Abstract: The sauropod dinosaur, Barapasaurus tagorei, is known from the Early Jurassic Kota Formation (Sinemurian to Pliensbachian) of India. The taxon is represented by c. 300 bones that were found associated with large fossilized tree trunks and were collected from the interface of sandstone and mudstone units covering an area of c. 276 m2. The collection includes one partial skeleton; most of the remainder of the bones were disarticulated, disassociated and dispersed, but taphonomic analysis permits recognition of associated elements comprising several individuals. Skeletal anatomy of Barapasaurus includes several teeth, vertebrae from the caudal cervicals rearward to the terminal caudals, and most elements of the appendicular skeleton. Barapasaurus is characterized by spoon-shaped teeth with bulbous bases and grooves on the anterolabial and posterolingual sides of the crown, coarse tubercles on the carina, acamerate cranial and dorsal vertebrae, lateral laminae of the middle and caudal dorsal neural spines composed of spinodiapophyseal and spinopostzygapophyseal laminae, neural canal of the mid-dorsal vertebrae opens dorsally through a narrow slit into a large cavity and sacrum with four co-ossified vertebrae. Phylogenetic analysis reveals that Barapasaurus is basal in comparison with Vulcanodon and is removed from Eusauropoda.
Sauropod dinosaurs appeared during the Late Triassic and diversified successfully in both the number of genera and species throughout the Mesozoic but became extinct by the end of Late Cretaceous. Fossil remains of these gigantic, terrestrial, herbivorous dinosaurs with small skull, simple teeth, long neck, long tail, quadrupedal gait and upright limbs are found in almost all the continents (McIntosh 1990; Upchurch et al. 2004). Nevertheless, records of early sauropods of Late Triassic or Early Jurassic age are still quite rare in comparison with their Cretaceous successors and are known from a very few places such as India, Thailand, China and Africa.
In India, early sauropod dinosaurs were first recorded in 1962 with the discovery of about 300 bones occurring just above the interface of sandstone and mudstone units of the Kota Formation of peninsular India (Jain et al. 1962). The Early Jurassic Kota Formation occurs in the Pranhita-Godavari basin, a Gondwana basin in Deccan, India (Text-fig. 1). This is the only continental Early Jurassic horizon in India and also one of the few in the world producing rich terrestrial fauna including fishes, reptiles, mammals, freshwater ostracodes, conchostrachans and land insects (Jain 1980). After the initial discovery, excavations in the following years produced a large number of bones including at least six skeletons in partial association. Study of these collections led to the establishment of a new sauropod dinosaur, Barapasaurus tagorei (Jain et al. 1975, 1979), one of the oldest sauropod dinosaurs of the world. The material included about six individuals ranging in ontogenetic development from juvenile to adult. Since the age of the Kota Formation is generally considered as Early to early Middle Jurassic (Bandyopadhyay and RoyChowdhury 1996; Bandyopadhyay and Sengupta 2006), this material constitutes an early population of this sauropod in India. Jain et al. (1975, 1979) presented preliminary osteological descriptions of some of the diagnostic elements in the hypodigm, concluding that certain aspects of the anatomy are primitive for a sauropod, while other aspects resembled those of the prosauropods.
According to the original diagnosis by Jain et al. (1975), Barapasaurus tagorei is a large sauropod with slender limbs, spoon-shaped teeth with coarse denticles on posterior and anterior keels, opisthocoelous cranial dorsal vertebrae, sacrum consisting of four co-ossified vertebrae and narrow width between the sacricostal yokes, deep medial wall of the ilium, and shallow curvature of the proximal part of the anterior border of the scapula. It bears similarity with the prosauropod dinosaurs based on slender limbs, distinctive fourth trochanter of the femur and small pelvic basin (Jain et al. 1979).
Earlier description and diagnosis of Barapasaurus was by Jain et al. (1975, 1979) during the time when sauropods were much less well known. Subsequently, information on several new basal sauropods came to light from different parts of the world including two more basal sauropods from the Pranhita-Godavari basin of India. Of them, Kotasaurus yamanpalliensisYadagiri, 1988 was also recovered from the Early Jurassic Kota Formation of the Pranhita-Godavari basin (Table 1) from a locality c. 40 km north of the type locality of B. tagorei (Text-fig. 1). The other basal sauropod Lamplughsaura dharmaramensis Kutty, Chatterjee, Galton and Upchurch, 2004 has been recovered from the upper part of the early Early Jurassic (Hettangian) Dharmaram Formation of the Pranhita-Godavari basin (Table 1) and has recently been described on the basis of a partially associated skeleton of a nearly adult individual (Kutty et al. 2007).
Table 1. A synopsis of the known basal sauropods.
Material, characteristic features
Lower part of the Kota Formation, P-G basin, India
Early Jurassic (Sinemurian to Pliensbachian)
About 12 individuals; except teeth no skull material; simple dorsal vertebrae without elaborated spinal laminae; absence of pneumatocoel on the base of neural arch opening into neural canal; long but low iliac blade with straight dorsal border; scapula with narrow proximal surface; limb bones relatively slender; femur with lesser trochanter, v-shaped chevrons with well-developed articular facets on the dorsolateral corners.
Upper part of the Dharmaram Formation, P-G basin, India
early Early Jurassic (Hettangian)
A heavily built quadrupedal taxon with a body length of 10 m; teeth with strongly emarginated distal edge; caudal cervical neural spines bearing a transversely expanded spine table and a vertically oriented ligamentous furrow on cranial and caudal surfaces; craniodorsally directed spur on the proximal caudal neural spines while a large process on midcaudal neural spine; caudal neural spines shorter than transverse process and consequently lost in tail; manual ungual I nontrenchant.
Sacrum with ?4 fused vertebrae; ischium longer than pubis, pubes forming a prominent anterior facing pubic apron, ilium with long pubic peduncle, amphicoelous caudals, straight columnar femur, forelimb 78% of the length of hindlimb, proximal end of metatarsal II transversely narrower than the other metatarsals, hallux ungual enlarged, distal nonungual pedal phalanges of digit III and IV stubby and wider, proximal articular surfaces of pedal digits unguals II and III dorsoventrally compressed resulting in wider breadth than height.
Associated skeleton with a prominent process (?cranial process) projecting laterally from the proximal end of the femur, making the lateral margin of the femoral shaft strongly concave in cranial view and a very prominent, acuminate, S-shaped fourth trochanter in the proximal half of the robust straight femur.
Partial disarticulated skeleton; dorsal neural spines flared transversely at their distal end; dorsal vertebrae with broad, triangular, hyposphene in caudal view; ventral ridge on the hyposphene of the caudal dorsal vertebrae; a deep sulcus adjacent to the lateral distal margin of the deltopectoral crest; and an extremely short, broad metacarpal I.
Lower part of the Fenghahe Formation of Zhonghe, Yunnan Province of Republic of China
One or more partial skeleton including a 12–13-m-long individual; the holotype material is a partial skeleton consisting of a nearly complete left dentary, one cervical vertebra, several dorsal vertebrae, cranial caudal vertebrae, both scapulae, incomplete pelvic girdle and the hind limbs.
Nearly complete skeleton except the skull; well-developed posterior process in the premaxilla; large, spoon-shaped teeth without any serration but with thick longitudinal striations on lingual and labial sides; amphiplatyan to amphicoelous vertebrae; three fused sacrals; dense internal structures in cranial sacral; caudal chevrons not bifurcated.
A dorsoventrally shallow maxilla with ten preserved teeth in various stages of eruption in 16 alveoli which gradually decrease in size posteriorly; spatulate teeth with wrinkled texture in the enamel; tooth crowns lingually concave and labially convex.
A partly articulated skeleton; thin bony plate present on the posterodorsal margin of the postorbital; 20 dentary teeth with denticulate crown margins; anterior end of dentary slightly expanded in comparison to the depth of dentary at midlength; quadrate without any posterior fossa; forked chevron; prominent crest on the lateral surface of the proximal end of the fibula; lesser trochanter occurring laterally on the femur; flat pubic apron and flattened planter surface of pedal unguals II.
The other known basal sauropods include Vulcanodon karibaensisRaath, 1972 from the Early Jurassic of Zimbabwe, Isanosaurus attavipachiBuffetaut, Suteethorn, Cuny, Tong, Le Loeuff, Khansubha, and Jongautchariyakul, 2000 from the northeastern Thailand, Antetonitrus ingenipesYates and Kitching, 2003 and Blikanasaurus cromptoniGalton and Van Heerden, 1985 from the Late Triassic of South Africa, Tazoudasaurus naimi Allain, Aquesbi, Dejax, Meyer, Monbaron, Montenat, Richir, Rochid, Russell and Taquet, 2004 from the Early Jurassic of the Moroccan High Atlas, and ‘melanorosaurids’ from Africa and China (Table 1). Apart from these Gondwanan basal sauropods, descriptions of four sauropod taxa from the People’s Republic of China establish a Laurasian distribution of basal sauropods in the Early Jurassic. These are Chinshakiangosaurus chunghoensis Ye videDong, 1992, Jingshanosaurus xinwaensis Zhang and Yang, 1994, Gongxianosaurus shibeinsis He, Wang, Liu, Zhou, Liu, Cai and Dai, 1998 from Sichuan and Yunnanosaurus robustusYoung, 1951 from Yunnan Province (Table 1).
However, Upchurch et al. (2007a) concluded that Barapasaurus, Kotasaurus and Lamplughsaura from India were basal Sauropoda and removed from the Eusauropoda, although Allain and Aquesbi (2008) considered Barapasaurus as a Eusauropod. It is evident that the relationships among all basal sauropods cannot be fully analysed without additional elucidation of osteology in all these genera. With the recent expansion of knowledge of early sauropods (e.g. Vulcanodon, Kotasaurus, Chinshakiangosaurus, Antetonitrus, Blikanasaurus, Tazaoudasaurus and Jingshanosaurus), a more extensive description of Barapasaurus became imperative. Hence, this article reviews the osteology of B. tagorei, with emphasis on specimens that have not been previously described.
The Gondwana succession of the Pranhita-Godavari basin (Text-fig. 1) occurs as a narrow, rectilinear outcrop trending NNW–SSE and is bordered on both sides by Proterozoic and/or Archaean rocks; the succession is overlain by Deccan Trap basalt of Cretaceous to Palaeocene age (69–63 Ma) (Pande 2002). The overall dip of the succession is 5–12 degrees north and north-west, while the general palaeocurrent direction is north (Sengupta 1970). The Gondwana sediments were deposited in glacial, fluvioglacial, fluviatile and lacustrine conditions and range in age from Permian to Cretaceous. The Kota Formation in the Upper Gondwana of the Pranhita-Godavari basin overlies the Dharmaram Formation and shows a more or less uniform lithology throughout the valley (Table 2). Rudra (1982) divided the Kota Formation into two lithological units. The lower unit includes 15–25 m of thick, hard, compact and coarse sandstone, which is pebbly in places. This sandstone becomes finer and grades both laterally and vertically into finer siltstone and mudstones. The upper part of the Kota Formation includes marl and 1–2 m thick limestone beds. This is followed upward by mudstone and ferruginous shale interbedded with sandstone. Rudra and Maulik (1994) suggested that a meandering river system deposited the Lower Kota, while a braided river system formed the upper part; the limestone facies was interpreted to be a lacustrine deposit.
Table 2. Gondwana stratigraphy of the Pranhita-Godavari basin.
Siltstone, ferruginous sandstone – pebbly at places
The Kota Formation has quite a rich vertebrate fauna whose remains come from two successive stratigraphic levels (Table 3). The mudstone of the dominantly fluviatile lower unit has yielded two sauropod dinosaurs and three mammals. Barapasaurus and Kotasaurus are the two basal sauropod dinosaurs. The osteology of Barapasaurus is treated in this article. Kotasaurus is characterized by simple dorsal vertebrae (without spinal laminae), low iliac blade, narrow proximal surface to the scapula, relatively slender limb bones, femur with lesser trochanter and ‘v’-shaped chevrons with well-developed articular facets on the dorsolateral corners (Yadagiri 1988, 2001). Some theropod teeth have also been recovered from this horizon. On the basis of isolated teeth, three mammals have been identified from this horizon – a kuehneotherid (Kotatherium) (Datta 1981) and a probable amphilestid (Indotherium) (Yadagiri 1984). Datta and Das (2001) recently described a molariform tooth Indozostrodon belonging to Megazostrodontidae.
Table 3. The vertebrate fauna of the Kota Formation.
Toarcian to ?Aalenian
Pholidophorus kingii, P. indicus
Gondtherium dattai, Godavariodon denisei
Barapasaurus tagorei, Kotasaurus yamanpalliensis
Sinemurian to Pliensbachian
The fauna of the upper unit of the Kota Formation, a limestone-dominated lacustrine deposit, includes three semionontids (Lepidotes, Paradapedium and Tetragonolepis) (Jain 1959, 1973, 1983), a pholidophorid (Pholidophorous) (Yadagiri and Prasad 1977) and a coelacanth (Indocoelacanthus) (Jain 1974a). Of the ‘holostean’ fishes, Lepidotes is the most common semionotid. Paradapedium, though very close to Dapedium, morphologically differs in skull and body proportions. Among the reptiles, there are a pterosaur (Campylognathoides) (Jain 1974b), a mesosuchian crocodylomorph (Bandyopadhyay and Roychowdhury 1996) and a cryptodiran turtle (Indochelys) (Datta et al. 2000). The turtle, Indochelys, shows similarity with the North American Kayentachelys. A scute and other fragmentary limb bones indicate the presence of another diapsid – a scelidosaurid dinosaur in this horizon (Bandyopadhyay and Roychowdhury 1996).
