triangle with its most
acute apex directed posterolaterally. About one-fifth of the facet
is on the astragalus. This facet is separated from the more pos-
terior tibial facet by the sharp ridge that runs from the lateral
edge of the ascending process to the posterolateral corner of the
calcaneum. This corner is the highest point of the calcaneum.
Behind it, the concave tibial facet contains two small lateral
foramina as well as two larger foramina situated just at the as-
tragalocalcaneal junction. The calcaneal portion of the tibial
facet forms a narrow triangle with its apex (the shallowest por-
tion) directed laterally.
Metatarsals—Metatarsals I and V remain unknown, but all
three central elements are preserved. They are typical of thero-
pod metatarsals in being tightly articulated proximally but diver-
FIGURE 8. Left fibula of Majungasaurus crenatissimus (UA 9077). A, Anterior view; B, lateral view; C, posterior view; D, medial view; E, proximal
view; F, distal view. Abbreviations:apc, contact for the astragalar ascending process; ifs, scar for M. iliofibularis; im, ridge for attachment of
interosseus membrane; mf, medial fossa; tf, tibial facet. Scale bars equals 5 cm.
SOCIETY OF VERTEBRATE PALEONTOLOGY, MEMOIR 8172

gent distally, particularly metatarsal IV (Figs. 10, 11). They have
similar general proportions to the metatarsals of non-
coelurosaur tetanurans such as Allosaurus, although there are
several important differences.
Metatarsal II is known from two specimens, FMNH PR 2278
and UA 9034, the former of which is complete (Fig. 10; Table 1).
The bone is almost straight, with a narrow shaft that widens
steadily from proximal to distal. The proximal articulation is
typical for a theropod, being flat and relatively featureless (Fig.
10A); its proximal profile is a narrowly rounded triangle with
the apex directed posteriorly. The distal articulation bears two
condyles for contact with phalanx II-1 (Fig. 10D). The lateral
condyle is much larger than the medial and is flattened on its
ventral surface, whereas the medial condyle is acuminate. The
two condyles are separated by a deep fossa. Both collateral liga-
ment pits are present, but the lateral is much deeper than the
medial, which is merely a faint fossa. The ‘hyperextensor’pit
(“oblique ligament fossa”of Snively et al., 2004) is extremely
shallow. The flat lateral surface of the shaft indicates that this
bone was closely appressed to that of metatarsal III for most of
its length. Indeed, articulating metatarsals II and III (FMNH PR
2278) indicates that only the distal one-third of metatarsal II is
divergent. The shaft is otherwise rounded in cross-section
around the anterior, medial, and posterior surfaces, forming a
D-shape overall. A longitudinally striated facet is present along
the middle third of the posterior surface, representing the inser-
tion for M. gastrocnemius pars medialis (Carrano and Hutchin-
son, 2002). A flat area on the medial surface, adjacent to the
proximal end, marks the likely articulation site for metatarsal I
(Tarsitano, 1983).
Metatarsal III is known only from FMNH PR 2278 (Fig. 10). It
is proportionally robust, being noticeably larger in all dimensions
than either metatarsal II or IV (Table 1). The straight shaft
shows a distinct lateral deviation along its distal third, in oppo-
sition to the ‘medial deflection’seen in many theropods (Snively
et al., 2004). The proximal articulation is rectangular but not
distinctly hourglass-shaped, lacking the deep lateral notch for
metatarsal IV present in tetanurans. Instead, the markedly
smaller proximal end of metatarsal II contacts within a broad
curved fossa on the medial edge of proximal metatarsal III. As
the posterior edge of this surface is damaged, it cannot be de-
termined whether metatarsal III is expanded in this direction as
in Elaphrosaurus and Ceratosaurus. The facet for metatarsal IV
is rugose and extends over a broadly concave surface. The proxi-
mal surface itself is damaged as well, but it appears to have been
convex posteriorly and flat anteriorly. The distal articulation is
FIGURE 9. Left astragalocalcaneum of Majungasaurus crenatissimus
(FMNH PR 2278). A, anterior view; B, lateral view; C, proximal view; D,
posterior view, showing approximate mediolateral dimensions of as-
tragalus (ast) and calcaneum (calc); E, medial view. Abbreviations:ahg,
anterior horizontal groove; ap, ascending process; f, fossa; ff, fibular
facet; tf, tibial facet. Scale bar equals 1 cm.
FIGURE 10. Articulated left metatarsals II-IV of Majungasaurus cre-
natissimus (FMNH PR 2278). A, proximal view (anterior toward the
bottom); B, anterior view; C, posterior view; D, distal view (anterior
toward the top). Abbreviations:II, metatarsal II; III, metatarsal III; IV,
metatarsal IV; dt3c, contact surface for distal tarsal 3; dt4c, contact sur-
face for distal tarsal 4; edls, fossa for insertion of M. extensor digitorum
longus; gs, scar for M. gastrocnemius. Scale bar equals 1 cm.
CARRANO—APPENDICULAR SKELETON OF MAJUNGASAURUS 173

broad and roller-like, with only a very faint central sulcus. Along
the medial shaft, a large, flat facet indicates the articulation with
metatarsal II; it occupies nearly two-thirds of the length of the
bone.
