- The Paleontological Society
The Upper Dharmaram Formation (Lower Jurassic, Sinemurian) of India has yielded three sauropodomorph dinosaurs, two new taxa and an indeterminate one. Lamplughsaura dharmaramensis n. gen. and sp., represented by several partial skeletons, is a heavily built quadrupedal form (body length ∼10 m). Autapomorphies include teeth with strongly emarginated distal edge; caudal cervical neural spines bearing a vertically oriented ligamentous furrow on cranial and caudal surfaces and a transversely expanded spine table; caudal neural spines bearing a craniodorsally directed spur (proximal caudal vertebrae) or a large process (midcaudal vertebrae); caudal neural spines shorter than transverse processes so former lost first in passing along tail; and a plesiomorphy that is the nontrenchant form of manual ungual I. The Indian dinosaurs were coded for two recent datamatrices for basal sauropodomorphs. The results of this preliminary analysis indicate that Lamplughsaura is either a basal Sauropoda or, less likely, based on Templeton's test, a stem sauropodomorph. The second large form, represented by the proximal half of a femur, is a sauropodomorph that is more derived than Saturnalia (Brazil) and Thecodontosaurus (Great Britain) from the Upper Triassic. This is also true for the smaller (body length ∼4 m as adult) Pradhania gracilis n. gen. and sp. which lies outside of the Sauropoda + Plateosauria clade, so it is definitely a stem sauropodomorph. Pradhania is known from fragmentary material; an autapomorphy is the very prominent medial longitudinal ridge on the maxilla.
Basal sauropodomorphs are small to large (maximum lengths approximately 2.5–15 m), bipedal, facultatively bipedal or quadrupedal saurischian dinosaurs. They were herbivores with small skulls, elongate necks and tails, and strong limbs that pioneered high browsing on the foliage of trees that other contemporary herbivores could not reach (Bakker, 1978; Parrish, 1998). They are usually the most common terrestrial vertebrates in the Upper Triassic and Lower Jurassic beds in which they occur. They have been found on all the major continents, including Antarctica (Hammer and Hickerson, 1994; Rich et al., 1997), but the Australian record of basal sauropodomorphs was based on Agrosaurus Seeley, 1891 that proves to be mislabeled material of Thecodontosaurus Riley and Stutchbury, 1836 from the Upper Triassic of England (Vickers-Rich et al., 1999; Galton, 2000). Basal sauropodomorphs first appeared in the early Carnian (∼230 Ma), became distributed globally in the Norian, endured the terminal Triassic extinction, rebounded in the Sinemurian and Pleisbachian, but finally disappeared at the end of the Toarcian (∼178 Ma). For nearly 50 million years, they were the dominant group of herbivorous dinosaurs in the terrestrial ecosystem of the Pangean world. For details on the basal sauropodomorphs traditionally referred to the Prosauropoda, see Galton and Upchurch (2004), and for basal sauropods see Upchurch et al. (2004) and discussion by Wilson (2005).
Records of basal sauropodomorphs from the Upper Triassic– Lower Jurassic beds of India are sparse. Huene (1940) reported two fragmentary vertebrae of a sauropodomorph from the Upper Triassic Maleri Formation of the Pranhita-Godavari Valley, Andhra Pradesh, but this identification is incorrect (Colbert, 1958; Roy Chowdhury, 1965). Kutty (1969) briefly reported the discovery of a basal sauropodomorph fauna from his newly defined Dharmaram Formation, which lies immediately above the Maleri Formation. He identified two taxa, a large plateosaurid and a small thecodontosaurid, but the material was never described. Later, Kutty and Sengupta (1989) discussed the Triassic faunal succession of the Pranhita-Godavari Valley and considered the sauropodomorph-bearing horizon, the upper Dharmaram Formation, to be coeval with the uppermost Rhaetian stage of the Triassic sequence of Germany. They also listed another small basal sauropodomorph from the upper Maleri Formation (Massospondylus sp. Kutty et al., 1987), which now appears to be a basal saurischian dinosaur similar to Guaibasaurus Bonaparte et al., 1999 from the Upper Triassic of Brazil (see Table 1). The material from the lower Dharmaram Formation listed by Kutty and Sengupta (1989) is fragmentary and nondiagnostic. In this paper, we describe the sauropodomorph material from the upper Dharmaram Formation (Lower Jurassic, Sinemurian) of India as two new taxa and a third indeterminate one, based on the plateosaurid and thecodontosaurid of Kutty (1969).
ISI, Geological Studies Unit of the Indian Statistical Institute, Calcutta, India; IVPP, Institute of Vertebrate Paleontology and Paleoanthropology, Beijing, Peoples Republic of China.
geologic and plate tectonic setting
The continental Gondwana formations, which are well exposed in the Pranhita-Godavari Valley of southern India (Fig. 1) and range in age from Permian to Cretaceous, have yielded several vertebrate faunal successions (Table 1). Among these successions, the Upper Triassic Maleri Formation is probably the best known with a variety of vertebrate remains, including metoposaurs, rhynchosaurs, and phytosaurs, which are comparable to the fauna from the Löwenstein Formation (as Lower and Middle Stubensandstein) of Germany (Roy Chowdhury, 1965; Chatterjee, 1974, 1978). Above the Maleri Formation, Kutty (1969) mapped a new formation, the Dharmaram Formation, an alternating series of sandstones and mudstones, originally thought to be coeval with the Trossingen Formation (as Upper Stubensandstein-Knollenmergel) of the Germanic Upper Triassic. Later, Kutty and Sengupta (1989) recognized four temporally successive vertebrate faunal zones in the Upper Triassic sequences of India: the Lower Maleri, Upper Maleri, Lower Dharmaram, and Upper Dharmaram. Each zone has a distinctive faunal assemblage (Table 1). The sauropodomorph material described here is from the Upper Dharmaram Zone. Recent analysis of the Upper Dharmaram fauna suggests the presence of a sphenosuchian similar to Dibothrosuchus Wu and Chatterjee, 1993 (Lower Jurassic Lufeng Group, China), and a large theropod, mainly represented by isolated teeth and limb fragments, that is similar to Dilophosaurus Welles, 1984 (Lower Jurassic Kayenta Formation, Arizona). Surprisingly, the typical Upper Triassic fauna of the Lower Dharmaram, such as phytosaurs and aetosaurs, are entirely absent in the Upper Dharmaram Formation, indicating that its age is younger than Norian. However, both basal sauropodomorphs and sphenosuchians are known to extend into the Early Jurassic period in several parts of the world. The boundary between the Lower and Upper Dharmaram Formations may coincide with the Triassic–Jurassic extinction event that occurred 208 million years ago. The beginning of the Jurassic age in the Upper Dharmaram horizon is marked by the proliferation of basal sauropodomorphs and the disappearance of the typical Triassic (Norian) fauna. The associated fauna of the Upper Dharmaram Formation is related to faunas from the Upper Elliot and Clarens Formations in southern Africa, the Kayenta Formation of Arizona, and the Lower Lufeng Group of China. Olsen and Galton (1977) correlated these groups on the basis of vertebrate remains and favored an early Jurassic assignment. The correctness of this assignment has been confirmed by several subsequent studies (e.g., Olson and Galton, 1984; Olsen and Sues, 1986; Shubin and Sues, 1991; Sues et al., 1994). Accordingly, we agree with Bandopadhyay and Roy Chowdhury (1996) who regard the Upper Dharmaram Formation as early Jurassic (Sinemurian) in age.
Pangea was intact during the late Triassic and early Jurassic times when sauropodomorphs were evolving and radiating into what would become different landmasses (Fig. 1.1). The early appearance of the most basal sauropodomorph Saturnalia Langer et al., 1999 in Brazil (Langer et al., 1999; Langer, 2004), plus other less well defined basal sauropodomorphs from Zimbabwe (Molteno Formation, proximal femur, Raath, 1996) and Madagascar [unnamed, Flynn et al. (1998); form not similar to Azendohsaurus Gauffre, 1993 from Morocco (Gauffre, 1993) that is probably nondinosaurian (Jalil and Knoll, 2002)], during the Carnian, indicates that Gondwana may have been the center for their origin (Fig. 1.1). The global distribution of basal sauropodomorphs indicates that geography and climate were probably not significant deterrents to their migration. The India-Madagascar-Seychelles block was a landmass in Gondwana during the evolution of basal sauropodomorphs (Chatterjee and Scotese, 1999). The first stage of the rifting of Pangea took place in the middle Jurassic (180 Ma) between North America and Africa after an episode of volcanic activity, leading to the separation of Gondwana and Laurasia. At that time India was still part of Gondwana.
material and methods
All the sauropodomorph material was collected from the uppermost mudstone unit of the Dharmaram Formation of early Jurassic age. The fossil localities lie between Krishnapur and Dharmaram villages in the Adilabad district, Andhra Pradesh (Fig. 1.2). The overbank deposits, consisting mostly of an extensive mudstone unit, contain abundant terrestrial vertebrate fossils. These bones are well preserved, usually with little abrasion but often coated with calcareous or hematitic encrustations, and were prepared mechanically using a pin-vice and brush. Three sauropodomorphs are recognized (although one of these forms cannot be diagnosed at the generic or specific level), and the emphasis of this paper is on the largest one, which is represented by excellent material that provides a wealth of anatomical information.
Dinosauria Owen, 1842
Sauropodomorpha Huene, 1932
?Sauropoda Marsh, 1878
Lamplughsaura new genus
Lamplughsaura dharmaramensis new genus and species
Lamplughsaura (pronounced Lam-plo-saura, feminine) is named in honor of the late Pamela Lamplugh Robinson of University College, University of London, England, who founded and guided the ISI; the specific name refers to the Dharmaram Formation.