Yadagiri (1986) reported the presence of a micro-vertebrate assemblage comprising an elasmobranch (Lissodus), a rhynchocephalian and fragments of a pleurodont dentition, identified as Paikasisaurus indicus of uncertain affinity. Later, Prasad and Arratia (2004) described two elasmobranchs ?Polyacrodus and Lissodus along with some other ‘holosteans’. Two sphenodontians, Rebbanasaurus and Godavarisaurus, and three dentary fragments and a partial maxilla of a probable pleurodont lepidosauromorph similar to basal rhynchocephalians have been described by Evans et al. (2001). Subsequently, Evans et al. (2002) described an acrodont lizard, Bharatagama and two indeterminate agamid lizards. Among the micro-mammals, two docodonts, Gondtherium and Godavariodon (Prasad 2003; Prasad and Manhas 2007) and a small lower molar of uncertain affinities, Dyskritodon (Prasad and Manhas 2002), two dryolestids and a probable ‘amphilestid’Paikasigudodon (Prasad and Manhas 1997, 2002) have been described from this horizon. Besides, there are a therian (Trishulotherium) (Yadagiri 1984), and a holotherian (Nakunodon) (Yadagiri 1985) of uncertain familial affinities (Averianov 2002).
On the basis of fishes, the Kota Formation has long been considered to be of Liassic age (King 1881; Robinson 1970). However, several workers noticed the similarity of Kota fishes with the European Toarcian fishes (Schaeffer and Patterson 1984). Patterson and Owen (1991) suggested that the major marine transgression during the Toarcian might have been instrumental for the invasion of the European Liassic fishes in the circum-tethyean continents including parts of Indian subcontinent. Recovery of Darwinula, an early Middle Jurassic ostracod, led Govindan (1975) to suggest a Middle Jurassic age for the Kota Formation. Analysing the faunas of the underlying Dharmaram Formation and of the lower and upper units of the Kota Formation and comparing them with faunas from coeval horizons, Bandyopadhyay and Roychowdhury (1996) and Bandyopadhyay and Sengupta (2006) suggested that the Lower Kota Formation has an age ranging from Sinemurian to Pliensbachian, while the age of the upper Kota is Toarcian and may even be extended to Middle Jurassic (?Aalenian).
Taphonomy of the Barapasaurus assemblage
A large number of dinosaur bones were discovered as surface finds from the Kota Formation of the P-G valley in 1958–59 (Jain et al. 1962). Proper excavation in 1960–61 at a site about 18 km south-east of Sironcha (18°51′N, 79°58′E) (Text-fig. 1), near a small village of Pochampalli (18°44′N, 80°05′E) (Gadchiroli district, Maharashtra) led to the recovery of a rich layer of sauropod bones, later described as Barapasaurus tagorei (Jain et al. 1975, 1979). These bones were found with fossilized large tree trunks just above a sandstone–mudstone interface (Jain et al. 1962). The rich dinosaur bone layer occurred in an area of about 276 m2. Only postcranial material and a few teeth of adult individuals totalling c. 300 bones were found from the site. On the basis of six left femora, it is estimated that there were at least six individuals (Text-fig. 2). Apart from one partially associated skeleton, the bones were disarticulated, disassociated and dispersed. However, the state of preservation of the majority of the fossil bones is remarkably good. Although orientation of the bones is polymodal (Bandyopadhyay et al. 2002, fig. 14), two strong modes were identified. The majority of the bones were oriented nearly NNW–SSE while another set of bones along with logs were oriented NE–SW. The disarticulated bones are all complete and well preserved. The more or less good shape of the vertebrae and the girdle bones indicate that the bones did not suffer much distortion. Some of the bones show long, fine desiccation cracks indicating brief surface exposure before burial, but do not show any other surface marks.
Bandyopadhyay et al. (2002) interpreted the bone assemblage of B. tagorei as an accumulation from mass mortality because of a catastrophic event. A flood might have been the cause of death of this herd of Barapasaurus, whose carcasses were transported by floodwater for a distance. The associated tree logs were uprooted in the flood event and were transported together with Barapasaurus. Subsequently, these carcasses became entangled with the tree trunks, decomposed and disarticulated. The sauropod skulls being fragile, light and with weak necks were fragmented and washed away, while the heavier postcranial bones were left behind along with the tree trunks. This bone assemblage was exposed on the surface for some time and subsequently buried by silt and clay material.
A major part of the material is in the mounted skeleton of B. tagorei, displayed at the Geology Museum of ISI, and the rest of the material is in the Palaeontological collection of the Geological Studies Unit, ISI. Most of the material (Material A) had been collected from the area near the village Pochampalli, about 18 km southeast of Sironcha (18°51′N, 79°58′E), Gadchiroli district, Maharashtra, India.
Material A. ISIR 50, sacrum (four co-ossified sacral vertebrae with sacricostal yoke); ISIR 51, right ilium (mounted); ISIR 52, right ilium; ISIR 53, right ischium (mounted); ISIR 54, left ischium; ISIR 55, right pubis (mounted); ISIR 56, left pubis (mounted); ISIR 57, right pubis; ISIR 58, right femur in two pieces (mounted); ISIR 59, left femur (mounted); ISIR 60, left femur in two pieces; ISIR 61, left tibia; ISIR 62, left tibia (mounted); ISIR 63, only distal end of left tibia; ISIR 64, left fibula; ISIR 68, left scapula (mounted); ISIR 69, left coracoid (mounted); ISIR 70, left humerus; ISIR 71, left radius; ISIR 72, left ulna; ISIR 74, twelfth dorsal vertebra (mounted); ISIR 79, first dorsal vertebra (mounted); ISIR 80 third cranial dorsal vertebra (mounted); ISIR 81, second dorsal vertebra (mounted); ISIR 83, ungual phalanx (digit I) of right pes (mounted); ISIR 84, ungual phalanx of digit III of right pes, (mounted); ISIR 85, right humerus; ISIR 86, proximal half of left humerus (mounted); ISIR 87, left humerus; ISIR 88, right humerus (mounted); ISIR 89, right radius; ISIR 90, right ulna; ISIR 91, left ulna (mounted); ISIR 92, right scapulo-coracoid; ISIR 94, metacarpal III; ISIR 95, metacarpal IV; ISIR 96, metacarpal V; ISIR 97–98, left femora; ISIR 99–100, right femora; ISIR 101, right tibia; ISIR 102, left tibia; ISIR 105, right fibula; ISIR 106, left fibula; ISIR 108, left metacarpal of digit I; ISIR 110, ungual phalanx of left digit I (juvenile); ISIR 111, large left ilium, ischiadic peduncle & iliac plate broken (mounted); ISIR 112, right large ilium; ischiadic peduncle and iliac plate broken; ISIR 114, left ischium (mounted); ISIR 115, right ischium; ISIR 116, left ischium; ISIR 117, left pubis; ISIR 118, left pubis; ISIR 121, tenth cervical vertebra (mounted); ISIR 122, fourth dorsal vertebra (mounted); ISIR 123, mid-dorsal vertebra (D5) (mounted); ISIR 124, mid-dorsal vertebra (D6) (mounted); ISIR 125, mid-dorsal vertebra (D7) (mounted); ISIR 126, mid-dorsal vertebra (D8) (mounted); ISIR 127, mid-dorsal vertebra (D9) (mounted); ISIR 128, mid-dorsal vertebra (D10) (mounted); ISIR 129, mid-dorsal vertebra (D11) (mounted); ISIR 133, posterior caudal vertebra: ISIR 134, posterior caudal (Ca43) (last in the mount); ISIR 700, mid- to caudal dorsal vertebra; ISIR 701, third dorsal vertebra; ISIR 702, isolated cervico-dorsal vertebra; ISIR 703, second dorsal vertebra; ISIR 717, complete right upper tooth; ISIR 718, small complete tooth; ISIR 719, tooth with complete root but incomplete crown; ISIR 720, tooth crown only; ISIR 721, tooth crown only; ISIR 722, tooth crown only; ISIR 723, distal chevron pair; ISIR 724, distal chevron pair; ISIR 725, single chevron co-ossified with the proximal part of the right femur of the mounted skeleton; ISIR 726, almost complete dorsal vertebra (D13) (mounted); ISIR 727, almost complete mid- to caudal dorsal vertebra with slit-like neural canal; ISIR 728, most elongate cervical vertebra (mounted); ISIR 733, distal caudal vertebra (Ca 34); ISIR 734, distal caudal vertebra (Ca 32); ISIR 737, isolated sacral vertebra ISIR 739, distal caudal; ISIR 740, pubis found near ISIR 739; ISIR 741, small right femur; ISIR 743, right calcaneum; ISIR 745, anterior caudal vertebra (Ca 1 series); ISIR 746–747, mid-caudal (Ca 13–14); ISIR 748, distal caudal (Ca 33); ISIR 749, large ungual phalanx of left digit I of manus; ISIR770, almost complete posterior dorsal vertebra (D14) (mounted).
Besides the first excavation during 1960–61, another small excavation was carried out in 1961–62 in north of the village Krishnapur (19°13′20″N, 79°31′18″E), Adilabad district, Andhra Pradesh, India (Material B).
Material B. ISIR 65, left metatarsal I (mounted); ISIR 66, left metatarsal II (mounted); ISIR 67, left metatarsal IV (mounted); ISIR 77, right tibia; ISIR 93, right humerus; ISIR 103, right tibia (mounted); ISIR 104, right fibula (mounted); ISIR 107, right astragalus.
A third excavation of very limited scope was attempted in 1964 in the same locality at Sironcha, and those collections (Material C) are referred as ‘Colbert excavation’ in memory of late Prof. Edwin H. Colbert’s participation (Colbert 1980, 1989).
Material C. Colbert collection. ISIR 113, left ilium; ISIR 120, cervical vertebra; ISIR 135, right femur; ISIR 136–143, co-ossified neural arches with spine; ISIR 144–149, caudal centra; ISIR 704, caudal centrum.
There is no evidence that this material includes more than one species.
Institutional Abbreviations. GSI, Geological Survey of India, Kolkata, India; ISI, Indian Statistical Institute, Kolkata, India (ISIR, prefix to specimen number indicating reptile collection).
Repository. Palaeontological collection, Geological Studies Unit, Indian Statistical Institute.
Locality and horizon. Pochampalli (18°44′N, 80°05′E), Gadchiroli district, Maharashtra, India; Early to early Middle Jurassic Kota Formation, Pranhita-Godavari basin, Deccan, India.
Remarks. The following amended diagnosis, representing a unique combination of characters including at least one autapomorphy, the slit-like opening of the neural canal of the dorsal vertebrae, distinguishes the species.
Amended diagnosis. Large sauropod with slender limbs; teeth spoon-shaped with bulbous base and with grooves on anterolabial and posterolingual sides of the crown; coarse tubercles mostly on posterior carina; cervical and cranial dorsal vertebrae opisthocoelous while others platycoelous; primitive acamerate vertebrae morphology in cervical and cranial dorsal vertebrae; intraprezygapophyseal laminae joined the prezygapophyses on the midline at the dorsal margin of the neural canal in the caudal cervicals; hyposphene-hypantrum articulation well-developed in middle and caudal dorsal vertebrae; lateral laminae of the middle and caudal dorsal neural spines composed of spinodiapophyseal and spinopostzygapophyseal laminae; dorsal neural spines flattened craniocaudally but wide transversely; neural canal in the mid-dorsal vertebrae open dorsally through a narrow slit-like opening into a large cavity; sacrum with four co-ossified vertebrae; sacral centra hour glass-shaped and amphiplatyan; sacral neural spines high; sacricostal yoke set close together; distal caudals spool-shaped with caudally inclined neural spine; y-shaped chevrons with fused cranial and caudal projections; scapula with tall narrow blade; coracoid subcircular with coracoid foramen; humerus with prominent deltopectoral crest, expanded at both ends; ulna with triradiate proximal end stouter than radius but slender in shaft; ilium with prominent preacetabular process; medial wall of acetabulum quite deep; ischiadic peduncle of ilium short, while pubic peduncle long and directed downward and a little cranially; pubis and ischium almost of same length; pubis with a large obturator foramen and the pubic apron articulated with its partner along nearly the full length of the midline resulting a narrow pelvic basin; ischium slender, straight and distally moderately expanded; the symphyseal contact narrow; femur long and slender with hemispherical femur head set at right angle to the straight, slender shaft; well-developed fourth trochanter projecting caudally a ridge-like process with an acuminate and declined tip; short, robust tibia with well-developed cnemial crest; fibula slender with weakly developed lateral trochanter; tibia articulating with prominently raised elliptical rugosity on the lateral side of the fibula; subtriangular astragalus bearing proximally a prominent ascending process and a medial depression; calcaneum quadrangular but proximally semi-trapezoidal; distal articular surface for the metatarsals with a fine mediolateral ridge.