Distal to this point, the marked lateral offset displaces the
distal end of metatarsal III away from that of metatarsal II. Here
a prominent ridge runs distally down the anteromedial edge of
the shaft. The lateral facet for metatarsal IV is similarly pro-
nounced but extends only to the midshaft. Posteriorly, two oval
rugosities may mark additional insertion points for Mm. gastroc-
nemii, adjacent to the articulations for metatarsals II and IV. The
collateral ligament pits are equally pronounced, and the ‘hyper-
extensor’pit is deep and mediolaterally wide.
Metatarsal IV, known from two specimens (FMNH PR 2278,
UA 9035), is approximately the same length as metatarsal II
(Fig. 10; Table 1). The proximal end is D-shaped, although the
medial edge is slightly concave where it contacts metatarsal III.
The weak projection at the posteromedial corner is smaller
than the extended process seen in most tetanurans. The proximal
articulation is concave, with thicker rims along the postero-
medial and lateral edges. A small triangular rugosity along the
posterolateral edge marks the contact with metatarsal V. The
distal end is convex and bears two unequal condyles for articu-
lation with phalanx IV-1; its shape is typical of most thero-
pods. The lateral surface bears a flat, faintly ridged contact for
metatarsal III that extends approximately to the midshaft. Here
the shaft diverges strongly laterally, a feature most clearly ob-
served in anterior view. The shaft is rounded along the other
edges, again forming a D. Two rugose, flat ridges run down the
length of the shaft along the posteromedial and posterior faces,
probably marking the insertion of M. gastrocnemius pars latera-
lis (Carrano and Hutchinson, 2002). The shaft narrows both me-
diolaterally and anteroposteriorly just before reaching the distal
articulation. Both collateral ligament pits are weakly present,
but the lateral one is particularly shallow, present only as a
fossa.
Phalanges—Most of the pedal phalanges are represented. Pre-
suming the typical theropod phalangeal formula of 2-3-4-5-x, this
collection includes all but I-1 and III-4 (Fig. 11). However, subtle
proportional differences can be detected among the phalanges,
suggesting that some dimorphism may have been present, as in
the limb bones of Masiakasaurus (Carrano, et al., 2002). Because
of these variations, similarly shaped phalanges can be difficult to
distinguish as isolated specimens. Thus some of the following
identifications should be considered tentative. This is especially
true for the digit III phalanges (e.g., III-2 and III-3).
Only the ungual phalanx (I-2) from digit I is known (UA Bv
532, 1658). Unlike the other pedal unguals, this element is nearly
symmetrical, straight, and strongly laterally compressed. The
proximal articular surface is tall, narrow, and concave but lacks
a central ridge. Ventrally there is no flexor tubercle, but rather a
shallow, flattened fossa. Vascular grooves are present but irregu-
lar on both lateral and medial surfaces, with numerous large
foramina and distinct grooves near the tip. A pair of vascular
grooves is apparent on one side only.
Digit II is known in its entirety. Both non-ungual phalanges
bear a large dorsal tubercle and two prominent, flattened ventral
heels. Of these, the lateral is approximately twice as large as the
medial and extends farther distally as an elongate ridge. Both are
rugose, although the medial heel is flatter. The remainder of the
phalangeal shaft is rounded but taller than wide. A distinct hy-
perextensor pit is visible on the dorsal surface. The distal articu-
lar surface bears two rounded condyles separated by a deep
sulcus. The condyles appear to diverge ventrally and terminate in
flat ventral heels: they are subequal in size, although the medial
is slightly larger. Both collateral ligament pits are deep, but the
lateral pit is larger and more rounded. II-1 (UA 9036, FMNH PR
2426; UA Bv 1260) is the longest and tallest pedal phalanx, with
a concave, dorsoventrally oval proximal articular surface. Pedal
phalanx II-2 (FMNH PR 2427, UA 9037) is about half as long as
II-1. Its proximal surface forms a rounded triangle, with two
facets separated by a weak vertical ridge. The medial facet is
slightly larger than the lateral, corresponding to the morphology
of the distal condyles of II-1. The hyperextensor pit is deeper and
more pronounced on II-1 than on II-2, bounded distally by the
dorsal margin of the distal articular surface.
The ungual phalanx of digit II (II-3; FMNH PR 2428, UA
9038) is markedly asymmetrical. On the triangular proximal sur-
face, the long axis is tilted so that the apex appears to lean
laterally. The two similarly sized articular facets are separated by
a wide, low ridge and bear numerous large vascular foramina.
The dorsal tubercle is pronounced, but ventrally there is no
FIGURE 11. Reconstructed left pes of Majungasaurus crenatissimus in
anterodorsal view. Metatarsals and phalanx IV-2 and is from FMNH PR
2278; the remaining phalanges are from the following specimens: UA
Bv-532 (I-2), FMNH PR 2426 (II-1), FMNH PR 2427 (II-2), UA 9038
(II-3), FMNH PR 2429 (III-1; reversed), UA 9042 (III-2), UA 9039
(III-3), FMNH PR 2430 (IV-1; reversed), FMNH PR 2431 (IV-3),
FMNH PR 2432 (IV-4), and FMNH PR 2434 (IV-5). This composite
demonstrates the general morphology and approximate proportions of
the pes; note that images of many phalanges have been scaled up to
correspond with the sizes of metatarsals II-IV. Abbreviations:I, digit I;
II, digit II; III, digit III; IV, digit IV. Dashed lines indicate approximate
size and position of metatarsal I and phalanges I-1 and III-4, based on
Aucasaurus garridoi.