Holotype, ISI R257, a partial associated postcranial skeleton of a nearly adult individual consisting of an almost complete neck, 8 dorsal neural arches and 4 centra, a sacral neural arch, 11 caudal vertebrae, 4 dorsal ribs, 5 chevrons, scapulae, left sternal plate, humeri, right ulna, 2 metacarpals, ischia, right femur, fibulae, left tibia, astragalus, calcaneum, metatarsal III (Figs. 2, 5, 9–15). Upper Dharmaram Formation, Lower Jurassic (Sinemurian, ∼200 Ma). North of Krishnapur village (79°32′, 19°15′), Adilabad District, Andhra Pradesh, southern India (Fig. 1.2).
Teeth with coarse denticles few or absent on mesial edge; distal edge has distinct concavity in side view; caudal cervical neural spines bear a vertically oriented ligamentous furrow on cranial and caudal surfaces and a transversely expanded spine table; caudal neural spines shorter than transverse processes so former disappear first passing along tail, caudal neural spines bear a craniodorsally directed spur (proximal caudal vertebrae) or a distinct process (midcaudal vertebrae); descending caudal flange of distal end of tibia covers 66% of transverse width of the astragalus; ungual phalanx of manual digit I plesiomorphic as it tapers gently distally and is not markedly recurved and trenchant.
With the exception of ISI R260 (see below), all the large bones from the same horizon and locality as the holotype (Fig. 1.2) are referred to Lamplughsaura dharmaramensis. These include the disarticulated partial skeletons of four individuals (Fig. 2), two of which have cranial bones. The referred specimens are shown in Table 2.
Isolated cranial elements are preserved in two individuals: ISI R258 contains a right premaxilla, a right nasal, a right frontal, a left prefrontal, an incomplete right postorbital, a right jugal, vomers, a right palatine, a right pterygoid, a right ectopterygoid, a left quadrate-quadratojugal, a right dentary, a fragmentary left dentary, a right prearticular, a right surangular, a supraoccipital and a right opisthotic (Figs. 3–7). ISI R259 includes a basioccipital-basisphenoid complex, a right prootic, and a fragmentary left maxilla (Figs. 4–7).
Although the cranial bones are disarticulated, indicating that these individuals were not fully adult, there is little sign of distortion. Many of the roofing and palatal bones are from the right side of the skull. To determine the spatial and topographic relationships, as well as the dimensions of the skull, several cranial bones of ISI R258 were reassembled in the regions of the skull roof, palate, braincase (with ISI R259, Figs. 4, 5, 7) and lower jaw. Several views of the skull and lower jaw are reconstructed from these elements with the aid of information from the skulls of other basal sauropodomorphs. The reconstructed skull (Fig. 6) is about 270 mm long, 156 mm wide across the postorbitals, and 140 mm high. It is proportionately larger and more massive in construction than that of Plateosaurus Meyer, 1837 (Upper Triassic, Germany; Galton, 1984a, 1985; Galton and Upchurch, 2004) but it is a good match for that of Riojasaurus Bonaparte, 1969 (skull length ∼250 mm; Upper Triassic, Argentina; Bonaparte and Pumares, 1995).
The right side of the premaxilla shows the vertical median symphysis at the rostral end (Figs. 3, 6). Behind the broken ascending rostral process, the aperture for the external naris is seen. The premaxilla has four alveoli with two teeth in position. This number of premaxillary alveoli probably represents the plesiomorphic state for sauropodomorphs (Galton and Upchurch, 2004). It is not clear whether the premaxilla possessed or lacked the lateral plate of bone that supports the labial tooth crown surfaces in sauropods (Upchurch, 1995; Upchurch et al., 2004). Dorsal to the alveolar margin, there is a lateral row of nutrient foramina (Fig. 6). The distal end of the ascending process of the premaxilla tapers to a slender point (the plesiomorphic state), and therefore lacks the derived transversely expanded condition present in Efraasia Galton, 1973, Coloradisaurus Lambert, 1983, Massospondylus Owen, 1854 and Plateosaurus Meyer, 1837 (Yates and Kitching, 2003; Galton and Upchurch, 2004).
The middle part of the left maxilla has a horizontal base and an ascending process (Figs. 3, 6). The base of the maxilla, which contains the alveolar margin, is rostrocaudally slightly concave laterally but convex medially. The first preserved alveolus is broken and shows the longitudinal cross-section of the socket with a small replacement tooth that lies considerably higher than the rest of the alveolar margin. The second alveolus shows an empty socket, followed caudally by an erupting tooth that is compressed labiolingually. From the base of the maxilla, the ascending process rises upwards and backwards, resembles the condition seen in sauropods rather than the relatively vertical process found in most basal sauropodomorphs (Gauffre, 1993; Galton and Upchurch, 2004). The rostral part of the medial wall of the antorbital fossa is represented by a sheet of bone extending backwards from the ascending process of the maxilla, the rostal margin of which was overlain by the broad nasal.
The nasal is the longest bone in the skull roof, sloping and tapering forward to meet the premaxilla (Figs. 3, 4, 6). Behind the premaxillary contact, the nasal is inflated and bifurcated with a ventrolateral branch that forms the caudal rim of the external naris. Caudally, this branch has a long sutural contact with the ascending process of the maxilla. Farther back, the nasal becomes narrower and receives the prefrontal laterally. Caudally, it has a complex, overlapping suture with the frontal. There is no median nasal depression, indicating that Lamplughsaura possesses the plesiomorphic state present in sauropods, rather than the derived state found in many basal sauropodomorphs (Yates and Kitching, 2003; Galton and Upchurch, 2004).
The frontal is a flat, squarish, horizontal bone on the skull roof where the rostral margin is strongly overlapped by the nasal (Figs. 3, 4, 6). Behind this contact, the frontal has a curved, rugose margin that is wrapped around by the prefrontal. The frontal reaches its greatest width near its caudal edge, where it receives the postorbital in a lateral notch. The transverse caudal edge receives the parietal and, rostral to this, the dorsal surface is somewhat excavated for the attachment of the M. pseudotemporalis. Thus, there is a supratemporal fossa surrounding the supratemporal fenestra, as occurs in many dinosaurs except eusauropods (Upchurch, 1998; Galton and Upchurch, 2004). Medially, each frontal contacts its fellow by a simple butt joint.
The parietal was not found.
The prefrontal is a complex bone that is L-shaped in side view with a flat, dorsal surface and a lateral, vertical plate (Figs. 3, 5, 6). The dorsal surface wraps around the sutured lateral margin of the frontal and nasal bones. The caudal margin of the prefrontal forms the rostral rim of the orbit. The lateral, vertical plate (Fig. 5) presumably passed medial to the lacrimal (not preserved) to form a bar with it between the orbit and the antorbital fenestra. Rostrally, the prefrontal gradually tapers into a narrow process that overlaps the nasal. The caudal process of the prefrontal is relatively short, so that the frontal forms a large part of the dorsal margin of the orbit.
The postorbital is incomplete at the dorsal, posterior, and ventral ends (Figs. 3, 4, 6). It is a T-shaped bone that separates the orbit from the upper and lower temporal openings. The dorsal horizontal bar connects to the frontal rostrally and to the squamosal caudally. The vertical bar, for the ascending process of the jugal, is not rostrocaudally compressed, unlike the derived state present in most sauropods (Wilson and Sereno, 1998).
The jugal is a triradiate bone with caudal, dorsal, and cranial processes (Figs. 3, 4, 6). At the base of the caudal process, there is a concave facet medially on the ventral part for the reception of the ectopterygoid. The dorsal process receives the postorbital at its caudal margin to form an obliquely oriented bar between the orbit and the lower temporal fenestra. Most of the rostral process, for articulation with the maxilla, is missing, so it is not possible to determine whether it reached the margin of the antorbital fenestra.
The quadratojugal is roughly Y-shaped with the vertical parts for the quadrate and a rostral branch (missing) for the jugal (Figs. 3, 5, 6). The vertical parts, which form a thin sheet of bone ending in a heel-like process, is strongly sutured to the lateral edge of the quadrate except for the part interrupted by the quadrate foramen (Fig. 5, not shown in Fig. 6). The two vomers are fused along the midline to form a bar between the choanae (Figs. 3, 6). The dorsal symphysis forms a ridge, but an excavation interrupts the ventral union at midlength, probably for the attachment of soft palatal tissue. Each vomer is a narrow vertical plate that becomes deeper rostrally to meet the premaxilla. The caudal extension for articulation with the palatine is not preserved.
The lateral maxillary base of the right palatine is preserved (Figs. 3, 6). The palatine bridges the gap between the maxilla and the pterygoid but it is notched rostrally by the choana and caudally by the postpalatine fenestra. The lateral base sutures against the maxilla and was perhaps overlapped slightly ventrally by the maxilla. There does not appear to have been a long maxillary process of the palatine, unlike the derived condition observed in eusauropods (Upchurch, 1998).
The pterygoid is a large and complex bone, differentiated into a medial palatal ramus, a lateral transverse flange for the ectopterygoid, and a caudal quadrate ramus (Figs. 3, 4, 6). The rostral extension of the palatal ramus for articulation with the palatine is not preserved. The rest of the palatal ramus is straight at the median symphysis, but caudally it gradually diverges from the midline around the braincase. Here it embraces the basipterygoid process of the basisphenoid in a socket on its dorsal surface to form a flexible joint. There is no ‘hook’ or ‘finger’-like process from this articulation that clasps the distal end of the basipterygoid process, unlike the pterygoids of several sauropods (Upchurch, 1998; Wilson and Sereno, 1998; Yates and Kitching, 2003). From this joint, the deep quadrate ramus extends backwards and outwards to overlap the quadrate. Between the quadrate ramus and the lateral flange, the central part of the pterygoid is thick and constricted. The lateral flange receives the ectopterygoid rostrally and dorsally along a long suture.