Most of the elements of Barapasaurus were not found in direct association. While examining the nature of the association of some specimens, partial associations of the bones could be established. The holotype sacrum (ISIR 50) of B. tagorei is associated with right and left ilium (ISIR 51 and ISIR 111) indicating the presence of a medium-sized individual (Text-fig. 2, Association A). One partial skeleton (ISIR 113, ISIR 120, ISIR 135–149) consisting of limb bones, pelvic bones and vertebrae is a nearly mature adult as indicated by ossification of neural arches. These were excavated in a condition of disarticulation, but in close association from the ‘Colbert excavation’. From this skeleton, proportions can be established for certain limb elements, and some of the vertebrae can be identified by region (Text-fig. 2, Association J).
Skeletal elements of a small individual are so far the most complete association in the collection. This includes left scapula (ISIR 68) and complete coracoid (ISIR 69), left humerus (ISIR 70) and left radius and ulna (ISIR 71 and ISIR 72). In the same association were found the right ilium (ISIR 52), right and left ischium (ISIR 115 and ISIR 54), right and left pubis (ISIR 57 and ISIR 117), left femur (ISIR 60), left tibia and fibula (ISIR 62 and ISIR 64) (Text-fig. 2, Association C).
Published descriptions of B. tagorei include Jain et al. (1975, 1979). The following osteological descriptions are based on all available material in the ISI collections. The side view of the skeleton of B. tagorei (Text-fig. 3) summarizes our present understanding of the anatomy of this dinosaur in a contemporary interpretation of stance and posture.
Skull and teeth
No skull bones are known. Several teeth were recovered with the B. tagorei skeleton. Jain et al. (1975) described these teeth as spoon-shaped, with anterior and posterior keels bearing coarse denticles. This dentition (Text-fig. 4) includes three nearly complete teeth and three crowns without roots.
The following description for the spoon-shaped teeth adopts the orientation terminology of Calvo (1994). Moreover, the outer ‘labial’ surface of the crown is described below as the distal surface, and the inner ‘lingual’ surfaces of the crown as the mesial surface. Grooves are situated on the anterolabial and posterolingual sides of the crown. The largest tooth (ISIR 717, Text-fig. 4A–E) consists of a nearly complete root, missing only its tip, and nearly complete crown, missing parts of the apex. It is 58 mm tall (root tip to apex diameter). Crown dimensions are 15 mm maximum anterior-posterior (carina-carina) diameter, and 24 mm maximum height. In overall shape, the tooth is only slightly curved in anterior aspect and straight in lingual aspect with slight asymmetrical bulge of the crown. The tapered root is subcircular in cross-sectional aspect, and is slightly constricted at its junction with the crown, which is also constricted at its base and subcircular. By Calvo’s (1994) definition of anterior-posterior, this tooth is right upper or left lower; its straight profile suggests this tooth is maxillary rather than mandibular.
Most of the enamel on the crown is weakly wrinkled, visible only under magnification, as in almost all basal sauropods (Barrett and Upchurch 2007). Along its apicobasal axis (root tip to apex), the labial surface of the crown is convex; the lingual surface along the same axis is weakly sigmoid, producing a spatulate profile in lateral or mesial aspect. Its labial surface is convex in overall cross-sectional shape, but with a weakly developed groove on the apical one-third of the crown near the anterior carina. The lingual surface is likewise convex in cross-section, but with a weakly developed groove near the opposite (posterior) carina. These labial and lingual grooves are primitively present in eusauropods (Barrett and Upchurch 2007). This tooth is missing part of the apex, but its bulbous base retains anterior and posterior carinae, with three tubercles on the posterior carina. These tubercles are coarse, each with an overall scalloped, asymmetrical outline, shallow profile proximally and steep profile apically. Upchurch et al. (2007a) reported in the discussion of their character state C86 that Barapasaurus possesses coarse denticles with a 45 degree orientation on the basis of a published figure (Kutty et al. 2007, fig. 8). However, a close look at ISIR 717 reveals that the denticles are actually set at higher angles with the long axis; the angle of the individual denticle varies from c. 60–70 degrees to the long axis (Text-fig. 4C–D). This tooth does not show evidence of an apical wear facet, perhaps because part of the apex is missing. The crenulated enamel in the labial and lingual grooves lacks any indication of occlusal wear.
The three crowns (ISIR 720–722) (Text-fig. 4F–K) are roughly the same sizes as ISIR 717 and possess nearly identical anatomy, with only slight variation. In all three, the carinae are incomplete. These teeth are spatulate, with bulbous base, and tapering profile in lingual and labial aspect. ISIR 721 has two weakly developed tubercles on its posterior carina. These tubercles, which are slightly raised surfaces of the enamel, extend onto the lingual surface as weakly expressed, linear wrinkles in the crenulations, but show no evidence of striation or wear. ISIR 720 has a slightly narrowed crown, but nearly complete apex, demonstrating the complete shape of the crown; it lacks evidence of wear facets, perhaps because the carinae are incomplete.
Tooth ISIR 719 (Text-fig. 4L–M) is approximately two-thirds the size of the previously described teeth. Most of its root and crown are intact, but the tooth is poorly preserved. A much smaller tooth (ISIR 718) (Text-fig. 4N–O) is nearly complete, with elongate root and reduced crown. In overall shape, this tooth resembles a mammalian incisor. The bulbous crown is truncated by a beveled surface, apparently a lingually inclined wear facet. This is probably the tooth of a juvenile individual.
In the two teeth that have tubercles, they occur only on the posterior carina. None have tubercles on the anterior carina, a condition that might be attributed to incomplete preservation.
Partial vertebral associations and a number of isolated vertebrae were collected. Three pairs of dorsal vertebrae, ISIR 123 and ISIR 124, ISIR 726 and ISIR 770 (Text-fig. 2, Association D) and ISIR 127 & ISIR 128 (Text-fig. 2, Association E), which are very likely to be adjacent ones, belong to such associations. Of the other two, one consists of eight articulated neural arches (ISIR 136, ISIR 137, ISIR 138, ISIR 139, ISIR 140, ISIR 141, ISIR 142 and ISIR 143) of a small individual from ‘Colbert excavation’ and a series of caudal centra (ISIR 144–ISIR 149) found detached from the neural arches but lying close to them (Text-fig. 2, Association J). These cover the region of the caudal cervicals and cranial dorsals. The third association includes five vertebrae (ISIR 728, ISIR 79, ISIR 81, ISIR 80 and ISIR 122) of a medium sized individual, disarticulated but lying close together and belongs to the cervico-dorsal region (Text-fig. 2, Association B).
Precise count of presacral vertebrae (cervicals and dorsals) is not possible to determine, as there was no completely associated vertebral column. However, it is assumed by analogy with other related forms (sauropods and prosauropods) that there were 26 presacral vertebrae. Except for the associated caudal cervical and the cranial dorsals, there are very few duplicate specimens in the collection of the presacral vertebrae. The material in the collection suggests that B. tagorei had at least 14 dorsal vertebrae as indicated by the cervico-dorsal transition determined on the basis of the available associations of this region. Moreover, the caudal cervicals show certain modification from the typical cervical structure. The available material does not allow a complete cervical count and it is assumed that there were at least eight cervical vertebrae posterior to the atlas-axis. Thus, this set of eight, the atlas-axis, and the two caudal cervicals give a minimum count of 12. We identify the next vertebra as the first dorsal, D1 rather than C13. According to Wilson and Sereno (1998), Upchurch (1995) and Upchurch et al. (2004) basal sauropods had 13 cervicals, 13 dorsals, and a minimum count of 26 presacrals. Because this count is not inconsistent with the typical presacral count in basal sauropods, we assume that B. tagorei had 12 cervicals and 14 dorsals making up the presacral region.
The arranging of the presacrals in their proper order is essentially ordering of the dorsals. There is no atlas or axis in the collection. The last cervical vertebra (C12) is ISIR 728 (Text-fig. 5, labeled ce-12), and the first four dorsals (D1–D4) are ISIR 79, ISIR 81, ISIR 80 and ISIR 122, respectively (Text-fig. 5, labeled 1–4). Lack of sufficient cranial cervical and mid-cervical vertebrae precludes proper ordering in this region. Among the collected cervicals, ISIR 121 has been separated as C10 from C12 (ISIR 728) but it is not possible to distinguish it from the other cranial or mid-cervicals. For the ordering of dorsals D5–D14, vertebrae from this region were first placed into groups on the basis of some broad characters. For example, on the basis of the structure of the neural canal and associated features on the neural arch ISIR 126 (D8), ISIR 127 (D9), ISIR 128 (D10) and ISIR 129 (D11) are placed in one group and ISIR 74 (D12), ISIR 726 (D13), ISIR770 (D14) in another. The latter is placed as the caudal set of dorsals, i.e., D12–D14, on the basis of its similarities with the vertebrae in the sacrum. ISIR123 and ISIR 124 are placed as D5 and D6 as indicated by their similarities with D4, especially in the position of the zygapophyses. Similarly, ISIR 125 (D7) based on its similarities with D6 on one hand and the ISIR 126 (D8) group on the other. Further ordering within the groups is based on details of individual vertebrae. In the following description of the vertebrae, the abbreviated form of the nomenclature of the vertebral laminae as proposed by Wilson (1999) have been used (for details, please see above).
Cervical vertebrae. Jain et al. (1975, 1979) described the cervical vertebrae as opisthocoelous, with centra probably a little less than twice the length of dorsal centra. The caudal cervicals increase in length rearward. Because only the caudal cervicals are known, this proportion applies only to the caudal-most cervical of Barapasaurus. (The exact positions of other cervical vertebrae in the collection are not clear). On the mounted skeleton, the caudal cervicals increase in length rearward.
Cervical vertebrae are deeply opisthocoelous with markedly convex cranial facets. The centra are elongate, longer than tall. They have a weakly developed double-keel construction ventrally, producing a flattened ventral surface and slightly squared cranial/caudal profiles of the articular facets. In lateral aspect, the vertebrae are constricted at mid-centrum, and a well-defined ridge is present along the ventrolateral edges; the diapophysis position is low. Facets for cervical ribs are indistinct.
Jain et al. (1979, pl. 98A) illustrated the largest cervical vertebra (ISIR 728), which we assign to position C12. Its centrum is approximately two times longer than tall. In lateral aspect, the cranial and caudal facets are not parallel; instead their extended outlines converge ventrally (Text-fig. 5C). This orientation indicates an arched profile in the caudal cervical series. Cervicodorsals display a similar condition, described below. The acdl extends from the diapophysis to the cranial part of neurocentral suture, whereas the pcdl connects the diapophysis to the caudal part of the neurocentral junction.
The parapophyses of the caudal cervicals occur on an outward and downward projecting buttress of the ventrolateral ridge just behind the rim of the cranial face of the centrum. The parapophyses on C12 (ISIR 728) are larger, project downward and outward and originate on the lateral sides of the dorsal to the ventrolateral ridge. The prezygapophyseal facets are projected more forward than outward from the cranial margin. The prezygapophyses are subcircular and fairly highly tilted. The cprl extend from the anterolateral margin of the centrum to the prezygapophyses. The prdl (for details, please see above) connect the lateral surfaces of the prezygapophyses with the cranial part of the transverse process. The transverse processes originate from anterior position low on the lateral surface of the neural arch.
The postzygapophyseal facets are shortened but not very distinct. The podl connect the diapophyses posterodorsally to the postzygapophyses and are almost parallel to the sprl which connect the prezygapophysis with the neural spine.
Part of the neural spine and neural arch is preserved in the tenth vertebra (ISIR 121). The neural spine appears to be short, craniocaudally elongated and laterally compressed near its base. At the mid-length, the spine expands upward, giving it a diamond-shaped outline for the top of the spine. The prespinal cavities are deeper than postspinal cavities; however, the depth reduces caudally. The cranial and caudal openings of the neural canal are triangular and low in position, with a relatively broad base in outline. The cranial and caudal faces of the neural arch are concave on both sides.
Cranial dorsal vertebrae (D1–D3). Jain et al. (1979) identified two cranial dorsal vertebrae as opisthocoelous. In the mounted skeleton, these two vertebrae (ISIR 79 and ISIR 81) are immediately posterior to the longest cervical (Text-fig. 5A, C). The cranial dorsal centra are the longest in the dorsal series. Their rib facets are situated high on the neural arch, and their neural spines are transversely expanded, indicating without doubt that they are dorsal vertebrae. Because their anatomy is distinctive, these two vertebrae represent separate vertebral positions, D1 and D2 respectively, as mounted.
The ventral surface of the opisthocoelous centrum of the cranialmost of the two mounted cranial dorsal vertebrae (ISIR 79) is craniocaudally concave, gently convex transversely, weakly double-keeled like the condition in the caudal cervical described above, and in contrast to succeeding dorsal vertebrae in which the ventral surfaces of the centra are flat. In lateral aspect, the projected planes of the centrum in this vertebra, like those of the caudal cervical described above, converge ventrally, continuing the arch of the cranio-dorsal region. Accordingly, this vertebra is assigned to the cranialmost position (D1) among the cranial dorsals. The centrum has deep cavities on the lateral surfaces; neither penetrates to the interior of the centrum; these depressions are not pleurocoels. Britt (1993; cf. Wedel et al. 2000) termed these lateral excavations as pneumatic fossae. Wedel et al. (2000) characterized these pneumatic fossae to be broad in contour but not enclosed by ostial margin to form a foramen. They suggested that Barapasaurus has primitive acamerate vertebrae morphology in which pneumatic fossae are present but do not significantly invade the centrum.