SOCIETY OF VERTEBRATE PALEONTOLOGY, MEMOIR 8174

flexor tubercle. Instead, a deep, longitudinal ventral fossa is
flanked by marked lateral and medial ridges. Numerous fo-
ramina within this fossa appear to lead to a distally directed
channel, and the fossa may also be confluent with the lateral and
medial vascular grooves. The medial surface is relatively flat,
with a distinct ventral groove that wraps onto the ventral surface
near the proximal end. Its ventral edge forms the larger of the
two grooves bounding the ventral fossa. The lateral surface is
more rounded, with a distinct triangular arrangement of grooves.
The ventral groove is the most distinct and connects with the
large fossa on the ventral surface. It contains several large fo-
ramina, the distalmost leading into a tunnel within the bone and
emerging again near the tip.
The first three phalanges of digit III are known in Majungas-
aurus. These non-ungual phalanges are remarkably short and
wide, with broad, flat articular condyles and correspondingly
shallow articular facets (Fig. 12). Indeed, the proximal articular
surface lacks a vertical ridge entirely, making it difficult to dis-
tinguish even III-1 from the remaining phalanges. As with digit
II phalanges, the medial condyle tends to be slightly larger than
the lateral. The ‘hyperextensor’pit is triangular, wide, and deep,
with a marked border where it abuts the distal articular surface.
Both collateral pits are deep and rounded, with little difference
in size and shape between lateral and medial. The proximal and
distal ends each bear a pair of distinct, flat ventral heels that are
separated by a shallow fossa. The lateral distal heel is abruptly
demarcated from the shaft at its proximal edge, creating a ridge
and furrow that run between the ventral and lateral surfaces. The
proportional breadth of these elements increases distally from
III-1 to III-3.
All of the phalanges of digit IV are known. They are remark-
ably short and blocky, with a very short shaft between the
proximal and distal ends. The proximal surface is taller than
wide, with nearly equal dorsal and ventral midline tubercles.
The distal articular surface is composed of two prominent,
rounded condyles that are separated by a deep sulcus. Because
the lateral distal condyle is smaller than the medial, the corre-
sponding proximal articular facets differ in size accordingly.
Whereas the two proximal ventral heels are flat and nearly con-
fluent with one another, the distal ventral heels bear dis-
tinct proximal edges and are widely separated by the articular
sulcus. A hyperextensor pit is present on all phalanges. In con-
trast to digit II, the medial collateral pit is markedly deeper than
the lateral pit on digit IV phalanges. The proximal surface of
IV-1 (FMNH PR 2430, UA 9040) is taller than wide, con-
cave, but shallow. The lateral collateral pit of IV-1 is elliptical,
whereas those of the other phalanges are rounded. These pha-
langes become progressively shorter up to IV-3 and IV-4, which
are little more than closely attached proximal and distal articular
surfaces.
The ungual phalanx of digit IV (IV-5; FMNH PR 2434, UA
9043; Fig. 13) is asymmetrical but in a manner opposite to that of
digit II, as with the phalanges. The proximal end is a rounded
triangle whose two facets are separated by a weak ridge. A ven-
tral tubercle is lacking, replaced instead by a deep fossa contain-
ing several foramina. The fossa appears to be confluent with the
ventral vascular grooves on the lateral and medial surfaces. A
triangular arrangement of grooves can be seen on the medial
surface, but on the lateral side only a ventral groove is apparent.
The lateral surface is flatter than the medial, terminating below
FIGURE 12. Right pedal phalanx III-1 of Majungasaurus crenatissimus (UA Bv-1265). A, medial view; B, dorsal view; C, ventral view; D, lateral
view; E, proximal view; F, distal view. Abbreviations:clp, collateral ligament pit; das, distal articular surface; hp,‘hyperextensor’pit; vh, ventral
‘heel.’Scale bar equals 1 cm.
CARRANO—APPENDICULAR SKELETON OF MAJUNGASAURUS 175

in a large ventral ridge that helps bound the fossa on the ventral
surface.
In general morphology, the pedal phalanges are quite similar
to, although more robust than, the corresponding bones in Au-
casaurus and Masiakasaurus. Specifically, these taxa share a rela-
tively broad digit III, pronounced collateral pits, flat ventral
heels, and remarkably shortened individual IV phalanges. The
pedal unguals are unusual in being highly asymmetrical (on II
and IV), lacking flexor tubercles, and bearing a triangular ar-
rangement of vascular grooves. These features are also seen in
isolated abelisaurid phalanges from India and South America
(Novas and Bandyopadhyay, 2001; Novas et al., 2004).
DISCUSSION
Forelimb Specializations of Abelisaurids
The unusual abelisaurid forelimb is one of the most distinctive
morphological hallmarks of this group. The remarkable parallels
between abelisaurids and tyrannosaurids in this regard have
been noted (Bonaparte et al., 1990), although close examination
reveals that the forelimbs in these two groups differ in certain
important details. In particular, abelisaurids, like other basal
theropods, apparently retained four functional manual digits
with primitive phalangeal size proportions (e.g., Carnotaurus,
Bonaparte et al., 1990; Aucasaurus, Coria et al., 2002).