The ectopterygoid is a small, transversely oriented bone with an inflated, hemispherical lateral process that sutures with the jugal (Figs. 3, 4, 6) as in most dinosaurs, rather than the maxilla as in neosauropods (Upchurch, 1998). The bone gradually tapers medially and overlaps the pterygoid dorsally to form the caudal margin of the postpalatine fenestra.
The quadrate is tall and narrow with winglike flanges spreading medially and laterally (Figs. 3–6). The bone is arched on its vertical axis in such a way that the caudal surface is highly concave in lateral view to form a deep otic notch for the attachment of the tympanum. The dorsal head terminates in a spherical peg that would have fitted into the articular socket of the squamosal (not preserved). Below the head, the quadrate is inflated medially and laterally. The lateral flange received the squamosal dorsally and the quadratojugal ventrally by straight, subvertical sutures; the quadrate foramen in the latter suture is deeply incised into the quadrate (Fig. 5, not shown in Fig. 6). The medial flange extends forwards and inwards and it is overlapped laterally by the quadrate ramus of the pterygoid. Ventrally, the quadrate terminates as two asymmetric mandibular condyles: the lateral one is spherical whereas the medial one is expanded fore and aft and projects more ventrally. These two condyles are separated by a deep longitudinal concavity. The jaw articulation is offset below the level of the maxillary tooth row, as occurs in most sauropodomorphs (Galton and Upchurch, 2004).
The braincase was reconstructed using the roof from ISI R258 and the floor and prootic from ISI R259 (Figs. 4–7). The basioccipital forms the floor of the endocranial cavity and is capped by the base of the exoccipital at its lateral edge around the foramen magnum (Figs. 4–7). From the condyle, the basioccipital narrows transversely in the rostral direction to a neck and then flares ventrally into a pair of basal tubera, the spherical ends of which are roughened for the attachment of the neck muscles. As in Riojasaurus, a shallow notch separates the tubera, which are less divergent than in Plateosaurus and Lufengosaurus (Young, 1951). At the tubera, the basioccipital is intimately fused to the basisphenoid at a vertical sutural surface. Dorsally, the basioccipital contains a transverse channel, the perilymphatic groove at its widest part, which forms the floor of the fenestra ovalis. The basioccipital condyle and basal tubera lie above the level of the basipterygoid processes: this is a derived condition that is also present in Plateosaurus, Coloradisaurus, and Lufengosaurus (Yates and Kitching, 2003; Galton and Upchurch, 2004).
The basisphenoid (Figs. 4–7) extends forwards from the basal tubera, narrows transversely, and is interrupted by a small foramen that is associated with the median pharyngeal system (Witmer, 1997). Unlike the condition in theropods, a pneumatic sinus does not permeate this pharyngeal cavity. Laterally, the basisphenoid broadens into two divergent basipterygoid processes for articulation with the pterygoids. Each basipterygoid process has a rough elliptical surface that fits into the socket of the pterygoid. The basipterygoid processes are slightly below the level of the basal tubera as in Riojasaurus. In many basal sauropodomorphs they project considerably ventrally (Galton and Upchurch, 2004). The lateral wall of the basisphenoid, rostrodorsal to the basal tubera, is deeply excavated by the rostral tympanic recess (Chatterjee, 1991). This recess extends considerably medially and may have a contralateral communication. The internal carotid artery entered through this recess via the vidian canal and extended medially and rostrally to reach the pituitary fossa. Rostrally, just above the base of the parasphenoid rostrum, the basisphenoid is excavated for the pituitary fossa. Behind this fossa lies the vertical wall of the dorsum sellae, which is pierced laterally by the foramen for the abducens (VI) nerve and, more dorsomedially, by the foramen for the internal carotid artery. Lateral to the base of the rostrum lies a pair of retractor fossae for the attachment of an eye muscle along with the M. protractor pterygoideus (Galton, 1974). The dorsal surface of the basisphenoid contains a deep fossa in the floor of the endocranial cavity that received the ventral prominence of the brain that was created by the pontine flexure.
The supraoccipital forms the upper border of the foramen magnum in the occiput (Figs. 4–7). In caudal view, this element has a derived semilunate and widened outline, as occurs in basal sauropodomorphs (Thecodontosaurus and Efraasia) and many sauropods (Yates and Kitching, 2003). The bone becomes narrow rostrally and dorsally to extend beneath the caudal wings of the parietal. It is deeply notched laterally to receive the parietal wing. Dorsally, there is a fontanelle between the supraoccipital and parietal, as is also seen in many basal sauropodomorphs (Yates and Kitching, 2003; Galton and Upchurch, 2004). In lateral aspect, the supraoccipital probably sloped at about 45° and its apex lies approximately above the rostral margin of the basal tubera: this resembles a derived state observed in several plateosaurian prosauropods, such as Plateosaurus and Coloradisaurus (Galton, 1990; Galton and Upchurch, 2004). The caudal surface on each side is perforated submarginally for the vena capitis medius. The corresponding cranial surface shows part of the floccular recess, surrounded by a C-shaped bony tube that houses the rostral vertical semicircular canal. The thickened lateral margin of the supraoccipital has a long and sinuous contact with the exoccipital-opisthotic complex.
The exoccipital-opisthotic bones are intimately fused with each other. The exoccipital is incompletely preserved at its base at the contact with the basioccipital (Figs. 4–7). Ventrally, the exoccipital is broad dorsolateral to the foramen magnum, and it is pierced by the two foramina for the rami of the hypoglossal nerve (XII). At the sidewall of the braincase, near the ventral edge of the paroccipital process, the exoccipital extends as a metotic strut and forms the caudal portion of the otic capsule. Here in lateral view, two large foramina are visible, separated by the crista interfenestralis of the opisthotic. The caudal one represents the metotic foramen for the exit of cranial nerves IX–XI and the caudal branch of the jugular vein. The rostral one is the fenestra ovalis that received the footplate of the stapes. The stout, horizontal paroccipital process of the opisthotic is directed caudolaterally, was overlapped by the squamosal at its expanded distal end. The prootic is represented by the ventral process that sits on the basisphenoid (Figs. 5,7). Dorsally, it is notched by the trigeminal nerve (V). The laterosphenoid and the parasphenoid rostrum are entirely missing.
The mandible is represented only by the dentary, prearticular, and surangular bones (Figs. 4, 6). The mandible is long and slender in the part where the jaw articulation set, below the dentary tooth row (Fig. 4; jaw shown too deep in Fig. 6). As a result, the upper margin of the surangular slants downwards caudally to the articular. The lateral surface of the jaw, below the caudal part of the dentary and the postdentary bones, is interrupted by the small and elliptical external mandibular fenestra (Fig. 4, shown too far back in Fig. 6).
The dentary in dorsal view is laterally bowed with a longitudinal ridge on its outer surface. This lateral ridge is present in most basal sauropodomorphs (e.g., Efraasia), but is generally absent in sauropods (Galton, 1990; Yates and Kitching, 2003; Galton and Upchurch, 2004). The rostral extremity of the dentary is curved medially to meet its counterpart at a flat, vertical, elliptical surface that formed a loose and weak symphysis (Figs. 4, 6). At the symphysis, the dentary is fairly deep, but it becomes shallow caudally and then deepens again into a thin vertical plate. Thus, Lamplughsaura has a vertically expanded symphysis, as also occurs in most sauropods (Upchurch, 1998; Upchurch and Barrett, 2000; Yates and Kitching, 2003). Medially the dentary forms much of the roof of the Meckelian canal. The dentary supports the marginal tooth row. There are 21 empty alveoli that are weakly partitioned by vertical ridges. The possession of more than 18 dentary teeth is common in many basal sauropodomorphs, whereas in sauropods (except the basal form Shunosaurus Dong et al., 1983) this number is reduced to fewer than 18 (Wilson and Sereno, 1998). The outer alveolar margin is considerably higher relative to the inner margin (Fig. 4): this feature is tentatively interpreted as the lateral plate that is commonly found in the premaxillae, maxillae, and dentaries of sauropods (Upchurch, 1995, 1998; Upchurch et al., 2004). In addition, the inner margin is lower due to the loss of the interdental plates (either separate or fused together as shown in the reconstruction but note that longitudinal groove and foramina for replacement teeth is ommited, Fig. 6). A similar loss of teeth and interdental plates is known in Shunosaurus (Chatterjee and Zheng, 2002).
The surangular is a thin vertical plate that lies above the angular and forms the caudal margin of the external mandibular fenestra (Figs. 4, 6). Behind this fenestra, the bone runs downwards and backwards, tapering in its course to wrap around the lateral and ventral margin of the articular. Here it joins with the articular to form the articular fossa for the quadrate. Behind this, the surangular forms the lateral surface of the retroarticular process that extends considerably behind the jaw articulation. Medially, the surangular forms the lateral wall of the adductor fossa.
The prearticular is a long crescentic bone on the inner surface of the jaw, and it forms the medial wall of the adductor fossa (Figs. 4, 6). For most of its length it lies above the angular and extends forwards to contact the splenial. It is directed caudally as a strap of bone to sheath the medial and ventral surfaces of the articular.
Most of the teeth are missing from the jaws, leaving the alveoli empty as is common in thecodont implantations, and the interdental plates are also missing, so it cannot be determined if the plates were fused to each other to form a single bone as occurs in many sauropods. In basal sauropodomorphs the upper jaw has a slightly higher tooth count than the lower one, so a few caudal maxillary teeth were unopposed by dentary teeth (Galton and Upchurch, 2004). The tooth count for the premaxilla was 4, for the dentary about 21 and probably the same count for the maxilla (Fig. 6).