The parapophyses are small and originate on the lateral face of the centrum with the ventral border still on the ventrolateral ridge as in the caudal cervicals. One sharp, well-defined dorsal ridge extends cranially and another caudally on the neurocentral suture. The rib facets are situated high on the neural arch, and their neural spines are transversely expanded. From the diapophysis, a lateral lamina, the acdl and another lamina, the pcdl, extend downward and join the cranial and the caudal part of the neurocentral contact. The prezygapophyses are slightly elongated craniocaudally and extend well beyond the plane of the cranial centrum facet. The prezygapophyses are joined on the midline at the dorsal margin of the neural canal by tprl. The prezygapophyseal process is supported by the prominent cprl connecting the prezygapophyses ventrolaterally with the cranial part of the centrum and by the prdl extending laterally from the prezygapophysis to the cranial part of the diapophysis which occurs above the neurocentral junction. The postzygapophyseal facets are large, subcircular, moderately tilted and are set well apart. A pair of laminae originates from the postzygapophysis; the podl connects the postzygapophysis to the diapophysis, while cpol connects the postzygapophysis with the caudal part of the neurocentral contact. The cranialmost dorsal vertebra (D1) has a coalesced neural spine, which is short, transversely expanded, convex on its cranial surface and concave on its rear surface. The neural arch arises from the cranial half of the centrum. The sprl flares from the caudal part of the prezygapophysis towards the cranial surface of the neural spine; the spol extends from the postzygapophysis to the caudal surface of the neural spine. On the ventral surface, a median keel and sharp ridges on the ventrolateral edges are well defined.
The second cranial dorsal vertebra (D2) (ISIR 81) in the mounted skeleton is taller, but its centrum length is shorter than the previous vertebra. An isolated vertebra (ISIR 703) (Text-fig. 6C–D) is similar to ISIR 81 with respect to its centrum proportions, but is somewhat smaller in overall dimensions. Because it is detached and can be examined from all sides, this vertebra adds considerable knowledge of this position in the vertebral column. These two vertebrae represent a position caudal to the first cranial dorsal, perhaps the next vertebra in succession and are assigned the position of second dorsal (D2). Like the first cranial dorsal, D2 is opisthocoelous. The centrum is deeply constricted, and the planes of the centrum facets are inclined forward with respect to the long axis of the centrum, the third vertebra in the cervicodorsal region to continue the arched profile in lateral aspect. This vertebra has shallow depressions on the lateral surface of the centrum, but they are not pleurocoels. Two laminae, the ppdl and the acpl originating from the parapophysis (which has moved upward compared to its position on D1), extend to the diapophysis and the cranial part of the centrum, respectively. The diapophysis is lower than on the succeeding dorsals. The pcdl extends from the diapophysis to the caudal part of the neurocentral junction.
The prezygapophysis is shortened, and the facets are more or less subcircular, less tilted and project more outward than forward. Like the first cranial dorsal vertebra, the tprl form wing-like expansions that buttress the prezygapophyseal process. The prdl and the cprl extend from the prezygapophysis to the diapophysis occurring above the neurocentral junction and the cranial part of the centrum, respectively. The postzygapophysis is shortened, and the facets are closer to the neural spine but away from each other. The podl extends from the postzygapophysis to the diapophysis. The prominent transverse processes are supported by the tprl. The neural arch arises on the cranial two-thirds of the centrum and is more erect than in the first cranial dorsal vertebra, and the diapophysis is lower than on succeeding dorsals. The rib facets, like those on the first cranial dorsal, are situated high on the neural arch. The neural spine is flat and transversely expanded dorsally. The sprl extends from the prezygapophysis to the cranial surface of the neural spine; the spol is not clearly preserved.
The third cranial dorsal vertebra (D3) in the mounted skeleton (ISIR 80) with its opisthocoelous centrum is quite similar to the previous vertebra barring its tall height and lower centrum length. The width of the centrum is narrow in the middle, less than half of the cranial face. The large parapophyses face more outward than downward. The acpl develops as a supporting ridge extending from the parapophyses to the centrum. The ppdl extends from the parapophysis to the diapophysis. The prezygapophyseal facets are narrow, less tilted and more transversely elongate than the previous vertebra and are projected more outward than forward from the neural spine. The tprl meet just dorsal to the neural canal opening and then descend as a single lamina. The prdl connects the prezygapophysis to the diapophysis above the neurocentral junction. The postzygapophyseal facets, however, are closer together. Accordingly, the tpol meet some way up from the neural canal opening and descend as a single lamina. The transverse process arises on the neural arch to a height roughly equalling the centrum height and it remains more or less in the same position in the subsequent dorsal vertebrae. It is directed outward and very slightly upward and located closer to the postzygapophyses than the cranial dorsals. The podl extends from the postzygapophysis to the diapophysis. The neural canal is tabular, with an oval cross-section in the middle and the height of the neural canal is relatively larger than its width. The floor of the canal is slightly sunk into the body of the centrum. The neural spine is flattened craniocaudally and wide transversely. The transverse width reduces considerably towards the base, more on the cranial face than on the caudal.
The dorsal vertebrae of Barapasaurus have hollow neural spines whose chambers communicate directly with the neural canal (Jain et al. 1979; Britt 1993; Wedel 2003). Wilson (1999) interpreted the vertebral laminae as primarily pneumatic in origin but also have a secondary function, and these laminae evolved initially to partition pneumatic diverticula. Wedel (2003) concluded that the presacral vertebrae of basal sauropods were probably pneumatized by diverticula of cervical air sacs. On the basis of presence of neural spine laminae and supramedullary chambers, he further suggested that the presacral vertebral pneumatic fossae of Barapasaurus are osteological correlates of a system of pneumatic diverticula.
The anatomy of another isolated opisthocoelous dorsal vertebra (ISIR 701) (Text-fig. 6A–B) assigned to position D3 differs from the previously described dorsals in several respects. The planes of the articular facets of the centrum are orthogonal with respect to the long axis of the centrum, and the neural arch is simple and arises from nearly the full length of the centrum. The depression on the dorsolateral surface of the centrum is broad and poorly defined. The centrum shape in lateral aspect is a modified spool, with broad ventrolateral ridges and a weak ventral keel.
Another cranial dorsal vertebra with an indeterminate position from the associated skeleton (ISIR 139) is a complete neural arch and neural spine. The transversely expanded spine is weakly convex cranially and weakly concave on its caudal surface. The neural spine is constricted at its base and expands upward to form a fan shape in cranial and caudal aspects.
Dorsal vertebrae (D4–D14). Jain et al. (1979) described the first two dorsal vertebrae (D1–D2) as opisthocoelous; the remaining dorsals (D3–D14) are essentially platycoelous. The mounted skeleton corresponds to the description of Jain et al. (1979) with 14 dorsal vertebrae (Text-fig. 5) of which the first two are opisthocoelous, and the rest are variously platycoelous or slightly biconcave and short. All the centra are real; neural arches and neural spines of the few cranial and some caudal dorsal vertebrae in the mounted skeleton are real; those in the middle are reconstructed. These dorsal vertebrae form an arch that rises gently rearward, where the top of the arch is established several positions cranial to the sacral complex. These vertebrae articulate in closed-pack condition in the mounted skeleton, and this arch appears to be natural. The centrum length in these vertebrae is remarkably uniform and shorter than the centra of the cranial dorsals. All centra have deep dorsolateral depressions, but none have pleurocoels. The centra are strongly constricted and spool-shaped.
The parapophyses in the fourth dorsal vertebra D4 (ISIR 122) are shared by the neural arch although they are mainly on the centrum. A vertical ridge supports each parapophysis ventrally on the lateral face of the centrum. In the succeeding dorsals, the position of the parapophyses gradually migrates upward and arises almost entirely on the neural arch thereafter to the caudal dorsals. The parapophyses from the mid-dorsals onwards are supported below by the acpl, which joins the ventral end of the cprl. A pcpl joins the parapophysis to the cranial and posterolateral part of the centrum. The ppdl in the mid-dorsals and caudal dorsals extend from the parapophyses to the diapophyses.
The prezygapophyses in the mid-dorsals are very low on the neural arch, with the development of a hypantrum in D5 (ISIR 123) and in the succeeding dorsals. The prezygapophyseal facets curve downward medially and slightly outward, forming additional articulation surfaces. The prezygapophyses change to a higher position from the mid-dorsal towards the caudal dorsal vertebrae. From the mid-dorsals rearward, the prdl connecting the prezygapophyses to the diapophyses moves further upward. The cprl extend from the prezygapophyses to the cranial part of the centra. The transverse processes in the dorsal vertebrae are constituted essentially by the combination of prdl, podl and the posteroventrally directed pcdl. The transverse processes are mostly directed laterally and are situated high in the caudal dorsals. The postzygapophyses become almost horizontal from mid-dorsals rearward, and the tilt of the facets becomes low and in the caudal dorsals, the facets are small. The tprl is present in the mid-dorsals and in the caudal dorsals. The sprl extends from the prezygapophysis to the cranial surface of the neural spine in the mid-dorsals. The tpol are also present in the dorsal series.
The cranial face of the neural arch in the cranial dorsals is concave on either side, but this face on the mid-dorsal vertebrae is flat, nearly vertical, and bends forward from the parapophyses. The neural canal in the cranial dorsals is tabular with an oval cross-section in the middle. The floor of the canal is slightly depressed on the centrum. The canal is larger, and there is a relative increase in the height of the openings compared to their width; caudally, the canal remains more or less the same except that the cranial opening is more subcircular in D4 (ISIR 122) but tall and oval in D7 (ISIR 125). The neural canal in the caudal dorsals is roughly triangular and taller than wide. In the mid-dorsals, the canal opens dorsally through a narrow slit-like opening into a large cavity. Below this slit-like opening, the canal may be narrow and deeply sunk on the centrum. The base of the neural spine forms the roof of the cavity, and the floor of the cavity is depressed on either side of the slit-like opening. The cranial opening of the neural canal is surrounded by a deep fossa between the cprl, the dorsal surface of the centrum and the tprl.
Neural spines are simple and subrectangular in cross-section. None of the dorsals in the collections has a divided neural spine. Hyposphene-hypantrum articulations in the middle and caudal dorsal vertebrae are well developed. The prsl and the posl covering the cranial aspect of the neural spine from the base to the top are present in the dorsals. The posl are prominent ridges that diverge upwards and fade away near the top of the spine. The neural spines in the cranial dorsals are flattened craniocaudally but wide transversely. This configuration is even more prominent in the cervicodorsal vertebrae of the Colbert collection. The cranio-caudal thickness of the neural spine increases caudally. The cranio-caudal and transverse widths of the neural spines in the mid-dorsals are roughly equal at the tip. The width of the cranial face of the neural spine becomes narrow towards the base of the spine. The caudal face of the spine is, however, wider than the cranial face. The lateral faces of the neural spines are gently concave in mid-dorsals, but the concavity disappears near the top of the spine. The spdl extending from the diapophysis to the lateral part of the neural spine is prominent from the mid-dorsals to the caudal dorsals; this lamina is joined by the spol at mid-spine forming a composite lateral lamina (Wilson and Sereno 1998; Wilson 1999).
Jain et al. (1979) described and figured a caudal dorsal vertebra D9 (ISIR 127) with unusual anatomy in the region of the neural canal and neural arch. This vertebra (Text-fig. 5B) is complete except for the neural spine. The neural canal is steeply triangular. The neural arch coalesces fore and aft in weak bridges of bone that enclose a rectangular depression, which deepens rearward, then communicates with the neural canal in an elongate slot about 2 cm wide.
One mid- to caudal dorsal vertebra (ISIR 700) (Text-fig. 6E–F) in the associated skeleton of Colbert collection is platycoelous and has a rectangular centrum with flattened ventral surface. The dorsolateral depression is broad and shallow, with no indication of a pleurocoel. The neural arch arises from the full length of the centrum, a feature difficult to establish on the mounted dorsals. This vertebra, like several others in the mounted skeleton, has an unusual anatomy in the neural canal. Jain et al. (1979) described this feature as a depression in the neural arch dorsal to the neural canal, and aptly called it the dorsal cavity. This intravertebral depression is apparently related to another modification of the neural canal, as described below.
Another mid to caudal dorsal vertebra (ISIR 727) (Text-fig. 6I–L) has similar anatomy, including the intravertebral dorsal cavity, and a second unusual feature. The expanded cranial face of the neural arch forms a distinctive, nearly hemispherical depression. This depression forms the caudal limit of an unidentified intervertebral expansion of the region of the neural canal. Several of the mounted vertebrae have this feature as well. This depression communicates horizontally with the neural canal and the intravertebral dorsal cavity. Jain et al. (1979) described these depressions as excavations on the cranial and caudal faces of the caudal dorsals, now in the mounted skeleton. A consequence of these excavations is confinement of the neural canal cavity within the neural arch. Thus, the intravertebral cavity for the neural canal is short and communicates dorsally through a small opening into the intravertebral dorsal cavity. The confined neural canal in these vertebrae extends horizontally through its cranial and caudal openings into the much larger intervertebral depressions. Jain et al. (1979) considered several explanations for the neural anatomy of the caudal dorsals. According to their description, sacral vertebrae (now mounted) possess these features as well.