Nonetheless, there are notable similarities to the indepen-
dently reduced forelimbs of tyrannosaurids (Carpenter and
Smith, 2001; Currie, 2003). In particular, although both groups
show manus reduction (digit number in tyrannosaurids, digit
length in abelisaurids), both also retain manus functionality
along with all of the primitive theropod pectoral and forelimb
musculature. It has been suggested that forelimb reduction is
largely an allometric effect visible in giant theropods (Currie,
2003), but differences in forelimb size and development between
abelisaurids and tyrannosaurids versus spinosaurids imply that
important functional convergences (and divergences) were also
at work.
The scapulocoracoid of Majungasaurus differs from that of
most other theropods in the broad, rounded shape of the cora-
coid and the pronounced dorsal lip of the glenoid. It is very
similar to the same element in Carnotaurus and Aucasaurus,as
well as those of Masiakasaurus,Elaphrosaurus, and Delta-
dromeus. The short humerus bears a globular head flanked
by small medial and lateral tubercles. The long deltopectoral
crest attached to a M. pectoralis that was likely large, which in
turn would have originated partly on the expansive ventral cora-
coid. The distal humeral condyles are unusually flattened, and
their articulation with the bizarre radius and ulna (as in Carno-
taurus and Aucasaurus) are not well understood. Still, the
rounded humeral head and distal radius/ulna suggest that both
the shoulder and elbow joints enjoyed a wide range of mobility
in abelisaurids.
The manual elements of Carnotaurus originally appeared to be
so specialized that their precise homologies and morphologies
were difficult to determine (Bonaparte, 1985; Bonaparte et al.,
1990). As a result, it has been exceedingly difficult to identify
potential manual elements from among the isolated epipodial
materials referred to Majungasaurus. A few unusual phalanges
probably pertain to the manus, but without adequate articulated
or comparative materials, more specific identifications are not
possible.
Nevertheless, the manus of Carnotaurus may not be as bizarre
as originally described. Both left and right manus are present,
although they are disarticulated, damaged, and incomplete. Re-
examination of MACN-CH 895 suggests that several typical
theropod manual elements are represented, including unguals.
These were probably arrayed among four digits, as in Aucasau-
rus,Ceratosaurus, and other primitive theropods, but beyond
this little can be said concerning their particular arrangement. At
present Carnotaurus alone does not provide convincing evidence
for anything radically unusual in the abelisaurid manus aside
from its extremely short length. Further preparation and descrip-
tion of the nearly complete Aucasaurus forelimb (Coria et al.,
2002) will undoubtedly shed light on the general morphology of
the abelisaurid manus.
Hind Limb Specializations of Abelisaurids
Not surprisingly, the abelisaurid hind limb shows far fewer
morphological deviations from the basic theropod design than
the forelimb. Nevertheless, the hind limb elements are somewhat
unusual in their relatively stocky proportions (Figs. 1, 7). In par-
FIGURE 13. Left pedal phalanx IV-5 of Majungasaurus crenatissimus
(FMNH PR 2434). A, lateral view; B, medial view; C, dorsal view. Ab-
breviations:dp, dorsal process; pas, proximal articular surface; vf, ven-
tral fossa; vg, vascular groove. Scale bar equals 1 cm.
SOCIETY OF VERTEBRATE PALEONTOLOGY, MEMOIR 8176

ticular, the tibia is short relative to other hind limb elements,
and the entire limb appears to be short relative to body length
(although the incomplete nature of most abelisaurid skele-
tons makes this difficult to determine precisely). It is interest-
ing to note that a subset of abelisaurids—notably Majunga-
saurus,Lametasaurus,Quilmesaurus, GSI K19/579, and GSI
K27/568—have stocky tibiae even compared to other mem-
bers of this clade. In contrast, Aucasaurus,Xenotarsosaurus, ISI
R91/1, and perhaps Carnotaurus appear to have tibial propor-
tions that are closer to those of other theropods (Novas et al.,
2004).
Where they can be identified, pelvic and hind limb muscle
attachment points appear to be consistent with those seen in
other basal theropods (e.g. Hutchinson, 2001). Like coelophy-
soids and some tetanurans, abelisaurids lack coelurosaur synapo-
morphies such as a marked accessory trochanter at the base of
the lesser trochanter (insertion of M. puboischiofemoralis inter-
nus 2), a well-defined fossa on the anterior pubic peduncle of the
ilium (origin of M. puboischiofemoralis internus 1), and a fibular
fossa restricted to the medial surface only (insertion of M. pop-
liteus). No shifts in muscle attachment positions associated with
these morphologies are therefore inferred to have occurred in
these taxa.
However, the abelisaurid hind limb is also derived relative to
those of coelophysoids and more primitive taxa. The inser-
tion for M. popliteus, composed largely of the medial fibular
fossa, has moved to occupy most of the medial surface of the
proximal fibula, unlike in coelophysoids where it originates from
a shallow posterior sulcus. The lesser trochanter is more elevated
and accompanied by a distinct bump for insertion of M. iliofemo-
ralis, indicating that the deep dorsal muscles were more similar
to the avian condition than to those of primitive theropods
(Hutchinson, 2001). In addition, the tarsus has achieved the
tetanuran condition, with a mediolaterally expanded distal
tibia that partially backs the fibula and articulates with the cal-
caneum.