A few replacing teeth are preserved in the alveoli, two each in the premaxilla, maxilla, and dentaries. In addition, several isolated teeth were found with the cranial elements. The shape of the teeth is different from those of other basal sauropodomorphs (Figs. 4, 8) but is reminiscent of basal sauropods and, as in sauropods (Wilson and Sereno, 1998), the enamel has a textured (‘wrinkled’) surface. The crown is spoon-shaped and curved lingually as seen in the sauropods Barapasaurus Jain et al., 1975 (Fig. 8), Shunosaurus (Upchurch, 1998; Chatterjee and Zheng, 2002) and Camarasaurus (Upchurch et al., 2004). Each tooth is compressed labiolingually, with the development of mesial and distal keels. In lingual view, the mesial and distal margins are not symmetrical: the mesial margin shows a broad convex curvature; the distal margin, on the other hand, exhibits a concavity or kink at the mid-height of the crown. Moreover, the mesial edge shows coarse denticles that, when present, range from 8 to12 per tooth. However, these mesial denticles are absent on premaxillary teeth and on several isolated teeth. The distal edge lacks serrations. In contrast, other basal sauropodomorphs show a larger number of coarse serrations on both mesial and distal edges and lack the spoon-shaped morphology (Fig. 8). These serrations in Lamplughsaura are set at an angle of 45° from the keel (as in other basal sauropodomorphs); they are located more towards the apex and gradually fade towards the base. Here the keel tends to curve lingually to enclose a vertical groove. This means that these teeth possess a lingual concavity, as in most sauropods (except derived groups such as Diplodocoidea and Titanosauria; Upchurch et al., 2004). The crown is separated from the cylindrical root by a distinct constriction.
The general vertebral count in articulated specimens of the basal sauropodomorphs Efraasia (Galton, 1973) and Plateosaurus (Huene, 1926) is 10 cervical vertebrae (including atlas), 15 dorsals, 3 sacrals, and about 50 caudals. The column is well represented in ISI R257 by an associated series of 8 cervical (C2–C9, found articulated), 8 dorsal neural arches and 4 centra, 1 sacral neural arch, and 11 caudal vertebrae (Figs. 5, 9– 10). The bones are well preserved and show little deformation (for measurements see App. 1.1).
The elongate cervical vertebrae, ranging from 70–90 mm (Figs. 5, 9–10) form a long neck. The axis (Figs. 5, 9) has an elongate, blade-like neural spine, the dorsal border of which is convex in lateral view. The cranial face of the centrum shows a larger superior depression, the site of attachment of the odontoid process of the atlas, and a ventral facet for articulation with the intercentrum (Fig. 5, not shown in Fig. 9). The prezygapophysis is small and oriented almost vertically; it does not project beyond the face of the centrum. The postzygapophysis is well developed, braced dorsally by the epipophysis, and extends further caudally: they extend beyond that caudal edge of the centrum (a plesiomorphic state also seen in sauropods, Sereno, 1999) (Fig. 8).
In the rest of the cervical series (Figs. 9–10) there is an increase in the length and size of the vertebrae proceeding backwards from the axis (App. 1.1). The centra of the succeeding cervical vertebrae are narrow, spool-shaped, and amphicoelous. Thus, Lamplughsaura lacks the derived opisthocoelous condition present in most sauropods (Upchurch, 1998; Upchurch et al., 2004). The cranial and caudal faces of the centra are obliquely oriented to the long axis of the vertebra such that in life the neck had a natural dorsal arch. There is a prominent ventral keel on the cranial half of each centrum in C5–C7 but it becomes weaker in more caudal cervical vertebrae, where the keel extends along the whole length of the centrum. The zygapophyseal facets slope at a small angle (∼30°) to the horizontal, indicating a wide range of lateral motion. Throughout the series, there is a prominent epipophysis above the postzygapophysis that is set more lateral relative to the prezygapophysis. There is a gradual shift of the rib attachment passing caudally along the cervical series. The parapophyses are directed outward and somewhat backward as they rise from the cranioventral rim of the centrum. In C4 the diapophysis is on the centrum whereas in C5 the diapophysis has shifted to the neural arch, being borne by a transverse process that is narrow and drooping. Farther back, the transverse process becomes progressively enlarged. Each transverse process lies well below the level of the zygapophysis and it is supported by strong lamellae cranially and caudally. Thus, Lamplughsaura lacks the derived condition found in many sauropods, in which the cervical transverse processes are supported from below by well-developed anterior and posterior centrodiapophyseal laminae (Upchurch, 1998; Wilson, 1999). There is a large, lateral excavation of the centrum ventral to the transverse process. The neural spine is low and narrow craniocaudally in the cervical series. However, the spine table becomes considerably wider transversely in the caudal cervical vertebrae (C7–C9), with a vertical cleft on the cranial surface for the attachment of nuchal ligaments (Figs. 9, 10). In lateral view, the width of the spine increases from the base upwards. On both the cranial and the caudal faces, two prominent lamellae, the spinopre-and spinopostzygapophyseal laminae, rise upwards on either side, converging at the spine table, making the transverse width of the spine narrower in the middle and wider dorsally. The surfaces of the spine between the lamellae are rugose. However, the spines do not show any sign of bifurcation.
There are seven disarticulated neural arches and two isolated centra of the dorsal series; comparisons with figures of the articulated dorsal series of Plateosaurus (Jaekel, 1913–14; Huene, 1926) indicate that they include cranial (∼D2, 5), mid- (∼D8), and caudal (∼D12, ∼14) dorsal vertebrae (Figs. 9, 10). The dorsal centra are more robust and larger than in the cervical vertebrae; they are rounded ventrally without a keel. The neural spines, instead of being elongate transversely as in the caudal cervical vertebrae, are compressed transversely and elongated craniocaudally. Another difference is the lack of fusion of the neural arches to the centra (also true of isolated sacral neural arch). This lack of fusion indicates that the holotypic individual (ISI R257) was not fully mature when it died, as is also indicated by the lack of fusion of most of the cranial sutures. The beginning of the dorsal series is clearly defined by several features, such as the short and transversely narrow neural spines, the absence of epipophyses, robust transverse processes directly upwards and outwards, and the migration of the parapophysis from the centrum to the base of the neural arch. By at least D8 (Figs. 9, 10) the parapophysis is on the base of the transverse processes that is held more horizontally. The transverse processes of the caudal dorsal vertebrae (∼D12, ∼D14) are directed more cranially and the two rib facets are adjacent to each to form a single attachment area (Fig. 10.14). The dorsal centra are shorter in the rostral dorsals but are longer in the caudal dorsal vertebrae, in which the neural spines are taller. As in other saurischians (Wilson, 1999), the transverse process lies at the level of the zygapophyses and is supported by cranial, ventral, and caudal laminae, enclosing various pneumatic fossae between them, which do not invade the centrum representing acamerate condition (Wedel, 2003). The spine table is long axially and compressed transversely. The accessory articulations between the zygapophyses that made the trunk region more rigid—the caudal hyposphene and the cranial hypantrum—are present in the preserved dorsal neural arches. The zygapophyses are nearly horizontal cranially but caudally the transverse processes are elliptical in lateral view, being directed upward and outward. The caudal dorsal neural spines are supported by relatively prominent spinopostzygapophyseal laminae, a derived state present in most sauropods (Upchurch, 1998; Upchurch et al., 2004).
The disarticulated neural arch of the sacral vertebra has the characteristic winglike transverse process that is expanded nearly vertically (Figs. 5, 10). A similar expanded rib facet is known in the sacral region of some basal sauropodomorphs, but the lack of the associated rib and centrum makes the precise placement of this neural arch difficult.
The neural arch and transverse processes are fused with the centrum in the 16 preserved vertebrae of the caudal series (Figs. 5, 10). The centrum of the isolated proximal caudal vertebra ISI R258 (Fig. 5) is higher than long (ratio ∼0.8) with obliquely inclined chevron facets ventrally, a small one proximally and a larger one distally (Fig. 5). Thus, Lamplughsaura possesses craniocaudally compressed proximal caudal centra, although the degree of this compression is not as great as that seen in most sauropods (Upchurch, 1998). The caudal centra possess a midline sulcus, which is a derived state present in several sauropods (Upchurch and Martin, 2003; Yates and Kitching, 2003). The neural spine, which is shorter than the incomplete transverse process, has a thin triangular-shaped and craniodorsally directed triangular sheet of bone or spur. In the most proximal (midcaudal, ∼22 out of estimated 40) of the 11 isolated caudal vertebrae from the distal half of the tail of ISI R257, there is a distinct craniodorsal process, almost as large as the neural spine itself, that gives an unusual concave curvature to the dorsal margin of the neural arch in lateral aspect (Figs. 5, 10). A ‘cranial spur’ is present on the front of the neural spine in mid-caudal vertebrae of several theropods, including Allosaurus (Rauhut, 2003), but it is not a process; this structure is not known in any other sauropodomorph. More distally (Figs. 5, 10), the centra become smaller and more slender, the transverse processes shorter, and the neural spines lower and more compressed transversely. The prezygapophyses extend more forwards in relation to the limited backward projection of the postzygapophyses. More distally (Figs. 5, 10), the vertebrae decrease considerably in size, the neural spine is atrophied, and the transverse process is small. The caudals are unique for a sauropodomorph in that the transverse processes are individually longer than the neural spines that are lost more proximally in the series. Usually the reverse is true, so in Plateosaurus the transverse processes and neural spines disappear at about vertebrae 27 and 35, respectively (Huene, 1926), and in sauropods the transverse processes disappear in the proximal part of the tail whereas spines can still be seen in middle and some distal caudals (Upchurch et al., 2004).
Ribs and chevrons
The cervical ribs were not co-ossified with the vertebrae and none are preserved. Three dorsal ribs are preserved in ISI R257 and four in ISI R258 (Fig. 10). Each rib has a long and curved shaft and an expanded proximal end. Cranial dorsal ribs have a capitulum and a tuberculum that are close together. Passing backwards, the ribs become progressively longer and stouter; the two rib heads are well differentiated and became Y-shaped. In the caudal dorsal series, the ribs are slim, decrease in length, and the two heads become confluent.