Jain et al. (1975) described the anatomy of the sacral vertebrae of B. tagorei. The mounted sacrum has four co-ossified sacral vertebrae with amphiplatyan centra. Neural spines are high, and the sacricostal yokes are set close together. For completeness, their measurements of this sacrum are repeated here: greatest length of the four conjoined centra, 705 mm; greatest width at first sacral, 460 mm. The sacral ribs are greatly expanded laterally where they coalesce to form the sacricostal yoke, for contact with the ilium; the yoke is long and curved. The fourth sacral vertebra has enormous sacral ribs that are taller than the centrum. The caudal facet of the centrum of the fourth sacral is slightly concave. The neural spines of the second and third sacral are co-ossified. The transversely expanded neural spines are tall and fan-shaped in cranial aspect. These features are illustrated with the ilium and pelvis, below.
A large, isolated sacral vertebra (ISIR 737), probably from the position of sacral 1, lacks the neural arch, which had not fused to the centrum. This vertebra is as large as those are on the mounted skeleton. Its lack of ossification indicates that individuals could have achieved considerably greater size than indicated by the mounted skeleton. The floor of the neural canal is represented by a narrow, continuous groove on the dorsal surface of the centrum.
In the mounted skeleton, the caudal series (Text-fig. 7A–B) is arranged by decreasing diameter of the centra. All caudal centra are biconcave, and the articular ends have subcircular outline. Caudal vertebrae lack pleurocoels, but have broad dorsolateral depressions on the centra. The proximal caudals are axially shorter than their diameter. Passing distally the centra becomes elongate or spool-shaped with the decrease in the diameter and with an increasing length of the centra. The distal caudals are more rod-like, and a couple of them are fused. Ventrally the centra are concave, smooth without any keel. Most of the caudals have weakly developed chevron facets ventrally, producing a somewhat squared profile in cranial and caudal aspect.
The first caudal vertebra in the mounted skeleton (ISIR 745) (Text-fig. 7A) has a complete, triangular neural arch and an erect neural spine. The transverse process expands laterally and forms a sturdy wing-like process. Because of this expansion, this vertebra might be considered a sacro-caudal, but there is no indication that it contributed to the support of the ilia in this stage of growth, or that it would have supported the ilia later in ontogeny. Its centrum is taller (dorsoventral axis) than long (cranio-caudal axis), whereas mid- and caudal centra are longer than tall. The caudal face of the centrum of the first caudal vertebra slopes a little caudoventrally at a level below the cranial face, indicating that the tail is slightly downturned in this region. However, this feature is not prominent in other cranial vertebrae of the caudal series. It seems that the tail droops a little behind the pelvis and then remains straight (Text-fig. 3).
In the first few caudals, the neural arches cover almost the total length of the centrum, whereas it is confined to the cranial half of the centrum in middle caudals, while the distal caudals lack neural arches. The transverse processes in the second caudal vertebra in the mounted series are expanded laterally, their relative size reduced rearward but quite prominent to the middle of the caudal series. The prezygapophyses, preserved in proximal caudals and Ca14 (ISIR 747) (in the mounted skeleton) project anterodorsally and bend forward. The postzygapophyses are not preserved. The neural canal in the first caudal is triangular but becomes rounded caudally. Caudal vertebrae in positions Ca13-14 (ISIR 746–747) and Ca32 (ISIR 734) include neural spines which are simple, laterally compressed and project dorsally. In Ca13-14 (ISIR 746–747), the spines are comparatively tall, while in Ca32 (ISIR 734) the spine is short and is inclined backward.
The vertebra in the position of Ca33 (ISIR 748) (Text-fig. 7A) in the mounted skeleton appears to be nearly complete. Its neural spine is inclined rearward and broadly expanded. This vertebra lacks zygapophyses. The last mounted caudal is in the position of Ca43 (ISIR 134); it possesses an elongate centrum. An isolated distal caudal (ISIR 739 found near the pubis, ISIR 740) with a complete neural arch is approximately the same size as the last several caudals in the mounted skeleton. Its enormous, rod-like neural spine is broadly oval in cross-section and elongate, extending caudally well beyond the caudal articular facet. Five isolated caudals (ISIR 704, ISIR 144, ISIR 145, ISIR 146 and ISIR 147) from the Colbert excavation have amphicoelous to amphyplatyan centra. They lack neural arches, which were separated at their growth lines on the centra, indicating subadult anatomy. Eleven distal caudals (ISIR 706–716) are similarly elongated, and considering their size, have unexpectedly robust neural arches. The centra are fused on the two smallest caudal vertebrae, perhaps representing the terminus of the tail. The centra are amphicoelous to amphiplatyan, with circular to ovoid cross-sections.
Three chevrons in the collection (ISIR 723, ISIR 724–725) permit limited characterization of the haemal arch anatomy. One (ISIR 725) (fused by preservation to the right femur, ISIR 58, on the mounted skeleton) is V-shaped with a closed proximal cross-bar (Text-fig. 7C); the position in the tail for this chevron is more caudal than the smaller two. The other two chevrons, ISIR 723 (Text-fig. 7D) and ISIR 724, are Y-shaped (forked), with fused cranial and caudal projections. In ISIR 723, the dorsal surface of the haemal canal is open, lacking a proximal cross-bar.
There are three scapula-coracoids in the collection: left scapula and coracoid (ISIR 68 and ISIR 69, mounted skeleton) and right scapulacoracoid (ISIR 92) (Text-fig. 8). The left scapula (ISIR 68) probably represents a smaller individual, where the scapula and coracoid are disarticulated, but it is fairly complete except for the cranial and caudal margins. The left coracoid (ISIR 69) and the right scapulacoracoid (ISIR 92) are more or less in close size range. In the coracoid, the cranio-ventral part is missing. The scapular blade is damaged in cranial and caudal margins in both the specimens. A major part of the scapular blade and the lower portion of the coracoid of ISIR 92 are missing. Although the scapula and coracoid of ISIR 69 and ISIR 92 are coalesced, the general curvature of the bones is well preserved. Jain et al. (1975, pl. 93A) briefly described the scapula and coracoid.
The scapula (ISIR 68) (Text-figs. 8A–E) on the mounted skeleton and isolated element (ISIR 92) is slender, with a modestly expanded blade showing a gentle dorsoventral convexity on its lateral surface. The medial surface is slightly concave. An oblique, faint ridge originates above the glenoid cavity and fades away laterally in the middle of the bone. Medially a similar but prominent ridge is present. The scapular blade is much thickened in the region of these two ridges. From its distal extremity, the cranial border curves gradually downward and expands to form a relatively weak acromium, producing a concave profile in lateral aspect. The caudal border is nearly straight in lateral profile, expanding only slightly where it contributes to the glenoid facet. The scapula thickens cranially towards the articulation with the coracoid. The scapula-coracoid articulation is weakly sigmoid in dorsal aspect. The caudal end is thick and rugose indicating the presence of suprascapular cartilage in life. Passing forward the blade gradually narrows in the mid-length and then flares again cranially. The cranial edge of the blade is very thin, forming a shallow curvature to meet ventrally with the thickened acromial outgrowth. The ventral edge is equally thin, narrows anteriorly and expands in the thickened glenoid region. Laterally, the fossa for supracoracoscapular muscle is shallow and forms an arc between the acromium and the glenoid rim.
The glenoid is semi-elliptical in outline; the articular surface is rugose facing craniodorsally with a central dip. In ISIR 68, the articular surface for coracoid is rough, pitted and considerably thickened. In ISIR 69 and ISIR 92, a faint ridge marks the contact between the two elements, where these are coalesced.
The coracoid is subcircular in outline, with a convex lateral surface and a fairly concave medial surface. The caudal border is nearly circular. The small coracoid foramen is situated cranial to the glenoid cavity near the scapulacoracoid articulation. Below the foramen, an antero-caudal ridge divides the bone into two sloping surfaces. The glenoid surface on the coracoid is subtriangular, rugose, facing somewhat laterally and caudally. Its inner half is slightly hollowed out. The ventro-medial edge of coracoid shows a smooth and longitudinal groove for probable contact with the sternal plate. The coracoid thickens caudally to contribute to the deeply concave glenoid cavity, which it shares with the scapula, for reception of the humerus.
The forelimb in B. tagorei is shorter than the hind limb; in the partial association of the small individual, the humerus is 78·5 per cent of the length of the femur, and the combined length of humerus + radius is 83·1 per cent of that of the femur + tibia. The proximal segment is distinctly longer than the distal; the length of the radius is 70 per cent of that of the humerus, and the length of the tibia is 60·5 per cent of that of the femur. A definitive quadrupedal pose for Barapasaurus has been inferred from its robust and elongate forelimb (Wilson 2005b).
Humerus. Humeri in the Barapasaurus collection are remarkably uniform despite pronounced differences in size. There are three left humeri (ISIR 70; Text-figs. 9A–F), (ISIR 86 and ISIR 87) and three right (ISIR 85, ISIR 88 and ISIR 93). The humeri ISIR 70 and ISIR 85 represent the smallest groups. These are about the same size and proportions and possibly belong to the same individual. ISIR 70 is complete, but the shaft is distorted, whereas the latter is deficient in the shaft region. ISIR 86 is a medium sized, least distorted humerus and is the basis for restoration along with ISIR 70. However, its proximal end is partly damaged. ISIR 87 is the proximal half of a large humerus, the shaft showing a central cavity. ISIR 88 is fairly complete except at the two ends. It is much less distorted and is intermediate in size between ISIR 86 and ISIR 87. ISIR 93 comprises four pieces which cannot be fitted properly. Its estimated length is comparable with ISIR 87. It is somewhat crushed antero-caudally, and the two ends bear well-marked rugosities.
The humerus is long, slender and expanded transversely at either end. The proximal expansion, with the prominent deltopectoral crest, is larger than the distal extremity. The two expansions, however, make a slight angle (15 degree) with each other. The bone is flattened cranio-caudally throughout its length. Its proximal articular surface is convexly rounded, triangular, with apex in caudal position; this end is marked by rugosities in the large humerus. Opposite the apex, on the cranial border of the articular facet, is a distinctive sulcus. The shaft is narrowest at its middle (approximately one-third of the proximal expansion) and expands distally to form slightly separated, convex articular surfaces for the radius and ulna; it is ovoid in cross-section and shows a central cavity in ISIR 87.
In lateral aspect, the humerus is sigmoid. Cranially the medial margin of the humerus is more deeply curved than the lateral margin. The deltopectoral crest is confined approximately to the proximal third of the humerus and is situated on the craniolateral margin of the surface. The crest does not form a sharp acute apex; it is produced into a thick and flat ridge and makes a shallow concavity with the adjacent cranial surface, which is otherwise almost flat. Apparently, this concavity was for the insertion of the supracoracoscapularis and coracobrchialis muscles. Laterally, a prominent ridge runs throughout the length of the bone. The distal end is slightly inclined cranially and is highly rugose and flat. The transverse diameter is twice the sagittal diameter. The radial and ulnar condyles are feebly developed and are separated by a longitudinal groove cranially. Caudally the anconeal fossa is moderate in depth.
Radius and ulna. Among the epipodials are one left radius (ISIR 71), one right radius (ISIR 89), two left ulnae (ISIR 72 and ISIR 91) and one right ulna (ISIR 90). From the field evidence, it seems that the radius ISIR 71, ulna ISIR 72 and humerus ISIR 70 belong to the left side of a small individual; the radius and the ulna were found on either side of the humerus in association. Similarly, the right epipodials of a larger individual are represented by the radius (ISIR 89) and ulna (ISIR 90), which were found side by side in the excavation. A left ulna (ISIR 91) was found two metres away from these two bones and is comparable in size (Text-fig. 2, Association G). However, no compatible humerus was recovered alongside, although a left humerus (ISIR 87) which seems to be the right side of the second association was obtained far away from this spot. ISIR 71 and ISIR 72 are fairly complete, well preserved and almost undistorted. The restoration of the radius and ulna are entirely based upon this material. In ISIR 89 and ISIR 90, the ends are worn and the shafts are deficient; ISIR 91 is better preserved and is almost complete. Its shaft and distal end show minor erosion.
The radius (Text-fig. 9M–R) is cylindrical, shorter and slimmer than the ulna and its length is 70 per cent of that of the humerus. The shaft is narrow, long, ovoid in cross-section, weakly bowed in cranial and caudal aspects, and straight in medial and lateral aspects. The shaft expands smoothly into an inflated distal extremity with rugose texture even more pronounced than on the proximal end. Proximally, the radius expands cranio-laterally to caudo-medially and shows rugosities. The proximal articular facet is convex and ovoid in profile. Laterally, a sinuous ridge runs lengthwise and terminates distally as a projection for the ligamentous connection to the ulna. Below this projection is a flat, triangular facet for contact with the ulna. The distal facet is also convex, pronouncedly rugose but is distinctly circular in profile.
The ulna (Text-fig. 9G–L) is much heavier and slightly longer than the radius. It has an enlarged and robust proximal end, a narrow shaft, and a slightly expanded distal end. The proximal end is triradiate with prominent craniolateral and craniomedial processes and less prominent caudal process. Its rugose triangular proximal facet has a moderately deep cranial groove formed by craniomedial and craniolateral processes to receive the proximal end of the radius. The cross-section of the bone is triangular for the proximal two-thirds of its length, beyond which the shaft becomes ovoid in cross-section. The radial groove on the medial surface is well marked and deeply set. The cranial apex of the proximal end is bluntly rounded, whereas the caudal one projects sharply as a thin flange. From it, a sharp ridge runs obliquely to the distal end. A corresponding tuberosity is present on this ridge, as in the radius, near the distal end, presumably for the ligament attachment to bind the radius. Below this tuberosity is a facet for reception of the radius. Thus, the two bones were locked essentially parallel to the each other in life without any supination. The shaft is straight, and distally the bone is expanded a little transversely; the distal facet is elongated, rugose and concave for reception of the carpus.