The pronounced cnemial crest is associated with a prominent
medial femoral epicondyle in other abelisauroids, and such a
structure may also have been present in Majungasaurus. This
marks part of the origin of Mm. femorotibiales, which contrib-
uted to the knee extensor tendon(s) that inserted on the cnemial
crest. The possible elaboration of Mm. femorotibiales, but not
other, more proximal knee extensors (such as Mm. iliotibiales
and M. ambiens) suggests enhancement of the knee extension
moment. The extensive lateral fossa on the proximal tibia may
have housed a large M. tibialis anterior as well.
Consolidation of the tarsus into a single, block-like structure
has been cited as a synapomorphy of Ceratosauria (sensu
Gauthier, 1986; Rowe, 1989), but recent phylogenetic work (e.g.,
Carrano et al., 2002; Rauhut, 2003) suggests that this is con-
vergent between coelophysoids and abelisauroids. Further dif-
ferences in detail between the fusion patterns in these groups
support this hypothesis (see below). Regardless, although some
degree of immobility obviously would have been conferred by
such fusion, it is not clear what effect this might have had on
tarsal function because little mobility is inferred between these
elements in theropods where they remain unfused.
The abelisauroid pes is also unusual in bearing markedly
asymmetrical unguals at the ends of relatively broad, shortened
digits. The digits themselves also appear to be more strongly
curved and bear pronounced, flattened ventral tendon attach-
ments. The metatarsals are fairly tightly appressed, not diverging
from one another until rather close to their distal ends. The
prominence of metatarsal III appears to be a common feature
of abelisauroids, although the particular breadth exhibited
by Majungasaurus, for example, may characterize only abelisau-
rids.
Phylogenetic Implications
The appendicular morphology of Majungasaurus, particularly
the tarsus, strongly suggests closer affinities with tetanurans than
with coelophysoids and other more primitive theropods. For ex-
ample, the distal tibia has a flat lateral malleolus that backs the
fibula, lodging within an elongate tarsal facet that overlaps the
dorsal surfaces of both the astragalus and calcaneum. The cal-
caneal facet for the tibia faces slightly medially. The tibial distal
end is mediolaterally elongate (Novas, 1996), with a centrally-
placed buttress marking the articulation with the astragalar as-
cending process (Molnar et al., 1996). This process is tall and
plate-like rather than triangular and peg-like (Sereno, 1999), and
sits superficially on the anterior tibia without being inset. In
these features the abelisaurid hind limb shows similarities with
tetanurans and not with coelophysoids. Likewise, the enlarged
femoral lesser trochanter (accompanied by a distinct M. ilio-
femoralis externus insertion) is more derived than that of coe-
lophysoids and other primitive theropods, resembling the condi-
tion in spinosauroids and other basal tetanurans (Hutchinson,
2001).
In other aspects, the appendicular morphology of Majungas-
aurus is more similar to those of more primitive theropods, in-
cluding coelophysoids. These features include an anteromedially
oriented femoral head (Bonaparte, 1991), a femoral lesser tro-
chanter well below the level of the femoral head (Novas, 1991),
flattened (rather than elliptical) unguals on pedal digits II-IV
(Russell and Dong, 1993), and a relatively broad scapular blade
(Gauthier, 1986). It is important to note, however, that these
represent symplesiomorphies and not synapomorphies of Majun-
gasaurus and coelophysoids. Only one appendicular feature—
fusion between the astragalus and calcaneum—can be observed
that appears to be synapomorphic (or homoplastic) between
these two taxa. Even this characteristic, however, shows some
differences in detail. In abelisauroids (Majungasaurus,Xenotar-
sosaurus,Masiakasaurus), the astragalus and calcaneum are
fused to each other, with the ascending process occasionally
fused to the medial fibula and the anterior tibia. By contrast, in
coelophysoids (Coelophysis,Syntarsus,Dilophosaurus,Lilien-
sternus), the astragalus and calcaneum tend to fuse together but
with little direct involvement of the tibia or fibula.
Thus, the morphology of Majungasaurus is consistent with the
results of previous phylogenetic analyses (Sampson et al., 2001;
Carrano et al., 2002; Rauhut, 2003; Wilson et al., 2003) in indi-
cating a closer relationship between abelisauroids (and Cerato-
saurus) and tetanurans than between abelisauroids and coelo-
physoids. Particular similarities in the tarsus suggest that many
‘tetanuran’morphological innovations actually diagnose a more
inclusive clade, and therefore must have appeared considerably
earlier in theropod evolution than had been previously appreci-
ated (Carrano et al., 2002; Rauhut, 2003).
CONCLUSIONS
The appendicular morphology of the abelisaurid theropod
Majungasaurus crenatissimus is described. Abelisaurid appen-
dicular materials have not been well documented, and several
relatively complete specimens of this Late Cretaceous Malagasy
theropod greatly clarify this region of the skeleton in these thero-
pods.
The forelimb of Majungasaurus is similar to those of other
abelisaurids and includes a short, highly modified humerus. Nu-
merous abelisaurid and abelisauroid synapomorphies are also
found in the pelvis and hind limb. In addition, the pelvis and hind
limb display a combination of features that strongly suggest a
close affinity between abelisauroids and tetanurans. Function-
ally, the abelisaurid forelimb remains obscure and its interpre-
tation must await more complete materials. The hind limb ap-
CARRANO—APPENDICULAR SKELETON OF MAJUNGASAURUS 177

pears to show specializations for strong knee extensors, as well as
unusual modifications of the metatarsus and pes that may have
additional locomotor implications.