There are several isolated chevrons (ISI R257/35, 257/55-257/ 58) of normal form for a basal sauropodomorph. The proximal end shows two distinct facets for articulation with the centra. Proximally, the bone encloses a large foramen below the articulation; the shaft is compressed transversely and curves backwards but rises slightly upwards at the distal end.
Both left (ISI R257/36) and right (ISI R257/ 37) scapulae are nearly complete and undistorted. No coracoid is preserved, but there is most of the left sternal plate (ISI R257/ 38) (for measurements see App. 1.2).
The scapula (Figs. 11–13) is tall, narrow, and slightly longer than the humerus. Both ends are expanded equally and linked by a narrow shaft. The dorsal end has a convex curvature in lateral aspect. It is rugose and slightly thickened for the attachment of the cartilaginous extension, the suprascapula. The main body of the blade is concave medially to conform to the contour of the rib cage. The ventral expansion consists of an acromial process at the cranial edge, a sinuous, coracoidal sutural surface in the middle, and a thick buttress supporting the glenoid at the caudal edge. This region of the scapula is not as expanded as it is in eusauropods (Upchurch, 1998; Upchurch et al., 2004). The glenoid surface is smooth and slightly concave. Medially, there is a longitudinal groove cranial to the glenoid fossa.
The sternum consists of two rectangular plates, each of which has a rounded cranial margin (Fig. 13). The plate is convex ventrally, concave dorsally, and about 33% of the scapular length. The bone is thickest near the cranial edge that was behind the coracoid, where it curves upwards, but it thins markedly at the medial symphysis. This element lacks the prominent ridge along the midline of its dorsal surface that is observed in some basal sauropods, such as Shunosaurus and Omeisaurus (Upchurch, 1998).
The different segments of the forelimb are preserved in two individuals, ISI R257 and the better-preserved ISI R261 (for measurements see App. 1.2). The humerus of ISI R261 is 77% the length of that of the holotype and the forearm and manus were found in natural articulation. Although some basal sauropodomorphs were bipedal or facultatively bipedal, the forelimb of Lamplughsaura was stout and long, about 74% of the length of the hindlimb (Appendix 2): this means that Lamplughsaura possessed the derived elongate forelimbs also seen in sauropods (Upchurch, 1998; Upchurch et al., 2004). The humerus is longer than the ulna that in turn is longer than the manus.
The proximal and distal expansions of the humerus lie in the same plane (Figs. 12–13). The more expanded proximal end is rugose and the articular head extends caudally as a spherical prominence and it is symmetrically placed in line with the long axis of the bone. The low deltopectoral crest is perpendicular to the proximal end and it is directed cranially as a triangular flange that rises to an apex at about one-third of the humeral length. In most basal sauropodomorphs, this apex extends farther distally past the midlength of the bone (Sereno, 1999; Galton and Upchurch, 2004). The cranial margin of the deltopectoral crest is sigmoid rather than straight: the former is an unusual condition that is present in a few other sauropodomorphs, such as Coloradisaurus and Lufengosaurus (Yates and Kitching, 2003). The shaft is slim and oval in cross section. The distal end has two small condyles for the radius and ulna. These two condyles are more pronounced in caudal aspect, indicating a more extended position of the forearm for a quadrupedal pose. There is no depression on the distal part of the cranial face of the humerus, between these condyles. Absence of this depression is a derived state found in sauropods (Yates and Kitching, 2003).
The radius (Fig. 12) is slimmer and slightly shorter than the ulna. The proximal end is expanded to form a concave articular surface for the radial condyle of the humerus. The shaft is long, straight, and cylindrical but begins to inflate modestly at the distal end that is slightly spherical and articulated with the radiale of the wrist.
The ulna is a stout bone with a highly expanded proximal end, but its olecranon process is weak (Figs. 12–13). The area lateral to the pyramid-shaped olecranal process is craniocaudally concave, forming a shallow recess for receipt of the radius (Fig. 12); distal to this process, the ulna gradually tapers to a thin bone. Thus, Lamplughsaura appears to possess an incipient version of the derived state present in sauropods, in which the proximal end of the ulna is triradiate because of the presence of two processes that clasp the proximal end of the radius (Wilson and Sereno, 1998). The shaft is flattened transversely and bowed away from the radius to form a large interosseal gap. The distal end is slightly expanded, with the plane of expansion at 50° to that of the proximal end.
The carpals and manus are articulated in ISI R261/3 (Figs. 12– 13), in which the left forearm, carpus, and manus were found in a highly flexed position. The right manus, lacking the wrist bones, is preserved in ISI R263/4 (Fig. 11). In addition, left metacarpals II and III are present in ISI R257 (Fig. 11).
There are two proximal carpals of the wrist, the intermedium and the ulnare (Fig. 12.6), which are slightly displaced from their original position, are closely held and show a complex morphology. Thus, the wrist in Lamplughsaura did not possess unossified (or absent) proximal carpals, unlike those of sauropods (Gauthier, 1986; Upchurch et al., 2004). The intermedium has a concave surface for the reception of the convex end of the radius. The ulnare is an irregular oval bone and it articulated with the ulna. Distal carpal 1 is missing, but distal carpals 2 and 3 are much smaller than the proximal carpals and are preserved, capping metacarpals II and III (Fig. 12).
The manus is almost complete (Figs. 12–13). Proximally, the ends of the metacarpals are triangular, closely packed, and show overlapping relationships from the medial to the lateral side to form a gentle arc, not the U-shaped arrangement or colonnade seen in sauropods (Wilson and Sereno, 1998). Metacarpal I is short, stout and wide, with a proximal width to length ratio of 0.5, and distally the lateral condyle is larger than the medial one. Similarly short, robust, and distally asymmetrical first metacarpals are found in most basal sauropodomorphs and the basal sauropod Antetonitrus (Yates and Kitching, 2003). Metacarpals II and III are longer than metacarpal I and each has a prominent extensor pit. Metacarpal IV is shorter and more slender than metacarpal III and metacarpal V is very short and slender. Thus, Lamplughsaura lacks the derived condition found in most sauropods in which metacarpal V is large (∼90% of metacarpal III length) and potentially weight bearing (Upchurch et al., 2004).
The phalangeal formula was probably 2-3-4-3-2, and so there is little indication of the extreme phalangeal reduction present in sauropods such as Shunosaurus, Omeisaurus, and neosauropods (Upchurch et al., 2004). The form of most of the phalanges is typical for a basal sauropodomorph (Galton and Upchurch, 2004). All the nonterminal phalanges are longer than wide, the first phalanx of digit I has a prominent heel and is twisted so the axes of the planes of articulation are at 45° (also in Antetonitrus, Yates and Kitching, 2003), and proximally the unguals are higher than wide and the first is the longest. However, ungual I is not highly recurved and trenchant, as is usually the case in basal sauropodomorphs (Fig. 17), but instead is slender and only gently tapered and recurved (Figs. 12–13).
The pelvic girdle is represented by the partial right ilium (ISI R259/3), left and right ischia (ISI R257/42 and R257/43), and the right pubis (ISI R258/27; also both pubes in ISI R262/1,2) (for measurements see App. 1.3). The pelvis is the typical triradiate basal sauropodomorph type and the acetabular region is deeply perforated (Fig. 14).
The ilium is incompletely preserved (Figs. 14–15). It shows a long pubic peduncle and a highly perforated acetabulum bordered by subequal pubic and ischiac peduncles. The iliac blade is thin but its dorsal part is not preserved so its depth and the extent of the cranial and caudal processes are not known.
The pubis is a stout and thick bone, directed forwards and downwards (Figs. 14–15). Proximally it expands to articulate with the ilium and ischium, forming a curved acetabular rim between them. Below the acetabulum, and cranial to the ischiac contact, the pubis becomes a thin, vertical plate that is perforated by a small obturator foramen. As is most other sauropodomorphs, there is no pubic tubercle on the lateral surface of the proximal plate (Yates and Kitching, 2003). The shaft is very thick on the craniolateral margin, which is concave in craniodorsal view, but becomes thin and expands medially to form a broad apronlike region with a symphysis for much of its length. The distal end is slightly expanded craniocaudally. The pelvic cavity bounded by the two pubes is narrower and shallower than that formed by the two ischia.
The ischium is a Y-shaped bone proximally, is much slimmer than the pubis, and is incomplete distally (Figs. 11, 14–15). The pubic and ischiadic articulations are separated by the shallowly incised acetabular margin. Distal to this, the shaft gradually tapers into a flat, slender rod, teardrop-shaped in cross section, where the caudal edge is rounded but the cranial edge is sharp. Medially it has a short and deep symphysis and the two bones form a large and wide pelvic basin. Distally the ischium is incomplete.
The femur is a strong and stout bone as in other basal sauropodomorphs. The right femur (Fig. 14.2, 14.3, ISI R257/44) is somewhat crushed proximally, especially in the articular region, as it is in another nearly complete left femur (Figs. 11.5, 15.6, 15.7, ISI R259/4) that is very slightly shorter (for measurements of hindlimb bones see App. 1.3). The femur is sigmoidal in side view (Figs. 11, 14–15) but in cranial and caudal views it is straight, rather than sigmoidal as in many basal sauropodomorphs, so both proximal and distal expansions lie in the same plane. In addition, the fourth trochanter is on the medial edge (Fig. 14), not set more centrally on the shaft as in most basal sauropodomorphs, including another proximal half of a femur from the same locality (Fig. 15.5, see below). The head is directed dorsomedially without any constriction. The greater trochanter is well developed and extends laterally as a trenchant flange. Below the greater trochanter, the cranial surface of the shaft shows a ridgelike lesser trochanter somewhat obliquely placed in relation to the long axis of the bone. Thus, although reduced, the lesser trochanter is more prominent than in most sauropods (Upchurch et al., 2004). The caudal surface of the shaft supports a pendant, distally pointing fourth trochanter for the attachment of the caudofemoralis muscle. Thus, the fourth trochanter is in a position similar to that observed in sauropods, but differs from these in not being reduced to a low rounded ridge. The base of the fourth trochanter continues down to the midlength of the bone. The shaft is gently oval in cross section with a medullary cavity. Distally, the femur is partially divided by a longitudinal groove into two condyles on the caudal aspect. The medial (tibial) condyle is more pronounced than the lateral (fibular) condyle.