Metacarpals. There are three metacarpal bones (Text-fig. 10) in the collection. These are the third (ISIR 94), fourth (ISIR 95) and the fifth (ISIR 96) of the left side. Except the long third metacarpal (ISIR 94), the metacarpals are subequal in length. They were collected as surface finds and may belong to a single individual. The third and fifth metacarpals are complete, whereas the fourth is represented by the proximal half only. In addition, an isolated element (ISIR 108) is probably the first metacarpal of left side.
The third metacarpal (ISIR 94) (Text-fig. 10E–H) has enlarged ends and a narrow shaft, where the greatest diameter of the proximal end makes a small angle with the transverse diameter of the distal one. The proximal end is subtriangular with lateral extension. The medial margin forms the base of the triangular shape, the lateral margin tapers into an apex, to overlap the medio-proximal corner of the fourth metacarpal. From the caudal apex of the triangular end, a blunt but pronounced ridge continues up to the mid-length of the bone. The summit of the ridge shows rough end surface. Beyond the ridge, the shaft is narrow and triangular in cross-section. The distal end is convex, with an asymmetrical trochlear facet, and is roughly rectangular in outline. On the lateral surface near the distal extremity, a longitudinal groove indicates the passage of a flexor tendon.
The proximal end of the fourth metacarpal (ISIR 95) is triangular, convex, and elongated transversely. The ventral margin forms the broad base of the proximal triangle. The cranial apex of the triangle is continued as a ridge downward, demarcating the two separate contact surfaces for Mc III and Mc V on either side. The caudal surface in the proximal half has a shallow concavity. The proximal end narrows abruptly to narrow shaft. The distal half is lacking.
The fifth metacarpal (ISIR 96) (Text-fig. 10I–M) is a short and squat bone, with expanded proximal extremity and a broad shaft. The flat proximal extremity is subtriangular and slopes laterally. The shaft is flattened cranio-caudally. Posteriorly the shaft is concave. The distal end is ovate and elongated transversely.
An isolated metacarpal (ISIR 108) (Text-fig. 10A–D), probably the first metacarpal of the left side, is long and stout with little expansion at the two ends. The proximal end is roughly subtriangular and is more expanded than the distal end. Laterally a blunt, hook-like projection occurs, below which the shaft is semicircular in cross-section, robust and broad and maintains an almost uniform width. The laterodistal corner is broken. The distal end is semi-rectangular with a median broad groove.
Phalanges. There are four ungual phalanges in the Barapasaurus collection, two from the manus and two from the pes (described below), ranging in size from large to small. Unguals ISIR 749 (Text-fig. 10N–Q) and ISIR 110 (Text-fig. 10R) are the ungual phalanges, digit I of the manus of different individuals. The surface texture of the larger of the two manual unguals (ISIR 749) is extremely rugose and pitted proximally, and smoother distally where the claw sheath covered the bone in life. The articular surface is irregular rather than smooth, indicating limited movement and weight bearing by this digit. The shaft is tightly curved, and the bone terminates in a blunt knob-like protuberance. The smaller ungual (ISIR 110) belonged to a young individual. It is similar to the larger ungual, but less heavily rugose and pitted. The proximal end is oval, recurved, blunt and concave with a median ridge and two symmetrical depressions beneath the proximal margin of the sheath. The articular surface is concave, smooth and bears a prominent rim of constriction just below this surface. The cranial surface is pitted, indicating a horny sheath of cover during life. The lateral and medial surfaces have a groove extending from the middle to the tip of the claw. The distal end is bluntly rounded. Barapasaurus might have a tubular manus as suggested by Upchurch et al. (2004). Although the collected bones of Barapasaurus lack a complete preserved manus, Wilson (2005b) inferred reduction of manual phalanges in this taxon.
The collection includes five ilia, five ischia and five pubes, of which two distinct size groups and associations can be separated. A right ilium (ISIR 52), one left (ISIR 54) and one right (ISIR 115) ischia, one left (ISIR 117) and one right (ISIR 57) pubes belong to a small individual (Text-fig. 2, Association C). One left (ISIR 111) and one right (ISIR 51) ilia, one left (ISIR 114) and one right (ISIR 53) ischia and one left (ISIR 56) and one right (ISIR 55) pubes belong to a large individual (Text-fig. 2, Association A). The sacrum (ISIR 50) was found along with the pelvis of the associated large individual; however, two ischia were slightly away from the locality. In addition to these partial associations, a few isolated pelvic bones were also collected: these include one right (ISIR 112) and one left ilia (ISIR 113), one left ischium (ISIR 116) and one left pubis (ISIR 118) (Text-fig. 2, Association F). All the pelvic bones occurred as disarticulated elements without any co-ossification between them.
Ilium. The ilium was described and illustrated by Jain et al. (1975, figs 3–4; Jain et al. (1979), pls 93–94) on the basis of ISIR 51 and ISIR 52. In the mounted skeleton, the ilia are ISIR 51 and ISIR 111. These ilia are generally deficient in the blade in the caudo-dorsal region, whereas the cranial process of the blade is well developed. Except ISIR 112, the ilia are not very crushed and are in a good state of preservation. The iliac profile (Text-fig. 11A–B) in lateral aspect is rounded dorsally with a gentle convexity. The curved and subtriangular iliac crest extends cranially to form the prominent preacetabular process. The pubic peduncle is likewise prominent and extends downward and slightly cranially. The ischiadic peduncle is short and subrectangular in lateral view. The acetabulum is deeply concave, and the bone is thickened medially. The sacricostal yoke does not contribute to the deep medial wall of the acetabulum. As mounted, the dorsal crests of the ilia are narrow and converge cranially. The ilia are firmly ankylosed to the sacral ribs (Text-fig. 5E).
Pubis. ISIR 57 (Text-fig. 12E–H) is the best-preserved pubis, which retains the twist below the proximal and distal regions. In other specimens, there is a variation of the amount of the twist because of different degree of crushing. All the specimens are deficient in the caudal margin, in the region between the two symphyses. Jain et al. (1975, 1979) described and discussed the implications of the unusual anatomy of the pubis based on ISIR 55, ISIR 56 and ISIR 57; to this list ISIR 117 may be added. They suggested that the pubis is somewhat shorter than this ischium and illustrated these two bones in profile with considerable difference in length. However, the pubis (ISIR 117) and ischium (ISIR 54) in the associated skeleton ‘Colbert excavation’ are almost exactly the same length. The pubic apron is developed but without substantial thickening and without obscuring the basic anatomy of the shaft. The proximal articulations with the ilium and ischium are borne by the stout iliac and ischiadic peduncles, respectively. The proximal part of the pubis is pierced by large obturator foramen, which is situated well within the pubic apron. The pubic apron articulated with its partner along nearly the full length of the midline. The resulting pelvic basin, which is the opening between the pubes beneath the sacrum, was consequently narrow, as recognized by Jain et al. (1979). The shaft of the pubis is straight in anterolateral profile and expands distally to form a greatly enlarged distal extremity to meet its partner at the midline. The pubic shaft in posterolateral aspect is steeply curved. The distal extremities of the pubes articulate along the midline, forming a cranioventrally sloping trough.
Ischium. Jain et al. (1975) described and illustrated the ischium of B. tagorei on the basis of two specimens, ISIR 53 and ISIR 54 (Text-fig. 12A–D). The curvature of the ischiadic shaft is preserved only in ISIR 54, which is fairly complete. The relatively delicate cranial flange was invariably damaged in all specimens. The shaft of the ischium is slender and straight, transversely compressed and longer than the pubis shaft. Its distal extremity is moderately expanded, subrounded, and the symphyseal contact is narrow. In articulation with the other bones of the pelvis, it is oriented downward and rearward, at roughly 45 degree angle with respect to the horizontal axis of the vertebral column. It is almost exactly the same length as the pubis (see above), rather than longer as originally published. Its articulation with the ilium is broad and robust where it contacts the iliac peduncle. Its pubic peduncle is deep and equally robust. The ischiadic contribution to the acetabulum is roughly equal to that of the pubis.
The Barapasaurus mounted skeleton includes the femora, tibiae, right fibula, right astragalus, three left metatarsals and two ungual phalanges. In addition, rear limb bones in the associated skeleton include both femora, left tibia and both fibulae.
Femur. There are seven well-preserved femora in the Barapasaurus collection; these include three right femora ISIR 58, ISIR 99 and ISIR 100; and four left femora, ISIR 59, ISIR 60, ISIR 97 and ISIR 98. The femora are fairly complete except ISIR 100 in which only the proximal half is retained. The lengths of the femora range from 875 to 1365 mm. The right femur on the mounted skeleton (ISIR 58) (Text-fig. 13E–F) has a length of 1167 mm. It appears that ISIR 58 and ISIR 59 represent a pair of femora of a juvenile individual. Most of the femora are compressed craniocaudally from burial; consequently, the shafts are flattened, and the condyles are also partly damaged.
The femur is long and slender with expanded ends and a straight shaft. The prominent, hemispherical head is set at a right angle to the straight, slender shaft. The proximal end is rugose and ovate, with two distinct levels. The higher one is more rugose with deep furrows forming the proper head, while lower part is less rugose and grades laterally into the greater trochanter. Laterally, a shallow depression occurs just below the greater trochanter. The shaft has a fairly uniform width, ovoid in cross-section, with a central cavity (ISIR 100). There is no lesser trochanter, but the fourth trochanter is prominent and forms a distinctive ridge (Text-fig. 13B). The fourth trochanter is situated on the caudo-medial margin, and its apex is slightly higher than the midlength. It projects caudally as a ridge-like process with an acuminate and declined tip. The medial surface of the fourth trochanter shows a rough surface.
Distally the robust tibial and fibular condyles are well developed and pronounced. The medial condyle is larger than the lateral condyle. The two condyles are separated by the intercondylar groove. The lateral condyle is further subdivided into an internal condyle and a lateral epicondyle by a lateral longitudinal grove (Janensch 1961) which might have provided the passage of a strong tendon. The medial and the internal condyle together receive the proximal end of the tibia while the lateral condyle articulates with the fibula.
Some morphological variations as a result of growth have been noted on the femora of B. tagorei. On the basis of the greatest length of the femora, three distinct size groups can be recognized; these are small (ISIR 59 = 875 mm and ISIR 60 = 885 mm), medium (ISIR 99 = 101 mm) and large (ISIR 58 = 122 mm; ISIR 97 = 1365 mm; ISIR 98 = 132 mm and ISIR 100 = 134 mm (estimated). The ends of the small femora do not have any rugosities, whereas the larger femora are marked by pronounced rugosity, and the medium sized femora show only some pitting. The head of the large femora is distinct, whereas in the medium and small femora the head cannot be demarcated from the rest of the proximal end.
The anatomy of the fourth trochanter resembles that described by Yadagiri (2001) for Kotasaurus, but is not sufficiently preserved to permit confident comparison.
Tibia and fibula. The epipodials include four left tibiae (ISIR 61, ISIR 62, ISIR 63 and ISIR 102), three right tibiae (ISIR 77, ISIR 101 and ISIR 103), two left fibulae (ISIR 64 and ISIR 105) and two right fibulae (ISIR 104 and ISIR 106). A partial association of a hind limb was recognized when the tibia, ISIR 62 was found along the side of the fibula, ISIR 64; these two were recovered near the femur, ISIR 60 (Text-fig. 2, Association C). The right tibia, ISIR 101 occurred nearby the right femur, ISIR 58 (Text-fig. 2, Association H). The size of ISIR 101 and ISIR 102 indicate a probable pair and their corresponding fibulae may be ISIR 105 and ISIR 106. The tibia, ISIR 103 and the fibula ISIR 104 may belong to the same side of a large individual. The tibia ISIR 62 is least damaged and retains the twist of the two ends. Although the tibia, ISIR 63 is represented by only the distal end, it possesses the notch for the astragalus. The fibulae ISIR 64 and ISIR 104 are fairly complete.
Jain et al. (1975) described the tibia of B. tagorei. The tibia length (ISIR 62 = 505 mm) is short relative to the length of the femora (865 mm and 867 mm) in the associated skeleton. The lengths of the right tibia (ISIR 103) and right femur (ISIR 58) in the mount are 714 and 1167 mm, respectively, approximately the same ratio (61 per cent) as in the associated skeleton (58 per cent). The tibia (Text-fig. 14A–E, L) is robust, with well-developed cnemial crest. The proximal articular surface is rugose, ovoid and interrupted by a notch. The cnemial crest is a thin vertically elongated flange protruding craniolaterally for the attachment of triceps femoris muscle. It extends up to one-fourth of the tibia length and is separated from the rest of the proximal end by a prominent notch in the caudolateral aspect. The remainder of the proximal end is differentiated into two subequal medial and lateral expansions caudally for the articulation of the femur. The lateral expansion and the cnemial crest converge downward forming a triangular surface for the contact with the fibula but with a notch between them. In cranial profile, the shaft is broad and straight. The shaft is semi-circular in cross-section and expands slightly towards the distal extremity. The distal articular facet is distinctly concave, triangular in outline and lacks separation between the caudoventral process and the articular surface for the ascending process. Distally the tibia articulates with the astragalus by a notch and a descending flange. The notch is located on the caudolateral corner for the reception of the ascending process of the astragalus. The descending flange on the caudomedial surface of the tibia projects ventrally to fit into the astragalus. A faint groove between the notch and the descending flange marks the passage of a tendon for the flexor muscle of the foot. Rugose marks occur along the periphery of the distal end.