ACKNOWLEDGMENTS
I thank D. W. Krause, C. A. Forster, and S. D. Sampson for
the invitation to work on the Mahajanga Basin Project, and spe-
cifically on the appendicular skeleton of Majungasaurus (then
Majungatholus). I also acknowledge all the members of the 1993-
2001 field expeditions for their help in recovering these materi-
als, and V. Heisey and M. Getty for skillfully preparing them. F.
E. Novas kindly provided a copy of several then-in-press manu-
scripts, M. A. Loewen supplied photographs of most of the pha-
langes, A. Farke provided the images used in Fig. 6A and B, and
R. Ridgley created the image used in Figure 1. Translations of
Bonaparte and Novas (1985), Depéret (1896a, b), Depe´ret and
Savornin (1925), Lavocat (1955), Martinez et al. (1986), and No-
vas (1991) are available from the Polyglot Paleontologist website
(http://ravenel.si.edu/paleo/paleoglot/). This paper benefited
greatly from the comments of J. A. Wilson, M. C. Lamanna, D. W.
Krause, and an anonymous reviewer. This research was sup-
ported by grants from the National Science Foundation (DEB-
9224396, DEB-9904045, EAR-9418816, EAR-9706302, EAR-
0106477, EAR-0116517, EAR-0446488), the Dinosaur Society
(1995), and the National Geographic Society (1999, 2001, 2004).
LITERATURE CITED
Bonaparte, J. F. 1985. A horned Cretaceous carnosaur from Patagonia.
National Geographic Research 1:149–151.
Bonaparte, J. F. 1991. The Gondwanian theropod families Abelisauridae
and Noasauridae. Historical Biology 5:1–25.
Bonaparte, J. F., and F. E. Novas. 1985. Abelisaurus comahuensis,n.g.,
n. sp., Carnosauria del Cretácico Tardio de Patagonia. Ameghiniana
21:259–265.
Bonaparte, J. F., F. E. Novas, and R. A. Coria. 1990. Carnotaurus sastrei
Bonaparte, the horned, lightly built carnosaur from the Middle Cre-
taceous of Patagonia. Contributions in Science, Natural History Mu-
seum of Los Angeles County 416:1–41.
Britt, B. B., K. C. Cloward, C. A. Miles, and J. H. Madsen, Jr. 1999. A
juvenile Ceratosaurus (Theropoda, Dinosauria) from Bone Cabin
Quarry West (Upper Jurassic, Morrison Formation), Wyoming.
Journal of Vertebrate Paleontology 19(3, Supplement):33A.
Britt, B. B., 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,
Supplement):32A.
Bryant, H. N., and K. L. Seymour. 1990. Observations and comments on
the reliability of muscle reconstruction in fossil vertebrates. Journal
of Morphology 206:109–117.
Carpenter, K., and M. Smith. 2001. Forelimb osteology and biomechanics
of Tyrannosaurus rex; pp. 90–116 in D. H. Tanke and K. Carpenter
(eds.), Mesozoic Vertebrate Life: New Research Inspired by the
Paleontology of Philip J. Currie. Indiana University Press, Bloom-
ington.
Carrano, M. T., and J. R. Hutchinson. 2002. The pelvic and hind limb
musculature of Tyrannosaurus rex (Dinosauria: Theropoda). Jour-
nal of Morphology 253:207–228.
Carrano, M. T., S. D. Sampson, and C. A. Forster. 2002. The osteology
of Masiakasaurus knopfleri, a small abelisauroid (Dinosauria:
Theropoda) from the Late Cretaceous of Madagascar. Journal of
Vertebrate Paleontology 22:510–534.
Carrano, M. T., and S. D. Sampson. 2004. New discoveries of Ma-
siakasaurus knopfleri and the morphology of the Noasauridae (Di-
nosauria: Theropoda). Journal of Vertebrate Paleontology 24(3,
Supplement):44A.
Chakravarti, D. K. 1934. On the systematic position of Lametasaurus
indicus. Proceedings of the 21st Indian Science Congress:352.
Chakravarti, D. K. 1935. Is Lametasaurus indicus an armored dinosaur?
American Journal of Science, series 5, 30:138–141.
Chatterjee, S. 1978. Indosuchus and Indosaurus, Cretaceous carnosaurs
from India. Journal of Paleontology 52:570–580.
Coria, R. A., L. M. Chiappe, and L. Dingus. 2002. A new close relative
of Carnotaurus sastrei Bonaparte 1985 (Theropoda: Abelisauridae)
from the Late Cretaceous of Patagonia. Journal of Vertebrate Pa-
leontology 22:460–465.
Currie, P. J. 2003. Allometric growth in tyrannosaurids (Dinosauria:
Theropoda) from the Upper Cretaceous of North America and
Asia. Canadian Journal of Earth Sciences 40:651–665.
Depéret, C. 1896a. Note sur les dinosauriens sauropodes & théropodes
du Crétacésupérieur de Madagascar. Bulletin de la Société
géologique de France, 3e série, 24:176–194.