The left tibia is completely preserved in ISI R257/46 (Fig. 15) in association with the fibula and the bones of the ankle. It is a robust bone, somewhat shorter than the femur (Fig. 15): Lamplughsaura therefore possesses the derived reduced tibia:femur length ratio observed in virtually all sauropods (McIntosh, 1990; Upchurch, 1995, 1998; Wilson, 2002; Upchurch et al., 2004). The proximal end is expanded craniocaudally, whereas the distal end is inflated transversely so the shaft is twisted with the expanded ends at about 45° to each other. Proximally, there is a prominent cnemial crest on the cranial aspect for the attachment of the iliotibialis muscle; this crest merges with the shaft more distally. Laterocaudal to this crest, there is a thin flange on the tibia for the reception of the fibula. This fibular crest projects more proximally than the opposite caudal edge. The shaft is twisted and flattened rostrocaudally to connect the two expanded ends. Distally, the tibia is deeply notched laterally to interlock with the ascending process of the astragalus. Its medial surface has a distal flange on the caudal aspect that sits on a shelf of the astragalus. Thus, the caudal fossa of the astragalus is covered by the tibia, a plesiomorphic state retained by basal sauropodomorphs (Wilson and Sereno, 1998). The articulation between the tibia and the astragalus is complex and close, prohibiting any motion between them (Fig. 15). Thus the ankle joint is typically mesotarsal, with the proximal tarsus forming an additional segment of the crus.
The fibula is represented by the left (ISI R257/47) and right (ISI R257/48) of the holotype individual (Fig. 15) and a left (ISI R258/28) of a smaller individual. Both proximal and distal ends are expanded fore and aft and connected by a slender shaft. The proximal end forms a crescent, and the concavity medial surface for attachment to the tibia bears a prominent but short vertical ridge. The shaft is flattened transversely, bows away from the tibia, and lacks the lateral muscle scar at midlength, a derived feature of eusauropods (Upchurch et al., 2004). The expanded distal end contacts the tibia medially and sits on the concave proximal surface of the calcaneum.
The tarsus is represented by the two proximal elements of the left side, the larger astragalus (ISI R257/48) and the smaller calcaneum (ISI R257/49). The astragalus is elongated transversely (Figs. 11, 14–15). In cranial aspect, the lateral part has an ascending process with a flat, horizontal surface on the proximal end that abuts against the tibia (Figs. 11, 15). This surface slopes medially into a concavity for further tibial articulation. There is a depression at the base of the ascending process on the cranial face of the astragalus, the primitive state, whereas its loss characterizes eusauropods (Wilson and Sereno, 1998). In caudal aspect, the ascending process has a vertical wall, somewhat hollowed out on either side of a vertical ridge (Fig. 11). This prominent vertical ridge is unusual for a basal sauropodomorph but also occurs in Melanorosaurus Haughton, 1924, Vulcanodon Raath, 1972, Barapasaurus, and Eusauropoda (Galton and Upchurch, 2004; Upchurch et al., 2004). The lateral surface of the astragalus is slightly concave to receive the convex surface of the calcaneum. The distal surface is hemicylindrical for articulation with the metatarsals; the ankle was preserved in articulation and there were no distal tarsals: this suggests that the absence of these distal tarsals represents a genuine character state, and indicates the Lamplughsaura probably possessed the derived state present in sauropods except Blikanasaurus Galton and Heerden, 1985 (Galton and Upchurch, 2004; Upchurch et al., 2004). The calcaneum is a small triangular bone, concave proximally for the reception of the fibula (Fig. 14.4). Its medial articulation with the astragalus is exact and close. The ventral surface is convex for articulation with the pes.
The right pes of ISI R262/4 was found articulated, except that digit V was missing (Figs. 14–15). Additional elements of the pes include the distal end of metatarsal I and a right digit II (Fig. 15.14; ISI R261/5), left metatarsals III and IV (Fig. 15, ISI R258/ 29-32), and a larger right metatarsal IV (ISI R257/50). The pes is strongly built with elongated metatarsals and a digitigrade pose (Fig. 14).
The form of the pes is typical of a basal sauropodomorph (Galton and Upchurch, 2004). Each metatarsal is expanded proximally, with a triangular to rectangular proximal end that overlaps its neighbor laterally. Metatarsal I is short and robust, with a length to proximal width ratio of 2.0, so it is less robust than that of eusauropods and much shorter than that of Vulcanodon that retains a plesiomorphic long metatarsal I (Upchurch et al., 2004). The proximal end of metatarsal II is hourglass shaped as in other sauropodomorphs. Metatarsals III is longer than II (but less wide) but is less than 40% of tibia length, a derived state found in sauropods (Yates and Kitching, 2003; Upchurch et al., 2004). Metatarsal IV is shorter but more expanded proximally than III (Fig. 15); the wide proximal end is a derived state present in basal sauropodomorphs and also in Antetonitrus (Yates and Kitching, 2003). The shaft is compressed dorsoventrally, and somewhat ovoid in cross-section. Metatarsal V is not known. Distally each metatarsal is slightly inflated and forms a rolling, hemicylindrical surface for ginglymic articulation with its proximal phalanx.
The phalangeal formula is 2-3-4-5-? (Fig. 14) and digit III is the longest as in other basal sauropodomorphs (Galton and Upchurch, 2004). All the nonterminal phalanges are longer than wide, the plesiomorphic condition, and none are wider than long as occurs in Melanorosaurus (Galton et al., 2005), Camelotia (Galton, 1998), and sauropods (Yates and Kitching, 2003; Galton and Upchurch, 2004). Digit I is forwardly directed and carries the longest ungual phalanx. Unguals I to IV decrease in length, with ungual I shorter than metatarsal I, and show slight changes in form (Figs. 14, 15). The unguals are of the normal form for basal sauropodomorphs (Galton and Upchurch, 2004), being directed forwards, roughly symmetrical and gently recurved with longitudinal claw grooves on each side. They are not strongly compressed transversely and directed forwards and laterally, so the articular surfaces are only visible in lateral view, as in sauropods (Wilson and Sereno, 1998; Yates and Kitching, 2003).
From the skeletal remains now known from specimens ISI R257–R259, R261 and R262, it is estimated that a near adult individual (femur length 630 mm with most cranial sutures still open and only neurocentral sutures of caudal vertebrae closed) of Lamplughsaura was about 10 m from snout to tail tip when stretched full length (Fig. 16), and would be slightly larger than that of Riojasaurus (femur length 600 mm; body length ∼10 m; Bonaparte and Pumares, 1995). The skeleton of Riojasaurus (Bonaparte, 1972) was helpful in restoring the skeleton of Lamplughsaura because of its similar anatomy and proportions. Seebacher (2001) listed the mass of a 6.5 m Plateosaurus as 1,072 kg, and the 10 m Riojasaurus at around 3,000 kg. Alexander et al. (1979) gave the allometric equation [D = 5.2W0.36] for estimating the mass of quadrupedal mammals, where D is the diameter of the femur (in mm) and W is the mass in kg. Using this equation, the estimated body mass of Lamplughsaura is 1,982 kg, much lower than that of Riojasaurus as calculated by Seebacher (2001). As restored, Lamplughsaura (Fig. 16) is a tall, massive, heavily built, quadrupedal sauropodomorph with a small head, a long and flexible neck, a rigid and arched trunk, deep rib cage, long, powerful vertical limbs, and a long tail. The humerus is proportionately large with a robust but low deltopectoral crest; the ulna is powerful, the manus is broad and short. The forelimb is stout and long, about 74% of the length of the hindlimb (within the range seen for sauropods), and was used for body support. The elbow was probably turned back and the knee forward to give a narrow trackway. The development of a straight shaft in both the humerus and the femur (at least in caudal view) in Lamplughsaura is linked to its graviportal posture, with the limbs providing more direct vertical support. The long and stout forelimb indicates that Lamplughsaura was probably an obligate quadruped, with graviportal parasagittal limbs and a digitigrade stance. Christian and Preuschoft (1996) inferred the posture of Plateosaurus from the bending moment in the sagittal plane of the vertebral column. They observed two distinct local maxima at the shoulders and hips that are consistent with a habitual quadrupedal posture for the animal.
Thulborn (1990) estimated the average walking speed of sauropodomorphs from several skeletons. For Anchisaurus Marsh, 1885, (Lower Jurassic, Connecticut Valley; Galton, 1976; Yates, 2004) the estimated speed was 1.6–2.74 km/h whereas for Plateosaurus it was 2.96–3.17/km. Both parameters of cursorial ability, the tibia/femur ratio (0.83) and the metatarsal III/tibia ratio (0.36), are low in Lamplughsaura, indicating that the animal was graviportal, where the limbs were adapted for weight bearing, but not a fast runner.
Sauropodomorpha incertae sedis
Pradhania new genus
Pradhania gracilis new genus and species
The genus is named in honor of Dhuiya Pradhan, a gifted fossil collector at ISI; the specific name refers to the gracile nature of the animal.