The fibula (ISIR 64, Text-fig. 14F–K) is slender and transversely compressed, expanded proximally and distally, and narrowest at the centre of the shaft. The weakly crescentic proximal end adjoins the crista lateralis of the tibia, and together these surfaces articulate with the lateral condyle of the femur. In lateral view, the fibular shaft is narrow and straight from its proximal extremity to near the distal extremity, where it expands somewhat to form the carpal facet and becomes sigmoid in profile. The lateral trochanter is weakly developed depicting the origin of the flexor digitorium longus muscle (Borsuk-Bialynicka 1977; Wilson and Sereno 1998), and the tibial articular surface is a narrow ridge parallel to the shaft proximally, but rather oblique near its distal extremity. A raised elliptical rugosity occurs on the lateral side of the mid-shaft of the fibula for the articulation of the tibia (Wilson and Sereno 1998); the muscle scars on the fibula are considered as a derived condition of Barapasaurus (Wilson and Sereno 1998). The articular facets are convex. The distal end is elongated and subrounded, highly rugose and fits to the lateral surface of the astragalus.
Tarsus. The tarsus is represented by a right astragalus (ISIR 107) which was found in association with tibia ISIR 103 and a right calcaneum (ISIR 743). There are three metatarsals of the right side in the collection; these include the metatarsal I (ISIR 65), metatarsal II (ISIR 66) and metatarsal IV (ISIR 67). These three were found associated with the astragalus (ISIR 107), tibia (ISIR 103) and fibula (ISIR 104). Metatarsal I and IV are complete and undistorted; metatarsal II is damaged in the shaft and proximal region. The ends of the metatarsals are rugose.
The right astragalus (Text-figs 14L, 15A–C) is flat and subtriangular. The proximal surface bears a prominent ascending process and a medial depression; the ascending process is convex and articulates with the distal extremity of the tibia. Medially, the base of the ascending process slopes downward gradually into the depression. Its convex distal surface is uniformly contoured, broad and rugose. The cranial margin is straight and rugose, with a gently sloping surface for the articulation of the metatarsals. Separate facets for individual metatarsals are not discernible. The calcaneal articular surface is indistinct. The caudal surface is less rugose and exhibits an upturned lip which superficially covers the tibia. The medial margin is rugose and tapers cranially. The lateral surface is quite deep; its upper half is quite smooth, probably for articulation with the calcaneum, and the lower half is rugose. Other anatomical features of this element cannot be discerned owing to its position in the mounted skeleton.
The calcaneum (ISIR 743) (Text-fig. 15D–G) is roughly quadrangular. Cranially, it has a convex outline. The proximal articular surface is semi-trapezoidal, slightly concave and smooth for reception of the fibula. The medial surface, which articulates with the astragalus, is gently convex and smooth. Laterally, the calcaneum is a little thickened but smooth and gently rounded. The caudal surface is flat. The distal articular surface for the metatarsals is convex, rough and has a fine mediolateral ridge on the caudodistal margin, which might have attached the fleshy pad (Bonnan 2000). Wilson (2005b) concurred that the pes of Barapasaurus was supported by heel pad. Bonnan mistakenly mentioned the presence of calcaneum in Barapasaurus (2000, 2005; Bonnan, pers. comm. 2007). It had not been reported earlier. The calcaneum in Text-fig. 15D–G would fit into the space between the distal extremity of the fibula and lateral metatarsals in the mounted skeleton (Text-fig. 16A–C).
Metatarsals. Right metatarsals I, II, and IV (Text-figs. 15A–C, 16A–D) are elongate, with proximodistal lengths of 178, 259 and 250 mm. Metatarsal I (ISIR 65) is the shortest, but heaviest of the three, and its shaft is twisted. The twist of the proximal expansion is 55 degrees with respect to the distal one. Its proximal articular surface is more greatly expanded than the distal end. The proximal end is elongated and ovoid, with its apex pointing laterally for overlapping the adjacent metatarsal II. The shaft is broad and compressed cranio-caudally; its lateral margin forms a sharp ridge terminating distally into a notch. The medial margin is flat and broad. The distal extremity forms an asymmetric ginglymus where the lateral condyle is larger than the medial one.
Metatarsal II (ISIR 66) is longer and slimmer than the first, but about the same length as metatarsal IV. The proximal end has a biconcave outline and is more expanded than the distal end. Medially, the proximal surface is produced into a concave flange for the reception of metatarsal I. The shaft is considerably more robust, flattened, oval in cross-section and shows a similar ridge and a notch on the lateral margin as in the metatarsal I. The distal end has a symmetric ginglymus surface with a deep median groove.
Metatarsal IV (ISIR 67) is long and slender. The proximal end is narrow, elongated and subtriangular with a sharp apex pointing medially. The shaft has ridges both laterally and medially and similar notches on the distolateral margin as in metatarsal I and II. The distal end is squared and anterocaudally thickened; no distinct grove is visible on the ginglymus surface.
In life, the pes was semiplantigrade as indicated by the presence of weakly developed lateral trochanter on the fibula and the long metatarsus. The twisted metatarsal I and robust condition of metatarsals I and II indicate that much of the weight borne by the pes was on the inner side of the toes. Wilson (2005b) suggested that the pedal unguals of Barapasaurus were deflected laterally relative to the long axis of each digit and of the foot itself. He considered this feature as a derived character of Barapasaurus.
Phalanges. As mentioned in the description of the manus, there are four dissociated ungual phalanges, two of which are pedal. ISIR 83 and ISIR 84 (Text-fig. 16D–L) belong to digit I and digit III of the right pes. These two ungual phalanges are similar in general build, recurved, somewhat blunt and pitted on the surface. The proximal end is oval, concave with a median ridge with two symmetric depressions on its side. The articular surface is extensive up to the cranial tip, allowing considerable flexion and extension. The cranial margin is longer that the caudal one.
Upchurch et al. (2007a) in their phylogenetic analysis diagnosed 292 characters for 34 taxa of basal sauropodomorphs and on the basis of published information (e.g. Jain et al. 1975, 1979; Wilson and Sereno 1998) compared Barapasaurus which scores 99 characters in their data matrix (Upchurch et al. 2007a). It may be mentioned that because of lack of skull, braincase and mandible, the first 85 characters could not compared. Hence, 33·9 per cent character scoring out of 292 characters was actually used. Yates (2007a) carried out another phylogenetic analysis of the basal sauropodomorphs and identified 351 characters from which Barapasaurus shares 96 characters and the first 106 characters from his data matrix belonged to skull, braincase and mandible. Hence, he used 27·53 per cent of the characters for Barapasaurus. Later Yates (2007b) added more characters in his previous phylogenetic analysis. Subsequently, Smith and Pol (2007) described a basal sauropodomorph from Antarctica and carried out cladistic analysis and added eight more characters to the character scoring of Yates’s (2007a, b) phylogenetic analyses; but according to them, Barapasaurus does not share any of those characters.
The present description of Barapasaurus improves the character scoring of Yates (2007a, b), Smith and Pol (2007) and Upchurch et al. (2007a). In the present work, the data matrix comprising 34 taxa and 292 morphological characters that was initially used by Upchurch et al. (2007a) is modified (Table 4), especially those character states that are related to Barapasaurus. However, several of their characters, especially those with ratios could not be used because of lack of associated/articulated bones. The present description improves the scoring of Barapasaurus with a total of 143 characters out of 292 and 48·97 per cent of character scoring can be used. It may be mentioned here that most of the character states for Vulcanodon are either missing/unknown (characters ranging from 1–138 and 179–208). The data matrices of Yates (2007a) and Smith and Pol (2007) deal mostly with prosauropods and were not used here.
Table 4. Revised data matrix of Barapasaurus tagoreiUpchurch et al. (2007a) that was used for the current phylogenetic analysis.
The modified data matrix is subjected to a maximum parsimony analysis using PAUP 4.0b10 (Swofford 2000). Multistate characters were treated as unordered, and all characters were weighted equally. Unknown/missing characters were coded as ‘?’. The heuristic search algorithm was used with default settings. The analysis recovered 47 most parsimonious trees (MPTs), each with tree lengths (TL) of 738 steps, a consistency index (CI) of 0·40 rescaled to 0·26, a retention index (RI) of 0·65, and a homoplasy index of 0·6. The topology of the strict consensus tree is shown in Text-figure 17, which is generally compatible with that of Upchurch et al. (2007a). Although Vulcanodon and Kotasaurus were considered as basal in comparison with Barapasaurus (Upchurch et al. 2007a), in the current work Barapasaurus was found to be more basal than Vulcanodon and is further removed from the Eusauropoda (sensuUpchurch et al. 2007a). A bootstrap analysis (1000 replicates) shows moderate to strong support (>60 per cent) for the fully resolved clades within the strict consensus tree. In addition, topological constraints were created to explore two separate hypotheses. These comprise: (1) Kotasaurus and Barapasaurus are sister taxa; and (2) Barapasaurus falls within Eusauropoda. In the current analysis, it was found that 21 MPTs (TL = 748) were generated when the first constraint was enforced showing that another 10 steps are required for Kotasaurus and Barapasaurus to be sister taxa. On the other hand, the second constraint resulted in the generation of 675 MPTs (TL = 741), showing that three more steps are necessary to incorporate Barapasaurus within Eusauropoda. In both the cases when the constraints were applied, the resulting topologies show that the Sauropoda constitutes a monophyletic clade, the Eusauropoda with Camarasaurus and Omeisaurus remaining as sister taxa, similar to that seen in the unconstrained MPTs.
Previous work on classification of Barapasaurus
In 1975, Jain et al. described B. tagorei under the Infraorder Sauropoda but did not assign it to a family. They drew attention to the similarity of some aspects of the anatomy of B. tagorei to that of the Prosauropoda, especially with respect to the sacrum. Gauthier (1986) gave an outline of the sauropod relationship where he considered Barapasaurus as one of the most primitive sauropods along with Vulcanodon and listed at least 20 synapomorphies uniting them with other sauropod taxa. Subsequent workers (Bonaparte 1986; Mcintosh 1990; Wilson and Sereno 1998; Upchurch et al. 2004, 2007a) also kept Barapasaurus within Sauropoda. Wilson and Sereno (1998) defined Sauropoda as all sauropodomorphs more closely related to Saltasaurus than to Plateosaurus and proposed 17 synapomorphies for Sauropoda among which Barapasaurus shares (1) quadrupedal posture with columnar limbs and short metapodial; (2) four sacral vertebrae; (3) low deltopectoral crest of humerus; (4) absence of olecranon process; (5) triradiating proximal end of ulna with deep radial fossa; (6) ilium with low and subrectangular ischial peduncle; and (7) ischial shaft equal to pubic shaft. The other proposed synapomorphies are difficult to ascertain because of non-availability of associated bones. Upchurch et al. (2004) defined Sauropoda as a stem-based taxon with 15 characteristic features some of which were earlier suggested by Wilson and Sereno (1998); Barapasaurus shares (1) triradiate proximal end of ulna; (2) middle and distal portions of the femoral shaft straight in cranial view. Upchurch et al. (2007a) produced a detailed phylogenetic analysis on basal sauropodomorph relationships and maintained Sauropoda as a stem-based taxon which includes Eusauropoda, Barapasaurus, Vulcanodon, Kotasaurus, Chinshakiangosaurus, Antetonitrus, Blikanasaurus, ‘melanorosaurids’ and Jingshanosaurus.
Wilson and Sereno (1998) defined Eusauropoda as a stem-based taxon and included all sauropods more closely related to Saltasaurus than to Vulcanodon and diagnosed 40 characters. Because of lack of skull/skull elements, out of 40 characters the first 13 cranial characters are not known in Barapasaurus, and from the rest, Barapasaurus shares (1) spatulate crown; (2) tooth enamel with wrinkled texture; (3) opisthocoelous cervical centra; (4) dorsal neural spines broader transversely than craniocaudally; (5) fork-shaped distal chevron; (6) fibular lateral trochanter present; (7) minimum shaft width of metatarsal I greater than that of metatarsal II–IV; and (8) pedal phalanges broader. Wilson and Sereno (1998) proposed another clade ‘Barapasaurus + Omeisaurus + Neosauropoda (sensuUpchurch 1995)’ and diagnosed seven characters of which Barapasaurus has 6 characters, including (1) neural arches of caudal cervical vertebrae and cranial dorsal vertebrae with interprezygapophyseal lamina; (2) opisthocoelous cranial dorsal centra; (3) neural arches of the middle and caudal dorsal vertebrae with composite lateral lamina; (4) presence of sacricostal yoke; (5) fibula with broad triangular scar for tibia; and (6) astragalar posterior fossa divided by crest.