Depéret, C. 1896b. Sur l’existence de Dinosauriens, Sauropodes et
Théropodes, dans le Crétacésupérieur de Madagascar. Comptes
Rendus Hebdomadaires des Seances de l’Academie des Sciences à
Paris 122:483–485.
Depéret, C., and J. Savornin. 1928. La faune de Reptiles et de Poissons
albiens de Timimoun (Sahara algérien). Bulletin de la Société
géologique de France, 4e série, 27:257–265.
Dilkes, D. W. 2000. Appendicular myology of the hadrosaurian dinosaur
Maiasaura peeblesorum from the Late Cretaceous (Campanian) of
Montana. Transactions of the Royal Society of Edinburgh: Earth
Sciences, 90:87–125.
Dilkes, D. W. 2001. An ontogenetic perspective on locomotion in the
Late Cretaceous dinosaur Maiasaura peeblesorum (Ornithischia:
Hadrosauridae). Canadian Journal of Earth Sciences, 38:1205–1227.
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 8, San Fran-
cisco.
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. 2001. The evolution of femoral osteology and soft
tissues on the line to extant birds (Neornithes). Zoological Journal
of the Linnean Society 131:169–197.
Jasinoski, S. C., A. P. Russell, and P. J. Currie. 2006. An integrative
phylogenetic and extrapolatory approach to the reconstruction of
dromaeosaur (Theropoda: Eumaniraptora) shoulder musculature.
Zoological Journal of the Linnean Society 146:301–344.
Kellner, A. W. A., and D. d. A. Campos. 2002. On a theropod dinosaur
(Abelisauria) from the continental Cretaceous of Brazil. Arquivos
do Museu Nacional, Rio de Janeiro 60:163–170.
Krause, D. W., J. H. Hartman, and N. A. Wells. 1997. Late Cretaceous
vertebrates from Madagascar: Implications for biotic change in deep
time; pp. 3–43 in S. D. Goodman and B. D. Patterson (eds.), Natural
Change and Human Impact in Madagascar. Smithsonian Institution
Press, Washington, D.C.
Krause, D. W., S. D. Sampson, M. T. Carrano, and P. M. O’Connor. 2007.
Overview of the history of discovery, taxonomy, phylogeny, and
biogeography of Majungasaurus crenatissimus (Theropoda: Abelis-
auridae) from the Late Cretaceous of Madagascar; pp. 1–20 in S. D.
Sampson and D. W. Krause (eds.), Majungasaurus crenatissimus
from the Late Cretaceous of Madagascar. Society of Vertebrate
Paleontology Memoir 8.
Lavocat, R. 1955. Sur une portion de mandibule de théropode provenant
du Crétacésupérieur de Madagascar. Bulletin du Muséum de
l’Histoire Naturelle, Paris, 2e série, 27:256–259.
Martínez, R., O. Giménez, J. Rodríguez, and G. Bochatey. 1986. Xeno-
tarsosaurus bonapartei nov. gen. et sp. (Carnosauria, Abelisauridae),
un nuevo Theropoda de la Formación Bajo Barreal, Chubut, Ar-
gentina. IV Congreso Argentino de Paleontología y Bioestratigrafía:
23–31.
Marsh, O. C. 1881. Principle characters of American Jurassic dinosaurs.
Part V. The American Journal of Science and Arts, Series 3, 21:
418–423
Marsh, O. C. 1884. The classification and affinities of dinosaurian rep-
tiles. Nature 31:68–69.
Matley, C. A. 1921. On the stratigraphy, fossils and geological relation-
ships of the Lameta beds of Jubbulpore. Records of the Geological
Survey of India 53:142–169.
Matley, C. A. 1924. Note on an armoured dinosaur from the Lameta beds
of Jubbulpore. Records of the Geological Survey of India 55:
105–109.
Molnar, R. E. 1990. Problematic Theropoda: “carnosaurs”; pp. 306–317
in D. B. Weishampel, P. Dodson, and H. Osmólska (eds.), The
Dinosauria. University of California Press, Berkeley.
SOCIETY OF VERTEBRATE PALEONTOLOGY, MEMOIR 8178

Molnar, R. E., A. L. Angriman, and Z. Gasparini. 1996. An Antarctic
Cretaceous theropod. Memoirs of the Queensland Museum 39:
669–674.
Novas, F. E. 1991. Los Tyrannosauridae, gigantescos dinosaurios celu-
rosaurios del Cretácico tardio de Laurasia. Ameghiniana 28:401.
Novas, F. E. 1996. Dinosaur monophyly. Journal of Vertebrate Paleon-
tology 16:723–741.
Novas, F. E., and S. Bandyopadhyay. 2001. Abelisaurid pedal unguals
from the Late Cretaceous of India. Asociación Paleontológica Ar-
gentina Publicación Especial, Buenos Aires:145–149.
Novas, F. E., F. L. Agnolin, and S. Bandyopadhyay. 2004. Cretaceous
theropods from India: a review of specimens described by Huene
and Matley (1933). Revista del Museo Argentino de Ciencias Natu-
rales, nuevo serie, 6:67–103.
Osmólska, H., E. Roniewicz, and R. Barsbold. 1972. A new dinosaur,
Gallimimus bullatus n. gen., n. sp. (Ornithomimidae) from the Up-
per Cretaceous of Mongolia. Palaeontologia Polonica 27:103–143.