Holotype, ISI R265, an incomplete skeleton of a nearly adult individual found in close association on the surface, consisting a few skull elements, vertebrae (two cervicals, one sacral), and a very incomplete left manus (Figs. 17, 18). Upper Dharmaram Formation, Lower Jurassic. Northwest of Krishnapur village (79°30′, 19°15′), Adilabad District, Andhra Pradesh, India.
Maxilla with a very prominent longitudinal ridge medially. Cranial cervicals extremely elongated so the central length is about 4.5 times its height (also in Massospondylus).
The maxilla (Figs. 17–18) has a triangular body but most of the dorsal edge is broken except caudally, where only a very small part of the base of the caudal end of the dorsal process is preserved. Consequently, the form of the process is unknown. Laterally, there are five nutrient foramina on the rostral part of the maxilla with four forming a ventral longitudinal row. The medial surface bears a very prominent longitudinal ridge that overhangs the alveolar margin. This ridge becomes very wide rostrally and it is separated from the premaxillary process by a deep, vertical groove. There are 20 alveoli, most of which are empty except for three in which small, erupting teeth are present. Such a large number of maxillary teeth also occur in Coloradisaurus (23 or 24; Upper Triassic, Argentina; Bonaparte, 1978), Lamplughsaura n. gen. (∼21), and Plateosaurus (24–30). It is not clear whether this specimen possessed a lateral plate on the maxilla, or the dentary, so it has been coded as state ‘?’ for this feature in the cladistic data matrices, see below.
The supraoccipital (Fig. 17) has been reconstructed from the preserved right half. It is narrow transversely and may have been more vertically placed than in Lamplughsaura. This suggests that Pradhania n. gen. lacks the derived semilunate and transversely widened supraoccipital observed in Lamplughsaura and many sauropods (see above). It shows a deep notch on the rostrolateral surface for the reception of the parietal. The caudal margin forms the dorsal rim of the foramen magnum.
The rostral part of the dentary (Fig. 17) shows the typical downward curvature of the ventral margin towards the symphysis as in many basal sauropodomorphs. The symphyseal surface is a narrow triangular surface, with an apex pointing ventrally, so it lacks the robust symphysis seen in most sauropods.
A few erupting teeth are preserved in the maxilla and dentary (Fig. 17). The teeth are spatulate and symmetrical in mesial view, transversely compressed, the mesial and distal edges contain coarse obliquely oriented denticles (17 or more), and the enamel is not wrinkled. There are no grooves or ridges on the labial and lingual surfaces of the crown that is constricted at its base. A small pit or window on the base of the lingual side of each alveolus provided access for the replacement tooth.
The neurocentral sutures are closed, indicating an adult individual. Two cervical vertebrae, possibly the third and fourth, are similar to those of Massospondylus (ISI casts of specimen in South African Museum). Each cervical is extremely elongated so the central length is about 4.5 times its height (Figs. 17–18), very elongate for a basal sauropodomorph, and rivaling the condition seen in several sauropods. This elongation could represent a local autapomorphy relative to most other basal sauropodomorphs. The ventral margin of the centrum has a prominent keel. There are no laminae supporting the transverse processes from below, the plesiomorphic condition, whereas such laminae represent the derived state that is present in most sauropods (see above). The neural spine is extremely low and caudally placed. The zygapophyses are short and project slightly beyond the faces of the centrum.
A sacral vertebra (Figs. 5, 17–18) has a rib with a prominent process directed cranioventrally to caudodorsally as occurs in sacral 2 and in the caudosacral. The cranial face of the centrum is rugose for attachment to the preceding sacral; the caudal face is smooth with a rounded margin.
Digit I of the left manus is well preserved (Fig. 18) and is of normal basal sauropodomorph form (Fig. 17). Metacarpal I is expanded proximally and overlaps the base of the proximally displaced metacarpal II (rest of digit represented only by distal half of phalanx 1). Distally it has a spherical end that articulated with the socket of the first phalanx 1 that is enlarged by a prominent heel (Fig. 17.15) as in most basal sauropodomorphs (Sereno, 1999; Galton and Upchurch, 2004). The first phalanx is twisted along its length, so the articular surfaces are set at an angle of about 45° as in other basal sauropodomorphs (Sereno, 1999; Galton and Upchurch, 2004). The large ungual phalanx has a prominent flexor tubercle near its proximoventral edge, is strongly recurved, and both sides are flanked by a longitudinal claw groove.
Pradhania n. gen. is smaller (adult body length ∼4 m) and more lightly built than Lamplughsaura n. gen. (at ∼10 m). Several characters, such as the prominent medial flange on the maxilla, spatulate teeth with denticles, elongate cervical vertebrae with low neural spines, and the trenchant form of the ungual of manual digit I, clearly indicate that Pradhania is distinct from Lamplughsaura.
Sauropodomorpha incertae sedis
Genus and species indeterminate
Among the specimens originally referred to the Upper Dharmaram plateosaurid from Northwest of Krishnapur Village (Fig. 1.2) by Kutty (1969) is the proximal half of a left femur (Fig. 19; ISI R260/1). This large proximal femur (∼size of ISI R257) resembles those of Lamplughsaura n. gen. except that the head is more medially directed and the fourth trochanter is set in the middle of the caudal surface. The trochanter is in the plesiomorphic position, well away from the medial edge, rather than on the medial edge in the derived position as in Lamplughsaura (Figs. 14.2, 15.6). This femur, with an estimated length of ∼630 mm (so total body length ∼10 m), is much too large to be referable to Pradhania n. gen. (adult body length ∼4 m), so this femur probably represents a third taxon of basal sauropodomorph from the Lower Jurassic of India.
In order to investigate the phylogenetic relationships of Lamplughsaura n. gen., Pradhania n. gen., and ISI R260, we have coded them for two recent data matrices for basal sauropodomorphs (Yates and Kitching, 2003; Galton and Upchurch, 2004). The modified data-matrices are available as Nexus files from PU on request, or can be downloaded from 〈http://www.ucl.ac.uk/~ucfbpup/docs/〉. The results of these analyses are described and discussed below.
The modified data matrix of Yates and Kitching (2003; see also Yates, 2004) consists of 212 characters for 22 ingroup sauropodomorphs and two outgroups (Herrerasaurus Reig, 1963 and Neotheropoda). Application of the Branch-and-Bound search in PAUP 4.0b10 (Swofford, 2004) yielded 30 most parsimonious trees (MPTs) (L = 499; CI = 0.489; RI = 0.693; RCI = 0.340). A strict consensus tree (SCC) does not resolve the relationships of Lamplughsaura, Pradhania, or ISI R260, except that these three specimens are more closely related to Sauropoda + Plateosauria than are Thecodontosaurus (Great Britain; Benton et al., 2000; Yates, 2003a) and Saturnalia (Brazil; Langer et al., 1999; Langer, 2004) from the Upper Triassic. The a posteriori deletion of ISI R260 results in a nearly fully resolved reduced consensus tree, apart from a polytomy between Blikanasaurus (South Africa; Galton and Heerden, 1998), Antetonitrus (South Africa; Yates and Kitching, 2003), and Isanosaurus (Thailand; Buffetaut et al., 2000) from the Upper Triassic and a clade containing the remaining sauropods (Fig. 20.1). In this topology, Lamplughsaura and Pradhania are sister taxa and are placed above Thecodontosaurus and below Sauropoda + Plateosauria. Thus, Lamplughsaura and Pradhania are stem sauropodomorphs according to this analysis.
A Permutation Tail-Probability Test (PTP) was applied to the data matrix using 10,000 random replicates in PAUP. The matrix passed this test with a P-value of <0.0001, indicating that it contains a strong phylogenetic signal. However, this result applies to the data set as a whole and does not necessarily indicate strong support for the positions of Lamplughsaura and Pradhania.
A bootstrap analysis, using 10,000 random replicates and the heuristic search in PAUP, found that the support values for the sister group relationship between Lamplughsaura and Pradhania, and the position of these taxa as stem sauropodomorphs, are less than 50%.
Incorporation of the three Indian specimens into Galton and Upchurch (2004) creates a data matrix of 137 characters for 26 ingroup taxa (plus the outgroup ‘Ancestor’). This new data set was analyzed using the heuristic search in PAUP 4.0b10 (Swofford, 2004). This yielded 1,204 MPTs (L = 301; CI = 0.504; RI = 0.613; RCI = 0.317). The SCC displays very little resolution, except that Lamplughsaura, Pradhania, and ISI R260 represent taxa that are more closely related to sauropods and most prosauropods than is Saturnalia; more precise relationships cannot be determined given the information in the current matrix. The resolution of the SCC does not increase when either ISI R260 or Pradhania are deleted a posteriori, but does increase substantially when both are deleted (i.e., the number of topologies is reduced from 1,204 to three). An SCC based on these three remaining topologies is shown (Fig. 20.2). In the reduced consensus trees, Lamplughsaura has a stable position within the basal Sauropoda, between Blikanasaurus and Kotasaurus Yadagari, 1988. (Lower Jurassic, India, Jain et al., 1975, 1979).
A PTP Test was applied to the data matrix using 10,000 random replicates in PAUP. The matrix passed this text with a P-value of <0.0001, indicating that it contains a strong phylogenetic signal. However, this result applies to the data set as a whole, and does not necessarily indicate strong support for the positions of Lamplughsaura and Pradhania.
A bootstrap analysis (10,000 replicates using the heuristic search in PAUP) found the support for the placement of Lamplughsaura within basal Sauropoda was less then 50%. However, after the deletion of Blikanasaurus, Pradhania, and ISI R260, this bootstrap value rises to 59%.
The two data matrices have generated alternative hypotheses for the phylogenetic relationships of Lamplughsaura; i.e., the data of Yates and Kitching (2003) suggest that Lamplughsaura is a basal sauropodomorph that lies outside of the Sauropoda + Plateosauria clade, whereas that of Galton and Upchurch (2004) places this genus as a very basal sauropod. Under these circumstances, it is informative to use Templeton's tests to explore the relative strengths of these two hypotheses given the data available.