According to Upchurch et al. (2004), Eusauropoda is a node-based taxon which defines the most recent common ancestor of Shunosaurus and Saltasaurus and all the descendants of that ancestor. They diagnosed 38 characters including some of Wilson and Sereno’s suggested character states. The additional common characters of Barapasaurus includes (1) caudal margins of caudal cervical neural spines sloping strongly forward in lateral view; (2) dorsal surfaces of sacral plates level with the dorsal margin of the ilium; (3) dorsal margin of the ilium gently convex in lateral view; (4) reduced ischial peduncle of ilium so that the long axis of the iliac blade slopes craniodorsally in lateral view; and (5) femoral cranial trochanter absent. Wilson (2005a) considered the distinctive spatulate shape and crowns of individual tooth with wrinkled enamel texture as characteristic features for Eusauropoda. Barapasaurus teeth were mildly spatulate and had weakly crenulated texture on the enamel, indicating its affinity towards Eusauropoda. Wedel et al. (2000) and Wedel (2003) also considered Barapasaurus as ‘Eusauropoda’ on the basis of presacral vertebrae which bear lateral pneumatic fossae on the centra, some of which are deeper than others but do not meet each other closely to produce a medial septum.
Upchurch et al. (2007a) carried out another phylogenetic analysis of basal Sauropoda and maintained their earlier definition of Eusauropoda (sensuUpchurch et al. 2004); their analysis removed Barapasaurus from Eusauropoda. According to their phylogenetic analysis, Sauropoda includes the two Indian sauropods Barapasaurus and Kotasaurus as well as Vulcanodon, Chinshakiangosaurus, Antetonitrus, Blikanasaurus, ‘melanorosaurids’ and Jingshanosaurus and Eusauropoda. Since Eusauropoda (sensuUpchurch et al. 2007a) is mainly defined by cranial characters, whether Barapasaurus may be included within Eusauropoda or not cannot be assessed because of missing cranial characters. On the other hand, Barapasaurus along with Eusauropoda shares the synapomorphies such as opisthocoelus cervical centra, greater height of the mid-cervical neural arches in comparison with the centrum height, presence of spinodiapophyseal lamina on middle and caudal dorsal vertebrae, presence of postzygapophyseal lamina on all dorsal vertebrae, forked middle and distal chevrons, strongly convex dorsal iliac margin, middle and distal portions of the pubis lying in the same plane as the proximal end, absence of lesser trochanter and laterally directed cnemial crest (Node U of Upchurch et al. 2007a). The suggestion by Upchurch et al. (2007a) that Barapasaurus was a basal sauropod lying outside the Eusauropoda is consistent with other phylogenetic analyses of Benton et al. (2000), Yates and Kitching (2003), and Galton and Upchurch (2004). However, Allain and Aquesbi (2008) included Barapasaurus within the Eusauropoda.
The current phylogenetic analysis based on the revised data matrix of Upchurch et al. (2007a) shows that the Sauropoda including the Indian forms Barapasaurus and Kotasaurus along with Antetonitrus, Camarasaurus, Chinshakiangosaurus, Omeisaurus, Jingshanosaurus, Shunosaurus and Vulcanodon is fully resolved (Text-fig. 17) with the nodes achieving values of more than 60 per cent. These taxa share the characters such as the elliptical or subrectangular transverse section of the femoral shaft, angle between the femoral head and transverse axis of the distal end is close to 0 degree, tibia smaller than femur (characters 249–250 of Upchurch et al. 2007a, subsequently quoted here as ‘C249’ etc.) and presence of extensor depression on the distal end of the femur (C252). Barapasaurus is further removed from Eusauropoda and is found to be more basal in comparison with Vulcanodon (Text-fig 17) based on the different parameters of the mid-caudal centra (C145, length of the mid caudal vertebrae compared to the height of cranial articulation is less than 2 in Kotasaurus and Vulcanodon but >2 in Barapasaurus), presence of caudal hyposphenal ridge, length of the base of the caudal neural spines (C147–148), cranial disappearance of caudal ribs (C150) and subtriangular outline of the distal end of the ischium (C234). Barapasaurus shows advanced traits in comparison with Kotasaurus in several characters. These include opisthocoelus cervical centra (C106), presence of spinodiapophyseal lamina on middle and caudal dorsal vertebrae (C124) and composite lateral laminae on dorsal neural spines (C130), shape of the scapular blade (C157), extent of the deltopectoral crest (C165), different parameters of the ilium (C207–C209), position and profile of the fourth trochanter (C241–C242), ratio between the tibial and femoral length (C251), laterally directed cnemial crest (C253) and presence of muscle scar on the lateral surface of the fibula. In contrast, Kotasaurus shows advanced traits based on mid-caudal centra (C145), caudal neural spines (C148), ratio between the radial and humeral length (C176) and metatarsal length: width ratio (C269).
Some comments on Barapasaurus and Kotasaurus
Until recently, only two sauropods had been known from the lower part of the Kota Formation of the Pranhita-Godavari basin of India. Apart from B. tagorei another species of basal sauropod, K. yamanpalliensis was collected by Yadagiri (1988, 2001) from the same stratigraphic horizon but from a different locality. The fossil locality of Kotasaurus is situated 5 km west of the village Yamanpalli which is 40 km north of Pochampalli village, the type fossil locality of Barapasaurus (Text-Fig. 1). Multiple taxa of sauropods in the same level of a stratigraphic horizon are known to occur in different parts of the world. Kotasaurus and Barapasaurus occur in the same stratigraphic level of the Early Jurassic Lower Kota Formation of Pranhita-Godavari basin. Although only two sauropods are now known from this horizon, more sauropod material will likely to be described in future from the Kota Formation.
Recently, questions have been raised on the validity of the genus Kotasaurus yamanpalliensis. Yates (2007a) mentioned in his article that Kotasaurus is a chimera (a view held by Oliver Rauhut, p. 30 inYates 2007a) and he did not use its characters in his phylogenetic analysis. Allain and Aquesbi (2008) also refrained from using the character states of Kotasaurus as the assigned taxon might belong to more than one species (a view held by Jeff Wilson, p. 401 inAllain and Aquesbi 2008). Rauhut recently opined (Rauhut and Lopez-Arbarello 2008, p. 561) that at least one more taxon of sauropod is included in Kotasaurus. Three authors (SB, SR and DPS) of the present article examined the mounted skeleton and other isolated material of Kotasaurus. It must be mentioned here that there are a few bones in both the mounted skeleton and the isolated ones, which bear some similarities with Barapasaurus and may belong to Barapasaurus. However, there are also major osteological dissimilarities between Kotasaurus and Barapasaurus which separate these two taxa. Some of the differences were mentioned by Yadagiri (2001). The vertebral morphology of the presacral series of the two taxa is quite different. Barapasaurus is characterized by the presence of acamerate vertebrae morphology, 16 vertebral laminae and hollow neural spines in the presacral series, whereas vertebrae of Kotasauras presacral series have ‘lateral depressions, which may be either deep and small, or shallow and large’ and 7 vertebral laminae (4 in cervical and 3 in dorsal vertebrae) (Yadagiri 2001, p. 246). Gigantism and elongation of neck are the major biomechanical problems of sauropods which were mostly resolved by complex vertebral pneumaticity, origination of vertebral laminae, high and hollow dorsal neural spines which lighten as well as increase the strength of the vertebral column. Although Barapasaurus does not have extensively subdivided internal structures in the centra of advanced sauropods such as Mamenchisaurus, diplodocids and brachiosaurids, the presence of pneumatic fossae, 16 vertebral laminae and hollow neural spines in Barapasaurus indicates the beginning of increasing body size and length of neck in sauropod evolutionary history, and it may be said that Barapasaurus marks the beginning of gigantic sauropods. The neural canal of the caudal mid- dorsals in Barapasaurus is a specialized feature; it becomes narrow and deeply sunk on the centrum ventrally but opens dorsally into a large cavity through a narrow slit-like opening. The base of the neural spine forms the roof of the cavity and the floor of the cavity is depressed on either side of the slit-like opening. In contrast the neural canal in Kotasaurus is a normal tubular form running almost for the whole length of the arch. Moreover, the transverse processes of Barapasaurus are mostly directed laterally while in Kotasaurus these are directed upward.
The second major difference between the two taxa is the sacrum. Four co-ossified sacral vertebrae have been recognized in Barapasaurus since its description (Jain et al. 1975). However, the presence of four co-ossified sacral vertebrae is questionable in Kotasaurus. Yadagiri (1988, p. 103) mentioned the presence of three co-ossified sacral vertebrae but in the same paper he wrote that there are two conjoined sacral vertebrae, while the third one is broken (p. 110). Later, Yadagiri (2001, p. 242) mentioned the presence of ‘19 sacrals (one with three fused centra, two with fused centra)’ but none with four fused centra. In the illustration of sacral vertebrae, there are four conjoined sacral vertebrae (Yadagiri 2001, p. 245). He further wrote, ‘The full restoration of the sacrum was also based on features of the incomplete specimens. The co-ossified sacrum consists of four centra’ (Yadagiri 2001, p. 246). In this publication, the length of the sacrum is also measured on the basis of the fourth sacral vertebra. Three of the present authors (SB, SR and DPS) checked the sacrum in the mounted skeleton of Kotasaurus and noted that there are actually three co-ossified sacral vertebrae, while a fourth loose sacral vertebra is attached to it. Besides the mounted skeleton, there is no other sacrum with four co-ossified centra in the collection. It appears that Kotasaurus probably had three co-ossified sacral vertebrae and a loose dorsosacral for strengthening of the sacrum, whereas a dorsal vertebra was added to the sacrum of Barapasaurus.
Among other differences, the Kotasaurus scapula is tall but significantly narrower than Barapasaurus whose scapular blade shows gentle dorsoventral convexity, larger proximal expansion and a relatively weak acromium. The ilia of the two taxa differ considerably. The iliac blade of Barapasaurus is comparatively high, and the dorsal margin is rounded and gently convex; the preacetabular process is quite prominent, curved and subtriangular (Text-fig. 11A–B), whereas the iliac blade of Kotasaurus is quite low, and as a result, the tall neural spine is visible in the mounted skeleton; the dorsal margin of the ilium is almost straight, and the preacetabular process extends comparatively farther and rises above the level of pubic peduncle. The caudal part of the ilium of Kotasaurus extends well beyond the ischiadic peduncle in the mounted skeleton, but it is broken caudally; an isolated specimen of the ilium in the collection, though small, reveals a blunt shape of the caudalmost part. The caudal part of the ilium of Kotasaurus is larger than that of Barapasaurus. The acetabular part of the ilium of Kotasaurus is wider than Barapasaurus. The obturator foramen of Kotasaurus is significantly larger than Barapasaurus. Another important difference is the lesser trochanter of the femur which is present in Kotasaurus but completely absent in Barapasaurus. The fibula of Barapasaurus has a broad triangular scar for the tibia which is not noted in Kotasaurus. The presence of prominent astragalar peg situated anteroventrally is characteristic of Kotasaurus but is not found in Barapasaurus. From the above comparative discussion, it is clear that Barapasaurus has its own unique characteristic features that are different and is more highly derived than Kotasaurus.
To conclude, the position of Barapasaurus within the phylogenetic tree of the sauropods is redefined. More scores on the different character states used by previous workers to build up the sauropod phylogeny can be now used for Barapasaurus through a detailed study of the described as well many undescribed material belonging to that taxon. The new phylogenetic tree suggests Barapasaurus is more advanced than Kotasaurus but is more basal in comparison with Vulcanodon. However, Barapasaurus is removed from Eusauropoda. Kotasaurus has been accepted here as a valid taxon, though there may be irregularities in the mounted skeleton. All the material assigned to Kotasaurus may not be of same taxon, but there is undoubtedly at least a second distinct sauropod present in the Kota Formation. Finally, it may be mentioned that the overall morphological features of B. tagorei such as quadrupedal posture, spatulate teeth with wrinkled enamel texture, shortening of the trunk, complex vertebral laminae and pneumatic fossae in the presacral series, articulation of hyposphene-hypantrum in the dorsal vertebrae, strengthening of sacrum by the addition of vertebra making four co-ossified sacral vertebrae, narrowness of the pubic apron, and slender, columnar limbs set the trend for future sauropod evolution.
Acknowledgements. The discovery of B. tagorei was made in 1958 by a team consisting of the late Pamela Lamplugh Robinson of University College, London, the late S. L. Jain and T. K. RoyChowdhury of the Indian Statistical Institute. T. S. Kutty joined the team during the excavation in 1961. Subsequent excavation (known as the ‘Colbert excavation’) in 1964 was carried out by the late Edwin. H. Colbert, late S. L. Jain and late P. L. Robinson. These scientists along with S. Chatterjee (presently at Texas Tech University) did the basic descriptions of B. tagorei. The expeditions were part of an integrated programme of the study of the Gondwana rocks of the Pranhita-Godavari basin which was sponsored by the Indian Statistical Institute and partly financed by the Royal Society, London. SB and DPS are thankful to T. K. RoyChowdhury for discussions and clarification of certain aspects of the anatomy of B. tagorei. The authors are grateful to T. K. RoyChowdhury for allowing them to use his unpublished geological map of the Pranhita-Godavari basin. Constructive criticism and suggestions by Jeff Wilson, Oliver Rauhut and another anonymous reviewer have improved the paper. We thank Victor Leshyk for the drawing in Text-figure 3. Thanks are due to Dr. K. Ayyasami of the Geological Survey of India, Hyderabad and Dr. B. G. Sidharth of the Birla Science Museum, Hyderabad for access to Kotasaurus material. The late Edwin H. Colbert encouraged this project and provided sage advice in its early stages. The Colbert Endowment of The Museum of Northern Arizona and the Indian Statistical Institute provided travel and research fund to DDG.