Ostrom, J. H. 1974. The pectoral girdle and forelimb function of Dei-
nonychus (Reptilia: Saurischia): a correction. Postilla 165:1–11.
Owen, R. 1842. Report on British fossil reptiles, part II. Report of the
British Association for the Advancement of Science 11:60–204.
Rauhut, O. W. M. 2003. The interrelationships and evolution of basal
theropod dinosaurs. Special Papers in Palaeontology 69:1–213.
Rogers, R. R., and J. H. Hartman. 1998. Revised age of the dinosaur-
bearing Maevarano Formation (Upper Cretaceous), Mahajanga Ba-
sin, Madagascar. Journal of African Earth Sciences 27:160–162.
Rogers, R. R., J. H. Hartman, and D. W. Krause. 2000. Stratigraphic
analysis of Upper Cretaceous rocks in the Mahajanga Basin, north-
western Madagascar: implications for ancient and modern faunas.
Journal of Geology 108:275–301.
Rogers, R. R., D. W. Krause, K. Curry Rogers, A. H. Rasoamiara-
manana, and L. Rahantarisoa. 2007. Paleoenvironment and pa-
leoecology of Majungasaurus crenatissimus (Theropoda: Abelisau-
ridae) from the Late Cretaceous of Madagascar; pp. 21–31 in S. D.
Sampson and D. W. Krause (eds.), Majungasaurus crenatissimus
(Theropoda: Abelisauridae) from the Late Cretaceous of Madagas-
car. Society of Vertebrate Paleontology Memoir 8.
Romer, A. S. 1923. The pelvic musculature of saurischian dinosaurs.
Bulletin of the American Museum of Natural History 48:605–617.
Rowe, T. B. 1989. A new species of the theropod dinosaur Syntarsus
from the Early Jurassic Kayenta Formation of Arizona. Journal of
Vertebrate Paleontology 9:125–136.
Rowe, T. B., and J. A. Gauthier. 1990. Ceratosauria; pp. 151–168 in D. B.
Weishampel, P. Dodson, and H. Osmólska (eds.), The Dinosauria.
University of California Press, Berkeley.
Russell, D. A., and Z. Dong. 1993. The affinities of a new theropod from
the Alxa Desert, Inner Mongolia, People’s Republic of China. Ca-
nadian Journal of Earth Sciences 30:2107–2127.
Sampson, S. D., and L. M. Witmer. 2007. Craniofacial anatomy of Majun-
gasaurus crenatissimus (Theropoda: Abelisauridae) from the Late
Cretaceous of Madagascar; pp. 32–102 in S. D. Sampson and D. W.
Krause (eds.), Majungasaurus crenatissimus (Theropoda: Abelisau-
ridae) from the Late Cretaceous of Madagascar. Journal of Verte-
brate Paleontology Memoir 8.
Sampson, S. D., M. T. Carrano, and C. A. Forster. 2001. A bizarre
predatory dinosaur from the Late Cretaceous of Madagascar. Na-
ture 409:504–506.
Sampson, S. D., D. W. Krause, P. Dodson, and C. A. Forster. 1996. The
premaxilla of Majungasaurus (Dinosauria: Theropoda), with impli-
cations for Gondwanan paleobiogeography. Journal of Vertebrate
Paleontology 16:601–605.
Sampson, S. D., 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 biogeog-
raphy of Gondwana. Science 280:1048–1051.
Seeley, H. G.. 1888. On the classification of the fossil animals commonly
named Dinosauria. Proceedings of the Royal Society of London
43:165–171
Sereno, P. C. 1999. The evolution of dinosaurs. Science 284:2137–2147.
Snively, E., A. P. Russell, and G. L. Powell. 2004. Evolutionary mor-
phology of the coelurosaurian arctometatarsus: descriptive, morpho-
metric and phylogenetic approaches. Zoological Journal of the Lin-
nean Society 142:525–553.
Sues, H.-D., and P. Taquet. 1979. A pachycephalosaurid dinosaur from
Madagascar and a Laurasia-Gondwanaland connection in the Cre-
taceous. Nature 279:633–635.
Tarsitano, S. F. 1983. Stance and gait in theropod dinosaurs. Acta Pal-
aeontologica Polonica 28:251–264.
Walker, A. D. 1964. Triassic reptiles from the Elgin area: Ornithosuchus
and the origin of carnosaurs. Philosophical Transactions of the
Royal Society of London B 248:53–134.
Walker, A. D. 1977. Evolution of the pelvis in birds and dinosaurs; pp.
319–358 in S. M. Andrews, R. S. Miles, and A. D. Walker (eds.),
Problems in Vertebrate Evolution. Linnean Society Symposium Se-
ries 4.
Welles, S. P. 1984. Dilophosaurus wetherilli (Dinosauria, Theropoda)
osteology and comparisons. Palaeontographica Abteilung A 185:
85–180.
Wilson, J. A., P. C. Sereno, S. Srivastava, D. K. Bhatt, A. Khosla, and A.
Sahni. 2003. A new abelisaurid (Dinosauria, Theropoda) from the
Lameta Formation (Cretaceous, Maastrichtian) of India. Contribu-
tions from the Museum of Paleontology, University of Michigan
31:1–42.
Submitted November 22, 2004; accepted January 5, 2007.
CARRANO—APPENDICULAR SKELETON OF MAJUNGASAURUS 179