The first test examined whether Lamplughsaura could reasonably be placed in the Sauropoda on the basis of the data set of Yates and Kitching (2003). We therefore employed a constraint in which Lamplughsaura is required to cluster with other sauropods (i.e., Anchisaurus, Melanorosaurus, Antetonitrus, Blikanasaurus, Isanosaurus, etc.). With this constraint enforced, the Branch-and-Bound search in PAUP found 10 shortest trees (L = 502 steps). These trees are only three steps longer than the unconstrained MPTs. Comparison of the constrained and unconstrained trees, using a Templeton's test, indicates that the placement of Lamplughsaura, within the Sauropoda is not a significantly worse explanation of the data of Yates and Kitching (2003) (P = 0.523–0.622).
The second test examined whether Lamplughsaura could reasonably be placed as a stem sauropodomorph (i.e., outside of a clade containing sauropods, Plateosauria, and Melanorosauridae) on the basis of the data set of Galton and Upchurch (2004). We therefore constructed a constraint topology in which Lamplughsaura is forced to lie outside of a clade containing all sauropodomorphs except Saturnalia, Thecodontosaurus, Pradhania, and ISI R260. With this constraint enforced, the Heuristic search in PAUP found 65 shortest trees (L = 344 steps). Thus, these constrained trees are 43 steps longer than the original unconstrained MPTs. Comparison of the constrained and unconstrained topologies, using a Templeton's test, indicates that the placement of Lamplughsaura outside of the Sauropoda is a significantly worse explanation of the data set from Galton and Upchurch (2004) (P < 0.0001).
In summary, it appears that the placement of Lamplughsaura within the basal Sauropoda is compatible with both Yates and Kitching (2003) and Galton and Upchurch (2004), whereas placement of this specimen within stem sauropodomorphs is strongly rejected by the latter data set.
discussion and conclusions
The Indian taxa
The results of the above phylogenetic analyses and statistical tests do not provide a definitive estimate of the relationships of the three Indian specimens. Nevertheless, some tentative conclusions can be reached. First, the relationships of ISI R260 and, to a lesser extent, Pradhania n. gen., are highly unstable. This probably reflects the large quantities of missing data for these specimens [e.g., ISI R260 lacks 97% and 90% of the data for Yates and Kitching (2003) and Galton and Upchurch (2004), respectively]. ISI R260 is probably more derived than Saturnalia and Thecodontosaurus, but no more precise statement about its relationships can be made at this stage. Pradhania is more derived than Saturnalia and Thecodontosaurus, and the Yates and Kitching (2003) result (see Fig. 20.1) suggests that it lies outside of the Sauropoda + Plateosauria clade. Thus, Pradhania is tentatively regarded as a very basal sauropodomorph.
Lamplughsaura n. gen. is a much more nearly complete specimen, lacking only 20% of the data for both Yates and Kitching (2003) and Galton and Upchurch (2004). It thus has a more stable position within the MPTs. However, the two data sets yield conflicting estimates of the relationships of Lamplughsaura: either it is a basal sauropodomorph lying outside the Sauropoda + Plateosauria clade (Yates and Kitching, 2003) or it is a basal sauropod (Galton and Upchurch, 2004). Here, we provisionally prefer the latter view for two reasons: 1) the bootstrap value supporting Lamplughsaura within Sauropoda is higher than that placing it as a basal sauropodomorph; and 2) the results of the Templeton's tests (see above) indicate that the data of Yates and Kitching (2003) do not decisively reject this hypothesis at this stage.
Sauropod characters of Lamplughsaura
To emphasize the above tentative conclusion, it may be useful to outline the character states that apparently support the placement of Lamplughsaura within the Sauropoda. The following list is based on a Delayed Transformation Optimization of the Galton and Upchurch (2004) characters onto the reduced consensus topologies found by the a posteriori deletion of ISI R260. Those characters marked by ‘*’ have unequivocal distributions (i.e., their status as synapomorphies would not be affected if a different optimization criterion were used). The main cause of ambiguity in the distributions of these characters is the extreme incompleteness of Blikanasaurus, which is only known from a distal hind limb and pes (Galton and van Heerden, 1998). Synapomorphies uniting Lamplughsaura with other sauropods include: 1) the rostral end of the dentary is wider dorsoventrally and more robust than the caudal portion; 2) strongly reduced external mandibular fenestra (may have occurred independently in Jingshanosaurus Zhang and Yang, 1995); 3) lingual surfaces of tooth crowns are concave; 4) hindlimb: trunk length ratio is >1.0; 5) cranial caudal vertebrae have centrum length: height ratios of <0.6; 6) the lateral margin of the pubis is concave in cranial view (occurs independently in “Gyposaurus” Young 1941, Lufengosaurus, and Massospondylus); 7) femoral head projects dorsomedially; 8) greatly reduced lesser trochanter on the femur; 9) femoral shaft is straight in cranial view (also occurs in Anchisaurus, Camelotia, “Gyposaurus,” Lufengosaurus, Melanorosaurus, Riojasaurus); 10) femoral fourth trochanter is located on the medial margin of the caudal surface of the shaft (also occurs in Anchisaurus, Camelotia, Lufengosaurus, Melanorosaurus, Mussaurus Bonaparte and Vince, 1979, and Riojasaurus); 11) femoral shaft is elliptical in horizontal cross section, with the long axis of this section oriented transversely (also occurs in Camelotia, Melanorosaurus, and Riojasaurus); *12) caudal fossa of the astragalus is divided into two portions by a ridge; and *13) distal tarsal absent or unossified. To these features could also be added the presence of a lateral plate on the dentary and wrinkled tooth enamel, both of which characterize sauropods (Upchurch, 1995; Wilson and Sereno, 1998).
The wider phylogenetic context
At present, the relationships of basal sauropodomorphs and the origins of sauropods represent two of the more confused and controversial aspects of dinosaur phylogeny (see Yates, 2003a; Wilson, 2005). Cladistic analyses have tended to support the view that Prosauropoda and Sauropoda are monophyletic sister taxa (Sereno, 1989, 1999; Galton and Upchurch, 2004). This view has created a morphological and stratigraphic ‘gap’ at the base of the Sauropoda. Several recent cladistic analyses, however, have found evidence that Prosauropoda is at least partially paraphyletic with regard to Sauropoda (Benton et al., 2000; Yates and Kitching, 2003; Yates, 2004). Although the latter studies recognize a monophyletic Plateosauria that contains many familiar prosauropod taxa (e.g., Plateosaurus and Lufengosaurus), they have also noted that Saturnalia, Thecodontosaurus, and Efraasia may lie basal to the plateosaur + sauropod divergence, and Anchisaurus and melanorosaurids may actually be basal sauropods. The new Indian basal sauropodomorph specimens are clearly relevant to this debate and, when better understood, may play a role in resolving the current controversy. If these taxa are very basal sauropodomorphs (Fig. 19.1), then they will substantially increase our knowledge of the diversity of taxa that lie just below the plateosaur + sauropod divergence: their anatomy will also necessitate some reevaluation of the character-state transformations that occurred in this very basal part of the sauropodomorph tree. Alternatively, if Lamplughsaura is a basal sauropod, then it has the potential to shed light on some key stages in the origin of this clade. For example, the Lamplughsaura dentary displays an unusual combination of apparently derived character states: the presence of a lateral plate and expanded symphysis are characteristic of sauropods, whereas the caudally placed lateral ridge [potentially associated with the possession of a fleshy cheek (Galton, 1984a)] has traditionally been regarded as a prosauropod synapomorphy (Galton, 1990; Yates and Kitching, 2003; Galton and Upchurch, 2004). Similarly, the first phalanx of manual digit I of Lamplughsaura has the characteristic twisted morphology that characterizes all prosauropods (Sereno, 1999; Galton and Upchurch, 2004): interpretation of the Indian form as a basal sauropod reinforces the conclusion (based on the anatomy of the manus of Antetonitrus) that this derived condition was also present in the most basal sauropods and was subsequently reversed to the plesiomorphic (untwisted) condition in Eusauropoda (Yates and Kitching, 2003). Thus, whichever set of relationships is preferred, we must accept some important convergence and/or reversal of characters during the early radiations of basal sauropodomorph and sauropod lineages.
Despite the potentially important insights that Lamplughsaura and Pradhania may eventually yield into sauropod origins, it should be noted that their inclusion in the cladistic analyses has not resulted in a decisive resolution of the prosauropod monophyly/paraphyly debate. The cladograms generated by the analyses of the modified Yates and Kitching (2003) and Galton and Upchurch (2004) data sets (Fig. 20) are essentially very similar to those found by the authors originally: the main differences are caused by the decreased resolution resulting from the addition of the Indian specimens. Clearly, a synthesis of the character sets proposed by Yates and Kitching (2003), Galton and Upchurch (2004), and other relevant data sets, is required before the full implications of these Indian taxa for tree topology and character transformations can be assessed with precision.
This work is a part of the continuing program of integrated geological research in the Pranhita-Godavari Valley under the auspices of the ISI. We thank A. Chaudhury, former Head of the Geology Unit of ISI, for his kind invitation and hospitality to one of us (SC) during the study of the new sauropodomorphs and Barapasaurus. SC also thanks D. Pradhan, for field assistance and the preparation of material, and Z. Dong for access to new material of Lufengosaurus and Shunosaurus. We are indebted to K. McQuilkin, J. Martz, and M. W. Nickell for the skilled execution of the finished illustrations. T. Lehman, J. McIntosh, J. Wilson, and A. Yates critically reviewed earlier versions of the manuscript and offered constructive suggestions. The Indian Statistical Institute and Texas Tech University supported the research.
- Accepted 6 September 2006.